diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,14065 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.17538354874909054,
+  "eval_steps": 500,
+  "global_step": 2000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 8.769177437454527e-05,
+      "grad_norm": 7.59375,
+      "learning_rate": 3e-06,
+      "loss": 4.1911,
+      "step": 1
+    },
+    {
+      "epoch": 0.00017538354874909053,
+      "grad_norm": 7.6875,
+      "learning_rate": 6e-06,
+      "loss": 4.1743,
+      "step": 2
+    },
+    {
+      "epoch": 0.0002630753231236358,
+      "grad_norm": 8.1875,
+      "learning_rate": 9e-06,
+      "loss": 4.1165,
+      "step": 3
+    },
+    {
+      "epoch": 0.00035076709749818106,
+      "grad_norm": 8.0,
+      "learning_rate": 1.2e-05,
+      "loss": 4.2652,
+      "step": 4
+    },
+    {
+      "epoch": 0.00043845887187272633,
+      "grad_norm": 7.9375,
+      "learning_rate": 1.5e-05,
+      "loss": 4.1691,
+      "step": 5
+    },
+    {
+      "epoch": 0.0005261506462472716,
+      "grad_norm": 8.0625,
+      "learning_rate": 1.8e-05,
+      "loss": 4.2089,
+      "step": 6
+    },
+    {
+      "epoch": 0.0006138424206218169,
+      "grad_norm": 8.25,
+      "learning_rate": 2.1000000000000002e-05,
+      "loss": 4.1677,
+      "step": 7
+    },
+    {
+      "epoch": 0.0007015341949963621,
+      "grad_norm": 7.8125,
+      "learning_rate": 2.4e-05,
+      "loss": 4.1564,
+      "step": 8
+    },
+    {
+      "epoch": 0.0007892259693709074,
+      "grad_norm": 7.75,
+      "learning_rate": 2.7e-05,
+      "loss": 4.1722,
+      "step": 9
+    },
+    {
+      "epoch": 0.0008769177437454527,
+      "grad_norm": 7.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.1554,
+      "step": 10
+    },
+    {
+      "epoch": 0.0009646095181199979,
+      "grad_norm": 7.53125,
+      "learning_rate": 3.2999999999999996e-05,
+      "loss": 4.1511,
+      "step": 11
+    },
+    {
+      "epoch": 0.0010523012924945432,
+      "grad_norm": 7.1875,
+      "learning_rate": 3.6e-05,
+      "loss": 4.152,
+      "step": 12
+    },
+    {
+      "epoch": 0.0011399930668690886,
+      "grad_norm": 6.8125,
+      "learning_rate": 3.9e-05,
+      "loss": 4.1581,
+      "step": 13
+    },
+    {
+      "epoch": 0.0012276848412436337,
+      "grad_norm": 6.625,
+      "learning_rate": 4.2000000000000004e-05,
+      "loss": 4.0485,
+      "step": 14
+    },
+    {
+      "epoch": 0.001315376615618179,
+      "grad_norm": 6.28125,
+      "learning_rate": 4.4999999999999996e-05,
+      "loss": 4.083,
+      "step": 15
+    },
+    {
+      "epoch": 0.0014030683899927242,
+      "grad_norm": 5.625,
+      "learning_rate": 4.8e-05,
+      "loss": 4.075,
+      "step": 16
+    },
+    {
+      "epoch": 0.0014907601643672696,
+      "grad_norm": 5.625,
+      "learning_rate": 5.1000000000000006e-05,
+      "loss": 4.0571,
+      "step": 17
+    },
+    {
+      "epoch": 0.0015784519387418148,
+      "grad_norm": 5.53125,
+      "learning_rate": 5.4e-05,
+      "loss": 4.0213,
+      "step": 18
+    },
+    {
+      "epoch": 0.0016661437131163602,
+      "grad_norm": 4.5625,
+      "learning_rate": 5.7e-05,
+      "loss": 4.0436,
+      "step": 19
+    },
+    {
+      "epoch": 0.0017538354874909053,
+      "grad_norm": 4.6875,
+      "learning_rate": 6e-05,
+      "loss": 4.0306,
+      "step": 20
+    },
+    {
+      "epoch": 0.0018415272618654507,
+      "grad_norm": 4.125,
+      "learning_rate": 6.3e-05,
+      "loss": 4.0364,
+      "step": 21
+    },
+    {
+      "epoch": 0.0019292190362399958,
+      "grad_norm": 3.875,
+      "learning_rate": 6.599999999999999e-05,
+      "loss": 3.9774,
+      "step": 22
+    },
+    {
+      "epoch": 0.002016910810614541,
+      "grad_norm": 3.421875,
+      "learning_rate": 6.9e-05,
+      "loss": 3.9513,
+      "step": 23
+    },
+    {
+      "epoch": 0.0021046025849890864,
+      "grad_norm": 3.21875,
+      "learning_rate": 7.2e-05,
+      "loss": 3.9453,
+      "step": 24
+    },
+    {
+      "epoch": 0.0021922943593636317,
+      "grad_norm": 2.734375,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 3.9338,
+      "step": 25
+    },
+    {
+      "epoch": 0.002279986133738177,
+      "grad_norm": 2.375,
+      "learning_rate": 7.8e-05,
+      "loss": 3.9318,
+      "step": 26
+    },
+    {
+      "epoch": 0.0023676779081127225,
+      "grad_norm": 1.921875,
+      "learning_rate": 8.1e-05,
+      "loss": 3.8758,
+      "step": 27
+    },
+    {
+      "epoch": 0.0024553696824872674,
+      "grad_norm": 1.828125,
+      "learning_rate": 8.400000000000001e-05,
+      "loss": 3.8885,
+      "step": 28
+    },
+    {
+      "epoch": 0.002543061456861813,
+      "grad_norm": 1.765625,
+      "learning_rate": 8.7e-05,
+      "loss": 3.8412,
+      "step": 29
+    },
+    {
+      "epoch": 0.002630753231236358,
+      "grad_norm": 1.65625,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 3.8114,
+      "step": 30
+    },
+    {
+      "epoch": 0.0027184450056109036,
+      "grad_norm": 1.5234375,
+      "learning_rate": 9.3e-05,
+      "loss": 3.778,
+      "step": 31
+    },
+    {
+      "epoch": 0.0028061367799854485,
+      "grad_norm": 1.515625,
+      "learning_rate": 9.6e-05,
+      "loss": 3.7355,
+      "step": 32
+    },
+    {
+      "epoch": 0.002893828554359994,
+      "grad_norm": 1.5,
+      "learning_rate": 9.900000000000001e-05,
+      "loss": 3.7612,
+      "step": 33
+    },
+    {
+      "epoch": 0.0029815203287345392,
+      "grad_norm": 1.390625,
+      "learning_rate": 0.00010200000000000001,
+      "loss": 3.7869,
+      "step": 34
+    },
+    {
+      "epoch": 0.0030692121031090846,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.00010500000000000002,
+      "loss": 3.6998,
+      "step": 35
+    },
+    {
+      "epoch": 0.0031569038774836296,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.000108,
+      "loss": 3.645,
+      "step": 36
+    },
+    {
+      "epoch": 0.003244595651858175,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.000111,
+      "loss": 3.6778,
+      "step": 37
+    },
+    {
+      "epoch": 0.0033322874262327203,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.000114,
+      "loss": 3.6011,
+      "step": 38
+    },
+    {
+      "epoch": 0.0034199792006072657,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.000117,
+      "loss": 3.6147,
+      "step": 39
+    },
+    {
+      "epoch": 0.0035076709749818106,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00012,
+      "loss": 3.5427,
+      "step": 40
+    },
+    {
+      "epoch": 0.003595362749356356,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.000123,
+      "loss": 3.5252,
+      "step": 41
+    },
+    {
+      "epoch": 0.0036830545237309014,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.000126,
+      "loss": 3.5946,
+      "step": 42
+    },
+    {
+      "epoch": 0.0037707462981054467,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.000129,
+      "loss": 3.498,
+      "step": 43
+    },
+    {
+      "epoch": 0.0038584380724799917,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.00013199999999999998,
+      "loss": 3.4045,
+      "step": 44
+    },
+    {
+      "epoch": 0.0039461298468545375,
+      "grad_norm": 1.46875,
+      "learning_rate": 0.000135,
+      "loss": 3.4418,
+      "step": 45
+    },
+    {
+      "epoch": 0.004033821621229082,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.000138,
+      "loss": 3.3624,
+      "step": 46
+    },
+    {
+      "epoch": 0.004121513395603627,
+      "grad_norm": 1.6328125,
+      "learning_rate": 0.000141,
+      "loss": 3.3476,
+      "step": 47
+    },
+    {
+      "epoch": 0.004209205169978173,
+      "grad_norm": 1.5078125,
+      "learning_rate": 0.000144,
+      "loss": 3.2416,
+      "step": 48
+    },
+    {
+      "epoch": 0.004296896944352718,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.000147,
+      "loss": 3.263,
+      "step": 49
+    },
+    {
+      "epoch": 0.0043845887187272635,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 3.2391,
+      "step": 50
+    },
+    {
+      "epoch": 0.004472280493101809,
+      "grad_norm": 0.8671875,
+      "learning_rate": 0.000153,
+      "loss": 3.1366,
+      "step": 51
+    },
+    {
+      "epoch": 0.004559972267476354,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.000156,
+      "loss": 3.103,
+      "step": 52
+    },
+    {
+      "epoch": 0.0046476640418509,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.000159,
+      "loss": 3.1938,
+      "step": 53
+    },
+    {
+      "epoch": 0.004735355816225445,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.000162,
+      "loss": 3.1503,
+      "step": 54
+    },
+    {
+      "epoch": 0.0048230475905999895,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.000165,
+      "loss": 3.1695,
+      "step": 55
+    },
+    {
+      "epoch": 0.004910739364974535,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.00016800000000000002,
+      "loss": 3.0348,
+      "step": 56
+    },
+    {
+      "epoch": 0.00499843113934908,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.000171,
+      "loss": 3.0016,
+      "step": 57
+    },
+    {
+      "epoch": 0.005086122913723626,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.000174,
+      "loss": 2.9536,
+      "step": 58
+    },
+    {
+      "epoch": 0.005173814688098171,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.000177,
+      "loss": 2.9378,
+      "step": 59
+    },
+    {
+      "epoch": 0.005261506462472716,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 2.9442,
+      "step": 60
+    },
+    {
+      "epoch": 0.005349198236847262,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.000183,
+      "loss": 2.8868,
+      "step": 61
+    },
+    {
+      "epoch": 0.005436890011221807,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.000186,
+      "loss": 2.8555,
+      "step": 62
+    },
+    {
+      "epoch": 0.005524581785596352,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.000189,
+      "loss": 2.8699,
+      "step": 63
+    },
+    {
+      "epoch": 0.005612273559970897,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.000192,
+      "loss": 2.9004,
+      "step": 64
+    },
+    {
+      "epoch": 0.005699965334345442,
+      "grad_norm": 0.625,
+      "learning_rate": 0.00019500000000000002,
+      "loss": 2.8415,
+      "step": 65
+    },
+    {
+      "epoch": 0.005787657108719988,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00019800000000000002,
+      "loss": 2.756,
+      "step": 66
+    },
+    {
+      "epoch": 0.005875348883094533,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.000201,
+      "loss": 2.8199,
+      "step": 67
+    },
+    {
+      "epoch": 0.0059630406574690785,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00020400000000000003,
+      "loss": 2.753,
+      "step": 68
+    },
+    {
+      "epoch": 0.006050732431843624,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00020700000000000002,
+      "loss": 2.677,
+      "step": 69
+    },
+    {
+      "epoch": 0.006138424206218169,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00021000000000000004,
+      "loss": 2.7099,
+      "step": 70
+    },
+    {
+      "epoch": 0.006226115980592715,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00021299999999999997,
+      "loss": 2.6848,
+      "step": 71
+    },
+    {
+      "epoch": 0.006313807754967259,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.000216,
+      "loss": 2.6327,
+      "step": 72
+    },
+    {
+      "epoch": 0.0064014995293418045,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.00021899999999999998,
+      "loss": 2.6438,
+      "step": 73
+    },
+    {
+      "epoch": 0.00648919130371635,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.000222,
+      "loss": 2.6331,
+      "step": 74
+    },
+    {
+      "epoch": 0.006576883078090895,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.000225,
+      "loss": 2.616,
+      "step": 75
+    },
+    {
+      "epoch": 0.006664574852465441,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.000228,
+      "loss": 2.4532,
+      "step": 76
+    },
+    {
+      "epoch": 0.006752266626839986,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.000231,
+      "loss": 2.4668,
+      "step": 77
+    },
+    {
+      "epoch": 0.006839958401214531,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.000234,
+      "loss": 2.4524,
+      "step": 78
+    },
+    {
+      "epoch": 0.006927650175589077,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00023700000000000001,
+      "loss": 2.4424,
+      "step": 79
+    },
+    {
+      "epoch": 0.007015341949963621,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.00024,
+      "loss": 2.403,
+      "step": 80
+    },
+    {
+      "epoch": 0.007103033724338167,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.00024300000000000002,
+      "loss": 2.4171,
+      "step": 81
+    },
+    {
+      "epoch": 0.007190725498712712,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.000246,
+      "loss": 2.4317,
+      "step": 82
+    },
+    {
+      "epoch": 0.007278417273087257,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.00024900000000000004,
+      "loss": 2.3678,
+      "step": 83
+    },
+    {
+      "epoch": 0.007366109047461803,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.000252,
+      "loss": 2.3322,
+      "step": 84
+    },
+    {
+      "epoch": 0.007453800821836348,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.000255,
+      "loss": 2.2875,
+      "step": 85
+    },
+    {
+      "epoch": 0.0075414925962108935,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.000258,
+      "loss": 2.2755,
+      "step": 86
+    },
+    {
+      "epoch": 0.007629184370585439,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.000261,
+      "loss": 2.245,
+      "step": 87
+    },
+    {
+      "epoch": 0.007716876144959983,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.00026399999999999997,
+      "loss": 2.1879,
+      "step": 88
+    },
+    {
+      "epoch": 0.007804567919334529,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.000267,
+      "loss": 2.263,
+      "step": 89
+    },
+    {
+      "epoch": 0.007892259693709075,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.00027,
+      "loss": 2.2339,
+      "step": 90
+    },
+    {
+      "epoch": 0.00797995146808362,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.000273,
+      "loss": 2.2315,
+      "step": 91
+    },
+    {
+      "epoch": 0.008067643242458164,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 0.000276,
+      "loss": 2.1785,
+      "step": 92
+    },
+    {
+      "epoch": 0.00815533501683271,
+      "grad_norm": 0.216796875,
+      "learning_rate": 0.000279,
+      "loss": 2.1301,
+      "step": 93
+    },
+    {
+      "epoch": 0.008243026791207255,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.000282,
+      "loss": 2.1555,
+      "step": 94
+    },
+    {
+      "epoch": 0.008330718565581801,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.000285,
+      "loss": 2.0664,
+      "step": 95
+    },
+    {
+      "epoch": 0.008418410339956345,
+      "grad_norm": 0.212890625,
+      "learning_rate": 0.000288,
+      "loss": 2.0808,
+      "step": 96
+    },
+    {
+      "epoch": 0.008506102114330892,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.000291,
+      "loss": 1.9987,
+      "step": 97
+    },
+    {
+      "epoch": 0.008593793888705436,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.000294,
+      "loss": 2.0368,
+      "step": 98
+    },
+    {
+      "epoch": 0.008681485663079982,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.000297,
+      "loss": 2.0081,
+      "step": 99
+    },
+    {
+      "epoch": 0.008769177437454527,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00030000000000000003,
+      "loss": 2.0451,
+      "step": 100
+    },
+    {
+      "epoch": 0.008856869211829071,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.00030300000000000005,
+      "loss": 1.9185,
+      "step": 101
+    },
+    {
+      "epoch": 0.008944560986203618,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.000306,
+      "loss": 1.9602,
+      "step": 102
+    },
+    {
+      "epoch": 0.009032252760578162,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.000309,
+      "loss": 1.9397,
+      "step": 103
+    },
+    {
+      "epoch": 0.009119944534952708,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.000312,
+      "loss": 1.9686,
+      "step": 104
+    },
+    {
+      "epoch": 0.009207636309327253,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.000315,
+      "loss": 1.903,
+      "step": 105
+    },
+    {
+      "epoch": 0.0092953280837018,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.000318,
+      "loss": 1.9059,
+      "step": 106
+    },
+    {
+      "epoch": 0.009383019858076344,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.000321,
+      "loss": 1.8646,
+      "step": 107
+    },
+    {
+      "epoch": 0.00947071163245089,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.000324,
+      "loss": 1.8592,
+      "step": 108
+    },
+    {
+      "epoch": 0.009558403406825434,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.000327,
+      "loss": 1.8891,
+      "step": 109
+    },
+    {
+      "epoch": 0.009646095181199979,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.00033,
+      "loss": 1.7582,
+      "step": 110
+    },
+    {
+      "epoch": 0.009733786955574525,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.000333,
+      "loss": 1.7959,
+      "step": 111
+    },
+    {
+      "epoch": 0.00982147872994907,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.00033600000000000004,
+      "loss": 1.8503,
+      "step": 112
+    },
+    {
+      "epoch": 0.009909170504323616,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.000339,
+      "loss": 1.8528,
+      "step": 113
+    },
+    {
+      "epoch": 0.00999686227869816,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.000342,
+      "loss": 1.7961,
+      "step": 114
+    },
+    {
+      "epoch": 0.010084554053072707,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.00034500000000000004,
+      "loss": 1.7692,
+      "step": 115
+    },
+    {
+      "epoch": 0.010172245827447251,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.000348,
+      "loss": 1.781,
+      "step": 116
+    },
+    {
+      "epoch": 0.010259937601821796,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.000351,
+      "loss": 1.7401,
+      "step": 117
+    },
+    {
+      "epoch": 0.010347629376196342,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.000354,
+      "loss": 1.7695,
+      "step": 118
+    },
+    {
+      "epoch": 0.010435321150570886,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.000357,
+      "loss": 1.744,
+      "step": 119
+    },
+    {
+      "epoch": 0.010523012924945433,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00035999999999999997,
+      "loss": 1.7182,
+      "step": 120
+    },
+    {
+      "epoch": 0.010610704699319977,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.000363,
+      "loss": 1.7426,
+      "step": 121
+    },
+    {
+      "epoch": 0.010698396473694523,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.000366,
+      "loss": 1.6804,
+      "step": 122
+    },
+    {
+      "epoch": 0.010786088248069068,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.000369,
+      "loss": 1.753,
+      "step": 123
+    },
+    {
+      "epoch": 0.010873780022443614,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.000372,
+      "loss": 1.7363,
+      "step": 124
+    },
+    {
+      "epoch": 0.010961471796818159,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.000375,
+      "loss": 1.7182,
+      "step": 125
+    },
+    {
+      "epoch": 0.011049163571192703,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.000378,
+      "loss": 1.6622,
+      "step": 126
+    },
+    {
+      "epoch": 0.01113685534556725,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.000381,
+      "loss": 1.6381,
+      "step": 127
+    },
+    {
+      "epoch": 0.011224547119941794,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.000384,
+      "loss": 1.6721,
+      "step": 128
+    },
+    {
+      "epoch": 0.01131223889431634,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00038700000000000003,
+      "loss": 1.6262,
+      "step": 129
+    },
+    {
+      "epoch": 0.011399930668690885,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.00039000000000000005,
+      "loss": 1.7461,
+      "step": 130
+    },
+    {
+      "epoch": 0.011487622443065431,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 0.000393,
+      "loss": 1.6759,
+      "step": 131
+    },
+    {
+      "epoch": 0.011575314217439975,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.00039600000000000003,
+      "loss": 1.6878,
+      "step": 132
+    },
+    {
+      "epoch": 0.011663005991814522,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00039900000000000005,
+      "loss": 1.6794,
+      "step": 133
+    },
+    {
+      "epoch": 0.011750697766189066,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.000402,
+      "loss": 1.6118,
+      "step": 134
+    },
+    {
+      "epoch": 0.01183838954056361,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.00040500000000000003,
+      "loss": 1.6834,
+      "step": 135
+    },
+    {
+      "epoch": 0.011926081314938157,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.00040800000000000005,
+      "loss": 1.6754,
+      "step": 136
+    },
+    {
+      "epoch": 0.012013773089312701,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.000411,
+      "loss": 1.6149,
+      "step": 137
+    },
+    {
+      "epoch": 0.012101464863687248,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.00041400000000000003,
+      "loss": 1.7056,
+      "step": 138
+    },
+    {
+      "epoch": 0.012189156638061792,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00041700000000000005,
+      "loss": 1.6293,
+      "step": 139
+    },
+    {
+      "epoch": 0.012276848412436338,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.00042000000000000007,
+      "loss": 1.6483,
+      "step": 140
+    },
+    {
+      "epoch": 0.012364540186810883,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.000423,
+      "loss": 1.6171,
+      "step": 141
+    },
+    {
+      "epoch": 0.01245223196118543,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.00042599999999999995,
+      "loss": 1.6129,
+      "step": 142
+    },
+    {
+      "epoch": 0.012539923735559974,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.00042899999999999997,
+      "loss": 1.6066,
+      "step": 143
+    },
+    {
+      "epoch": 0.012627615509934518,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.000432,
+      "loss": 1.5817,
+      "step": 144
+    },
+    {
+      "epoch": 0.012715307284309064,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.000435,
+      "loss": 1.666,
+      "step": 145
+    },
+    {
+      "epoch": 0.012802999058683609,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.00043799999999999997,
+      "loss": 1.6477,
+      "step": 146
+    },
+    {
+      "epoch": 0.012890690833058155,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.000441,
+      "loss": 1.5617,
+      "step": 147
+    },
+    {
+      "epoch": 0.0129783826074327,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.000444,
+      "loss": 1.5307,
+      "step": 148
+    },
+    {
+      "epoch": 0.013066074381807246,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.00044699999999999997,
+      "loss": 1.5576,
+      "step": 149
+    },
+    {
+      "epoch": 0.01315376615618179,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00045,
+      "loss": 1.5612,
+      "step": 150
+    },
+    {
+      "epoch": 0.013241457930556335,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.000453,
+      "loss": 1.5933,
+      "step": 151
+    },
+    {
+      "epoch": 0.013329149704930881,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.000456,
+      "loss": 1.5505,
+      "step": 152
+    },
+    {
+      "epoch": 0.013416841479305426,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.000459,
+      "loss": 1.4969,
+      "step": 153
+    },
+    {
+      "epoch": 0.013504533253679972,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.000462,
+      "loss": 1.5668,
+      "step": 154
+    },
+    {
+      "epoch": 0.013592225028054516,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.000465,
+      "loss": 1.5287,
+      "step": 155
+    },
+    {
+      "epoch": 0.013679916802429063,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.000468,
+      "loss": 1.545,
+      "step": 156
+    },
+    {
+      "epoch": 0.013767608576803607,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.000471,
+      "loss": 1.615,
+      "step": 157
+    },
+    {
+      "epoch": 0.013855300351178153,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.00047400000000000003,
+      "loss": 1.5476,
+      "step": 158
+    },
+    {
+      "epoch": 0.013942992125552698,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.000477,
+      "loss": 1.5439,
+      "step": 159
+    },
+    {
+      "epoch": 0.014030683899927242,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.00048,
+      "loss": 1.5647,
+      "step": 160
+    },
+    {
+      "epoch": 0.014118375674301789,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.00048300000000000003,
+      "loss": 1.6254,
+      "step": 161
+    },
+    {
+      "epoch": 0.014206067448676333,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.00048600000000000005,
+      "loss": 1.4983,
+      "step": 162
+    },
+    {
+      "epoch": 0.01429375922305088,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.0004890000000000001,
+      "loss": 1.5548,
+      "step": 163
+    },
+    {
+      "epoch": 0.014381450997425424,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.000492,
+      "loss": 1.4867,
+      "step": 164
+    },
+    {
+      "epoch": 0.01446914277179997,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.000495,
+      "loss": 1.6054,
+      "step": 165
+    },
+    {
+      "epoch": 0.014556834546174515,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.0004980000000000001,
+      "loss": 1.553,
+      "step": 166
+    },
+    {
+      "epoch": 0.014644526320549061,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.000501,
+      "loss": 1.568,
+      "step": 167
+    },
+    {
+      "epoch": 0.014732218094923605,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.000504,
+      "loss": 1.5792,
+      "step": 168
+    },
+    {
+      "epoch": 0.01481990986929815,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0005070000000000001,
+      "loss": 1.5449,
+      "step": 169
+    },
+    {
+      "epoch": 0.014907601643672696,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.00051,
+      "loss": 1.5593,
+      "step": 170
+    },
+    {
+      "epoch": 0.01499529341804724,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.000513,
+      "loss": 1.4883,
+      "step": 171
+    },
+    {
+      "epoch": 0.015082985192421787,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.000516,
+      "loss": 1.529,
+      "step": 172
+    },
+    {
+      "epoch": 0.015170676966796331,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0005189999999999999,
+      "loss": 1.466,
+      "step": 173
+    },
+    {
+      "epoch": 0.015258368741170878,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.000522,
+      "loss": 1.4907,
+      "step": 174
+    },
+    {
+      "epoch": 0.015346060515545422,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.000525,
+      "loss": 1.5739,
+      "step": 175
+    },
+    {
+      "epoch": 0.015433752289919967,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.0005279999999999999,
+      "loss": 1.5142,
+      "step": 176
+    },
+    {
+      "epoch": 0.015521444064294513,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.000531,
+      "loss": 1.4573,
+      "step": 177
+    },
+    {
+      "epoch": 0.015609135838669057,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.000534,
+      "loss": 1.5181,
+      "step": 178
+    },
+    {
+      "epoch": 0.015696827613043602,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.000537,
+      "loss": 1.4548,
+      "step": 179
+    },
+    {
+      "epoch": 0.01578451938741815,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.00054,
+      "loss": 1.4154,
+      "step": 180
+    },
+    {
+      "epoch": 0.015872211161792694,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.000543,
+      "loss": 1.4346,
+      "step": 181
+    },
+    {
+      "epoch": 0.01595990293616724,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.000546,
+      "loss": 1.6055,
+      "step": 182
+    },
+    {
+      "epoch": 0.016047594710541783,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.000549,
+      "loss": 1.4598,
+      "step": 183
+    },
+    {
+      "epoch": 0.016135286484916328,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.000552,
+      "loss": 1.4608,
+      "step": 184
+    },
+    {
+      "epoch": 0.016222978259290876,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.000555,
+      "loss": 1.5092,
+      "step": 185
+    },
+    {
+      "epoch": 0.01631067003366542,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.000558,
+      "loss": 1.5338,
+      "step": 186
+    },
+    {
+      "epoch": 0.016398361808039965,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.000561,
+      "loss": 1.4916,
+      "step": 187
+    },
+    {
+      "epoch": 0.01648605358241451,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.000564,
+      "loss": 1.4815,
+      "step": 188
+    },
+    {
+      "epoch": 0.016573745356789057,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.000567,
+      "loss": 1.4502,
+      "step": 189
+    },
+    {
+      "epoch": 0.016661437131163602,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.00057,
+      "loss": 1.5195,
+      "step": 190
+    },
+    {
+      "epoch": 0.016749128905538146,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.000573,
+      "loss": 1.545,
+      "step": 191
+    },
+    {
+      "epoch": 0.01683682067991269,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.000576,
+      "loss": 1.5517,
+      "step": 192
+    },
+    {
+      "epoch": 0.016924512454287235,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.000579,
+      "loss": 1.5469,
+      "step": 193
+    },
+    {
+      "epoch": 0.017012204228661783,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.000582,
+      "loss": 1.4311,
+      "step": 194
+    },
+    {
+      "epoch": 0.017099896003036328,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.000585,
+      "loss": 1.5292,
+      "step": 195
+    },
+    {
+      "epoch": 0.017187587777410872,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.000588,
+      "loss": 1.4326,
+      "step": 196
+    },
+    {
+      "epoch": 0.017275279551785417,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.000591,
+      "loss": 1.4556,
+      "step": 197
+    },
+    {
+      "epoch": 0.017362971326159965,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.000594,
+      "loss": 1.4719,
+      "step": 198
+    },
+    {
+      "epoch": 0.01745066310053451,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.0005970000000000001,
+      "loss": 1.5113,
+      "step": 199
+    },
+    {
+      "epoch": 0.017538354874909054,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.0006000000000000001,
+      "loss": 1.4525,
+      "step": 200
+    },
+    {
+      "epoch": 0.0176260466492836,
+      "grad_norm": 0.0888671875,
+      "learning_rate": 0.000603,
+      "loss": 1.4616,
+      "step": 201
+    },
+    {
+      "epoch": 0.017713738423658143,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0006060000000000001,
+      "loss": 1.4662,
+      "step": 202
+    },
+    {
+      "epoch": 0.01780143019803269,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.0006090000000000001,
+      "loss": 1.4347,
+      "step": 203
+    },
+    {
+      "epoch": 0.017889121972407235,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.000612,
+      "loss": 1.5131,
+      "step": 204
+    },
+    {
+      "epoch": 0.01797681374678178,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.000615,
+      "loss": 1.4848,
+      "step": 205
+    },
+    {
+      "epoch": 0.018064505521156324,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.000618,
+      "loss": 1.4201,
+      "step": 206
+    },
+    {
+      "epoch": 0.018152197295530872,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.000621,
+      "loss": 1.4722,
+      "step": 207
+    },
+    {
+      "epoch": 0.018239889069905417,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.000624,
+      "loss": 1.345,
+      "step": 208
+    },
+    {
+      "epoch": 0.01832758084427996,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.000627,
+      "loss": 1.3573,
+      "step": 209
+    },
+    {
+      "epoch": 0.018415272618654506,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.00063,
+      "loss": 1.4277,
+      "step": 210
+    },
+    {
+      "epoch": 0.01850296439302905,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 0.000633,
+      "loss": 1.4629,
+      "step": 211
+    },
+    {
+      "epoch": 0.0185906561674036,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.000636,
+      "loss": 1.4655,
+      "step": 212
+    },
+    {
+      "epoch": 0.018678347941778143,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.000639,
+      "loss": 1.4401,
+      "step": 213
+    },
+    {
+      "epoch": 0.018766039716152687,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.000642,
+      "loss": 1.4852,
+      "step": 214
+    },
+    {
+      "epoch": 0.018853731490527232,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.000645,
+      "loss": 1.3802,
+      "step": 215
+    },
+    {
+      "epoch": 0.01894142326490178,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.000648,
+      "loss": 1.4288,
+      "step": 216
+    },
+    {
+      "epoch": 0.019029115039276324,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.000651,
+      "loss": 1.467,
+      "step": 217
+    },
+    {
+      "epoch": 0.01911680681365087,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.000654,
+      "loss": 1.4569,
+      "step": 218
+    },
+    {
+      "epoch": 0.019204498588025413,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.000657,
+      "loss": 1.4554,
+      "step": 219
+    },
+    {
+      "epoch": 0.019292190362399958,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00066,
+      "loss": 1.4055,
+      "step": 220
+    },
+    {
+      "epoch": 0.019379882136774506,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.0006630000000000001,
+      "loss": 1.4994,
+      "step": 221
+    },
+    {
+      "epoch": 0.01946757391114905,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.000666,
+      "loss": 1.4716,
+      "step": 222
+    },
+    {
+      "epoch": 0.019555265685523595,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.000669,
+      "loss": 1.3835,
+      "step": 223
+    },
+    {
+      "epoch": 0.01964295745989814,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.0006720000000000001,
+      "loss": 1.3807,
+      "step": 224
+    },
+    {
+      "epoch": 0.019730649234272687,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.000675,
+      "loss": 1.4361,
+      "step": 225
+    },
+    {
+      "epoch": 0.019818341008647232,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.000678,
+      "loss": 1.4103,
+      "step": 226
+    },
+    {
+      "epoch": 0.019906032783021776,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.0006810000000000001,
+      "loss": 1.4801,
+      "step": 227
+    },
+    {
+      "epoch": 0.01999372455739632,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.000684,
+      "loss": 1.4361,
+      "step": 228
+    },
+    {
+      "epoch": 0.020081416331770865,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.000687,
+      "loss": 1.4731,
+      "step": 229
+    },
+    {
+      "epoch": 0.020169108106145413,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.0006900000000000001,
+      "loss": 1.4505,
+      "step": 230
+    },
+    {
+      "epoch": 0.020256799880519958,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.000693,
+      "loss": 1.3999,
+      "step": 231
+    },
+    {
+      "epoch": 0.020344491654894502,
+      "grad_norm": 0.0859375,
+      "learning_rate": 0.000696,
+      "loss": 1.4207,
+      "step": 232
+    },
+    {
+      "epoch": 0.020432183429269047,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0006990000000000001,
+      "loss": 1.4174,
+      "step": 233
+    },
+    {
+      "epoch": 0.02051987520364359,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.000702,
+      "loss": 1.4266,
+      "step": 234
+    },
+    {
+      "epoch": 0.02060756697801814,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.000705,
+      "loss": 1.3834,
+      "step": 235
+    },
+    {
+      "epoch": 0.020695258752392684,
+      "grad_norm": 0.09326171875,
+      "learning_rate": 0.000708,
+      "loss": 1.3796,
+      "step": 236
+    },
+    {
+      "epoch": 0.02078295052676723,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0007109999999999999,
+      "loss": 1.4036,
+      "step": 237
+    },
+    {
+      "epoch": 0.020870642301141773,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.000714,
+      "loss": 1.3561,
+      "step": 238
+    },
+    {
+      "epoch": 0.02095833407551632,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 0.000717,
+      "loss": 1.4045,
+      "step": 239
+    },
+    {
+      "epoch": 0.021046025849890865,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0007199999999999999,
+      "loss": 1.41,
+      "step": 240
+    },
+    {
+      "epoch": 0.02113371762426541,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.000723,
+      "loss": 1.4192,
+      "step": 241
+    },
+    {
+      "epoch": 0.021221409398639954,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.000726,
+      "loss": 1.3841,
+      "step": 242
+    },
+    {
+      "epoch": 0.0213091011730145,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.000729,
+      "loss": 1.3593,
+      "step": 243
+    },
+    {
+      "epoch": 0.021396792947389047,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.000732,
+      "loss": 1.3946,
+      "step": 244
+    },
+    {
+      "epoch": 0.02148448472176359,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.000735,
+      "loss": 1.4397,
+      "step": 245
+    },
+    {
+      "epoch": 0.021572176496138136,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.000738,
+      "loss": 1.3494,
+      "step": 246
+    },
+    {
+      "epoch": 0.02165986827051268,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.000741,
+      "loss": 1.4703,
+      "step": 247
+    },
+    {
+      "epoch": 0.02174756004488723,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.000744,
+      "loss": 1.3589,
+      "step": 248
+    },
+    {
+      "epoch": 0.021835251819261773,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.000747,
+      "loss": 1.3954,
+      "step": 249
+    },
+    {
+      "epoch": 0.021922943593636317,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.00075,
+      "loss": 1.3917,
+      "step": 250
+    },
+    {
+      "epoch": 0.022010635368010862,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.000753,
+      "loss": 1.3834,
+      "step": 251
+    },
+    {
+      "epoch": 0.022098327142385406,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.000756,
+      "loss": 1.4159,
+      "step": 252
+    },
+    {
+      "epoch": 0.022186018916759954,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.000759,
+      "loss": 1.4357,
+      "step": 253
+    },
+    {
+      "epoch": 0.0222737106911345,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.000762,
+      "loss": 1.4194,
+      "step": 254
+    },
+    {
+      "epoch": 0.022361402465509043,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.0007650000000000001,
+      "loss": 1.3669,
+      "step": 255
+    },
+    {
+      "epoch": 0.022449094239883588,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.000768,
+      "loss": 1.4522,
+      "step": 256
+    },
+    {
+      "epoch": 0.022536786014258136,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.000771,
+      "loss": 1.3672,
+      "step": 257
+    },
+    {
+      "epoch": 0.02262447778863268,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.0007740000000000001,
+      "loss": 1.3274,
+      "step": 258
+    },
+    {
+      "epoch": 0.022712169563007225,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.000777,
+      "loss": 1.3584,
+      "step": 259
+    },
+    {
+      "epoch": 0.02279986133738177,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.0007800000000000001,
+      "loss": 1.3933,
+      "step": 260
+    },
+    {
+      "epoch": 0.022887553111756314,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0007830000000000001,
+      "loss": 1.4353,
+      "step": 261
+    },
+    {
+      "epoch": 0.022975244886130862,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.000786,
+      "loss": 1.3784,
+      "step": 262
+    },
+    {
+      "epoch": 0.023062936660505406,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.0007890000000000001,
+      "loss": 1.4389,
+      "step": 263
+    },
+    {
+      "epoch": 0.02315062843487995,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.0007920000000000001,
+      "loss": 1.3353,
+      "step": 264
+    },
+    {
+      "epoch": 0.023238320209254495,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.000795,
+      "loss": 1.375,
+      "step": 265
+    },
+    {
+      "epoch": 0.023326011983629043,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 0.0007980000000000001,
+      "loss": 1.3568,
+      "step": 266
+    },
+    {
+      "epoch": 0.023413703758003588,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0008010000000000001,
+      "loss": 1.3975,
+      "step": 267
+    },
+    {
+      "epoch": 0.023501395532378132,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.000804,
+      "loss": 1.4104,
+      "step": 268
+    },
+    {
+      "epoch": 0.023589087306752677,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.0008070000000000001,
+      "loss": 1.4314,
+      "step": 269
+    },
+    {
+      "epoch": 0.02367677908112722,
+      "grad_norm": 0.08935546875,
+      "learning_rate": 0.0008100000000000001,
+      "loss": 1.3977,
+      "step": 270
+    },
+    {
+      "epoch": 0.02376447085550177,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.000813,
+      "loss": 1.4509,
+      "step": 271
+    },
+    {
+      "epoch": 0.023852162629876314,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0008160000000000001,
+      "loss": 1.4231,
+      "step": 272
+    },
+    {
+      "epoch": 0.02393985440425086,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.0008190000000000001,
+      "loss": 1.381,
+      "step": 273
+    },
+    {
+      "epoch": 0.024027546178625403,
+      "grad_norm": 0.224609375,
+      "learning_rate": 0.000822,
+      "loss": 1.423,
+      "step": 274
+    },
+    {
+      "epoch": 0.02411523795299995,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0008250000000000001,
+      "loss": 1.4066,
+      "step": 275
+    },
+    {
+      "epoch": 0.024202929727374495,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.0008280000000000001,
+      "loss": 1.374,
+      "step": 276
+    },
+    {
+      "epoch": 0.02429062150174904,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0008310000000000001,
+      "loss": 1.3806,
+      "step": 277
+    },
+    {
+      "epoch": 0.024378313276123584,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.0008340000000000001,
+      "loss": 1.4127,
+      "step": 278
+    },
+    {
+      "epoch": 0.02446600505049813,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.0008370000000000001,
+      "loss": 1.3966,
+      "step": 279
+    },
+    {
+      "epoch": 0.024553696824872677,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 0.0008400000000000001,
+      "loss": 1.401,
+      "step": 280
+    },
+    {
+      "epoch": 0.02464138859924722,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0008430000000000001,
+      "loss": 1.4582,
+      "step": 281
+    },
+    {
+      "epoch": 0.024729080373621766,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.000846,
+      "loss": 1.3452,
+      "step": 282
+    },
+    {
+      "epoch": 0.02481677214799631,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.0008489999999999999,
+      "loss": 1.3579,
+      "step": 283
+    },
+    {
+      "epoch": 0.02490446392237086,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0008519999999999999,
+      "loss": 1.3803,
+      "step": 284
+    },
+    {
+      "epoch": 0.024992155696745403,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.000855,
+      "loss": 1.3182,
+      "step": 285
+    },
+    {
+      "epoch": 0.025079847471119947,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.0008579999999999999,
+      "loss": 1.4524,
+      "step": 286
+    },
+    {
+      "epoch": 0.025167539245494492,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.000861,
+      "loss": 1.3664,
+      "step": 287
+    },
+    {
+      "epoch": 0.025255231019869036,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.000864,
+      "loss": 1.4199,
+      "step": 288
+    },
+    {
+      "epoch": 0.025342922794243584,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0008669999999999999,
+      "loss": 1.3351,
+      "step": 289
+    },
+    {
+      "epoch": 0.02543061456861813,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.00087,
+      "loss": 1.3552,
+      "step": 290
+    },
+    {
+      "epoch": 0.025518306342992673,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.000873,
+      "loss": 1.3583,
+      "step": 291
+    },
+    {
+      "epoch": 0.025605998117367218,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0008759999999999999,
+      "loss": 1.3731,
+      "step": 292
+    },
+    {
+      "epoch": 0.025693689891741762,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.000879,
+      "loss": 1.3517,
+      "step": 293
+    },
+    {
+      "epoch": 0.02578138166611631,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.000882,
+      "loss": 1.4209,
+      "step": 294
+    },
+    {
+      "epoch": 0.025869073440490855,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0008849999999999999,
+      "loss": 1.3834,
+      "step": 295
+    },
+    {
+      "epoch": 0.0259567652148654,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.000888,
+      "loss": 1.4558,
+      "step": 296
+    },
+    {
+      "epoch": 0.026044456989239944,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.000891,
+      "loss": 1.4401,
+      "step": 297
+    },
+    {
+      "epoch": 0.026132148763614492,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0008939999999999999,
+      "loss": 1.3981,
+      "step": 298
+    },
+    {
+      "epoch": 0.026219840537989036,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.000897,
+      "loss": 1.3309,
+      "step": 299
+    },
+    {
+      "epoch": 0.02630753231236358,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.0009,
+      "loss": 1.34,
+      "step": 300
+    },
+    {
+      "epoch": 0.026395224086738125,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.0009029999999999999,
+      "loss": 1.3813,
+      "step": 301
+    },
+    {
+      "epoch": 0.02648291586111267,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 0.000906,
+      "loss": 1.4285,
+      "step": 302
+    },
+    {
+      "epoch": 0.026570607635487218,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.000909,
+      "loss": 1.3756,
+      "step": 303
+    },
+    {
+      "epoch": 0.026658299409861762,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.000912,
+      "loss": 1.394,
+      "step": 304
+    },
+    {
+      "epoch": 0.026745991184236307,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.000915,
+      "loss": 1.4289,
+      "step": 305
+    },
+    {
+      "epoch": 0.02683368295861085,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.000918,
+      "loss": 1.3585,
+      "step": 306
+    },
+    {
+      "epoch": 0.0269213747329854,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.000921,
+      "loss": 1.3685,
+      "step": 307
+    },
+    {
+      "epoch": 0.027009066507359944,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.000924,
+      "loss": 1.3597,
+      "step": 308
+    },
+    {
+      "epoch": 0.02709675828173449,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 0.000927,
+      "loss": 1.3464,
+      "step": 309
+    },
+    {
+      "epoch": 0.027184450056109033,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.00093,
+      "loss": 1.4275,
+      "step": 310
+    },
+    {
+      "epoch": 0.027272141830483577,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.000933,
+      "loss": 1.3538,
+      "step": 311
+    },
+    {
+      "epoch": 0.027359833604858125,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.000936,
+      "loss": 1.3189,
+      "step": 312
+    },
+    {
+      "epoch": 0.02744752537923267,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.0009390000000000001,
+      "loss": 1.337,
+      "step": 313
+    },
+    {
+      "epoch": 0.027535217153607214,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.000942,
+      "loss": 1.3509,
+      "step": 314
+    },
+    {
+      "epoch": 0.02762290892798176,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.000945,
+      "loss": 1.4155,
+      "step": 315
+    },
+    {
+      "epoch": 0.027710600702356307,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0009480000000000001,
+      "loss": 1.3987,
+      "step": 316
+    },
+    {
+      "epoch": 0.02779829247673085,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.000951,
+      "loss": 1.3961,
+      "step": 317
+    },
+    {
+      "epoch": 0.027885984251105396,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.000954,
+      "loss": 1.3766,
+      "step": 318
+    },
+    {
+      "epoch": 0.02797367602547994,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0009570000000000001,
+      "loss": 1.3738,
+      "step": 319
+    },
+    {
+      "epoch": 0.028061367799854485,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.00096,
+      "loss": 1.4111,
+      "step": 320
+    },
+    {
+      "epoch": 0.028149059574229033,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.000963,
+      "loss": 1.344,
+      "step": 321
+    },
+    {
+      "epoch": 0.028236751348603577,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.0009660000000000001,
+      "loss": 1.3448,
+      "step": 322
+    },
+    {
+      "epoch": 0.028324443122978122,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.000969,
+      "loss": 1.4107,
+      "step": 323
+    },
+    {
+      "epoch": 0.028412134897352666,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.0009720000000000001,
+      "loss": 1.4008,
+      "step": 324
+    },
+    {
+      "epoch": 0.028499826671727214,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.0009750000000000001,
+      "loss": 1.3442,
+      "step": 325
+    },
+    {
+      "epoch": 0.02858751844610176,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.0009780000000000001,
+      "loss": 1.4229,
+      "step": 326
+    },
+    {
+      "epoch": 0.028675210220476303,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.000981,
+      "loss": 1.3201,
+      "step": 327
+    },
+    {
+      "epoch": 0.028762901994850848,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.000984,
+      "loss": 1.4149,
+      "step": 328
+    },
+    {
+      "epoch": 0.028850593769225392,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.000987,
+      "loss": 1.3379,
+      "step": 329
+    },
+    {
+      "epoch": 0.02893828554359994,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.00099,
+      "loss": 1.3845,
+      "step": 330
+    },
+    {
+      "epoch": 0.029025977317974485,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.0009930000000000002,
+      "loss": 1.3366,
+      "step": 331
+    },
+    {
+      "epoch": 0.02911366909234903,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0009960000000000001,
+      "loss": 1.4185,
+      "step": 332
+    },
+    {
+      "epoch": 0.029201360866723574,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.000999,
+      "loss": 1.3812,
+      "step": 333
+    },
+    {
+      "epoch": 0.029289052641098122,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.001002,
+      "loss": 1.3699,
+      "step": 334
+    },
+    {
+      "epoch": 0.029376744415472666,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.001005,
+      "loss": 1.3294,
+      "step": 335
+    },
+    {
+      "epoch": 0.02946443618984721,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.001008,
+      "loss": 1.3399,
+      "step": 336
+    },
+    {
+      "epoch": 0.029552127964221755,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.0010110000000000002,
+      "loss": 1.3641,
+      "step": 337
+    },
+    {
+      "epoch": 0.0296398197385963,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0010140000000000001,
+      "loss": 1.3034,
+      "step": 338
+    },
+    {
+      "epoch": 0.029727511512970848,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.0010170000000000001,
+      "loss": 1.4665,
+      "step": 339
+    },
+    {
+      "epoch": 0.029815203287345392,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.00102,
+      "loss": 1.362,
+      "step": 340
+    },
+    {
+      "epoch": 0.029902895061719937,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.001023,
+      "loss": 1.4033,
+      "step": 341
+    },
+    {
+      "epoch": 0.02999058683609448,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.001026,
+      "loss": 1.3324,
+      "step": 342
+    },
+    {
+      "epoch": 0.030078278610469026,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0010290000000000002,
+      "loss": 1.3564,
+      "step": 343
+    },
+    {
+      "epoch": 0.030165970384843574,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.001032,
+      "loss": 1.4015,
+      "step": 344
+    },
+    {
+      "epoch": 0.03025366215921812,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.001035,
+      "loss": 1.4183,
+      "step": 345
+    },
+    {
+      "epoch": 0.030341353933592663,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.0010379999999999999,
+      "loss": 1.402,
+      "step": 346
+    },
+    {
+      "epoch": 0.030429045707967207,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.001041,
+      "loss": 1.4356,
+      "step": 347
+    },
+    {
+      "epoch": 0.030516737482341755,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.001044,
+      "loss": 1.3444,
+      "step": 348
+    },
+    {
+      "epoch": 0.0306044292567163,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.001047,
+      "loss": 1.2408,
+      "step": 349
+    },
+    {
+      "epoch": 0.030692121031090844,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.00105,
+      "loss": 1.3412,
+      "step": 350
+    },
+    {
+      "epoch": 0.03077981280546539,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.001053,
+      "loss": 1.3695,
+      "step": 351
+    },
+    {
+      "epoch": 0.030867504579839933,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.0010559999999999999,
+      "loss": 1.2612,
+      "step": 352
+    },
+    {
+      "epoch": 0.03095519635421448,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.001059,
+      "loss": 1.3209,
+      "step": 353
+    },
+    {
+      "epoch": 0.031042888128589026,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.001062,
+      "loss": 1.3619,
+      "step": 354
+    },
+    {
+      "epoch": 0.03113057990296357,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.001065,
+      "loss": 1.3898,
+      "step": 355
+    },
+    {
+      "epoch": 0.031218271677338115,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.001068,
+      "loss": 1.3756,
+      "step": 356
+    },
+    {
+      "epoch": 0.03130596345171266,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.001071,
+      "loss": 1.3883,
+      "step": 357
+    },
+    {
+      "epoch": 0.031393655226087204,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.001074,
+      "loss": 1.3457,
+      "step": 358
+    },
+    {
+      "epoch": 0.031481347000461755,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.001077,
+      "loss": 1.3499,
+      "step": 359
+    },
+    {
+      "epoch": 0.0315690387748363,
+      "grad_norm": 0.125,
+      "learning_rate": 0.00108,
+      "loss": 1.385,
+      "step": 360
+    },
+    {
+      "epoch": 0.031656730549210844,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.001083,
+      "loss": 1.3921,
+      "step": 361
+    },
+    {
+      "epoch": 0.03174442232358539,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.001086,
+      "loss": 1.3615,
+      "step": 362
+    },
+    {
+      "epoch": 0.03183211409795993,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.001089,
+      "loss": 1.3125,
+      "step": 363
+    },
+    {
+      "epoch": 0.03191980587233448,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.001092,
+      "loss": 1.3186,
+      "step": 364
+    },
+    {
+      "epoch": 0.03200749764670902,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.001095,
+      "loss": 1.364,
+      "step": 365
+    },
+    {
+      "epoch": 0.03209518942108357,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.001098,
+      "loss": 1.4273,
+      "step": 366
+    },
+    {
+      "epoch": 0.03218288119545811,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.001101,
+      "loss": 1.3585,
+      "step": 367
+    },
+    {
+      "epoch": 0.032270572969832656,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.001104,
+      "loss": 1.3809,
+      "step": 368
+    },
+    {
+      "epoch": 0.03235826474420721,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.001107,
+      "loss": 1.4134,
+      "step": 369
+    },
+    {
+      "epoch": 0.03244595651858175,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.00111,
+      "loss": 1.3046,
+      "step": 370
+    },
+    {
+      "epoch": 0.032533648292956296,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.001113,
+      "loss": 1.44,
+      "step": 371
+    },
+    {
+      "epoch": 0.03262134006733084,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.001116,
+      "loss": 1.3767,
+      "step": 372
+    },
+    {
+      "epoch": 0.032709031841705385,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.001119,
+      "loss": 1.3568,
+      "step": 373
+    },
+    {
+      "epoch": 0.03279672361607993,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.001122,
+      "loss": 1.2876,
+      "step": 374
+    },
+    {
+      "epoch": 0.032884415390454474,
+      "grad_norm": 0.09326171875,
+      "learning_rate": 0.0011250000000000001,
+      "loss": 1.3378,
+      "step": 375
+    },
+    {
+      "epoch": 0.03297210716482902,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.001128,
+      "loss": 1.2907,
+      "step": 376
+    },
+    {
+      "epoch": 0.033059798939203563,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.001131,
+      "loss": 1.3901,
+      "step": 377
+    },
+    {
+      "epoch": 0.033147490713578115,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.001134,
+      "loss": 1.3821,
+      "step": 378
+    },
+    {
+      "epoch": 0.03323518248795266,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.001137,
+      "loss": 1.3682,
+      "step": 379
+    },
+    {
+      "epoch": 0.033322874262327204,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.00114,
+      "loss": 1.3609,
+      "step": 380
+    },
+    {
+      "epoch": 0.03341056603670175,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.0011430000000000001,
+      "loss": 1.3734,
+      "step": 381
+    },
+    {
+      "epoch": 0.03349825781107629,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.001146,
+      "loss": 1.426,
+      "step": 382
+    },
+    {
+      "epoch": 0.03358594958545084,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.001149,
+      "loss": 1.3942,
+      "step": 383
+    },
+    {
+      "epoch": 0.03367364135982538,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.001152,
+      "loss": 1.3952,
+      "step": 384
+    },
+    {
+      "epoch": 0.033761333134199926,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.001155,
+      "loss": 1.4302,
+      "step": 385
+    },
+    {
+      "epoch": 0.03384902490857447,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.001158,
+      "loss": 1.3961,
+      "step": 386
+    },
+    {
+      "epoch": 0.03393671668294902,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0011610000000000001,
+      "loss": 1.3277,
+      "step": 387
+    },
+    {
+      "epoch": 0.03402440845732357,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.001164,
+      "loss": 1.2633,
+      "step": 388
+    },
+    {
+      "epoch": 0.03411210023169811,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.001167,
+      "loss": 1.3971,
+      "step": 389
+    },
+    {
+      "epoch": 0.034199792006072656,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.00117,
+      "loss": 1.3659,
+      "step": 390
+    },
+    {
+      "epoch": 0.0342874837804472,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.001173,
+      "loss": 1.4147,
+      "step": 391
+    },
+    {
+      "epoch": 0.034375175554821745,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.001176,
+      "loss": 1.3504,
+      "step": 392
+    },
+    {
+      "epoch": 0.03446286732919629,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0011790000000000001,
+      "loss": 1.3842,
+      "step": 393
+    },
+    {
+      "epoch": 0.034550559103570834,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 0.001182,
+      "loss": 1.4277,
+      "step": 394
+    },
+    {
+      "epoch": 0.03463825087794538,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.001185,
+      "loss": 1.3052,
+      "step": 395
+    },
+    {
+      "epoch": 0.03472594265231993,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.001188,
+      "loss": 1.41,
+      "step": 396
+    },
+    {
+      "epoch": 0.034813634426694474,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.001191,
+      "loss": 1.3605,
+      "step": 397
+    },
+    {
+      "epoch": 0.03490132620106902,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0011940000000000002,
+      "loss": 1.3762,
+      "step": 398
+    },
+    {
+      "epoch": 0.03498901797544356,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.0011970000000000001,
+      "loss": 1.2598,
+      "step": 399
+    },
+    {
+      "epoch": 0.03507670974981811,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 0.0012000000000000001,
+      "loss": 1.352,
+      "step": 400
+    },
+    {
+      "epoch": 0.03516440152419265,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.001203,
+      "loss": 1.3672,
+      "step": 401
+    },
+    {
+      "epoch": 0.0352520932985672,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.001206,
+      "loss": 1.4493,
+      "step": 402
+    },
+    {
+      "epoch": 0.03533978507294174,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 0.001209,
+      "loss": 1.3753,
+      "step": 403
+    },
+    {
+      "epoch": 0.035427476847316286,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.0012120000000000002,
+      "loss": 1.3092,
+      "step": 404
+    },
+    {
+      "epoch": 0.03551516862169084,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0012150000000000002,
+      "loss": 1.3083,
+      "step": 405
+    },
+    {
+      "epoch": 0.03560286039606538,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0012180000000000001,
+      "loss": 1.2666,
+      "step": 406
+    },
+    {
+      "epoch": 0.035690552170439926,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.0012209999999999999,
+      "loss": 1.31,
+      "step": 407
+    },
+    {
+      "epoch": 0.03577824394481447,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.001224,
+      "loss": 1.3723,
+      "step": 408
+    },
+    {
+      "epoch": 0.035865935719189015,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.001227,
+      "loss": 1.296,
+      "step": 409
+    },
+    {
+      "epoch": 0.03595362749356356,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00123,
+      "loss": 1.3209,
+      "step": 410
+    },
+    {
+      "epoch": 0.036041319267938104,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.001233,
+      "loss": 1.3439,
+      "step": 411
+    },
+    {
+      "epoch": 0.03612901104231265,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.001236,
+      "loss": 1.361,
+      "step": 412
+    },
+    {
+      "epoch": 0.036216702816687193,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0012389999999999999,
+      "loss": 1.3845,
+      "step": 413
+    },
+    {
+      "epoch": 0.036304394591061745,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.001242,
+      "loss": 1.3486,
+      "step": 414
+    },
+    {
+      "epoch": 0.03639208636543629,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.001245,
+      "loss": 1.3611,
+      "step": 415
+    },
+    {
+      "epoch": 0.036479778139810834,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.001248,
+      "loss": 1.3968,
+      "step": 416
+    },
+    {
+      "epoch": 0.03656746991418538,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.001251,
+      "loss": 1.3516,
+      "step": 417
+    },
+    {
+      "epoch": 0.03665516168855992,
+      "grad_norm": 0.216796875,
+      "learning_rate": 0.001254,
+      "loss": 1.3711,
+      "step": 418
+    },
+    {
+      "epoch": 0.03674285346293447,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.0012569999999999999,
+      "loss": 1.3854,
+      "step": 419
+    },
+    {
+      "epoch": 0.03683054523730901,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00126,
+      "loss": 1.3558,
+      "step": 420
+    },
+    {
+      "epoch": 0.036918237011683556,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.001263,
+      "loss": 1.2976,
+      "step": 421
+    },
+    {
+      "epoch": 0.0370059287860581,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.001266,
+      "loss": 1.3582,
+      "step": 422
+    },
+    {
+      "epoch": 0.03709362056043265,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.001269,
+      "loss": 1.3893,
+      "step": 423
+    },
+    {
+      "epoch": 0.0371813123348072,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.001272,
+      "loss": 1.2884,
+      "step": 424
+    },
+    {
+      "epoch": 0.03726900410918174,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.001275,
+      "loss": 1.3434,
+      "step": 425
+    },
+    {
+      "epoch": 0.037356695883556286,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.001278,
+      "loss": 1.3381,
+      "step": 426
+    },
+    {
+      "epoch": 0.03744438765793083,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.001281,
+      "loss": 1.3903,
+      "step": 427
+    },
+    {
+      "epoch": 0.037532079432305375,
+      "grad_norm": 0.08935546875,
+      "learning_rate": 0.001284,
+      "loss": 1.3506,
+      "step": 428
+    },
+    {
+      "epoch": 0.03761977120667992,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.001287,
+      "loss": 1.2621,
+      "step": 429
+    },
+    {
+      "epoch": 0.037707462981054464,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.00129,
+      "loss": 1.3424,
+      "step": 430
+    },
+    {
+      "epoch": 0.03779515475542901,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.001293,
+      "loss": 1.4801,
+      "step": 431
+    },
+    {
+      "epoch": 0.03788284652980356,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.001296,
+      "loss": 1.3269,
+      "step": 432
+    },
+    {
+      "epoch": 0.037970538304178104,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.001299,
+      "loss": 1.3656,
+      "step": 433
+    },
+    {
+      "epoch": 0.03805823007855265,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.001302,
+      "loss": 1.3883,
+      "step": 434
+    },
+    {
+      "epoch": 0.03814592185292719,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.001305,
+      "loss": 1.3636,
+      "step": 435
+    },
+    {
+      "epoch": 0.03823361362730174,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.001308,
+      "loss": 1.3846,
+      "step": 436
+    },
+    {
+      "epoch": 0.03832130540167628,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.001311,
+      "loss": 1.3506,
+      "step": 437
+    },
+    {
+      "epoch": 0.03840899717605083,
+      "grad_norm": 0.0859375,
+      "learning_rate": 0.001314,
+      "loss": 1.3023,
+      "step": 438
+    },
+    {
+      "epoch": 0.03849668895042537,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.001317,
+      "loss": 1.3817,
+      "step": 439
+    },
+    {
+      "epoch": 0.038584380724799916,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00132,
+      "loss": 1.3687,
+      "step": 440
+    },
+    {
+      "epoch": 0.03867207249917447,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.001323,
+      "loss": 1.3563,
+      "step": 441
+    },
+    {
+      "epoch": 0.03875976427354901,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.0013260000000000001,
+      "loss": 1.3338,
+      "step": 442
+    },
+    {
+      "epoch": 0.038847456047923556,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.001329,
+      "loss": 1.3922,
+      "step": 443
+    },
+    {
+      "epoch": 0.0389351478222981,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.001332,
+      "loss": 1.2568,
+      "step": 444
+    },
+    {
+      "epoch": 0.039022839596672645,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.001335,
+      "loss": 1.3628,
+      "step": 445
+    },
+    {
+      "epoch": 0.03911053137104719,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.001338,
+      "loss": 1.3751,
+      "step": 446
+    },
+    {
+      "epoch": 0.039198223145421734,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.001341,
+      "loss": 1.3495,
+      "step": 447
+    },
+    {
+      "epoch": 0.03928591491979628,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0013440000000000001,
+      "loss": 1.317,
+      "step": 448
+    },
+    {
+      "epoch": 0.039373606694170823,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.001347,
+      "loss": 1.3535,
+      "step": 449
+    },
+    {
+      "epoch": 0.039461298468545375,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.00135,
+      "loss": 1.2891,
+      "step": 450
+    },
+    {
+      "epoch": 0.03954899024291992,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.001353,
+      "loss": 1.3959,
+      "step": 451
+    },
+    {
+      "epoch": 0.039636682017294464,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.001356,
+      "loss": 1.3192,
+      "step": 452
+    },
+    {
+      "epoch": 0.03972437379166901,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.001359,
+      "loss": 1.4324,
+      "step": 453
+    },
+    {
+      "epoch": 0.03981206556604355,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.0013620000000000001,
+      "loss": 1.3246,
+      "step": 454
+    },
+    {
+      "epoch": 0.0398997573404181,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.0013650000000000001,
+      "loss": 1.2848,
+      "step": 455
+    },
+    {
+      "epoch": 0.03998744911479264,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.001368,
+      "loss": 1.3973,
+      "step": 456
+    },
+    {
+      "epoch": 0.040075140889167186,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.001371,
+      "loss": 1.3794,
+      "step": 457
+    },
+    {
+      "epoch": 0.04016283266354173,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.001374,
+      "loss": 1.2708,
+      "step": 458
+    },
+    {
+      "epoch": 0.04025052443791628,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0013770000000000002,
+      "loss": 1.3587,
+      "step": 459
+    },
+    {
+      "epoch": 0.04033821621229083,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0013800000000000002,
+      "loss": 1.2727,
+      "step": 460
+    },
+    {
+      "epoch": 0.04042590798666537,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0013830000000000001,
+      "loss": 1.3551,
+      "step": 461
+    },
+    {
+      "epoch": 0.040513599761039916,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.001386,
+      "loss": 1.3897,
+      "step": 462
+    },
+    {
+      "epoch": 0.04060129153541446,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.001389,
+      "loss": 1.3319,
+      "step": 463
+    },
+    {
+      "epoch": 0.040688983309789005,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.001392,
+      "loss": 1.2967,
+      "step": 464
+    },
+    {
+      "epoch": 0.04077667508416355,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.0013950000000000002,
+      "loss": 1.3376,
+      "step": 465
+    },
+    {
+      "epoch": 0.040864366858538094,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0013980000000000002,
+      "loss": 1.3628,
+      "step": 466
+    },
+    {
+      "epoch": 0.04095205863291264,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.0014010000000000001,
+      "loss": 1.2868,
+      "step": 467
+    },
+    {
+      "epoch": 0.04103975040728718,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.001404,
+      "loss": 1.3618,
+      "step": 468
+    },
+    {
+      "epoch": 0.041127442181661734,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 0.001407,
+      "loss": 1.3419,
+      "step": 469
+    },
+    {
+      "epoch": 0.04121513395603628,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.00141,
+      "loss": 1.3019,
+      "step": 470
+    },
+    {
+      "epoch": 0.04130282573041082,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.001413,
+      "loss": 1.4217,
+      "step": 471
+    },
+    {
+      "epoch": 0.04139051750478537,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.001416,
+      "loss": 1.3689,
+      "step": 472
+    },
+    {
+      "epoch": 0.04147820927915991,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.001419,
+      "loss": 1.3036,
+      "step": 473
+    },
+    {
+      "epoch": 0.04156590105353446,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.0014219999999999999,
+      "loss": 1.3835,
+      "step": 474
+    },
+    {
+      "epoch": 0.041653592827909,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.001425,
+      "loss": 1.2816,
+      "step": 475
+    },
+    {
+      "epoch": 0.041741284602283546,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.001428,
+      "loss": 1.3424,
+      "step": 476
+    },
+    {
+      "epoch": 0.04182897637665809,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.001431,
+      "loss": 1.3255,
+      "step": 477
+    },
+    {
+      "epoch": 0.04191666815103264,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.001434,
+      "loss": 1.358,
+      "step": 478
+    },
+    {
+      "epoch": 0.042004359925407186,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.001437,
+      "loss": 1.3379,
+      "step": 479
+    },
+    {
+      "epoch": 0.04209205169978173,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.0014399999999999999,
+      "loss": 1.2963,
+      "step": 480
+    },
+    {
+      "epoch": 0.042179743474156275,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.001443,
+      "loss": 1.2964,
+      "step": 481
+    },
+    {
+      "epoch": 0.04226743524853082,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.001446,
+      "loss": 1.2804,
+      "step": 482
+    },
+    {
+      "epoch": 0.042355127022905364,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.001449,
+      "loss": 1.353,
+      "step": 483
+    },
+    {
+      "epoch": 0.04244281879727991,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.001452,
+      "loss": 1.2805,
+      "step": 484
+    },
+    {
+      "epoch": 0.042530510571654453,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.001455,
+      "loss": 1.3353,
+      "step": 485
+    },
+    {
+      "epoch": 0.042618202346029,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.001458,
+      "loss": 1.4368,
+      "step": 486
+    },
+    {
+      "epoch": 0.04270589412040355,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.001461,
+      "loss": 1.355,
+      "step": 487
+    },
+    {
+      "epoch": 0.042793585894778094,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.001464,
+      "loss": 1.3571,
+      "step": 488
+    },
+    {
+      "epoch": 0.04288127766915264,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.001467,
+      "loss": 1.3144,
+      "step": 489
+    },
+    {
+      "epoch": 0.04296896944352718,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.00147,
+      "loss": 1.3431,
+      "step": 490
+    },
+    {
+      "epoch": 0.04305666121790173,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.001473,
+      "loss": 1.3331,
+      "step": 491
+    },
+    {
+      "epoch": 0.04314435299227627,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.001476,
+      "loss": 1.3873,
+      "step": 492
+    },
+    {
+      "epoch": 0.043232044766650816,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.001479,
+      "loss": 1.3456,
+      "step": 493
+    },
+    {
+      "epoch": 0.04331973654102536,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.001482,
+      "loss": 1.354,
+      "step": 494
+    },
+    {
+      "epoch": 0.043407428315399905,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.001485,
+      "loss": 1.3155,
+      "step": 495
+    },
+    {
+      "epoch": 0.04349512008977446,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.001488,
+      "loss": 1.3318,
+      "step": 496
+    },
+    {
+      "epoch": 0.043582811864149,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.001491,
+      "loss": 1.3467,
+      "step": 497
+    },
+    {
+      "epoch": 0.043670503638523546,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.001494,
+      "loss": 1.33,
+      "step": 498
+    },
+    {
+      "epoch": 0.04375819541289809,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.001497,
+      "loss": 1.3274,
+      "step": 499
+    },
+    {
+      "epoch": 0.043845887187272635,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.0015,
+      "loss": 1.3365,
+      "step": 500
+    },
+    {
+      "epoch": 0.043845887187272635,
+      "eval_loss": 1.3519667387008667,
+      "eval_runtime": 427.9384,
+      "eval_samples_per_second": 33.76,
+      "eval_steps_per_second": 8.44,
+      "step": 500
+    },
+    {
+      "epoch": 0.04393357896164718,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.001503,
+      "loss": 1.3827,
+      "step": 501
+    },
+    {
+      "epoch": 0.044021270736021724,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.001506,
+      "loss": 1.3601,
+      "step": 502
+    },
+    {
+      "epoch": 0.04410896251039627,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.0015090000000000001,
+      "loss": 1.3613,
+      "step": 503
+    },
+    {
+      "epoch": 0.04419665428477081,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.001512,
+      "loss": 1.3476,
+      "step": 504
+    },
+    {
+      "epoch": 0.044284346059145364,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.001515,
+      "loss": 1.382,
+      "step": 505
+    },
+    {
+      "epoch": 0.04437203783351991,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.001518,
+      "loss": 1.3764,
+      "step": 506
+    },
+    {
+      "epoch": 0.04445972960789445,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.001521,
+      "loss": 1.265,
+      "step": 507
+    },
+    {
+      "epoch": 0.044547421382269,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.001524,
+      "loss": 1.3309,
+      "step": 508
+    },
+    {
+      "epoch": 0.04463511315664354,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.0015270000000000001,
+      "loss": 1.3333,
+      "step": 509
+    },
+    {
+      "epoch": 0.04472280493101809,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.0015300000000000001,
+      "loss": 1.3741,
+      "step": 510
+    },
+    {
+      "epoch": 0.04481049670539263,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.001533,
+      "loss": 1.3363,
+      "step": 511
+    },
+    {
+      "epoch": 0.044898188479767176,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.001536,
+      "loss": 1.3045,
+      "step": 512
+    },
+    {
+      "epoch": 0.04498588025414172,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.001539,
+      "loss": 1.4171,
+      "step": 513
+    },
+    {
+      "epoch": 0.04507357202851627,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 0.001542,
+      "loss": 1.3343,
+      "step": 514
+    },
+    {
+      "epoch": 0.045161263802890816,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.0015450000000000001,
+      "loss": 1.3433,
+      "step": 515
+    },
+    {
+      "epoch": 0.04524895557726536,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.0015480000000000001,
+      "loss": 1.281,
+      "step": 516
+    },
+    {
+      "epoch": 0.045336647351639905,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.001551,
+      "loss": 1.3229,
+      "step": 517
+    },
+    {
+      "epoch": 0.04542433912601445,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.001554,
+      "loss": 1.3284,
+      "step": 518
+    },
+    {
+      "epoch": 0.045512030900388994,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.001557,
+      "loss": 1.3185,
+      "step": 519
+    },
+    {
+      "epoch": 0.04559972267476354,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.0015600000000000002,
+      "loss": 1.4157,
+      "step": 520
+    },
+    {
+      "epoch": 0.045687414449138083,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0015630000000000002,
+      "loss": 1.3225,
+      "step": 521
+    },
+    {
+      "epoch": 0.04577510622351263,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0015660000000000001,
+      "loss": 1.2768,
+      "step": 522
+    },
+    {
+      "epoch": 0.04586279799788718,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.001569,
+      "loss": 1.3378,
+      "step": 523
+    },
+    {
+      "epoch": 0.045950489772261724,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.001572,
+      "loss": 1.3312,
+      "step": 524
+    },
+    {
+      "epoch": 0.04603818154663627,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.001575,
+      "loss": 1.3473,
+      "step": 525
+    },
+    {
+      "epoch": 0.04612587332101081,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0015780000000000002,
+      "loss": 1.3193,
+      "step": 526
+    },
+    {
+      "epoch": 0.04621356509538536,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.0015810000000000002,
+      "loss": 1.3617,
+      "step": 527
+    },
+    {
+      "epoch": 0.0463012568697599,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0015840000000000001,
+      "loss": 1.4205,
+      "step": 528
+    },
+    {
+      "epoch": 0.046388948644134446,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 0.001587,
+      "loss": 1.3412,
+      "step": 529
+    },
+    {
+      "epoch": 0.04647664041850899,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.00159,
+      "loss": 1.3731,
+      "step": 530
+    },
+    {
+      "epoch": 0.046564332192883535,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.001593,
+      "loss": 1.3481,
+      "step": 531
+    },
+    {
+      "epoch": 0.04665202396725809,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.0015960000000000002,
+      "loss": 1.3438,
+      "step": 532
+    },
+    {
+      "epoch": 0.04673971574163263,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.0015990000000000002,
+      "loss": 1.3893,
+      "step": 533
+    },
+    {
+      "epoch": 0.046827407516007176,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0016020000000000001,
+      "loss": 1.3365,
+      "step": 534
+    },
+    {
+      "epoch": 0.04691509929038172,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.001605,
+      "loss": 1.3429,
+      "step": 535
+    },
+    {
+      "epoch": 0.047002791064756265,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.001608,
+      "loss": 1.3629,
+      "step": 536
+    },
+    {
+      "epoch": 0.04709048283913081,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.0016110000000000002,
+      "loss": 1.2756,
+      "step": 537
+    },
+    {
+      "epoch": 0.047178174613505354,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.0016140000000000002,
+      "loss": 1.3064,
+      "step": 538
+    },
+    {
+      "epoch": 0.0472658663878799,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0016170000000000002,
+      "loss": 1.3871,
+      "step": 539
+    },
+    {
+      "epoch": 0.04735355816225444,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.0016200000000000001,
+      "loss": 1.2368,
+      "step": 540
+    },
+    {
+      "epoch": 0.047441249936628994,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.001623,
+      "loss": 1.376,
+      "step": 541
+    },
+    {
+      "epoch": 0.04752894171100354,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.001626,
+      "loss": 1.3061,
+      "step": 542
+    },
+    {
+      "epoch": 0.04761663348537808,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0016290000000000002,
+      "loss": 1.2628,
+      "step": 543
+    },
+    {
+      "epoch": 0.04770432525975263,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.0016320000000000002,
+      "loss": 1.2792,
+      "step": 544
+    },
+    {
+      "epoch": 0.04779201703412717,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0016350000000000002,
+      "loss": 1.285,
+      "step": 545
+    },
+    {
+      "epoch": 0.04787970880850172,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.0016380000000000001,
+      "loss": 1.3308,
+      "step": 546
+    },
+    {
+      "epoch": 0.04796740058287626,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.001641,
+      "loss": 1.3833,
+      "step": 547
+    },
+    {
+      "epoch": 0.048055092357250806,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.001644,
+      "loss": 1.3428,
+      "step": 548
+    },
+    {
+      "epoch": 0.04814278413162535,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.0016470000000000002,
+      "loss": 1.3376,
+      "step": 549
+    },
+    {
+      "epoch": 0.0482304759059999,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.0016500000000000002,
+      "loss": 1.4305,
+      "step": 550
+    },
+    {
+      "epoch": 0.048318167680374446,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0016530000000000002,
+      "loss": 1.3828,
+      "step": 551
+    },
+    {
+      "epoch": 0.04840585945474899,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.0016560000000000001,
+      "loss": 1.323,
+      "step": 552
+    },
+    {
+      "epoch": 0.048493551229123535,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.001659,
+      "loss": 1.3733,
+      "step": 553
+    },
+    {
+      "epoch": 0.04858124300349808,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.0016620000000000003,
+      "loss": 1.3101,
+      "step": 554
+    },
+    {
+      "epoch": 0.048668934777872624,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.0016650000000000002,
+      "loss": 1.3806,
+      "step": 555
+    },
+    {
+      "epoch": 0.04875662655224717,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.0016680000000000002,
+      "loss": 1.321,
+      "step": 556
+    },
+    {
+      "epoch": 0.04884431832662171,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0016710000000000002,
+      "loss": 1.3756,
+      "step": 557
+    },
+    {
+      "epoch": 0.04893201010099626,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0016740000000000001,
+      "loss": 1.3487,
+      "step": 558
+    },
+    {
+      "epoch": 0.04901970187537081,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.001677,
+      "loss": 1.2843,
+      "step": 559
+    },
+    {
+      "epoch": 0.049107393649745354,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.0016800000000000003,
+      "loss": 1.3796,
+      "step": 560
+    },
+    {
+      "epoch": 0.0491950854241199,
+      "grad_norm": 0.08056640625,
+      "learning_rate": 0.0016830000000000003,
+      "loss": 1.3371,
+      "step": 561
+    },
+    {
+      "epoch": 0.04928277719849444,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0016860000000000002,
+      "loss": 1.3571,
+      "step": 562
+    },
+    {
+      "epoch": 0.04937046897286899,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.001689,
+      "loss": 1.3039,
+      "step": 563
+    },
+    {
+      "epoch": 0.04945816074724353,
+      "grad_norm": 0.0888671875,
+      "learning_rate": 0.001692,
+      "loss": 1.3606,
+      "step": 564
+    },
+    {
+      "epoch": 0.049545852521618076,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.001695,
+      "loss": 1.348,
+      "step": 565
+    },
+    {
+      "epoch": 0.04963354429599262,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0016979999999999999,
+      "loss": 1.291,
+      "step": 566
+    },
+    {
+      "epoch": 0.049721236070367165,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0017009999999999998,
+      "loss": 1.3581,
+      "step": 567
+    },
+    {
+      "epoch": 0.04980892784474172,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.0017039999999999998,
+      "loss": 1.311,
+      "step": 568
+    },
+    {
+      "epoch": 0.04989661961911626,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.001707,
+      "loss": 1.3596,
+      "step": 569
+    },
+    {
+      "epoch": 0.049984311393490806,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.00171,
+      "loss": 1.3427,
+      "step": 570
+    },
+    {
+      "epoch": 0.05007200316786535,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.001713,
+      "loss": 1.3852,
+      "step": 571
+    },
+    {
+      "epoch": 0.050159694942239895,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0017159999999999999,
+      "loss": 1.3147,
+      "step": 572
+    },
+    {
+      "epoch": 0.05024738671661444,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.0017189999999999998,
+      "loss": 1.3285,
+      "step": 573
+    },
+    {
+      "epoch": 0.050335078490988984,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.001722,
+      "loss": 1.3213,
+      "step": 574
+    },
+    {
+      "epoch": 0.05042277026536353,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.001725,
+      "loss": 1.2918,
+      "step": 575
+    },
+    {
+      "epoch": 0.05051046203973807,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.001728,
+      "loss": 1.3342,
+      "step": 576
+    },
+    {
+      "epoch": 0.05059815381411262,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.001731,
+      "loss": 1.3351,
+      "step": 577
+    },
+    {
+      "epoch": 0.05068584558848717,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0017339999999999999,
+      "loss": 1.4707,
+      "step": 578
+    },
+    {
+      "epoch": 0.05077353736286171,
+      "grad_norm": 0.125,
+      "learning_rate": 0.0017369999999999998,
+      "loss": 1.3766,
+      "step": 579
+    },
+    {
+      "epoch": 0.05086122913723626,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00174,
+      "loss": 1.3022,
+      "step": 580
+    },
+    {
+      "epoch": 0.0509489209116108,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.001743,
+      "loss": 1.3365,
+      "step": 581
+    },
+    {
+      "epoch": 0.05103661268598535,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.001746,
+      "loss": 1.2805,
+      "step": 582
+    },
+    {
+      "epoch": 0.05112430446035989,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.001749,
+      "loss": 1.3641,
+      "step": 583
+    },
+    {
+      "epoch": 0.051211996234734436,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0017519999999999999,
+      "loss": 1.3591,
+      "step": 584
+    },
+    {
+      "epoch": 0.05129968800910898,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.0017549999999999998,
+      "loss": 1.2795,
+      "step": 585
+    },
+    {
+      "epoch": 0.051387379783483525,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.001758,
+      "loss": 1.4044,
+      "step": 586
+    },
+    {
+      "epoch": 0.051475071557858076,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.001761,
+      "loss": 1.2848,
+      "step": 587
+    },
+    {
+      "epoch": 0.05156276333223262,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.001764,
+      "loss": 1.3604,
+      "step": 588
+    },
+    {
+      "epoch": 0.051650455106607165,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.001767,
+      "loss": 1.265,
+      "step": 589
+    },
+    {
+      "epoch": 0.05173814688098171,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.0017699999999999999,
+      "loss": 1.3369,
+      "step": 590
+    },
+    {
+      "epoch": 0.051825838655356254,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.001773,
+      "loss": 1.3202,
+      "step": 591
+    },
+    {
+      "epoch": 0.0519135304297308,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.001776,
+      "loss": 1.3488,
+      "step": 592
+    },
+    {
+      "epoch": 0.05200122220410534,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.001779,
+      "loss": 1.3653,
+      "step": 593
+    },
+    {
+      "epoch": 0.05208891397847989,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.001782,
+      "loss": 1.3817,
+      "step": 594
+    },
+    {
+      "epoch": 0.05217660575285443,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.001785,
+      "loss": 1.3422,
+      "step": 595
+    },
+    {
+      "epoch": 0.052264297527228984,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0017879999999999999,
+      "loss": 1.3877,
+      "step": 596
+    },
+    {
+      "epoch": 0.05235198930160353,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.001791,
+      "loss": 1.3137,
+      "step": 597
+    },
+    {
+      "epoch": 0.05243968107597807,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.001794,
+      "loss": 1.3798,
+      "step": 598
+    },
+    {
+      "epoch": 0.05252737285035262,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.001797,
+      "loss": 1.3497,
+      "step": 599
+    },
+    {
+      "epoch": 0.05261506462472716,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.0018,
+      "loss": 1.3417,
+      "step": 600
+    },
+    {
+      "epoch": 0.052702756399101706,
+      "grad_norm": 0.125,
+      "learning_rate": 0.001803,
+      "loss": 1.3298,
+      "step": 601
+    },
+    {
+      "epoch": 0.05279044817347625,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.0018059999999999999,
+      "loss": 1.3585,
+      "step": 602
+    },
+    {
+      "epoch": 0.052878139947850795,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 0.001809,
+      "loss": 1.331,
+      "step": 603
+    },
+    {
+      "epoch": 0.05296583172222534,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.001812,
+      "loss": 1.2992,
+      "step": 604
+    },
+    {
+      "epoch": 0.05305352349659989,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.001815,
+      "loss": 1.2915,
+      "step": 605
+    },
+    {
+      "epoch": 0.053141215270974436,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.001818,
+      "loss": 1.32,
+      "step": 606
+    },
+    {
+      "epoch": 0.05322890704534898,
+      "grad_norm": 0.0771484375,
+      "learning_rate": 0.001821,
+      "loss": 1.3601,
+      "step": 607
+    },
+    {
+      "epoch": 0.053316598819723525,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.001824,
+      "loss": 1.3336,
+      "step": 608
+    },
+    {
+      "epoch": 0.05340429059409807,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.001827,
+      "loss": 1.347,
+      "step": 609
+    },
+    {
+      "epoch": 0.053491982368472614,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00183,
+      "loss": 1.3347,
+      "step": 610
+    },
+    {
+      "epoch": 0.05357967414284716,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.001833,
+      "loss": 1.3217,
+      "step": 611
+    },
+    {
+      "epoch": 0.0536673659172217,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.001836,
+      "loss": 1.2653,
+      "step": 612
+    },
+    {
+      "epoch": 0.05375505769159625,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.001839,
+      "loss": 1.3131,
+      "step": 613
+    },
+    {
+      "epoch": 0.0538427494659708,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.001842,
+      "loss": 1.3646,
+      "step": 614
+    },
+    {
+      "epoch": 0.05393044124034534,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.001845,
+      "loss": 1.3012,
+      "step": 615
+    },
+    {
+      "epoch": 0.05401813301471989,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.001848,
+      "loss": 1.2882,
+      "step": 616
+    },
+    {
+      "epoch": 0.05410582478909443,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.001851,
+      "loss": 1.2802,
+      "step": 617
+    },
+    {
+      "epoch": 0.05419351656346898,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.001854,
+      "loss": 1.3446,
+      "step": 618
+    },
+    {
+      "epoch": 0.05428120833784352,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.001857,
+      "loss": 1.3352,
+      "step": 619
+    },
+    {
+      "epoch": 0.054368900112218066,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00186,
+      "loss": 1.3795,
+      "step": 620
+    },
+    {
+      "epoch": 0.05445659188659261,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.001863,
+      "loss": 1.3184,
+      "step": 621
+    },
+    {
+      "epoch": 0.054544283660967155,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.001866,
+      "loss": 1.306,
+      "step": 622
+    },
+    {
+      "epoch": 0.054631975435341706,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.001869,
+      "loss": 1.3334,
+      "step": 623
+    },
+    {
+      "epoch": 0.05471966720971625,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.001872,
+      "loss": 1.3268,
+      "step": 624
+    },
+    {
+      "epoch": 0.054807358984090795,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.001875,
+      "loss": 1.3805,
+      "step": 625
+    },
+    {
+      "epoch": 0.05489505075846534,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.0018780000000000001,
+      "loss": 1.4231,
+      "step": 626
+    },
+    {
+      "epoch": 0.054982742532839884,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.001881,
+      "loss": 1.3484,
+      "step": 627
+    },
+    {
+      "epoch": 0.05507043430721443,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.001884,
+      "loss": 1.3057,
+      "step": 628
+    },
+    {
+      "epoch": 0.05515812608158897,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.001887,
+      "loss": 1.3342,
+      "step": 629
+    },
+    {
+      "epoch": 0.05524581785596352,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.00189,
+      "loss": 1.3539,
+      "step": 630
+    },
+    {
+      "epoch": 0.05533350963033806,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0018930000000000002,
+      "loss": 1.3593,
+      "step": 631
+    },
+    {
+      "epoch": 0.055421201404712614,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0018960000000000001,
+      "loss": 1.3262,
+      "step": 632
+    },
+    {
+      "epoch": 0.05550889317908716,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.001899,
+      "loss": 1.3629,
+      "step": 633
+    },
+    {
+      "epoch": 0.0555965849534617,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.001902,
+      "loss": 1.313,
+      "step": 634
+    },
+    {
+      "epoch": 0.05568427672783625,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.001905,
+      "loss": 1.3779,
+      "step": 635
+    },
+    {
+      "epoch": 0.05577196850221079,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.001908,
+      "loss": 1.3722,
+      "step": 636
+    },
+    {
+      "epoch": 0.055859660276585336,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0019110000000000002,
+      "loss": 1.2903,
+      "step": 637
+    },
+    {
+      "epoch": 0.05594735205095988,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.0019140000000000001,
+      "loss": 1.2875,
+      "step": 638
+    },
+    {
+      "epoch": 0.056035043825334425,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.001917,
+      "loss": 1.3276,
+      "step": 639
+    },
+    {
+      "epoch": 0.05612273559970897,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.00192,
+      "loss": 1.2509,
+      "step": 640
+    },
+    {
+      "epoch": 0.05621042737408352,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.001923,
+      "loss": 1.3149,
+      "step": 641
+    },
+    {
+      "epoch": 0.056298119148458066,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.001926,
+      "loss": 1.3123,
+      "step": 642
+    },
+    {
+      "epoch": 0.05638581092283261,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.0019290000000000002,
+      "loss": 1.3152,
+      "step": 643
+    },
+    {
+      "epoch": 0.056473502697207155,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0019320000000000001,
+      "loss": 1.337,
+      "step": 644
+    },
+    {
+      "epoch": 0.0565611944715817,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.001935,
+      "loss": 1.3222,
+      "step": 645
+    },
+    {
+      "epoch": 0.056648886245956244,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.001938,
+      "loss": 1.2985,
+      "step": 646
+    },
+    {
+      "epoch": 0.05673657802033079,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.001941,
+      "loss": 1.4031,
+      "step": 647
+    },
+    {
+      "epoch": 0.05682426979470533,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.0019440000000000002,
+      "loss": 1.2429,
+      "step": 648
+    },
+    {
+      "epoch": 0.05691196156907988,
+      "grad_norm": 0.08154296875,
+      "learning_rate": 0.0019470000000000002,
+      "loss": 1.3375,
+      "step": 649
+    },
+    {
+      "epoch": 0.05699965334345443,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 0.0019500000000000001,
+      "loss": 1.3173,
+      "step": 650
+    },
+    {
+      "epoch": 0.05708734511782897,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.001953,
+      "loss": 1.3144,
+      "step": 651
+    },
+    {
+      "epoch": 0.05717503689220352,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.0019560000000000003,
+      "loss": 1.4079,
+      "step": 652
+    },
+    {
+      "epoch": 0.05726272866657806,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0019590000000000002,
+      "loss": 1.4005,
+      "step": 653
+    },
+    {
+      "epoch": 0.05735042044095261,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.001962,
+      "loss": 1.4134,
+      "step": 654
+    },
+    {
+      "epoch": 0.05743811221532715,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.001965,
+      "loss": 1.3191,
+      "step": 655
+    },
+    {
+      "epoch": 0.057525803989701696,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.001968,
+      "loss": 1.3546,
+      "step": 656
+    },
+    {
+      "epoch": 0.05761349576407624,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.001971,
+      "loss": 1.3522,
+      "step": 657
+    },
+    {
+      "epoch": 0.057701187538450785,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.001974,
+      "loss": 1.318,
+      "step": 658
+    },
+    {
+      "epoch": 0.057788879312825336,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.001977,
+      "loss": 1.3126,
+      "step": 659
+    },
+    {
+      "epoch": 0.05787657108719988,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.00198,
+      "loss": 1.3112,
+      "step": 660
+    },
+    {
+      "epoch": 0.057964262861574425,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.001983,
+      "loss": 1.3347,
+      "step": 661
+    },
+    {
+      "epoch": 0.05805195463594897,
+      "grad_norm": 0.125,
+      "learning_rate": 0.0019860000000000004,
+      "loss": 1.384,
+      "step": 662
+    },
+    {
+      "epoch": 0.058139646410323514,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 0.0019890000000000003,
+      "loss": 1.3456,
+      "step": 663
+    },
+    {
+      "epoch": 0.05822733818469806,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.0019920000000000003,
+      "loss": 1.3589,
+      "step": 664
+    },
+    {
+      "epoch": 0.0583150299590726,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0019950000000000002,
+      "loss": 1.3744,
+      "step": 665
+    },
+    {
+      "epoch": 0.05840272173344715,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.001998,
+      "loss": 1.3614,
+      "step": 666
+    },
+    {
+      "epoch": 0.05849041350782169,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.002001,
+      "loss": 1.3101,
+      "step": 667
+    },
+    {
+      "epoch": 0.058578105282196244,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.002004,
+      "loss": 1.3496,
+      "step": 668
+    },
+    {
+      "epoch": 0.05866579705657079,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.002007,
+      "loss": 1.3225,
+      "step": 669
+    },
+    {
+      "epoch": 0.05875348883094533,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.00201,
+      "loss": 1.3426,
+      "step": 670
+    },
+    {
+      "epoch": 0.05884118060531988,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.002013,
+      "loss": 1.2791,
+      "step": 671
+    },
+    {
+      "epoch": 0.05892887237969442,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 0.002016,
+      "loss": 1.2956,
+      "step": 672
+    },
+    {
+      "epoch": 0.059016564154068966,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.002019,
+      "loss": 1.2956,
+      "step": 673
+    },
+    {
+      "epoch": 0.05910425592844351,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.0020220000000000004,
+      "loss": 1.3212,
+      "step": 674
+    },
+    {
+      "epoch": 0.059191947702818055,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.0020250000000000003,
+      "loss": 1.3192,
+      "step": 675
+    },
+    {
+      "epoch": 0.0592796394771926,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.0020280000000000003,
+      "loss": 1.373,
+      "step": 676
+    },
+    {
+      "epoch": 0.059367331251567144,
+      "grad_norm": 0.09423828125,
+      "learning_rate": 0.0020310000000000003,
+      "loss": 1.4266,
+      "step": 677
+    },
+    {
+      "epoch": 0.059455023025941696,
+      "grad_norm": 0.08447265625,
+      "learning_rate": 0.0020340000000000002,
+      "loss": 1.3394,
+      "step": 678
+    },
+    {
+      "epoch": 0.05954271480031624,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.002037,
+      "loss": 1.3182,
+      "step": 679
+    },
+    {
+      "epoch": 0.059630406574690785,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.00204,
+      "loss": 1.3112,
+      "step": 680
+    },
+    {
+      "epoch": 0.05971809834906533,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.002043,
+      "loss": 1.2881,
+      "step": 681
+    },
+    {
+      "epoch": 0.059805790123439874,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.002046,
+      "loss": 1.3234,
+      "step": 682
+    },
+    {
+      "epoch": 0.05989348189781442,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.002049,
+      "loss": 1.3557,
+      "step": 683
+    },
+    {
+      "epoch": 0.05998117367218896,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.002052,
+      "loss": 1.3126,
+      "step": 684
+    },
+    {
+      "epoch": 0.06006886544656351,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.0020550000000000004,
+      "loss": 1.3457,
+      "step": 685
+    },
+    {
+      "epoch": 0.06015655722093805,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.0020580000000000004,
+      "loss": 1.295,
+      "step": 686
+    },
+    {
+      "epoch": 0.0602442489953126,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.0020610000000000003,
+      "loss": 1.3832,
+      "step": 687
+    },
+    {
+      "epoch": 0.06033194076968715,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.002064,
+      "loss": 1.338,
+      "step": 688
+    },
+    {
+      "epoch": 0.06041963254406169,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.002067,
+      "loss": 1.3339,
+      "step": 689
+    },
+    {
+      "epoch": 0.06050732431843624,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.00207,
+      "loss": 1.3436,
+      "step": 690
+    },
+    {
+      "epoch": 0.06059501609281078,
+      "grad_norm": 0.09423828125,
+      "learning_rate": 0.0020729999999999998,
+      "loss": 1.3324,
+      "step": 691
+    },
+    {
+      "epoch": 0.060682707867185326,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0020759999999999997,
+      "loss": 1.3587,
+      "step": 692
+    },
+    {
+      "epoch": 0.06077039964155987,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.0020789999999999997,
+      "loss": 1.3409,
+      "step": 693
+    },
+    {
+      "epoch": 0.060858091415934415,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.002082,
+      "loss": 1.358,
+      "step": 694
+    },
+    {
+      "epoch": 0.06094578319030896,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.002085,
+      "loss": 1.2721,
+      "step": 695
+    },
+    {
+      "epoch": 0.06103347496468351,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.002088,
+      "loss": 1.3178,
+      "step": 696
+    },
+    {
+      "epoch": 0.061121166739058055,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.002091,
+      "loss": 1.2921,
+      "step": 697
+    },
+    {
+      "epoch": 0.0612088585134326,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.002094,
+      "loss": 1.2661,
+      "step": 698
+    },
+    {
+      "epoch": 0.061296550287807144,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.002097,
+      "loss": 1.3789,
+      "step": 699
+    },
+    {
+      "epoch": 0.06138424206218169,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.0021,
+      "loss": 1.3957,
+      "step": 700
+    },
+    {
+      "epoch": 0.06147193383655623,
+      "grad_norm": 0.125,
+      "learning_rate": 0.002103,
+      "loss": 1.348,
+      "step": 701
+    },
+    {
+      "epoch": 0.06155962561093078,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.002106,
+      "loss": 1.3111,
+      "step": 702
+    },
+    {
+      "epoch": 0.06164731738530532,
+      "grad_norm": 0.08251953125,
+      "learning_rate": 0.0021089999999999998,
+      "loss": 1.321,
+      "step": 703
+    },
+    {
+      "epoch": 0.06173500915967987,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.0021119999999999997,
+      "loss": 1.4042,
+      "step": 704
+    },
+    {
+      "epoch": 0.06182270093405442,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.002115,
+      "loss": 1.2891,
+      "step": 705
+    },
+    {
+      "epoch": 0.06191039270842896,
+      "grad_norm": 0.125,
+      "learning_rate": 0.002118,
+      "loss": 1.2994,
+      "step": 706
+    },
+    {
+      "epoch": 0.06199808448280351,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.002121,
+      "loss": 1.4005,
+      "step": 707
+    },
+    {
+      "epoch": 0.06208577625717805,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.002124,
+      "loss": 1.3539,
+      "step": 708
+    },
+    {
+      "epoch": 0.062173468031552596,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.002127,
+      "loss": 1.3263,
+      "step": 709
+    },
+    {
+      "epoch": 0.06226115980592714,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.00213,
+      "loss": 1.3272,
+      "step": 710
+    },
+    {
+      "epoch": 0.062348851580301685,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.002133,
+      "loss": 1.3914,
+      "step": 711
+    },
+    {
+      "epoch": 0.06243654335467623,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.002136,
+      "loss": 1.3741,
+      "step": 712
+    },
+    {
+      "epoch": 0.06252423512905078,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.002139,
+      "loss": 1.3691,
+      "step": 713
+    },
+    {
+      "epoch": 0.06261192690342532,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.002142,
+      "loss": 1.3387,
+      "step": 714
+    },
+    {
+      "epoch": 0.06269961867779987,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.0021449999999999998,
+      "loss": 1.3177,
+      "step": 715
+    },
+    {
+      "epoch": 0.06278731045217441,
+      "grad_norm": 0.08447265625,
+      "learning_rate": 0.002148,
+      "loss": 1.3291,
+      "step": 716
+    },
+    {
+      "epoch": 0.06287500222654896,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.002151,
+      "loss": 1.3442,
+      "step": 717
+    },
+    {
+      "epoch": 0.06296269400092351,
+      "grad_norm": 0.08154296875,
+      "learning_rate": 0.002154,
+      "loss": 1.352,
+      "step": 718
+    },
+    {
+      "epoch": 0.06305038577529805,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.002157,
+      "loss": 1.3209,
+      "step": 719
+    },
+    {
+      "epoch": 0.0631380775496726,
+      "grad_norm": 0.080078125,
+      "learning_rate": 0.00216,
+      "loss": 1.3168,
+      "step": 720
+    },
+    {
+      "epoch": 0.06322576932404714,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.002163,
+      "loss": 1.3624,
+      "step": 721
+    },
+    {
+      "epoch": 0.06331346109842169,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.002166,
+      "loss": 1.3289,
+      "step": 722
+    },
+    {
+      "epoch": 0.06340115287279623,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.002169,
+      "loss": 1.3062,
+      "step": 723
+    },
+    {
+      "epoch": 0.06348884464717078,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.002172,
+      "loss": 1.2927,
+      "step": 724
+    },
+    {
+      "epoch": 0.06357653642154532,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.002175,
+      "loss": 1.3349,
+      "step": 725
+    },
+    {
+      "epoch": 0.06366422819591987,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.002178,
+      "loss": 1.4218,
+      "step": 726
+    },
+    {
+      "epoch": 0.0637519199702944,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.0021809999999999998,
+      "loss": 1.4004,
+      "step": 727
+    },
+    {
+      "epoch": 0.06383961174466896,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.002184,
+      "loss": 1.3236,
+      "step": 728
+    },
+    {
+      "epoch": 0.06392730351904351,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.002187,
+      "loss": 1.2873,
+      "step": 729
+    },
+    {
+      "epoch": 0.06401499529341804,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.00219,
+      "loss": 1.3388,
+      "step": 730
+    },
+    {
+      "epoch": 0.0641026870677926,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.002193,
+      "loss": 1.3179,
+      "step": 731
+    },
+    {
+      "epoch": 0.06419037884216713,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.002196,
+      "loss": 1.3483,
+      "step": 732
+    },
+    {
+      "epoch": 0.06427807061654169,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.002199,
+      "loss": 1.3194,
+      "step": 733
+    },
+    {
+      "epoch": 0.06436576239091622,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.002202,
+      "loss": 1.3404,
+      "step": 734
+    },
+    {
+      "epoch": 0.06445345416529077,
+      "grad_norm": 0.125,
+      "learning_rate": 0.002205,
+      "loss": 1.2977,
+      "step": 735
+    },
+    {
+      "epoch": 0.06454114593966531,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.002208,
+      "loss": 1.362,
+      "step": 736
+    },
+    {
+      "epoch": 0.06462883771403986,
+      "grad_norm": 0.09423828125,
+      "learning_rate": 0.002211,
+      "loss": 1.3174,
+      "step": 737
+    },
+    {
+      "epoch": 0.06471652948841441,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.002214,
+      "loss": 1.3242,
+      "step": 738
+    },
+    {
+      "epoch": 0.06480422126278895,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0022170000000000002,
+      "loss": 1.296,
+      "step": 739
+    },
+    {
+      "epoch": 0.0648919130371635,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.00222,
+      "loss": 1.3848,
+      "step": 740
+    },
+    {
+      "epoch": 0.06497960481153804,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.002223,
+      "loss": 1.3586,
+      "step": 741
+    },
+    {
+      "epoch": 0.06506729658591259,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.002226,
+      "loss": 1.2866,
+      "step": 742
+    },
+    {
+      "epoch": 0.06515498836028713,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.002229,
+      "loss": 1.2886,
+      "step": 743
+    },
+    {
+      "epoch": 0.06524268013466168,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.002232,
+      "loss": 1.3764,
+      "step": 744
+    },
+    {
+      "epoch": 0.06533037190903622,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.002235,
+      "loss": 1.3943,
+      "step": 745
+    },
+    {
+      "epoch": 0.06541806368341077,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.002238,
+      "loss": 1.3645,
+      "step": 746
+    },
+    {
+      "epoch": 0.06550575545778532,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.002241,
+      "loss": 1.2752,
+      "step": 747
+    },
+    {
+      "epoch": 0.06559344723215986,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.002244,
+      "loss": 1.2695,
+      "step": 748
+    },
+    {
+      "epoch": 0.06568113900653441,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.002247,
+      "loss": 1.3008,
+      "step": 749
+    },
+    {
+      "epoch": 0.06576883078090895,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.0022500000000000003,
+      "loss": 1.2855,
+      "step": 750
+    },
+    {
+      "epoch": 0.0658565225552835,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.0022530000000000002,
+      "loss": 1.3534,
+      "step": 751
+    },
+    {
+      "epoch": 0.06594421432965804,
+      "grad_norm": 0.08642578125,
+      "learning_rate": 0.002256,
+      "loss": 1.3676,
+      "step": 752
+    },
+    {
+      "epoch": 0.06603190610403259,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.002259,
+      "loss": 1.2932,
+      "step": 753
+    },
+    {
+      "epoch": 0.06611959787840713,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.002262,
+      "loss": 1.3035,
+      "step": 754
+    },
+    {
+      "epoch": 0.06620728965278168,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.002265,
+      "loss": 1.2848,
+      "step": 755
+    },
+    {
+      "epoch": 0.06629498142715623,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.002268,
+      "loss": 1.2856,
+      "step": 756
+    },
+    {
+      "epoch": 0.06638267320153077,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.002271,
+      "loss": 1.3616,
+      "step": 757
+    },
+    {
+      "epoch": 0.06647036497590532,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.002274,
+      "loss": 1.3047,
+      "step": 758
+    },
+    {
+      "epoch": 0.06655805675027986,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.002277,
+      "loss": 1.4104,
+      "step": 759
+    },
+    {
+      "epoch": 0.06664574852465441,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.00228,
+      "loss": 1.2926,
+      "step": 760
+    },
+    {
+      "epoch": 0.06673344029902895,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.002283,
+      "loss": 1.3173,
+      "step": 761
+    },
+    {
+      "epoch": 0.0668211320734035,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0022860000000000003,
+      "loss": 1.3172,
+      "step": 762
+    },
+    {
+      "epoch": 0.06690882384777803,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.0022890000000000002,
+      "loss": 1.3202,
+      "step": 763
+    },
+    {
+      "epoch": 0.06699651562215259,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.002292,
+      "loss": 1.3341,
+      "step": 764
+    },
+    {
+      "epoch": 0.06708420739652714,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.002295,
+      "loss": 1.3316,
+      "step": 765
+    },
+    {
+      "epoch": 0.06717189917090167,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.002298,
+      "loss": 1.3074,
+      "step": 766
+    },
+    {
+      "epoch": 0.06725959094527623,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.002301,
+      "loss": 1.3714,
+      "step": 767
+    },
+    {
+      "epoch": 0.06734728271965076,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.002304,
+      "loss": 1.3505,
+      "step": 768
+    },
+    {
+      "epoch": 0.06743497449402532,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.002307,
+      "loss": 1.2795,
+      "step": 769
+    },
+    {
+      "epoch": 0.06752266626839985,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.00231,
+      "loss": 1.2834,
+      "step": 770
+    },
+    {
+      "epoch": 0.0676103580427744,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.002313,
+      "loss": 1.2901,
+      "step": 771
+    },
+    {
+      "epoch": 0.06769804981714894,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.002316,
+      "loss": 1.2928,
+      "step": 772
+    },
+    {
+      "epoch": 0.0677857415915235,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.0023190000000000003,
+      "loss": 1.3493,
+      "step": 773
+    },
+    {
+      "epoch": 0.06787343336589804,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0023220000000000003,
+      "loss": 1.3889,
+      "step": 774
+    },
+    {
+      "epoch": 0.06796112514027258,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.0023250000000000002,
+      "loss": 1.2773,
+      "step": 775
+    },
+    {
+      "epoch": 0.06804881691464713,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.002328,
+      "loss": 1.3119,
+      "step": 776
+    },
+    {
+      "epoch": 0.06813650868902167,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.002331,
+      "loss": 1.3411,
+      "step": 777
+    },
+    {
+      "epoch": 0.06822420046339622,
+      "grad_norm": 0.0771484375,
+      "learning_rate": 0.002334,
+      "loss": 1.3557,
+      "step": 778
+    },
+    {
+      "epoch": 0.06831189223777076,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.002337,
+      "loss": 1.3282,
+      "step": 779
+    },
+    {
+      "epoch": 0.06839958401214531,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.00234,
+      "loss": 1.3127,
+      "step": 780
+    },
+    {
+      "epoch": 0.06848727578651985,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.002343,
+      "loss": 1.3505,
+      "step": 781
+    },
+    {
+      "epoch": 0.0685749675608944,
+      "grad_norm": 0.08447265625,
+      "learning_rate": 0.002346,
+      "loss": 1.368,
+      "step": 782
+    },
+    {
+      "epoch": 0.06866265933526895,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.002349,
+      "loss": 1.3141,
+      "step": 783
+    },
+    {
+      "epoch": 0.06875035110964349,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.002352,
+      "loss": 1.3129,
+      "step": 784
+    },
+    {
+      "epoch": 0.06883804288401804,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.0023550000000000003,
+      "loss": 1.3506,
+      "step": 785
+    },
+    {
+      "epoch": 0.06892573465839258,
+      "grad_norm": 0.0888671875,
+      "learning_rate": 0.0023580000000000003,
+      "loss": 1.3198,
+      "step": 786
+    },
+    {
+      "epoch": 0.06901342643276713,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.0023610000000000003,
+      "loss": 1.3936,
+      "step": 787
+    },
+    {
+      "epoch": 0.06910111820714167,
+      "grad_norm": 0.09326171875,
+      "learning_rate": 0.002364,
+      "loss": 1.3044,
+      "step": 788
+    },
+    {
+      "epoch": 0.06918880998151622,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.002367,
+      "loss": 1.3279,
+      "step": 789
+    },
+    {
+      "epoch": 0.06927650175589076,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.00237,
+      "loss": 1.3941,
+      "step": 790
+    },
+    {
+      "epoch": 0.06936419353026531,
+      "grad_norm": 0.08349609375,
+      "learning_rate": 0.002373,
+      "loss": 1.3437,
+      "step": 791
+    },
+    {
+      "epoch": 0.06945188530463986,
+      "grad_norm": 0.0859375,
+      "learning_rate": 0.002376,
+      "loss": 1.4024,
+      "step": 792
+    },
+    {
+      "epoch": 0.0695395770790144,
+      "grad_norm": 0.08642578125,
+      "learning_rate": 0.002379,
+      "loss": 1.3566,
+      "step": 793
+    },
+    {
+      "epoch": 0.06962726885338895,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.002382,
+      "loss": 1.2291,
+      "step": 794
+    },
+    {
+      "epoch": 0.06971496062776349,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 0.002385,
+      "loss": 1.3785,
+      "step": 795
+    },
+    {
+      "epoch": 0.06980265240213804,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.0023880000000000004,
+      "loss": 1.3573,
+      "step": 796
+    },
+    {
+      "epoch": 0.06989034417651258,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0023910000000000003,
+      "loss": 1.3114,
+      "step": 797
+    },
+    {
+      "epoch": 0.06997803595088713,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.0023940000000000003,
+      "loss": 1.3111,
+      "step": 798
+    },
+    {
+      "epoch": 0.07006572772526166,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.0023970000000000003,
+      "loss": 1.416,
+      "step": 799
+    },
+    {
+      "epoch": 0.07015341949963622,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.0024000000000000002,
+      "loss": 1.2953,
+      "step": 800
+    },
+    {
+      "epoch": 0.07024111127401077,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.002403,
+      "loss": 1.3164,
+      "step": 801
+    },
+    {
+      "epoch": 0.0703288030483853,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.002406,
+      "loss": 1.44,
+      "step": 802
+    },
+    {
+      "epoch": 0.07041649482275986,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.002409,
+      "loss": 1.3285,
+      "step": 803
+    },
+    {
+      "epoch": 0.0705041865971344,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.002412,
+      "loss": 1.2978,
+      "step": 804
+    },
+    {
+      "epoch": 0.07059187837150895,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.002415,
+      "loss": 1.3385,
+      "step": 805
+    },
+    {
+      "epoch": 0.07067957014588348,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.002418,
+      "loss": 1.2502,
+      "step": 806
+    },
+    {
+      "epoch": 0.07076726192025803,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.0024210000000000004,
+      "loss": 1.2764,
+      "step": 807
+    },
+    {
+      "epoch": 0.07085495369463257,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.0024240000000000004,
+      "loss": 1.306,
+      "step": 808
+    },
+    {
+      "epoch": 0.07094264546900712,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.0024270000000000003,
+      "loss": 1.2523,
+      "step": 809
+    },
+    {
+      "epoch": 0.07103033724338167,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0024300000000000003,
+      "loss": 1.3436,
+      "step": 810
+    },
+    {
+      "epoch": 0.07111802901775621,
+      "grad_norm": 0.076171875,
+      "learning_rate": 0.0024330000000000003,
+      "loss": 1.329,
+      "step": 811
+    },
+    {
+      "epoch": 0.07120572079213076,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0024360000000000002,
+      "loss": 1.3534,
+      "step": 812
+    },
+    {
+      "epoch": 0.0712934125665053,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.0024389999999999998,
+      "loss": 1.2848,
+      "step": 813
+    },
+    {
+      "epoch": 0.07138110434087985,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.0024419999999999997,
+      "loss": 1.3431,
+      "step": 814
+    },
+    {
+      "epoch": 0.07146879611525439,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.0024449999999999997,
+      "loss": 1.2798,
+      "step": 815
+    },
+    {
+      "epoch": 0.07155648788962894,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.002448,
+      "loss": 1.3048,
+      "step": 816
+    },
+    {
+      "epoch": 0.07164417966400348,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.002451,
+      "loss": 1.2944,
+      "step": 817
+    },
+    {
+      "epoch": 0.07173187143837803,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.002454,
+      "loss": 1.4416,
+      "step": 818
+    },
+    {
+      "epoch": 0.07181956321275258,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.002457,
+      "loss": 1.3192,
+      "step": 819
+    },
+    {
+      "epoch": 0.07190725498712712,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 0.00246,
+      "loss": 1.3287,
+      "step": 820
+    },
+    {
+      "epoch": 0.07199494676150167,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.002463,
+      "loss": 1.3128,
+      "step": 821
+    },
+    {
+      "epoch": 0.07208263853587621,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.002466,
+      "loss": 1.3772,
+      "step": 822
+    },
+    {
+      "epoch": 0.07217033031025076,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.002469,
+      "loss": 1.3326,
+      "step": 823
+    },
+    {
+      "epoch": 0.0722580220846253,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.002472,
+      "loss": 1.3595,
+      "step": 824
+    },
+    {
+      "epoch": 0.07234571385899985,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0024749999999999998,
+      "loss": 1.3271,
+      "step": 825
+    },
+    {
+      "epoch": 0.07243340563337439,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0024779999999999997,
+      "loss": 1.3399,
+      "step": 826
+    },
+    {
+      "epoch": 0.07252109740774894,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.002481,
+      "loss": 1.3322,
+      "step": 827
+    },
+    {
+      "epoch": 0.07260878918212349,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.002484,
+      "loss": 1.3868,
+      "step": 828
+    },
+    {
+      "epoch": 0.07269648095649803,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.002487,
+      "loss": 1.3419,
+      "step": 829
+    },
+    {
+      "epoch": 0.07278417273087258,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.00249,
+      "loss": 1.3186,
+      "step": 830
+    },
+    {
+      "epoch": 0.07287186450524712,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.002493,
+      "loss": 1.2891,
+      "step": 831
+    },
+    {
+      "epoch": 0.07295955627962167,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.002496,
+      "loss": 1.3262,
+      "step": 832
+    },
+    {
+      "epoch": 0.0730472480539962,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 0.002499,
+      "loss": 1.3283,
+      "step": 833
+    },
+    {
+      "epoch": 0.07313493982837076,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.002502,
+      "loss": 1.3316,
+      "step": 834
+    },
+    {
+      "epoch": 0.0732226316027453,
+      "grad_norm": 0.0888671875,
+      "learning_rate": 0.002505,
+      "loss": 1.324,
+      "step": 835
+    },
+    {
+      "epoch": 0.07331032337711985,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.002508,
+      "loss": 1.3424,
+      "step": 836
+    },
+    {
+      "epoch": 0.0733980151514944,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.0025109999999999998,
+      "loss": 1.3264,
+      "step": 837
+    },
+    {
+      "epoch": 0.07348570692586893,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.0025139999999999997,
+      "loss": 1.3647,
+      "step": 838
+    },
+    {
+      "epoch": 0.07357339870024349,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.002517,
+      "loss": 1.3417,
+      "step": 839
+    },
+    {
+      "epoch": 0.07366109047461802,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.00252,
+      "loss": 1.2623,
+      "step": 840
+    },
+    {
+      "epoch": 0.07374878224899258,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.002523,
+      "loss": 1.3707,
+      "step": 841
+    },
+    {
+      "epoch": 0.07383647402336711,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.002526,
+      "loss": 1.398,
+      "step": 842
+    },
+    {
+      "epoch": 0.07392416579774166,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.002529,
+      "loss": 1.319,
+      "step": 843
+    },
+    {
+      "epoch": 0.0740118575721162,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.002532,
+      "loss": 1.324,
+      "step": 844
+    },
+    {
+      "epoch": 0.07409954934649075,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.002535,
+      "loss": 1.2863,
+      "step": 845
+    },
+    {
+      "epoch": 0.0741872411208653,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.002538,
+      "loss": 1.3849,
+      "step": 846
+    },
+    {
+      "epoch": 0.07427493289523984,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.002541,
+      "loss": 1.3583,
+      "step": 847
+    },
+    {
+      "epoch": 0.0743626246696144,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.002544,
+      "loss": 1.383,
+      "step": 848
+    },
+    {
+      "epoch": 0.07445031644398893,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.002547,
+      "loss": 1.2858,
+      "step": 849
+    },
+    {
+      "epoch": 0.07453800821836348,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.00255,
+      "loss": 1.3304,
+      "step": 850
+    },
+    {
+      "epoch": 0.07462569999273802,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.002553,
+      "loss": 1.4514,
+      "step": 851
+    },
+    {
+      "epoch": 0.07471339176711257,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.002556,
+      "loss": 1.3373,
+      "step": 852
+    },
+    {
+      "epoch": 0.07480108354148711,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.002559,
+      "loss": 1.4162,
+      "step": 853
+    },
+    {
+      "epoch": 0.07488877531586166,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.002562,
+      "loss": 1.3607,
+      "step": 854
+    },
+    {
+      "epoch": 0.07497646709023621,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.002565,
+      "loss": 1.3378,
+      "step": 855
+    },
+    {
+      "epoch": 0.07506415886461075,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.002568,
+      "loss": 1.3712,
+      "step": 856
+    },
+    {
+      "epoch": 0.0751518506389853,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.002571,
+      "loss": 1.3485,
+      "step": 857
+    },
+    {
+      "epoch": 0.07523954241335984,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.002574,
+      "loss": 1.3551,
+      "step": 858
+    },
+    {
+      "epoch": 0.07532723418773439,
+      "grad_norm": 0.072265625,
+      "learning_rate": 0.002577,
+      "loss": 1.3312,
+      "step": 859
+    },
+    {
+      "epoch": 0.07541492596210893,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.00258,
+      "loss": 1.3454,
+      "step": 860
+    },
+    {
+      "epoch": 0.07550261773648348,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.0025830000000000002,
+      "loss": 1.3705,
+      "step": 861
+    },
+    {
+      "epoch": 0.07559030951085802,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.002586,
+      "loss": 1.2941,
+      "step": 862
+    },
+    {
+      "epoch": 0.07567800128523257,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.002589,
+      "loss": 1.3392,
+      "step": 863
+    },
+    {
+      "epoch": 0.07576569305960712,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.002592,
+      "loss": 1.3258,
+      "step": 864
+    },
+    {
+      "epoch": 0.07585338483398166,
+      "grad_norm": 0.07568359375,
+      "learning_rate": 0.002595,
+      "loss": 1.288,
+      "step": 865
+    },
+    {
+      "epoch": 0.07594107660835621,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.002598,
+      "loss": 1.4316,
+      "step": 866
+    },
+    {
+      "epoch": 0.07602876838273075,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.002601,
+      "loss": 1.3078,
+      "step": 867
+    },
+    {
+      "epoch": 0.0761164601571053,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.002604,
+      "loss": 1.2932,
+      "step": 868
+    },
+    {
+      "epoch": 0.07620415193147984,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.002607,
+      "loss": 1.3165,
+      "step": 869
+    },
+    {
+      "epoch": 0.07629184370585439,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.00261,
+      "loss": 1.3193,
+      "step": 870
+    },
+    {
+      "epoch": 0.07637953548022892,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.002613,
+      "loss": 1.3277,
+      "step": 871
+    },
+    {
+      "epoch": 0.07646722725460348,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.002616,
+      "loss": 1.2847,
+      "step": 872
+    },
+    {
+      "epoch": 0.07655491902897803,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.0026190000000000002,
+      "loss": 1.335,
+      "step": 873
+    },
+    {
+      "epoch": 0.07664261080335256,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.002622,
+      "loss": 1.2934,
+      "step": 874
+    },
+    {
+      "epoch": 0.07673030257772712,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.002625,
+      "loss": 1.3118,
+      "step": 875
+    },
+    {
+      "epoch": 0.07681799435210165,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.002628,
+      "loss": 1.268,
+      "step": 876
+    },
+    {
+      "epoch": 0.0769056861264762,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.002631,
+      "loss": 1.3718,
+      "step": 877
+    },
+    {
+      "epoch": 0.07699337790085074,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.002634,
+      "loss": 1.347,
+      "step": 878
+    },
+    {
+      "epoch": 0.0770810696752253,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.002637,
+      "loss": 1.321,
+      "step": 879
+    },
+    {
+      "epoch": 0.07716876144959983,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.00264,
+      "loss": 1.3698,
+      "step": 880
+    },
+    {
+      "epoch": 0.07725645322397438,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.002643,
+      "loss": 1.3034,
+      "step": 881
+    },
+    {
+      "epoch": 0.07734414499834893,
+      "grad_norm": 0.08251953125,
+      "learning_rate": 0.002646,
+      "loss": 1.3008,
+      "step": 882
+    },
+    {
+      "epoch": 0.07743183677272347,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.002649,
+      "loss": 1.3692,
+      "step": 883
+    },
+    {
+      "epoch": 0.07751952854709802,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0026520000000000003,
+      "loss": 1.3591,
+      "step": 884
+    },
+    {
+      "epoch": 0.07760722032147256,
+      "grad_norm": 0.0732421875,
+      "learning_rate": 0.0026550000000000002,
+      "loss": 1.2839,
+      "step": 885
+    },
+    {
+      "epoch": 0.07769491209584711,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.002658,
+      "loss": 1.3315,
+      "step": 886
+    },
+    {
+      "epoch": 0.07778260387022165,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.002661,
+      "loss": 1.306,
+      "step": 887
+    },
+    {
+      "epoch": 0.0778702956445962,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.002664,
+      "loss": 1.3655,
+      "step": 888
+    },
+    {
+      "epoch": 0.07795798741897074,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.002667,
+      "loss": 1.3113,
+      "step": 889
+    },
+    {
+      "epoch": 0.07804567919334529,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.00267,
+      "loss": 1.3352,
+      "step": 890
+    },
+    {
+      "epoch": 0.07813337096771984,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.002673,
+      "loss": 1.2859,
+      "step": 891
+    },
+    {
+      "epoch": 0.07822106274209438,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.002676,
+      "loss": 1.3539,
+      "step": 892
+    },
+    {
+      "epoch": 0.07830875451646893,
+      "grad_norm": 0.234375,
+      "learning_rate": 0.002679,
+      "loss": 1.3543,
+      "step": 893
+    },
+    {
+      "epoch": 0.07839644629084347,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.002682,
+      "loss": 1.3276,
+      "step": 894
+    },
+    {
+      "epoch": 0.07848413806521802,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0026850000000000003,
+      "loss": 1.3294,
+      "step": 895
+    },
+    {
+      "epoch": 0.07857182983959256,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.0026880000000000003,
+      "loss": 1.2468,
+      "step": 896
+    },
+    {
+      "epoch": 0.07865952161396711,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.0026910000000000002,
+      "loss": 1.3238,
+      "step": 897
+    },
+    {
+      "epoch": 0.07874721338834165,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.002694,
+      "loss": 1.3838,
+      "step": 898
+    },
+    {
+      "epoch": 0.0788349051627162,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.002697,
+      "loss": 1.3141,
+      "step": 899
+    },
+    {
+      "epoch": 0.07892259693709075,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0027,
+      "loss": 1.2963,
+      "step": 900
+    },
+    {
+      "epoch": 0.07901028871146529,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.002703,
+      "loss": 1.256,
+      "step": 901
+    },
+    {
+      "epoch": 0.07909798048583984,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.002706,
+      "loss": 1.3802,
+      "step": 902
+    },
+    {
+      "epoch": 0.07918567226021438,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.002709,
+      "loss": 1.2523,
+      "step": 903
+    },
+    {
+      "epoch": 0.07927336403458893,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.002712,
+      "loss": 1.2966,
+      "step": 904
+    },
+    {
+      "epoch": 0.07936105580896347,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.002715,
+      "loss": 1.2996,
+      "step": 905
+    },
+    {
+      "epoch": 0.07944874758333802,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.002718,
+      "loss": 1.3653,
+      "step": 906
+    },
+    {
+      "epoch": 0.07953643935771255,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0027210000000000003,
+      "loss": 1.3168,
+      "step": 907
+    },
+    {
+      "epoch": 0.0796241311320871,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.0027240000000000003,
+      "loss": 1.4055,
+      "step": 908
+    },
+    {
+      "epoch": 0.07971182290646166,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0027270000000000003,
+      "loss": 1.3193,
+      "step": 909
+    },
+    {
+      "epoch": 0.0797995146808362,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.0027300000000000002,
+      "loss": 1.3323,
+      "step": 910
+    },
+    {
+      "epoch": 0.07988720645521075,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.002733,
+      "loss": 1.361,
+      "step": 911
+    },
+    {
+      "epoch": 0.07997489822958528,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.002736,
+      "loss": 1.3238,
+      "step": 912
+    },
+    {
+      "epoch": 0.08006259000395984,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.002739,
+      "loss": 1.3205,
+      "step": 913
+    },
+    {
+      "epoch": 0.08015028177833437,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.002742,
+      "loss": 1.2774,
+      "step": 914
+    },
+    {
+      "epoch": 0.08023797355270892,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.002745,
+      "loss": 1.3301,
+      "step": 915
+    },
+    {
+      "epoch": 0.08032566532708346,
+      "grad_norm": 0.09423828125,
+      "learning_rate": 0.002748,
+      "loss": 1.3214,
+      "step": 916
+    },
+    {
+      "epoch": 0.08041335710145801,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.002751,
+      "loss": 1.3142,
+      "step": 917
+    },
+    {
+      "epoch": 0.08050104887583256,
+      "grad_norm": 0.07470703125,
+      "learning_rate": 0.0027540000000000004,
+      "loss": 1.3766,
+      "step": 918
+    },
+    {
+      "epoch": 0.0805887406502071,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.0027570000000000003,
+      "loss": 1.2802,
+      "step": 919
+    },
+    {
+      "epoch": 0.08067643242458165,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.0027600000000000003,
+      "loss": 1.3042,
+      "step": 920
+    },
+    {
+      "epoch": 0.08076412419895619,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.0027630000000000003,
+      "loss": 1.3302,
+      "step": 921
+    },
+    {
+      "epoch": 0.08085181597333074,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0027660000000000002,
+      "loss": 1.3524,
+      "step": 922
+    },
+    {
+      "epoch": 0.08093950774770528,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.002769,
+      "loss": 1.229,
+      "step": 923
+    },
+    {
+      "epoch": 0.08102719952207983,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.002772,
+      "loss": 1.334,
+      "step": 924
+    },
+    {
+      "epoch": 0.08111489129645437,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.002775,
+      "loss": 1.3263,
+      "step": 925
+    },
+    {
+      "epoch": 0.08120258307082892,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.002778,
+      "loss": 1.3455,
+      "step": 926
+    },
+    {
+      "epoch": 0.08129027484520347,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 0.002781,
+      "loss": 1.3536,
+      "step": 927
+    },
+    {
+      "epoch": 0.08137796661957801,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.002784,
+      "loss": 1.3306,
+      "step": 928
+    },
+    {
+      "epoch": 0.08146565839395256,
+      "grad_norm": 0.0771484375,
+      "learning_rate": 0.0027870000000000004,
+      "loss": 1.3462,
+      "step": 929
+    },
+    {
+      "epoch": 0.0815533501683271,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0027900000000000004,
+      "loss": 1.2984,
+      "step": 930
+    },
+    {
+      "epoch": 0.08164104194270165,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0027930000000000003,
+      "loss": 1.3302,
+      "step": 931
+    },
+    {
+      "epoch": 0.08172873371707619,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.0027960000000000003,
+      "loss": 1.356,
+      "step": 932
+    },
+    {
+      "epoch": 0.08181642549145074,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.0027990000000000003,
+      "loss": 1.3102,
+      "step": 933
+    },
+    {
+      "epoch": 0.08190411726582528,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0028020000000000002,
+      "loss": 1.4226,
+      "step": 934
+    },
+    {
+      "epoch": 0.08199180904019983,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.002805,
+      "loss": 1.3052,
+      "step": 935
+    },
+    {
+      "epoch": 0.08207950081457437,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.002808,
+      "loss": 1.3552,
+      "step": 936
+    },
+    {
+      "epoch": 0.08216719258894892,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.002811,
+      "loss": 1.3334,
+      "step": 937
+    },
+    {
+      "epoch": 0.08225488436332347,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.002814,
+      "loss": 1.3238,
+      "step": 938
+    },
+    {
+      "epoch": 0.082342576137698,
+      "grad_norm": 0.08154296875,
+      "learning_rate": 0.002817,
+      "loss": 1.2994,
+      "step": 939
+    },
+    {
+      "epoch": 0.08243026791207256,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.00282,
+      "loss": 1.2673,
+      "step": 940
+    },
+    {
+      "epoch": 0.0825179596864471,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.002823,
+      "loss": 1.2656,
+      "step": 941
+    },
+    {
+      "epoch": 0.08260565146082165,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.002826,
+      "loss": 1.3713,
+      "step": 942
+    },
+    {
+      "epoch": 0.08269334323519618,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.002829,
+      "loss": 1.3689,
+      "step": 943
+    },
+    {
+      "epoch": 0.08278103500957074,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.002832,
+      "loss": 1.3078,
+      "step": 944
+    },
+    {
+      "epoch": 0.08286872678394527,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.002835,
+      "loss": 1.3463,
+      "step": 945
+    },
+    {
+      "epoch": 0.08295641855831982,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.002838,
+      "loss": 1.2728,
+      "step": 946
+    },
+    {
+      "epoch": 0.08304411033269438,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0028409999999999998,
+      "loss": 1.2866,
+      "step": 947
+    },
+    {
+      "epoch": 0.08313180210706891,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.0028439999999999997,
+      "loss": 1.3144,
+      "step": 948
+    },
+    {
+      "epoch": 0.08321949388144347,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.002847,
+      "loss": 1.2829,
+      "step": 949
+    },
+    {
+      "epoch": 0.083307185655818,
+      "grad_norm": 0.07763671875,
+      "learning_rate": 0.00285,
+      "loss": 1.3705,
+      "step": 950
+    },
+    {
+      "epoch": 0.08339487743019255,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.002853,
+      "loss": 1.3597,
+      "step": 951
+    },
+    {
+      "epoch": 0.08348256920456709,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.002856,
+      "loss": 1.3528,
+      "step": 952
+    },
+    {
+      "epoch": 0.08357026097894164,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.002859,
+      "loss": 1.3411,
+      "step": 953
+    },
+    {
+      "epoch": 0.08365795275331618,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.002862,
+      "loss": 1.3656,
+      "step": 954
+    },
+    {
+      "epoch": 0.08374564452769073,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.002865,
+      "loss": 1.3907,
+      "step": 955
+    },
+    {
+      "epoch": 0.08383333630206528,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.002868,
+      "loss": 1.3499,
+      "step": 956
+    },
+    {
+      "epoch": 0.08392102807643982,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.002871,
+      "loss": 1.326,
+      "step": 957
+    },
+    {
+      "epoch": 0.08400871985081437,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.002874,
+      "loss": 1.2958,
+      "step": 958
+    },
+    {
+      "epoch": 0.08409641162518891,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.002877,
+      "loss": 1.2943,
+      "step": 959
+    },
+    {
+      "epoch": 0.08418410339956346,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0028799999999999997,
+      "loss": 1.2968,
+      "step": 960
+    },
+    {
+      "epoch": 0.084271795173938,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.002883,
+      "loss": 1.3424,
+      "step": 961
+    },
+    {
+      "epoch": 0.08435948694831255,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.002886,
+      "loss": 1.326,
+      "step": 962
+    },
+    {
+      "epoch": 0.08444717872268709,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.002889,
+      "loss": 1.3498,
+      "step": 963
+    },
+    {
+      "epoch": 0.08453487049706164,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.002892,
+      "loss": 1.2791,
+      "step": 964
+    },
+    {
+      "epoch": 0.08462256227143619,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 0.002895,
+      "loss": 1.3582,
+      "step": 965
+    },
+    {
+      "epoch": 0.08471025404581073,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.002898,
+      "loss": 1.3123,
+      "step": 966
+    },
+    {
+      "epoch": 0.08479794582018528,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.002901,
+      "loss": 1.3389,
+      "step": 967
+    },
+    {
+      "epoch": 0.08488563759455982,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.002904,
+      "loss": 1.3073,
+      "step": 968
+    },
+    {
+      "epoch": 0.08497332936893437,
+      "grad_norm": 0.388671875,
+      "learning_rate": 0.002907,
+      "loss": 1.3919,
+      "step": 969
+    },
+    {
+      "epoch": 0.08506102114330891,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.00291,
+      "loss": 1.3619,
+      "step": 970
+    },
+    {
+      "epoch": 0.08514871291768346,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.002913,
+      "loss": 1.2457,
+      "step": 971
+    },
+    {
+      "epoch": 0.085236404692058,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.002916,
+      "loss": 1.3531,
+      "step": 972
+    },
+    {
+      "epoch": 0.08532409646643255,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.002919,
+      "loss": 1.3493,
+      "step": 973
+    },
+    {
+      "epoch": 0.0854117882408071,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.002922,
+      "loss": 1.3564,
+      "step": 974
+    },
+    {
+      "epoch": 0.08549948001518164,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.002925,
+      "loss": 1.3556,
+      "step": 975
+    },
+    {
+      "epoch": 0.08558717178955619,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.002928,
+      "loss": 1.3064,
+      "step": 976
+    },
+    {
+      "epoch": 0.08567486356393073,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.002931,
+      "loss": 1.3248,
+      "step": 977
+    },
+    {
+      "epoch": 0.08576255533830528,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.002934,
+      "loss": 1.2805,
+      "step": 978
+    },
+    {
+      "epoch": 0.08585024711267981,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.002937,
+      "loss": 1.3346,
+      "step": 979
+    },
+    {
+      "epoch": 0.08593793888705437,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.00294,
+      "loss": 1.2495,
+      "step": 980
+    },
+    {
+      "epoch": 0.0860256306614289,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.002943,
+      "loss": 1.3576,
+      "step": 981
+    },
+    {
+      "epoch": 0.08611332243580345,
+      "grad_norm": 0.0751953125,
+      "learning_rate": 0.002946,
+      "loss": 1.2991,
+      "step": 982
+    },
+    {
+      "epoch": 0.086201014210178,
+      "grad_norm": 0.08935546875,
+      "learning_rate": 0.0029490000000000002,
+      "loss": 1.4072,
+      "step": 983
+    },
+    {
+      "epoch": 0.08628870598455254,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.002952,
+      "loss": 1.3707,
+      "step": 984
+    },
+    {
+      "epoch": 0.0863763977589271,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.002955,
+      "loss": 1.3455,
+      "step": 985
+    },
+    {
+      "epoch": 0.08646408953330163,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.002958,
+      "loss": 1.3252,
+      "step": 986
+    },
+    {
+      "epoch": 0.08655178130767618,
+      "grad_norm": 0.08935546875,
+      "learning_rate": 0.002961,
+      "loss": 1.3285,
+      "step": 987
+    },
+    {
+      "epoch": 0.08663947308205072,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.002964,
+      "loss": 1.4067,
+      "step": 988
+    },
+    {
+      "epoch": 0.08672716485642527,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.002967,
+      "loss": 1.321,
+      "step": 989
+    },
+    {
+      "epoch": 0.08681485663079981,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.00297,
+      "loss": 1.3814,
+      "step": 990
+    },
+    {
+      "epoch": 0.08690254840517436,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.002973,
+      "loss": 1.3265,
+      "step": 991
+    },
+    {
+      "epoch": 0.08699024017954891,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.002976,
+      "loss": 1.4072,
+      "step": 992
+    },
+    {
+      "epoch": 0.08707793195392345,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.002979,
+      "loss": 1.3561,
+      "step": 993
+    },
+    {
+      "epoch": 0.087165623728298,
+      "grad_norm": 0.08349609375,
+      "learning_rate": 0.002982,
+      "loss": 1.3047,
+      "step": 994
+    },
+    {
+      "epoch": 0.08725331550267254,
+      "grad_norm": 0.0888671875,
+      "learning_rate": 0.0029850000000000002,
+      "loss": 1.2992,
+      "step": 995
+    },
+    {
+      "epoch": 0.08734100727704709,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.002988,
+      "loss": 1.3093,
+      "step": 996
+    },
+    {
+      "epoch": 0.08742869905142163,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.002991,
+      "loss": 1.3291,
+      "step": 997
+    },
+    {
+      "epoch": 0.08751639082579618,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.002994,
+      "loss": 1.4177,
+      "step": 998
+    },
+    {
+      "epoch": 0.08760408260017072,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.002997,
+      "loss": 1.2768,
+      "step": 999
+    },
+    {
+      "epoch": 0.08769177437454527,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.003,
+      "loss": 1.3108,
+      "step": 1000
+    },
+    {
+      "epoch": 0.08769177437454527,
+      "eval_loss": 1.3424164056777954,
+      "eval_runtime": 429.1223,
+      "eval_samples_per_second": 33.666,
+      "eval_steps_per_second": 8.417,
+      "step": 1000
+    },
+    {
+      "epoch": 0.08777946614891982,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0029999999384417424,
+      "loss": 1.3823,
+      "step": 1001
+    },
+    {
+      "epoch": 0.08786715792329436,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0029999997537669756,
+      "loss": 1.4031,
+      "step": 1002
+    },
+    {
+      "epoch": 0.08795484969766891,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.002999999445975716,
+      "loss": 1.3088,
+      "step": 1003
+    },
+    {
+      "epoch": 0.08804254147204345,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0029999990150679926,
+      "loss": 1.3848,
+      "step": 1004
+    },
+    {
+      "epoch": 0.088130233246418,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.002999998461043843,
+      "loss": 1.3547,
+      "step": 1005
+    },
+    {
+      "epoch": 0.08821792502079254,
+      "grad_norm": 0.08642578125,
+      "learning_rate": 0.0029999977839033198,
+      "loss": 1.2681,
+      "step": 1006
+    },
+    {
+      "epoch": 0.08830561679516709,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0029999969836464833,
+      "loss": 1.3275,
+      "step": 1007
+    },
+    {
+      "epoch": 0.08839330856954163,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.002999996060273407,
+      "loss": 1.3565,
+      "step": 1008
+    },
+    {
+      "epoch": 0.08848100034391618,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.0029999950137841744,
+      "loss": 1.2566,
+      "step": 1009
+    },
+    {
+      "epoch": 0.08856869211829073,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.002999993844178882,
+      "loss": 1.3053,
+      "step": 1010
+    },
+    {
+      "epoch": 0.08865638389266527,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.002999992551457636,
+      "loss": 1.307,
+      "step": 1011
+    },
+    {
+      "epoch": 0.08874407566703982,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.002999991135620554,
+      "loss": 1.3178,
+      "step": 1012
+    },
+    {
+      "epoch": 0.08883176744141436,
+      "grad_norm": 0.0810546875,
+      "learning_rate": 0.0029999895966677658,
+      "loss": 1.2969,
+      "step": 1013
+    },
+    {
+      "epoch": 0.0889194592157889,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.0029999879345994113,
+      "loss": 1.378,
+      "step": 1014
+    },
+    {
+      "epoch": 0.08900715099016344,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.002999986149415642,
+      "loss": 1.3091,
+      "step": 1015
+    },
+    {
+      "epoch": 0.089094842764538,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.002999984241116621,
+      "loss": 1.3379,
+      "step": 1016
+    },
+    {
+      "epoch": 0.08918253453891253,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.0029999822097025223,
+      "loss": 1.3691,
+      "step": 1017
+    },
+    {
+      "epoch": 0.08927022631328708,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.0029999800551735304,
+      "loss": 1.3491,
+      "step": 1018
+    },
+    {
+      "epoch": 0.08935791808766164,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.002999977777529843,
+      "loss": 1.3031,
+      "step": 1019
+    },
+    {
+      "epoch": 0.08944560986203617,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.002999975376771667,
+      "loss": 1.3721,
+      "step": 1020
+    },
+    {
+      "epoch": 0.08953330163641073,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.0029999728528992214,
+      "loss": 1.3064,
+      "step": 1021
+    },
+    {
+      "epoch": 0.08962099341078526,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.002999970205912737,
+      "loss": 1.3724,
+      "step": 1022
+    },
+    {
+      "epoch": 0.08970868518515981,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.002999967435812455,
+      "loss": 1.3757,
+      "step": 1023
+    },
+    {
+      "epoch": 0.08979637695953435,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0029999645425986265,
+      "loss": 1.3566,
+      "step": 1024
+    },
+    {
+      "epoch": 0.0898840687339089,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.0029999615262715183,
+      "loss": 1.393,
+      "step": 1025
+    },
+    {
+      "epoch": 0.08997176050828344,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0029999583868314025,
+      "loss": 1.3272,
+      "step": 1026
+    },
+    {
+      "epoch": 0.09005945228265799,
+      "grad_norm": 0.224609375,
+      "learning_rate": 0.0029999551242785674,
+      "loss": 1.3077,
+      "step": 1027
+    },
+    {
+      "epoch": 0.09014714405703254,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0029999517386133097,
+      "loss": 1.2973,
+      "step": 1028
+    },
+    {
+      "epoch": 0.09023483583140708,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.0029999482298359386,
+      "loss": 1.3414,
+      "step": 1029
+    },
+    {
+      "epoch": 0.09032252760578163,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.002999944597946773,
+      "loss": 1.2904,
+      "step": 1030
+    },
+    {
+      "epoch": 0.09041021938015617,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.0029999408429461456,
+      "loss": 1.3405,
+      "step": 1031
+    },
+    {
+      "epoch": 0.09049791115453072,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.0029999369648343976,
+      "loss": 1.3186,
+      "step": 1032
+    },
+    {
+      "epoch": 0.09058560292890526,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.0029999329636118837,
+      "loss": 1.2944,
+      "step": 1033
+    },
+    {
+      "epoch": 0.09067329470327981,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0029999288392789686,
+      "loss": 1.3552,
+      "step": 1034
+    },
+    {
+      "epoch": 0.09076098647765435,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.002999924591836028,
+      "loss": 1.3493,
+      "step": 1035
+    },
+    {
+      "epoch": 0.0908486782520289,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.002999920221283449,
+      "loss": 1.3785,
+      "step": 1036
+    },
+    {
+      "epoch": 0.09093637002640345,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.002999915727621631,
+      "loss": 1.341,
+      "step": 1037
+    },
+    {
+      "epoch": 0.09102406180077799,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0029999111108509834,
+      "loss": 1.3182,
+      "step": 1038
+    },
+    {
+      "epoch": 0.09111175357515254,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0029999063709719278,
+      "loss": 1.3108,
+      "step": 1039
+    },
+    {
+      "epoch": 0.09119944534952708,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.002999901507984895,
+      "loss": 1.3294,
+      "step": 1040
+    },
+    {
+      "epoch": 0.09128713712390163,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0029998965218903297,
+      "loss": 1.2971,
+      "step": 1041
+    },
+    {
+      "epoch": 0.09137482889827617,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.0029998914126886864,
+      "loss": 1.3484,
+      "step": 1042
+    },
+    {
+      "epoch": 0.09146252067265072,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.002999886180380431,
+      "loss": 1.3318,
+      "step": 1043
+    },
+    {
+      "epoch": 0.09155021244702526,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0029998808249660407,
+      "loss": 1.2998,
+      "step": 1044
+    },
+    {
+      "epoch": 0.09163790422139981,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.0029998753464460038,
+      "loss": 1.3707,
+      "step": 1045
+    },
+    {
+      "epoch": 0.09172559599577436,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.0029998697448208205,
+      "loss": 1.348,
+      "step": 1046
+    },
+    {
+      "epoch": 0.0918132877701489,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0029998640200910006,
+      "loss": 1.3921,
+      "step": 1047
+    },
+    {
+      "epoch": 0.09190097954452345,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.002999858172257067,
+      "loss": 1.368,
+      "step": 1048
+    },
+    {
+      "epoch": 0.09198867131889799,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.0029998522013195525,
+      "loss": 1.3864,
+      "step": 1049
+    },
+    {
+      "epoch": 0.09207636309327254,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0029998461072790017,
+      "loss": 1.3486,
+      "step": 1050
+    },
+    {
+      "epoch": 0.09216405486764707,
+      "grad_norm": 0.08203125,
+      "learning_rate": 0.002999839890135971,
+      "loss": 1.3157,
+      "step": 1051
+    },
+    {
+      "epoch": 0.09225174664202163,
+      "grad_norm": 0.072265625,
+      "learning_rate": 0.0029998335498910263,
+      "loss": 1.3175,
+      "step": 1052
+    },
+    {
+      "epoch": 0.09233943841639616,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.002999827086544747,
+      "loss": 1.4045,
+      "step": 1053
+    },
+    {
+      "epoch": 0.09242713019077071,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.0029998205000977213,
+      "loss": 1.3117,
+      "step": 1054
+    },
+    {
+      "epoch": 0.09251482196514527,
+      "grad_norm": 0.08056640625,
+      "learning_rate": 0.0029998137905505513,
+      "loss": 1.3054,
+      "step": 1055
+    },
+    {
+      "epoch": 0.0926025137395198,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.0029998069579038476,
+      "loss": 1.342,
+      "step": 1056
+    },
+    {
+      "epoch": 0.09269020551389436,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.0029998000021582345,
+      "loss": 1.2806,
+      "step": 1057
+    },
+    {
+      "epoch": 0.09277789728826889,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.002999792923314345,
+      "loss": 1.3156,
+      "step": 1058
+    },
+    {
+      "epoch": 0.09286558906264344,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.0029997857213728257,
+      "loss": 1.3024,
+      "step": 1059
+    },
+    {
+      "epoch": 0.09295328083701798,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.0029997783963343332,
+      "loss": 1.3808,
+      "step": 1060
+    },
+    {
+      "epoch": 0.09304097261139253,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.002999770948199535,
+      "loss": 1.3611,
+      "step": 1061
+    },
+    {
+      "epoch": 0.09312866438576707,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.002999763376969111,
+      "loss": 1.3256,
+      "step": 1062
+    },
+    {
+      "epoch": 0.09321635616014162,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.002999755682643751,
+      "loss": 1.3321,
+      "step": 1063
+    },
+    {
+      "epoch": 0.09330404793451617,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.0029997478652241576,
+      "loss": 1.372,
+      "step": 1064
+    },
+    {
+      "epoch": 0.09339173970889071,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.002999739924711043,
+      "loss": 1.3459,
+      "step": 1065
+    },
+    {
+      "epoch": 0.09347943148326526,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.002999731861105132,
+      "loss": 1.3769,
+      "step": 1066
+    },
+    {
+      "epoch": 0.0935671232576398,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.002999723674407159,
+      "loss": 1.3364,
+      "step": 1067
+    },
+    {
+      "epoch": 0.09365481503201435,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0029997153646178715,
+      "loss": 1.3105,
+      "step": 1068
+    },
+    {
+      "epoch": 0.09374250680638889,
+      "grad_norm": 0.08251953125,
+      "learning_rate": 0.0029997069317380274,
+      "loss": 1.3385,
+      "step": 1069
+    },
+    {
+      "epoch": 0.09383019858076344,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.002999698375768395,
+      "loss": 1.3305,
+      "step": 1070
+    },
+    {
+      "epoch": 0.09391789035513798,
+      "grad_norm": 0.08642578125,
+      "learning_rate": 0.002999689696709755,
+      "loss": 1.3452,
+      "step": 1071
+    },
+    {
+      "epoch": 0.09400558212951253,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.0029996808945628984,
+      "loss": 1.3774,
+      "step": 1072
+    },
+    {
+      "epoch": 0.09409327390388708,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.002999671969328629,
+      "loss": 1.3165,
+      "step": 1073
+    },
+    {
+      "epoch": 0.09418096567826162,
+      "grad_norm": 0.0673828125,
+      "learning_rate": 0.00299966292100776,
+      "loss": 1.3589,
+      "step": 1074
+    },
+    {
+      "epoch": 0.09426865745263617,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0029996537496011166,
+      "loss": 1.3527,
+      "step": 1075
+    },
+    {
+      "epoch": 0.09435634922701071,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.0029996444551095352,
+      "loss": 1.3123,
+      "step": 1076
+    },
+    {
+      "epoch": 0.09444404100138526,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.002999635037533864,
+      "loss": 1.3793,
+      "step": 1077
+    },
+    {
+      "epoch": 0.0945317327757598,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.0029996254968749614,
+      "loss": 1.3356,
+      "step": 1078
+    },
+    {
+      "epoch": 0.09461942455013435,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.002999615833133697,
+      "loss": 1.36,
+      "step": 1079
+    },
+    {
+      "epoch": 0.09470711632450889,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.0029996060463109535,
+      "loss": 1.2863,
+      "step": 1080
+    },
+    {
+      "epoch": 0.09479480809888344,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.002999596136407622,
+      "loss": 1.3104,
+      "step": 1081
+    },
+    {
+      "epoch": 0.09488249987325799,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.002999586103424607,
+      "loss": 1.2665,
+      "step": 1082
+    },
+    {
+      "epoch": 0.09497019164763253,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0029995759473628227,
+      "loss": 1.3407,
+      "step": 1083
+    },
+    {
+      "epoch": 0.09505788342200708,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.0029995656682231964,
+      "loss": 1.3338,
+      "step": 1084
+    },
+    {
+      "epoch": 0.09514557519638162,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0029995552660066656,
+      "loss": 1.3139,
+      "step": 1085
+    },
+    {
+      "epoch": 0.09523326697075617,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.0029995447407141777,
+      "loss": 1.3548,
+      "step": 1086
+    },
+    {
+      "epoch": 0.0953209587451307,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.0029995340923466935,
+      "loss": 1.268,
+      "step": 1087
+    },
+    {
+      "epoch": 0.09540865051950526,
+      "grad_norm": 0.08642578125,
+      "learning_rate": 0.0029995233209051835,
+      "loss": 1.3708,
+      "step": 1088
+    },
+    {
+      "epoch": 0.0954963422938798,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.002999512426390631,
+      "loss": 1.2914,
+      "step": 1089
+    },
+    {
+      "epoch": 0.09558403406825434,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.0029995014088040287,
+      "loss": 1.3179,
+      "step": 1090
+    },
+    {
+      "epoch": 0.0956717258426289,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0029994902681463815,
+      "loss": 1.3853,
+      "step": 1091
+    },
+    {
+      "epoch": 0.09575941761700343,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.0029994790044187056,
+      "loss": 1.2732,
+      "step": 1092
+    },
+    {
+      "epoch": 0.09584710939137799,
+      "grad_norm": 0.0751953125,
+      "learning_rate": 0.002999467617622028,
+      "loss": 1.2913,
+      "step": 1093
+    },
+    {
+      "epoch": 0.09593480116575252,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.002999456107757388,
+      "loss": 1.399,
+      "step": 1094
+    },
+    {
+      "epoch": 0.09602249294012707,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.0029994444748258344,
+      "loss": 1.364,
+      "step": 1095
+    },
+    {
+      "epoch": 0.09611018471450161,
+      "grad_norm": 0.08203125,
+      "learning_rate": 0.0029994327188284276,
+      "loss": 1.363,
+      "step": 1096
+    },
+    {
+      "epoch": 0.09619787648887616,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.002999420839766241,
+      "loss": 1.3517,
+      "step": 1097
+    },
+    {
+      "epoch": 0.0962855682632507,
+      "grad_norm": 0.08203125,
+      "learning_rate": 0.002999408837640357,
+      "loss": 1.3214,
+      "step": 1098
+    },
+    {
+      "epoch": 0.09637326003762525,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.0029993967124518706,
+      "loss": 1.2575,
+      "step": 1099
+    },
+    {
+      "epoch": 0.0964609518119998,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0029993844642018873,
+      "loss": 1.2762,
+      "step": 1100
+    },
+    {
+      "epoch": 0.09654864358637434,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.0029993720928915245,
+      "loss": 1.3552,
+      "step": 1101
+    },
+    {
+      "epoch": 0.09663633536074889,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.0029993595985219105,
+      "loss": 1.3962,
+      "step": 1102
+    },
+    {
+      "epoch": 0.09672402713512343,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.0029993469810941837,
+      "loss": 1.3563,
+      "step": 1103
+    },
+    {
+      "epoch": 0.09681171890949798,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.0029993342406094965,
+      "loss": 1.2809,
+      "step": 1104
+    },
+    {
+      "epoch": 0.09689941068387252,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.002999321377069009,
+      "loss": 1.3422,
+      "step": 1105
+    },
+    {
+      "epoch": 0.09698710245824707,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.0029993083904738954,
+      "loss": 1.333,
+      "step": 1106
+    },
+    {
+      "epoch": 0.09707479423262161,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00299929528082534,
+      "loss": 1.2935,
+      "step": 1107
+    },
+    {
+      "epoch": 0.09716248600699616,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0029992820481245385,
+      "loss": 1.3219,
+      "step": 1108
+    },
+    {
+      "epoch": 0.09725017778137071,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.0029992686923726963,
+      "loss": 1.3029,
+      "step": 1109
+    },
+    {
+      "epoch": 0.09733786955574525,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.0029992552135710334,
+      "loss": 1.265,
+      "step": 1110
+    },
+    {
+      "epoch": 0.0974255613301198,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 0.0029992416117207775,
+      "loss": 1.3016,
+      "step": 1111
+    },
+    {
+      "epoch": 0.09751325310449434,
+      "grad_norm": 0.08154296875,
+      "learning_rate": 0.00299922788682317,
+      "loss": 1.3026,
+      "step": 1112
+    },
+    {
+      "epoch": 0.09760094487886889,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0029992140388794626,
+      "loss": 1.3753,
+      "step": 1113
+    },
+    {
+      "epoch": 0.09768863665324343,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.0029992000678909173,
+      "loss": 1.3631,
+      "step": 1114
+    },
+    {
+      "epoch": 0.09777632842761798,
+      "grad_norm": 0.08935546875,
+      "learning_rate": 0.002999185973858809,
+      "loss": 1.3922,
+      "step": 1115
+    },
+    {
+      "epoch": 0.09786402020199252,
+      "grad_norm": 0.08447265625,
+      "learning_rate": 0.0029991717567844226,
+      "loss": 1.2998,
+      "step": 1116
+    },
+    {
+      "epoch": 0.09795171197636707,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.002999157416669055,
+      "loss": 1.2946,
+      "step": 1117
+    },
+    {
+      "epoch": 0.09803940375074162,
+      "grad_norm": 0.08203125,
+      "learning_rate": 0.002999142953514014,
+      "loss": 1.3215,
+      "step": 1118
+    },
+    {
+      "epoch": 0.09812709552511616,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.002999128367320618,
+      "loss": 1.407,
+      "step": 1119
+    },
+    {
+      "epoch": 0.09821478729949071,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.002999113658090198,
+      "loss": 1.3248,
+      "step": 1120
+    },
+    {
+      "epoch": 0.09830247907386525,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.002999098825824095,
+      "loss": 1.3874,
+      "step": 1121
+    },
+    {
+      "epoch": 0.0983901708482398,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.0029990838705236614,
+      "loss": 1.3076,
+      "step": 1122
+    },
+    {
+      "epoch": 0.09847786262261433,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.002999068792190262,
+      "loss": 1.3557,
+      "step": 1123
+    },
+    {
+      "epoch": 0.09856555439698889,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.0029990535908252713,
+      "loss": 1.2909,
+      "step": 1124
+    },
+    {
+      "epoch": 0.09865324617136342,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.0029990382664300754,
+      "loss": 1.2819,
+      "step": 1125
+    },
+    {
+      "epoch": 0.09874093794573797,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0029990228190060722,
+      "loss": 1.3976,
+      "step": 1126
+    },
+    {
+      "epoch": 0.09882862972011253,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.0029990072485546705,
+      "loss": 1.3246,
+      "step": 1127
+    },
+    {
+      "epoch": 0.09891632149448706,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.00299899155507729,
+      "loss": 1.4496,
+      "step": 1128
+    },
+    {
+      "epoch": 0.09900401326886162,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.002998975738575362,
+      "loss": 1.3328,
+      "step": 1129
+    },
+    {
+      "epoch": 0.09909170504323615,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.002998959799050329,
+      "loss": 1.3249,
+      "step": 1130
+    },
+    {
+      "epoch": 0.0991793968176107,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0029989437365036454,
+      "loss": 1.2959,
+      "step": 1131
+    },
+    {
+      "epoch": 0.09926708859198524,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.0029989275509367747,
+      "loss": 1.3706,
+      "step": 1132
+    },
+    {
+      "epoch": 0.0993547803663598,
+      "grad_norm": 0.09423828125,
+      "learning_rate": 0.0029989112423511933,
+      "loss": 1.312,
+      "step": 1133
+    },
+    {
+      "epoch": 0.09944247214073433,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.0029988948107483896,
+      "loss": 1.3564,
+      "step": 1134
+    },
+    {
+      "epoch": 0.09953016391510888,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.0029988782561298608,
+      "loss": 1.3321,
+      "step": 1135
+    },
+    {
+      "epoch": 0.09961785568948343,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.002998861578497117,
+      "loss": 1.3139,
+      "step": 1136
+    },
+    {
+      "epoch": 0.09970554746385797,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.00299884477785168,
+      "loss": 1.2867,
+      "step": 1137
+    },
+    {
+      "epoch": 0.09979323923823252,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.0029988278541950805,
+      "loss": 1.2831,
+      "step": 1138
+    },
+    {
+      "epoch": 0.09988093101260706,
+      "grad_norm": 0.08544921875,
+      "learning_rate": 0.002998810807528863,
+      "loss": 1.3432,
+      "step": 1139
+    },
+    {
+      "epoch": 0.09996862278698161,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.0029987936378545813,
+      "loss": 1.3169,
+      "step": 1140
+    },
+    {
+      "epoch": 0.10005631456135615,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 0.0029987763451738026,
+      "loss": 1.3212,
+      "step": 1141
+    },
+    {
+      "epoch": 0.1001440063357307,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.0029987589294881026,
+      "loss": 1.382,
+      "step": 1142
+    },
+    {
+      "epoch": 0.10023169811010524,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.0029987413907990703,
+      "loss": 1.2966,
+      "step": 1143
+    },
+    {
+      "epoch": 0.10031938988447979,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0029987237291083046,
+      "loss": 1.2973,
+      "step": 1144
+    },
+    {
+      "epoch": 0.10040708165885433,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.002998705944417417,
+      "loss": 1.3291,
+      "step": 1145
+    },
+    {
+      "epoch": 0.10049477343322888,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.002998688036728028,
+      "loss": 1.3089,
+      "step": 1146
+    },
+    {
+      "epoch": 0.10058246520760343,
+      "grad_norm": 0.076171875,
+      "learning_rate": 0.0029986700060417723,
+      "loss": 1.2835,
+      "step": 1147
+    },
+    {
+      "epoch": 0.10067015698197797,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.0029986518523602936,
+      "loss": 1.3405,
+      "step": 1148
+    },
+    {
+      "epoch": 0.10075784875635252,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.0029986335756852474,
+      "loss": 1.3233,
+      "step": 1149
+    },
+    {
+      "epoch": 0.10084554053072706,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.002998615176018301,
+      "loss": 1.282,
+      "step": 1150
+    },
+    {
+      "epoch": 0.10093323230510161,
+      "grad_norm": 0.08203125,
+      "learning_rate": 0.0029985966533611313,
+      "loss": 1.3426,
+      "step": 1151
+    },
+    {
+      "epoch": 0.10102092407947615,
+      "grad_norm": 0.0859375,
+      "learning_rate": 0.0029985780077154286,
+      "loss": 1.2735,
+      "step": 1152
+    },
+    {
+      "epoch": 0.1011086158538507,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.002998559239082893,
+      "loss": 1.3627,
+      "step": 1153
+    },
+    {
+      "epoch": 0.10119630762822523,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.002998540347465236,
+      "loss": 1.2934,
+      "step": 1154
+    },
+    {
+      "epoch": 0.10128399940259979,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.0029985213328641803,
+      "loss": 1.3252,
+      "step": 1155
+    },
+    {
+      "epoch": 0.10137169117697434,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.0029985021952814604,
+      "loss": 1.2824,
+      "step": 1156
+    },
+    {
+      "epoch": 0.10145938295134888,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0029984829347188212,
+      "loss": 1.356,
+      "step": 1157
+    },
+    {
+      "epoch": 0.10154707472572343,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00299846355117802,
+      "loss": 1.3524,
+      "step": 1158
+    },
+    {
+      "epoch": 0.10163476650009796,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.0029984440446608235,
+      "loss": 1.3321,
+      "step": 1159
+    },
+    {
+      "epoch": 0.10172245827447252,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 0.0029984244151690116,
+      "loss": 1.3648,
+      "step": 1160
+    },
+    {
+      "epoch": 0.10181015004884705,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.0029984046627043737,
+      "loss": 1.324,
+      "step": 1161
+    },
+    {
+      "epoch": 0.1018978418232216,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0029983847872687114,
+      "loss": 1.3589,
+      "step": 1162
+    },
+    {
+      "epoch": 0.10198553359759614,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.002998364788863837,
+      "loss": 1.3402,
+      "step": 1163
+    },
+    {
+      "epoch": 0.1020732253719707,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.002998344667491575,
+      "loss": 1.3162,
+      "step": 1164
+    },
+    {
+      "epoch": 0.10216091714634525,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.00299832442315376,
+      "loss": 1.3257,
+      "step": 1165
+    },
+    {
+      "epoch": 0.10224860892071978,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.0029983040558522384,
+      "loss": 1.4214,
+      "step": 1166
+    },
+    {
+      "epoch": 0.10233630069509433,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0029982835655888674,
+      "loss": 1.348,
+      "step": 1167
+    },
+    {
+      "epoch": 0.10242399246946887,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.002998262952365516,
+      "loss": 1.2412,
+      "step": 1168
+    },
+    {
+      "epoch": 0.10251168424384342,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0029982422161840636,
+      "loss": 1.3103,
+      "step": 1169
+    },
+    {
+      "epoch": 0.10259937601821796,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.0029982213570464017,
+      "loss": 1.3279,
+      "step": 1170
+    },
+    {
+      "epoch": 0.10268706779259251,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.0029982003749544324,
+      "loss": 1.34,
+      "step": 1171
+    },
+    {
+      "epoch": 0.10277475956696705,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.0029981792699100692,
+      "loss": 1.3723,
+      "step": 1172
+    },
+    {
+      "epoch": 0.1028624513413416,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.0029981580419152372,
+      "loss": 1.2719,
+      "step": 1173
+    },
+    {
+      "epoch": 0.10295014311571615,
+      "grad_norm": 0.06982421875,
+      "learning_rate": 0.0029981366909718715,
+      "loss": 1.2699,
+      "step": 1174
+    },
+    {
+      "epoch": 0.10303783489009069,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.0029981152170819206,
+      "loss": 1.3496,
+      "step": 1175
+    },
+    {
+      "epoch": 0.10312552666446524,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.0029980936202473416,
+      "loss": 1.3897,
+      "step": 1176
+    },
+    {
+      "epoch": 0.10321321843883978,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.002998071900470104,
+      "loss": 1.3579,
+      "step": 1177
+    },
+    {
+      "epoch": 0.10330091021321433,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.00299805005775219,
+      "loss": 1.2813,
+      "step": 1178
+    },
+    {
+      "epoch": 0.10338860198758887,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.00299802809209559,
+      "loss": 1.2954,
+      "step": 1179
+    },
+    {
+      "epoch": 0.10347629376196342,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.002998006003502308,
+      "loss": 1.4056,
+      "step": 1180
+    },
+    {
+      "epoch": 0.10356398553633796,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0029979837919743586,
+      "loss": 1.3866,
+      "step": 1181
+    },
+    {
+      "epoch": 0.10365167731071251,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0029979614575137673,
+      "loss": 1.3485,
+      "step": 1182
+    },
+    {
+      "epoch": 0.10373936908508706,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.00299793900012257,
+      "loss": 1.3949,
+      "step": 1183
+    },
+    {
+      "epoch": 0.1038270608594616,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.002997916419802817,
+      "loss": 1.391,
+      "step": 1184
+    },
+    {
+      "epoch": 0.10391475263383615,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0029978937165565656,
+      "loss": 1.3314,
+      "step": 1185
+    },
+    {
+      "epoch": 0.10400244440821069,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.002997870890385886,
+      "loss": 1.3823,
+      "step": 1186
+    },
+    {
+      "epoch": 0.10409013618258524,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.002997847941292861,
+      "loss": 1.391,
+      "step": 1187
+    },
+    {
+      "epoch": 0.10417782795695978,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.0029978248692795837,
+      "loss": 1.2936,
+      "step": 1188
+    },
+    {
+      "epoch": 0.10426551973133433,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0029978016743481576,
+      "loss": 1.2837,
+      "step": 1189
+    },
+    {
+      "epoch": 0.10435321150570886,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.0029977783565006983,
+      "loss": 1.2655,
+      "step": 1190
+    },
+    {
+      "epoch": 0.10444090328008342,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.0029977549157393316,
+      "loss": 1.2235,
+      "step": 1191
+    },
+    {
+      "epoch": 0.10452859505445797,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.0029977313520661956,
+      "loss": 1.3755,
+      "step": 1192
+    },
+    {
+      "epoch": 0.1046162868288325,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.0029977076654834406,
+      "loss": 1.3078,
+      "step": 1193
+    },
+    {
+      "epoch": 0.10470397860320706,
+      "grad_norm": 0.07763671875,
+      "learning_rate": 0.0029976838559932248,
+      "loss": 1.3872,
+      "step": 1194
+    },
+    {
+      "epoch": 0.1047916703775816,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.0029976599235977206,
+      "loss": 1.3248,
+      "step": 1195
+    },
+    {
+      "epoch": 0.10487936215195615,
+      "grad_norm": 0.08251953125,
+      "learning_rate": 0.0029976358682991104,
+      "loss": 1.2766,
+      "step": 1196
+    },
+    {
+      "epoch": 0.10496705392633068,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.002997611690099587,
+      "loss": 1.3499,
+      "step": 1197
+    },
+    {
+      "epoch": 0.10505474570070523,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.0029975873890013575,
+      "loss": 1.3031,
+      "step": 1198
+    },
+    {
+      "epoch": 0.10514243747507977,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0029975629650066367,
+      "loss": 1.3624,
+      "step": 1199
+    },
+    {
+      "epoch": 0.10523012924945432,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.0029975384181176513,
+      "loss": 1.3101,
+      "step": 1200
+    },
+    {
+      "epoch": 0.10531782102382888,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0029975137483366416,
+      "loss": 1.2886,
+      "step": 1201
+    },
+    {
+      "epoch": 0.10540551279820341,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.0029974889556658568,
+      "loss": 1.2838,
+      "step": 1202
+    },
+    {
+      "epoch": 0.10549320457257796,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.0029974640401075575,
+      "loss": 1.3332,
+      "step": 1203
+    },
+    {
+      "epoch": 0.1055808963469525,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.002997439001664016,
+      "loss": 1.3389,
+      "step": 1204
+    },
+    {
+      "epoch": 0.10566858812132705,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.002997413840337516,
+      "loss": 1.309,
+      "step": 1205
+    },
+    {
+      "epoch": 0.10575627989570159,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0029973885561303524,
+      "loss": 1.2586,
+      "step": 1206
+    },
+    {
+      "epoch": 0.10584397167007614,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0029973631490448306,
+      "loss": 1.3604,
+      "step": 1207
+    },
+    {
+      "epoch": 0.10593166344445068,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.0029973376190832674,
+      "loss": 1.3604,
+      "step": 1208
+    },
+    {
+      "epoch": 0.10601935521882523,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.002997311966247992,
+      "loss": 1.3153,
+      "step": 1209
+    },
+    {
+      "epoch": 0.10610704699319978,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0029972861905413436,
+      "loss": 1.3669,
+      "step": 1210
+    },
+    {
+      "epoch": 0.10619473876757432,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.002997260291965672,
+      "loss": 1.2343,
+      "step": 1211
+    },
+    {
+      "epoch": 0.10628243054194887,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 0.0029972342705233404,
+      "loss": 1.2207,
+      "step": 1212
+    },
+    {
+      "epoch": 0.10637012231632341,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.0029972081262167197,
+      "loss": 1.3529,
+      "step": 1213
+    },
+    {
+      "epoch": 0.10645781409069796,
+      "grad_norm": 0.08203125,
+      "learning_rate": 0.0029971818590481974,
+      "loss": 1.3519,
+      "step": 1214
+    },
+    {
+      "epoch": 0.1065455058650725,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0029971554690201665,
+      "loss": 1.305,
+      "step": 1215
+    },
+    {
+      "epoch": 0.10663319763944705,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0029971289561350344,
+      "loss": 1.3518,
+      "step": 1216
+    },
+    {
+      "epoch": 0.10672088941382159,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.0029971023203952192,
+      "loss": 1.3848,
+      "step": 1217
+    },
+    {
+      "epoch": 0.10680858118819614,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.00299707556180315,
+      "loss": 1.3416,
+      "step": 1218
+    },
+    {
+      "epoch": 0.10689627296257069,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.0029970486803612673,
+      "loss": 1.3402,
+      "step": 1219
+    },
+    {
+      "epoch": 0.10698396473694523,
+      "grad_norm": 0.07568359375,
+      "learning_rate": 0.002997021676072022,
+      "loss": 1.3151,
+      "step": 1220
+    },
+    {
+      "epoch": 0.10707165651131978,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.002996994548937877,
+      "loss": 1.3481,
+      "step": 1221
+    },
+    {
+      "epoch": 0.10715934828569432,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.002996967298961307,
+      "loss": 1.2862,
+      "step": 1222
+    },
+    {
+      "epoch": 0.10724704006006887,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.0029969399261447955,
+      "loss": 1.2572,
+      "step": 1223
+    },
+    {
+      "epoch": 0.1073347318344434,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.002996912430490841,
+      "loss": 1.284,
+      "step": 1224
+    },
+    {
+      "epoch": 0.10742242360881796,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 0.002996884812001949,
+      "loss": 1.3279,
+      "step": 1225
+    },
+    {
+      "epoch": 0.1075101153831925,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.002996857070680639,
+      "loss": 1.3086,
+      "step": 1226
+    },
+    {
+      "epoch": 0.10759780715756705,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.002996829206529442,
+      "loss": 1.3585,
+      "step": 1227
+    },
+    {
+      "epoch": 0.1076854989319416,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0029968012195508978,
+      "loss": 1.3435,
+      "step": 1228
+    },
+    {
+      "epoch": 0.10777319070631614,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.0029967731097475586,
+      "loss": 1.3828,
+      "step": 1229
+    },
+    {
+      "epoch": 0.10786088248069069,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.002996744877121989,
+      "loss": 1.3482,
+      "step": 1230
+    },
+    {
+      "epoch": 0.10794857425506522,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.0029967165216767634,
+      "loss": 1.3508,
+      "step": 1231
+    },
+    {
+      "epoch": 0.10803626602943978,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 0.0029966880434144673,
+      "loss": 1.2748,
+      "step": 1232
+    },
+    {
+      "epoch": 0.10812395780381431,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0029966594423376973,
+      "loss": 1.2977,
+      "step": 1233
+    },
+    {
+      "epoch": 0.10821164957818886,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.002996630718449064,
+      "loss": 1.3324,
+      "step": 1234
+    },
+    {
+      "epoch": 0.1082993413525634,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.0029966018717511845,
+      "loss": 1.2831,
+      "step": 1235
+    },
+    {
+      "epoch": 0.10838703312693795,
+      "grad_norm": 0.06884765625,
+      "learning_rate": 0.002996572902246691,
+      "loss": 1.2907,
+      "step": 1236
+    },
+    {
+      "epoch": 0.1084747249013125,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.0029965438099382245,
+      "loss": 1.3001,
+      "step": 1237
+    },
+    {
+      "epoch": 0.10856241667568704,
+      "grad_norm": 0.0673828125,
+      "learning_rate": 0.0029965145948284387,
+      "loss": 1.3221,
+      "step": 1238
+    },
+    {
+      "epoch": 0.1086501084500616,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.0029964852569199984,
+      "loss": 1.2634,
+      "step": 1239
+    },
+    {
+      "epoch": 0.10873780022443613,
+      "grad_norm": 0.072265625,
+      "learning_rate": 0.002996455796215578,
+      "loss": 1.294,
+      "step": 1240
+    },
+    {
+      "epoch": 0.10882549199881068,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.0029964262127178654,
+      "loss": 1.3118,
+      "step": 1241
+    },
+    {
+      "epoch": 0.10891318377318522,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.0029963965064295573,
+      "loss": 1.27,
+      "step": 1242
+    },
+    {
+      "epoch": 0.10900087554755977,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.002996366677353364,
+      "loss": 1.3144,
+      "step": 1243
+    },
+    {
+      "epoch": 0.10908856732193431,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 0.0029963367254920055,
+      "loss": 1.339,
+      "step": 1244
+    },
+    {
+      "epoch": 0.10917625909630886,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.002996306650848213,
+      "loss": 1.3353,
+      "step": 1245
+    },
+    {
+      "epoch": 0.10926395087068341,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0029962764534247287,
+      "loss": 1.2875,
+      "step": 1246
+    },
+    {
+      "epoch": 0.10935164264505795,
+      "grad_norm": 0.076171875,
+      "learning_rate": 0.0029962461332243085,
+      "loss": 1.3335,
+      "step": 1247
+    },
+    {
+      "epoch": 0.1094393344194325,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0029962156902497154,
+      "loss": 1.2532,
+      "step": 1248
+    },
+    {
+      "epoch": 0.10952702619380704,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.002996185124503727,
+      "loss": 1.2625,
+      "step": 1249
+    },
+    {
+      "epoch": 0.10961471796818159,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.0029961544359891302,
+      "loss": 1.3298,
+      "step": 1250
+    },
+    {
+      "epoch": 0.10970240974255613,
+      "grad_norm": 0.0712890625,
+      "learning_rate": 0.0029961236247087234,
+      "loss": 1.2993,
+      "step": 1251
+    },
+    {
+      "epoch": 0.10979010151693068,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.0029960926906653185,
+      "loss": 1.3254,
+      "step": 1252
+    },
+    {
+      "epoch": 0.10987779329130522,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.0029960616338617335,
+      "loss": 1.3134,
+      "step": 1253
+    },
+    {
+      "epoch": 0.10996548506567977,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.0029960304543008034,
+      "loss": 1.2123,
+      "step": 1254
+    },
+    {
+      "epoch": 0.11005317684005432,
+      "grad_norm": 0.0654296875,
+      "learning_rate": 0.00299599915198537,
+      "loss": 1.3634,
+      "step": 1255
+    },
+    {
+      "epoch": 0.11014086861442886,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.002995967726918289,
+      "loss": 1.3638,
+      "step": 1256
+    },
+    {
+      "epoch": 0.11022856038880341,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.002995936179102426,
+      "loss": 1.2941,
+      "step": 1257
+    },
+    {
+      "epoch": 0.11031625216317795,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0029959045085406573,
+      "loss": 1.3817,
+      "step": 1258
+    },
+    {
+      "epoch": 0.1104039439375525,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.002995872715235873,
+      "loss": 1.241,
+      "step": 1259
+    },
+    {
+      "epoch": 0.11049163571192704,
+      "grad_norm": 0.07080078125,
+      "learning_rate": 0.0029958407991909704,
+      "loss": 1.3193,
+      "step": 1260
+    },
+    {
+      "epoch": 0.11057932748630159,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0029958087604088617,
+      "loss": 1.3237,
+      "step": 1261
+    },
+    {
+      "epoch": 0.11066701926067612,
+      "grad_norm": 0.08349609375,
+      "learning_rate": 0.002995776598892468,
+      "loss": 1.303,
+      "step": 1262
+    },
+    {
+      "epoch": 0.11075471103505068,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.0029957443146447224,
+      "loss": 1.3359,
+      "step": 1263
+    },
+    {
+      "epoch": 0.11084240280942523,
+      "grad_norm": 0.125,
+      "learning_rate": 0.0029957119076685695,
+      "loss": 1.292,
+      "step": 1264
+    },
+    {
+      "epoch": 0.11093009458379977,
+      "grad_norm": 0.0693359375,
+      "learning_rate": 0.002995679377966965,
+      "loss": 1.3467,
+      "step": 1265
+    },
+    {
+      "epoch": 0.11101778635817432,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0029956467255428743,
+      "loss": 1.2914,
+      "step": 1266
+    },
+    {
+      "epoch": 0.11110547813254885,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0029956139503992765,
+      "loss": 1.3085,
+      "step": 1267
+    },
+    {
+      "epoch": 0.1111931699069234,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.0029955810525391603,
+      "loss": 1.2696,
+      "step": 1268
+    },
+    {
+      "epoch": 0.11128086168129794,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.002995548031965526,
+      "loss": 1.3236,
+      "step": 1269
+    },
+    {
+      "epoch": 0.1113685534556725,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 0.0029955148886813836,
+      "loss": 1.3616,
+      "step": 1270
+    },
+    {
+      "epoch": 0.11145624523004703,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.0029954816226897573,
+      "loss": 1.3317,
+      "step": 1271
+    },
+    {
+      "epoch": 0.11154393700442158,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.0029954482339936803,
+      "loss": 1.3282,
+      "step": 1272
+    },
+    {
+      "epoch": 0.11163162877879614,
+      "grad_norm": 0.0859375,
+      "learning_rate": 0.002995414722596198,
+      "loss": 1.3022,
+      "step": 1273
+    },
+    {
+      "epoch": 0.11171932055317067,
+      "grad_norm": 0.08544921875,
+      "learning_rate": 0.0029953810885003655,
+      "loss": 1.3222,
+      "step": 1274
+    },
+    {
+      "epoch": 0.11180701232754522,
+      "grad_norm": 0.07568359375,
+      "learning_rate": 0.0029953473317092514,
+      "loss": 1.3184,
+      "step": 1275
+    },
+    {
+      "epoch": 0.11189470410191976,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0029953134522259337,
+      "loss": 1.2974,
+      "step": 1276
+    },
+    {
+      "epoch": 0.11198239587629431,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.0029952794500535014,
+      "loss": 1.3494,
+      "step": 1277
+    },
+    {
+      "epoch": 0.11207008765066885,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.0029952453251950563,
+      "loss": 1.2715,
+      "step": 1278
+    },
+    {
+      "epoch": 0.1121577794250434,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0029952110776537105,
+      "loss": 1.2665,
+      "step": 1279
+    },
+    {
+      "epoch": 0.11224547119941794,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.002995176707432587,
+      "loss": 1.3398,
+      "step": 1280
+    },
+    {
+      "epoch": 0.11233316297379249,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.0029951422145348206,
+      "loss": 1.2775,
+      "step": 1281
+    },
+    {
+      "epoch": 0.11242085474816704,
+      "grad_norm": 0.059814453125,
+      "learning_rate": 0.0029951075989635566,
+      "loss": 1.2788,
+      "step": 1282
+    },
+    {
+      "epoch": 0.11250854652254158,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.0029950728607219513,
+      "loss": 1.3765,
+      "step": 1283
+    },
+    {
+      "epoch": 0.11259623829691613,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.0029950379998131744,
+      "loss": 1.2966,
+      "step": 1284
+    },
+    {
+      "epoch": 0.11268393007129067,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.0029950030162404035,
+      "loss": 1.2517,
+      "step": 1285
+    },
+    {
+      "epoch": 0.11277162184566522,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0029949679100068297,
+      "loss": 1.3251,
+      "step": 1286
+    },
+    {
+      "epoch": 0.11285931362003976,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.002994932681115655,
+      "loss": 1.3349,
+      "step": 1287
+    },
+    {
+      "epoch": 0.11294700539441431,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0029948973295700907,
+      "loss": 1.3208,
+      "step": 1288
+    },
+    {
+      "epoch": 0.11303469716878885,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.002994861855373362,
+      "loss": 1.4167,
+      "step": 1289
+    },
+    {
+      "epoch": 0.1131223889431634,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.002994826258528705,
+      "loss": 1.2385,
+      "step": 1290
+    },
+    {
+      "epoch": 0.11321008071753795,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.0029947905390393637,
+      "loss": 1.2785,
+      "step": 1291
+    },
+    {
+      "epoch": 0.11329777249191249,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.002994754696908597,
+      "loss": 1.3461,
+      "step": 1292
+    },
+    {
+      "epoch": 0.11338546426628704,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.0029947187321396735,
+      "loss": 1.3384,
+      "step": 1293
+    },
+    {
+      "epoch": 0.11347315604066158,
+      "grad_norm": 0.0732421875,
+      "learning_rate": 0.002994682644735873,
+      "loss": 1.3497,
+      "step": 1294
+    },
+    {
+      "epoch": 0.11356084781503613,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.0029946464347004867,
+      "loss": 1.3835,
+      "step": 1295
+    },
+    {
+      "epoch": 0.11364853958941067,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.002994610102036817,
+      "loss": 1.3218,
+      "step": 1296
+    },
+    {
+      "epoch": 0.11373623136378522,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.0029945736467481767,
+      "loss": 1.2858,
+      "step": 1297
+    },
+    {
+      "epoch": 0.11382392313815975,
+      "grad_norm": 0.0859375,
+      "learning_rate": 0.002994537068837891,
+      "loss": 1.2978,
+      "step": 1298
+    },
+    {
+      "epoch": 0.1139116149125343,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.002994500368309295,
+      "loss": 1.3046,
+      "step": 1299
+    },
+    {
+      "epoch": 0.11399930668690886,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0029944635451657365,
+      "loss": 1.2858,
+      "step": 1300
+    },
+    {
+      "epoch": 0.1140869984612834,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.002994426599410574,
+      "loss": 1.3892,
+      "step": 1301
+    },
+    {
+      "epoch": 0.11417469023565795,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.0029943895310471755,
+      "loss": 1.3058,
+      "step": 1302
+    },
+    {
+      "epoch": 0.11426238201003248,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.0029943523400789225,
+      "loss": 1.3764,
+      "step": 1303
+    },
+    {
+      "epoch": 0.11435007378440704,
+      "grad_norm": 0.09326171875,
+      "learning_rate": 0.0029943150265092067,
+      "loss": 1.3253,
+      "step": 1304
+    },
+    {
+      "epoch": 0.11443776555878157,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.0029942775903414307,
+      "loss": 1.3296,
+      "step": 1305
+    },
+    {
+      "epoch": 0.11452545733315612,
+      "grad_norm": 0.0888671875,
+      "learning_rate": 0.0029942400315790085,
+      "loss": 1.3896,
+      "step": 1306
+    },
+    {
+      "epoch": 0.11461314910753066,
+      "grad_norm": 0.0654296875,
+      "learning_rate": 0.0029942023502253657,
+      "loss": 1.3069,
+      "step": 1307
+    },
+    {
+      "epoch": 0.11470084088190521,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.0029941645462839388,
+      "loss": 1.2887,
+      "step": 1308
+    },
+    {
+      "epoch": 0.11478853265627977,
+      "grad_norm": 0.0693359375,
+      "learning_rate": 0.0029941266197581755,
+      "loss": 1.3141,
+      "step": 1309
+    },
+    {
+      "epoch": 0.1148762244306543,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.0029940885706515336,
+      "loss": 1.3429,
+      "step": 1310
+    },
+    {
+      "epoch": 0.11496391620502885,
+      "grad_norm": 0.0703125,
+      "learning_rate": 0.002994050398967484,
+      "loss": 1.3575,
+      "step": 1311
+    },
+    {
+      "epoch": 0.11505160797940339,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.002994012104709508,
+      "loss": 1.2967,
+      "step": 1312
+    },
+    {
+      "epoch": 0.11513929975377794,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.002993973687881097,
+      "loss": 1.4125,
+      "step": 1313
+    },
+    {
+      "epoch": 0.11522699152815248,
+      "grad_norm": 0.08056640625,
+      "learning_rate": 0.0029939351484857555,
+      "loss": 1.2555,
+      "step": 1314
+    },
+    {
+      "epoch": 0.11531468330252703,
+      "grad_norm": 0.08447265625,
+      "learning_rate": 0.0029938964865269977,
+      "loss": 1.2359,
+      "step": 1315
+    },
+    {
+      "epoch": 0.11540237507690157,
+      "grad_norm": 0.0810546875,
+      "learning_rate": 0.0029938577020083503,
+      "loss": 1.3448,
+      "step": 1316
+    },
+    {
+      "epoch": 0.11549006685127612,
+      "grad_norm": 0.0703125,
+      "learning_rate": 0.0029938187949333484,
+      "loss": 1.2827,
+      "step": 1317
+    },
+    {
+      "epoch": 0.11557775862565067,
+      "grad_norm": 0.07470703125,
+      "learning_rate": 0.0029937797653055423,
+      "loss": 1.3329,
+      "step": 1318
+    },
+    {
+      "epoch": 0.11566545040002521,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.00299374061312849,
+      "loss": 1.2784,
+      "step": 1319
+    },
+    {
+      "epoch": 0.11575314217439976,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.002993701338405763,
+      "loss": 1.3145,
+      "step": 1320
+    },
+    {
+      "epoch": 0.1158408339487743,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.002993661941140943,
+      "loss": 1.3254,
+      "step": 1321
+    },
+    {
+      "epoch": 0.11592852572314885,
+      "grad_norm": 0.06298828125,
+      "learning_rate": 0.002993622421337622,
+      "loss": 1.3961,
+      "step": 1322
+    },
+    {
+      "epoch": 0.11601621749752339,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.0029935827789994048,
+      "loss": 1.3328,
+      "step": 1323
+    },
+    {
+      "epoch": 0.11610390927189794,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.002993543014129907,
+      "loss": 1.2933,
+      "step": 1324
+    },
+    {
+      "epoch": 0.11619160104627248,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.0029935031267327543,
+      "loss": 1.3627,
+      "step": 1325
+    },
+    {
+      "epoch": 0.11627929282064703,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.0029934631168115847,
+      "loss": 1.272,
+      "step": 1326
+    },
+    {
+      "epoch": 0.11636698459502158,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.002993422984370047,
+      "loss": 1.288,
+      "step": 1327
+    },
+    {
+      "epoch": 0.11645467636939612,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.002993382729411801,
+      "loss": 1.3332,
+      "step": 1328
+    },
+    {
+      "epoch": 0.11654236814377067,
+      "grad_norm": 0.25,
+      "learning_rate": 0.0029933423519405184,
+      "loss": 1.3108,
+      "step": 1329
+    },
+    {
+      "epoch": 0.1166300599181452,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.002993301851959881,
+      "loss": 1.2691,
+      "step": 1330
+    },
+    {
+      "epoch": 0.11671775169251976,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.002993261229473582,
+      "loss": 1.3041,
+      "step": 1331
+    },
+    {
+      "epoch": 0.1168054434668943,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0029932204844853273,
+      "loss": 1.3623,
+      "step": 1332
+    },
+    {
+      "epoch": 0.11689313524126885,
+      "grad_norm": 0.06640625,
+      "learning_rate": 0.0029931796169988313,
+      "loss": 1.29,
+      "step": 1333
+    },
+    {
+      "epoch": 0.11698082701564338,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.002993138627017822,
+      "loss": 1.3212,
+      "step": 1334
+    },
+    {
+      "epoch": 0.11706851879001794,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 0.002993097514546037,
+      "loss": 1.3031,
+      "step": 1335
+    },
+    {
+      "epoch": 0.11715621056439249,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.0029930562795872258,
+      "loss": 1.2398,
+      "step": 1336
+    },
+    {
+      "epoch": 0.11724390233876703,
+      "grad_norm": 0.0703125,
+      "learning_rate": 0.002993014922145149,
+      "loss": 1.3357,
+      "step": 1337
+    },
+    {
+      "epoch": 0.11733159411314158,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.002992973442223578,
+      "loss": 1.3013,
+      "step": 1338
+    },
+    {
+      "epoch": 0.11741928588751611,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.0029929318398262965,
+      "loss": 1.3366,
+      "step": 1339
+    },
+    {
+      "epoch": 0.11750697766189067,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.002992890114957098,
+      "loss": 1.3721,
+      "step": 1340
+    },
+    {
+      "epoch": 0.1175946694362652,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.0029928482676197872,
+      "loss": 1.2499,
+      "step": 1341
+    },
+    {
+      "epoch": 0.11768236121063975,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.002992806297818181,
+      "loss": 1.3645,
+      "step": 1342
+    },
+    {
+      "epoch": 0.11777005298501429,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.002992764205556107,
+      "loss": 1.2865,
+      "step": 1343
+    },
+    {
+      "epoch": 0.11785774475938884,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0029927219908374037,
+      "loss": 1.3024,
+      "step": 1344
+    },
+    {
+      "epoch": 0.1179454365337634,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.002992679653665921,
+      "loss": 1.3191,
+      "step": 1345
+    },
+    {
+      "epoch": 0.11803312830813793,
+      "grad_norm": 0.0693359375,
+      "learning_rate": 0.0029926371940455204,
+      "loss": 1.3488,
+      "step": 1346
+    },
+    {
+      "epoch": 0.11812082008251248,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.0029925946119800737,
+      "loss": 1.3971,
+      "step": 1347
+    },
+    {
+      "epoch": 0.11820851185688702,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0029925519074734635,
+      "loss": 1.3229,
+      "step": 1348
+    },
+    {
+      "epoch": 0.11829620363126157,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.002992509080529586,
+      "loss": 1.3529,
+      "step": 1349
+    },
+    {
+      "epoch": 0.11838389540563611,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.002992466131152345,
+      "loss": 1.4095,
+      "step": 1350
+    },
+    {
+      "epoch": 0.11847158718001066,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.0029924230593456596,
+      "loss": 1.2716,
+      "step": 1351
+    },
+    {
+      "epoch": 0.1185592789543852,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.002992379865113456,
+      "loss": 1.2908,
+      "step": 1352
+    },
+    {
+      "epoch": 0.11864697072875975,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.0029923365484596743,
+      "loss": 1.2965,
+      "step": 1353
+    },
+    {
+      "epoch": 0.11873466250313429,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.0029922931093882645,
+      "loss": 1.3475,
+      "step": 1354
+    },
+    {
+      "epoch": 0.11882235427750884,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.002992249547903188,
+      "loss": 1.2868,
+      "step": 1355
+    },
+    {
+      "epoch": 0.11891004605188339,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.002992205864008418,
+      "loss": 1.3291,
+      "step": 1356
+    },
+    {
+      "epoch": 0.11899773782625793,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.002992162057707938,
+      "loss": 1.3299,
+      "step": 1357
+    },
+    {
+      "epoch": 0.11908542960063248,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0029921181290057425,
+      "loss": 1.3613,
+      "step": 1358
+    },
+    {
+      "epoch": 0.11917312137500702,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.0029920740779058393,
+      "loss": 1.3172,
+      "step": 1359
+    },
+    {
+      "epoch": 0.11926081314938157,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.0029920299044122445,
+      "loss": 1.3173,
+      "step": 1360
+    },
+    {
+      "epoch": 0.11934850492375611,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.0029919856085289863,
+      "loss": 1.3222,
+      "step": 1361
+    },
+    {
+      "epoch": 0.11943619669813066,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.0029919411902601055,
+      "loss": 1.3778,
+      "step": 1362
+    },
+    {
+      "epoch": 0.1195238884725052,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.002991896649609652,
+      "loss": 1.3441,
+      "step": 1363
+    },
+    {
+      "epoch": 0.11961158024687975,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0029918519865816885,
+      "loss": 1.3452,
+      "step": 1364
+    },
+    {
+      "epoch": 0.1196992720212543,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.0029918072011802872,
+      "loss": 1.2676,
+      "step": 1365
+    },
+    {
+      "epoch": 0.11978696379562884,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.002991762293409534,
+      "loss": 1.2948,
+      "step": 1366
+    },
+    {
+      "epoch": 0.11987465557000339,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.002991717263273523,
+      "loss": 1.2419,
+      "step": 1367
+    },
+    {
+      "epoch": 0.11996234734437793,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0029916721107763606,
+      "loss": 1.2408,
+      "step": 1368
+    },
+    {
+      "epoch": 0.12005003911875248,
+      "grad_norm": 0.08544921875,
+      "learning_rate": 0.0029916268359221655,
+      "loss": 1.3819,
+      "step": 1369
+    },
+    {
+      "epoch": 0.12013773089312701,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.002991581438715066,
+      "loss": 1.2392,
+      "step": 1370
+    },
+    {
+      "epoch": 0.12022542266750157,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.0029915359191592036,
+      "loss": 1.2534,
+      "step": 1371
+    },
+    {
+      "epoch": 0.1203131144418761,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.002991490277258728,
+      "loss": 1.2898,
+      "step": 1372
+    },
+    {
+      "epoch": 0.12040080621625066,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0029914445130178016,
+      "loss": 1.2815,
+      "step": 1373
+    },
+    {
+      "epoch": 0.1204884979906252,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.002991398626440599,
+      "loss": 1.3045,
+      "step": 1374
+    },
+    {
+      "epoch": 0.12057618976499974,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0029913526175313052,
+      "loss": 1.2907,
+      "step": 1375
+    },
+    {
+      "epoch": 0.1206638815393743,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.0029913064862941144,
+      "loss": 1.3564,
+      "step": 1376
+    },
+    {
+      "epoch": 0.12075157331374883,
+      "grad_norm": 0.076171875,
+      "learning_rate": 0.002991260232733235,
+      "loss": 1.2927,
+      "step": 1377
+    },
+    {
+      "epoch": 0.12083926508812338,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0029912138568528846,
+      "loss": 1.3473,
+      "step": 1378
+    },
+    {
+      "epoch": 0.12092695686249792,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.002991167358657293,
+      "loss": 1.3737,
+      "step": 1379
+    },
+    {
+      "epoch": 0.12101464863687247,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.0029911207381507,
+      "loss": 1.2672,
+      "step": 1380
+    },
+    {
+      "epoch": 0.12110234041124701,
+      "grad_norm": 0.08349609375,
+      "learning_rate": 0.0029910739953373584,
+      "loss": 1.267,
+      "step": 1381
+    },
+    {
+      "epoch": 0.12119003218562156,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.0029910271302215304,
+      "loss": 1.2971,
+      "step": 1382
+    },
+    {
+      "epoch": 0.12127772395999611,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.0029909801428074896,
+      "loss": 1.3148,
+      "step": 1383
+    },
+    {
+      "epoch": 0.12136541573437065,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.0029909330330995213,
+      "loss": 1.3299,
+      "step": 1384
+    },
+    {
+      "epoch": 0.1214531075087452,
+      "grad_norm": 0.06591796875,
+      "learning_rate": 0.0029908858011019226,
+      "loss": 1.2774,
+      "step": 1385
+    },
+    {
+      "epoch": 0.12154079928311974,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.002990838446819,
+      "loss": 1.3522,
+      "step": 1386
+    },
+    {
+      "epoch": 0.12162849105749429,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.002990790970255072,
+      "loss": 1.2872,
+      "step": 1387
+    },
+    {
+      "epoch": 0.12171618283186883,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.0029907433714144696,
+      "loss": 1.2608,
+      "step": 1388
+    },
+    {
+      "epoch": 0.12180387460624338,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.0029906956503015325,
+      "loss": 1.2954,
+      "step": 1389
+    },
+    {
+      "epoch": 0.12189156638061792,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.002990647806920613,
+      "loss": 1.3108,
+      "step": 1390
+    },
+    {
+      "epoch": 0.12197925815499247,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.0029905998412760744,
+      "loss": 1.3155,
+      "step": 1391
+    },
+    {
+      "epoch": 0.12206694992936702,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.002990551753372291,
+      "loss": 1.274,
+      "step": 1392
+    },
+    {
+      "epoch": 0.12215464170374156,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0029905035432136484,
+      "loss": 1.2705,
+      "step": 1393
+    },
+    {
+      "epoch": 0.12224233347811611,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.0029904552108045426,
+      "loss": 1.3025,
+      "step": 1394
+    },
+    {
+      "epoch": 0.12233002525249065,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0029904067561493824,
+      "loss": 1.3554,
+      "step": 1395
+    },
+    {
+      "epoch": 0.1224177170268652,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0029903581792525866,
+      "loss": 1.2629,
+      "step": 1396
+    },
+    {
+      "epoch": 0.12250540880123974,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0029903094801185847,
+      "loss": 1.3087,
+      "step": 1397
+    },
+    {
+      "epoch": 0.12259310057561429,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.002990260658751818,
+      "loss": 1.2803,
+      "step": 1398
+    },
+    {
+      "epoch": 0.12268079234998883,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0029902117151567394,
+      "loss": 1.3107,
+      "step": 1399
+    },
+    {
+      "epoch": 0.12276848412436338,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.002990162649337812,
+      "loss": 1.3432,
+      "step": 1400
+    },
+    {
+      "epoch": 0.12285617589873793,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.00299011346129951,
+      "loss": 1.3475,
+      "step": 1401
+    },
+    {
+      "epoch": 0.12294386767311247,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.00299006415104632,
+      "loss": 1.3691,
+      "step": 1402
+    },
+    {
+      "epoch": 0.12303155944748702,
+      "grad_norm": 0.08203125,
+      "learning_rate": 0.00299001471858274,
+      "loss": 1.3435,
+      "step": 1403
+    },
+    {
+      "epoch": 0.12311925122186156,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.002989965163913276,
+      "loss": 1.2595,
+      "step": 1404
+    },
+    {
+      "epoch": 0.12320694299623611,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.002989915487042448,
+      "loss": 1.2682,
+      "step": 1405
+    },
+    {
+      "epoch": 0.12329463477061064,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.002989865687974787,
+      "loss": 1.3321,
+      "step": 1406
+    },
+    {
+      "epoch": 0.1233823265449852,
+      "grad_norm": 0.07861328125,
+      "learning_rate": 0.0029898157667148334,
+      "loss": 1.315,
+      "step": 1407
+    },
+    {
+      "epoch": 0.12347001831935973,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0029897657232671408,
+      "loss": 1.3025,
+      "step": 1408
+    },
+    {
+      "epoch": 0.12355771009373429,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.002989715557636273,
+      "loss": 1.3442,
+      "step": 1409
+    },
+    {
+      "epoch": 0.12364540186810884,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.0029896652698268057,
+      "loss": 1.3382,
+      "step": 1410
+    },
+    {
+      "epoch": 0.12373309364248337,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.0029896148598433227,
+      "loss": 1.2784,
+      "step": 1411
+    },
+    {
+      "epoch": 0.12382078541685793,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.0029895643276904235,
+      "loss": 1.2851,
+      "step": 1412
+    },
+    {
+      "epoch": 0.12390847719123246,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.0029895136733727157,
+      "loss": 1.2747,
+      "step": 1413
+    },
+    {
+      "epoch": 0.12399616896560701,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.0029894628968948184,
+      "loss": 1.3125,
+      "step": 1414
+    },
+    {
+      "epoch": 0.12408386073998155,
+      "grad_norm": 0.07080078125,
+      "learning_rate": 0.002989411998261363,
+      "loss": 1.3735,
+      "step": 1415
+    },
+    {
+      "epoch": 0.1241715525143561,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.0029893609774769908,
+      "loss": 1.3115,
+      "step": 1416
+    },
+    {
+      "epoch": 0.12425924428873064,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0029893098345463547,
+      "loss": 1.3742,
+      "step": 1417
+    },
+    {
+      "epoch": 0.12434693606310519,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.0029892585694741196,
+      "loss": 1.3222,
+      "step": 1418
+    },
+    {
+      "epoch": 0.12443462783747974,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.00298920718226496,
+      "loss": 1.3342,
+      "step": 1419
+    },
+    {
+      "epoch": 0.12452231961185428,
+      "grad_norm": 0.09326171875,
+      "learning_rate": 0.0029891556729235626,
+      "loss": 1.3557,
+      "step": 1420
+    },
+    {
+      "epoch": 0.12461001138622883,
+      "grad_norm": 0.0791015625,
+      "learning_rate": 0.002989104041454625,
+      "loss": 1.3099,
+      "step": 1421
+    },
+    {
+      "epoch": 0.12469770316060337,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.002989052287862856,
+      "loss": 1.3465,
+      "step": 1422
+    },
+    {
+      "epoch": 0.12478539493497792,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.002989000412152975,
+      "loss": 1.3353,
+      "step": 1423
+    },
+    {
+      "epoch": 0.12487308670935246,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.0029889484143297124,
+      "loss": 1.318,
+      "step": 1424
+    },
+    {
+      "epoch": 0.12496077848372701,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0029888962943978113,
+      "loss": 1.2814,
+      "step": 1425
+    },
+    {
+      "epoch": 0.12504847025810156,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.002988844052362024,
+      "loss": 1.3305,
+      "step": 1426
+    },
+    {
+      "epoch": 0.1251361620324761,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.002988791688227116,
+      "loss": 1.3228,
+      "step": 1427
+    },
+    {
+      "epoch": 0.12522385380685064,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.002988739201997862,
+      "loss": 1.3273,
+      "step": 1428
+    },
+    {
+      "epoch": 0.1253115455812252,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.002988686593679048,
+      "loss": 1.2191,
+      "step": 1429
+    },
+    {
+      "epoch": 0.12539923735559974,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.0029886338632754733,
+      "loss": 1.2992,
+      "step": 1430
+    },
+    {
+      "epoch": 0.12548692912997428,
+      "grad_norm": 0.08642578125,
+      "learning_rate": 0.0029885810107919456,
+      "loss": 1.3225,
+      "step": 1431
+    },
+    {
+      "epoch": 0.12557462090434882,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.0029885280362332853,
+      "loss": 1.3324,
+      "step": 1432
+    },
+    {
+      "epoch": 0.12566231267872338,
+      "grad_norm": 0.09423828125,
+      "learning_rate": 0.0029884749396043237,
+      "loss": 1.3346,
+      "step": 1433
+    },
+    {
+      "epoch": 0.12575000445309792,
+      "grad_norm": 0.080078125,
+      "learning_rate": 0.0029884217209099023,
+      "loss": 1.2389,
+      "step": 1434
+    },
+    {
+      "epoch": 0.12583769622747246,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0029883683801548754,
+      "loss": 1.3179,
+      "step": 1435
+    },
+    {
+      "epoch": 0.12592538800184702,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.0029883149173441075,
+      "loss": 1.2994,
+      "step": 1436
+    },
+    {
+      "epoch": 0.12601307977622156,
+      "grad_norm": 0.08349609375,
+      "learning_rate": 0.0029882613324824737,
+      "loss": 1.2993,
+      "step": 1437
+    },
+    {
+      "epoch": 0.1261007715505961,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.002988207625574861,
+      "loss": 1.2902,
+      "step": 1438
+    },
+    {
+      "epoch": 0.12618846332497063,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0029881537966261673,
+      "loss": 1.295,
+      "step": 1439
+    },
+    {
+      "epoch": 0.1262761550993452,
+      "grad_norm": 0.0703125,
+      "learning_rate": 0.0029880998456413023,
+      "loss": 1.3192,
+      "step": 1440
+    },
+    {
+      "epoch": 0.12636384687371974,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.002988045772625185,
+      "loss": 1.2559,
+      "step": 1441
+    },
+    {
+      "epoch": 0.12645153864809427,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.0029879915775827483,
+      "loss": 1.2742,
+      "step": 1442
+    },
+    {
+      "epoch": 0.1265392304224688,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.002987937260518933,
+      "loss": 1.3096,
+      "step": 1443
+    },
+    {
+      "epoch": 0.12662692219684338,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.0029878828214386934,
+      "loss": 1.3017,
+      "step": 1444
+    },
+    {
+      "epoch": 0.12671461397121792,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.0029878282603469945,
+      "loss": 1.3181,
+      "step": 1445
+    },
+    {
+      "epoch": 0.12680230574559245,
+      "grad_norm": 0.08251953125,
+      "learning_rate": 0.0029877735772488117,
+      "loss": 1.3489,
+      "step": 1446
+    },
+    {
+      "epoch": 0.12688999751996702,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.0029877187721491323,
+      "loss": 1.2814,
+      "step": 1447
+    },
+    {
+      "epoch": 0.12697768929434156,
+      "grad_norm": 0.0859375,
+      "learning_rate": 0.002987663845052954,
+      "loss": 1.2852,
+      "step": 1448
+    },
+    {
+      "epoch": 0.1270653810687161,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.002987608795965286,
+      "loss": 1.3262,
+      "step": 1449
+    },
+    {
+      "epoch": 0.12715307284309063,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.002987553624891149,
+      "loss": 1.312,
+      "step": 1450
+    },
+    {
+      "epoch": 0.1272407646174652,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0029874983318355745,
+      "loss": 1.334,
+      "step": 1451
+    },
+    {
+      "epoch": 0.12732845639183973,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.0029874429168036047,
+      "loss": 1.3343,
+      "step": 1452
+    },
+    {
+      "epoch": 0.12741614816621427,
+      "grad_norm": 1.8125,
+      "learning_rate": 0.0029873873798002934,
+      "loss": 1.3057,
+      "step": 1453
+    },
+    {
+      "epoch": 0.1275038399405888,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 0.0029873317208307056,
+      "loss": 1.35,
+      "step": 1454
+    },
+    {
+      "epoch": 0.12759153171496337,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.002987275939899917,
+      "loss": 1.3545,
+      "step": 1455
+    },
+    {
+      "epoch": 0.1276792234893379,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.002987220037013015,
+      "loss": 1.2747,
+      "step": 1456
+    },
+    {
+      "epoch": 0.12776691526371245,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.002987164012175098,
+      "loss": 1.3802,
+      "step": 1457
+    },
+    {
+      "epoch": 0.12785460703808701,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.0029871078653912744,
+      "loss": 1.3492,
+      "step": 1458
+    },
+    {
+      "epoch": 0.12794229881246155,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.0029870515966666654,
+      "loss": 1.3027,
+      "step": 1459
+    },
+    {
+      "epoch": 0.1280299905868361,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.002986995206006402,
+      "loss": 1.3125,
+      "step": 1460
+    },
+    {
+      "epoch": 0.12811768236121063,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.0029869386934156276,
+      "loss": 1.2989,
+      "step": 1461
+    },
+    {
+      "epoch": 0.1282053741355852,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.0029868820588994957,
+      "loss": 1.2363,
+      "step": 1462
+    },
+    {
+      "epoch": 0.12829306590995973,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.0029868253024631715,
+      "loss": 1.3255,
+      "step": 1463
+    },
+    {
+      "epoch": 0.12838075768433427,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.00298676842411183,
+      "loss": 1.317,
+      "step": 1464
+    },
+    {
+      "epoch": 0.12846844945870883,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.00298671142385066,
+      "loss": 1.2981,
+      "step": 1465
+    },
+    {
+      "epoch": 0.12855614123308337,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0029866543016848577,
+      "loss": 1.3597,
+      "step": 1466
+    },
+    {
+      "epoch": 0.1286438330074579,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.002986597057619634,
+      "loss": 1.2674,
+      "step": 1467
+    },
+    {
+      "epoch": 0.12873152478183245,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.0029865396916602094,
+      "loss": 1.3773,
+      "step": 1468
+    },
+    {
+      "epoch": 0.128819216556207,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.002986482203811815,
+      "loss": 1.331,
+      "step": 1469
+    },
+    {
+      "epoch": 0.12890690833058155,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.002986424594079694,
+      "loss": 1.3114,
+      "step": 1470
+    },
+    {
+      "epoch": 0.1289946001049561,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.0029863668624690995,
+      "loss": 1.3031,
+      "step": 1471
+    },
+    {
+      "epoch": 0.12908229187933062,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.0029863090089852972,
+      "loss": 1.2871,
+      "step": 1472
+    },
+    {
+      "epoch": 0.1291699836537052,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0029862510336335635,
+      "loss": 1.3051,
+      "step": 1473
+    },
+    {
+      "epoch": 0.12925767542807973,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.002986192936419184,
+      "loss": 1.3369,
+      "step": 1474
+    },
+    {
+      "epoch": 0.12934536720245426,
+      "grad_norm": 0.0810546875,
+      "learning_rate": 0.0029861347173474584,
+      "loss": 1.3769,
+      "step": 1475
+    },
+    {
+      "epoch": 0.12943305897682883,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0029860763764236963,
+      "loss": 1.3722,
+      "step": 1476
+    },
+    {
+      "epoch": 0.12952075075120337,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0029860179136532174,
+      "loss": 1.3819,
+      "step": 1477
+    },
+    {
+      "epoch": 0.1296084425255779,
+      "grad_norm": 0.08154296875,
+      "learning_rate": 0.0029859593290413535,
+      "loss": 1.3006,
+      "step": 1478
+    },
+    {
+      "epoch": 0.12969613429995244,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.002985900622593448,
+      "loss": 1.2738,
+      "step": 1479
+    },
+    {
+      "epoch": 0.129783826074327,
+      "grad_norm": 0.07373046875,
+      "learning_rate": 0.0029858417943148534,
+      "loss": 1.3179,
+      "step": 1480
+    },
+    {
+      "epoch": 0.12987151784870155,
+      "grad_norm": 0.072265625,
+      "learning_rate": 0.0029857828442109366,
+      "loss": 1.2314,
+      "step": 1481
+    },
+    {
+      "epoch": 0.12995920962307608,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.0029857237722870724,
+      "loss": 1.3601,
+      "step": 1482
+    },
+    {
+      "epoch": 0.13004690139745065,
+      "grad_norm": 0.0654296875,
+      "learning_rate": 0.002985664578548648,
+      "loss": 1.2792,
+      "step": 1483
+    },
+    {
+      "epoch": 0.13013459317182519,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.0029856052630010625,
+      "loss": 1.2767,
+      "step": 1484
+    },
+    {
+      "epoch": 0.13022228494619972,
+      "grad_norm": 0.07861328125,
+      "learning_rate": 0.002985545825649725,
+      "loss": 1.336,
+      "step": 1485
+    },
+    {
+      "epoch": 0.13030997672057426,
+      "grad_norm": 0.08544921875,
+      "learning_rate": 0.0029854862665000554,
+      "loss": 1.3138,
+      "step": 1486
+    },
+    {
+      "epoch": 0.13039766849494883,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.002985426585557486,
+      "loss": 1.252,
+      "step": 1487
+    },
+    {
+      "epoch": 0.13048536026932336,
+      "grad_norm": 0.07177734375,
+      "learning_rate": 0.0029853667828274593,
+      "loss": 1.2989,
+      "step": 1488
+    },
+    {
+      "epoch": 0.1305730520436979,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.0029853068583154292,
+      "loss": 1.3,
+      "step": 1489
+    },
+    {
+      "epoch": 0.13066074381807244,
+      "grad_norm": 0.08349609375,
+      "learning_rate": 0.0029852468120268607,
+      "loss": 1.3237,
+      "step": 1490
+    },
+    {
+      "epoch": 0.130748435592447,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.00298518664396723,
+      "loss": 1.3524,
+      "step": 1491
+    },
+    {
+      "epoch": 0.13083612736682154,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.0029851263541420246,
+      "loss": 1.3396,
+      "step": 1492
+    },
+    {
+      "epoch": 0.13092381914119608,
+      "grad_norm": 0.08349609375,
+      "learning_rate": 0.0029850659425567413,
+      "loss": 1.3305,
+      "step": 1493
+    },
+    {
+      "epoch": 0.13101151091557064,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.0029850054092168915,
+      "loss": 1.3134,
+      "step": 1494
+    },
+    {
+      "epoch": 0.13109920268994518,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.002984944754127994,
+      "loss": 1.3054,
+      "step": 1495
+    },
+    {
+      "epoch": 0.13118689446431972,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0029848839772955815,
+      "loss": 1.2187,
+      "step": 1496
+    },
+    {
+      "epoch": 0.13127458623869426,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.002984823078725196,
+      "loss": 1.3308,
+      "step": 1497
+    },
+    {
+      "epoch": 0.13136227801306882,
+      "grad_norm": 0.08349609375,
+      "learning_rate": 0.0029847620584223917,
+      "loss": 1.2468,
+      "step": 1498
+    },
+    {
+      "epoch": 0.13144996978744336,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.0029847009163927333,
+      "loss": 1.2942,
+      "step": 1499
+    },
+    {
+      "epoch": 0.1315376615618179,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.0029846396526417964,
+      "loss": 1.2777,
+      "step": 1500
+    },
+    {
+      "epoch": 0.1315376615618179,
+      "eval_loss": 1.3182919025421143,
+      "eval_runtime": 429.272,
+      "eval_samples_per_second": 33.655,
+      "eval_steps_per_second": 8.414,
+      "step": 1500
+    },
+    {
+      "epoch": 0.13162535333619246,
+      "grad_norm": 0.076171875,
+      "learning_rate": 0.002984578267175169,
+      "loss": 1.284,
+      "step": 1501
+    },
+    {
+      "epoch": 0.131713045110567,
+      "grad_norm": 0.08447265625,
+      "learning_rate": 0.0029845167599984487,
+      "loss": 1.3167,
+      "step": 1502
+    },
+    {
+      "epoch": 0.13180073688494154,
+      "grad_norm": 0.0791015625,
+      "learning_rate": 0.002984455131117245,
+      "loss": 1.2918,
+      "step": 1503
+    },
+    {
+      "epoch": 0.13188842865931608,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.0029843933805371783,
+      "loss": 1.2734,
+      "step": 1504
+    },
+    {
+      "epoch": 0.13197612043369064,
+      "grad_norm": 0.0712890625,
+      "learning_rate": 0.00298433150826388,
+      "loss": 1.3092,
+      "step": 1505
+    },
+    {
+      "epoch": 0.13206381220806518,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.0029842695143029925,
+      "loss": 1.2861,
+      "step": 1506
+    },
+    {
+      "epoch": 0.13215150398243972,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 0.0029842073986601696,
+      "loss": 1.2986,
+      "step": 1507
+    },
+    {
+      "epoch": 0.13223919575681425,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.0029841451613410765,
+      "loss": 1.3754,
+      "step": 1508
+    },
+    {
+      "epoch": 0.13232688753118882,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.0029840828023513888,
+      "loss": 1.2943,
+      "step": 1509
+    },
+    {
+      "epoch": 0.13241457930556336,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0029840203216967933,
+      "loss": 1.256,
+      "step": 1510
+    },
+    {
+      "epoch": 0.1325022710799379,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.002983957719382988,
+      "loss": 1.3142,
+      "step": 1511
+    },
+    {
+      "epoch": 0.13258996285431246,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.0029838949954156826,
+      "loss": 1.3331,
+      "step": 1512
+    },
+    {
+      "epoch": 0.132677654628687,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.002983832149800597,
+      "loss": 1.3206,
+      "step": 1513
+    },
+    {
+      "epoch": 0.13276534640306153,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0029837691825434624,
+      "loss": 1.4007,
+      "step": 1514
+    },
+    {
+      "epoch": 0.13285303817743607,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.0029837060936500214,
+      "loss": 1.3109,
+      "step": 1515
+    },
+    {
+      "epoch": 0.13294072995181064,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.002983642883126028,
+      "loss": 1.3233,
+      "step": 1516
+    },
+    {
+      "epoch": 0.13302842172618518,
+      "grad_norm": 0.07373046875,
+      "learning_rate": 0.002983579550977246,
+      "loss": 1.3065,
+      "step": 1517
+    },
+    {
+      "epoch": 0.1331161135005597,
+      "grad_norm": 0.08154296875,
+      "learning_rate": 0.002983516097209452,
+      "loss": 1.3287,
+      "step": 1518
+    },
+    {
+      "epoch": 0.13320380527493428,
+      "grad_norm": 0.076171875,
+      "learning_rate": 0.002983452521828432,
+      "loss": 1.3336,
+      "step": 1519
+    },
+    {
+      "epoch": 0.13329149704930882,
+      "grad_norm": 0.06298828125,
+      "learning_rate": 0.0029833888248399845,
+      "loss": 1.3169,
+      "step": 1520
+    },
+    {
+      "epoch": 0.13337918882368335,
+      "grad_norm": 0.08349609375,
+      "learning_rate": 0.0029833250062499188,
+      "loss": 1.3375,
+      "step": 1521
+    },
+    {
+      "epoch": 0.1334668805980579,
+      "grad_norm": 0.06591796875,
+      "learning_rate": 0.0029832610660640536,
+      "loss": 1.3951,
+      "step": 1522
+    },
+    {
+      "epoch": 0.13355457237243246,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.0029831970042882216,
+      "loss": 1.2699,
+      "step": 1523
+    },
+    {
+      "epoch": 0.133642264146807,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0029831328209282645,
+      "loss": 1.2956,
+      "step": 1524
+    },
+    {
+      "epoch": 0.13372995592118153,
+      "grad_norm": 0.08154296875,
+      "learning_rate": 0.0029830685159900347,
+      "loss": 1.3326,
+      "step": 1525
+    },
+    {
+      "epoch": 0.13381764769555607,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.0029830040894793983,
+      "loss": 1.3339,
+      "step": 1526
+    },
+    {
+      "epoch": 0.13390533946993063,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.0029829395414022303,
+      "loss": 1.3356,
+      "step": 1527
+    },
+    {
+      "epoch": 0.13399303124430517,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.0029828748717644167,
+      "loss": 1.3342,
+      "step": 1528
+    },
+    {
+      "epoch": 0.1340807230186797,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.0029828100805718554,
+      "loss": 1.2823,
+      "step": 1529
+    },
+    {
+      "epoch": 0.13416841479305427,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0029827451678304555,
+      "loss": 1.2867,
+      "step": 1530
+    },
+    {
+      "epoch": 0.1342561065674288,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.002982680133546137,
+      "loss": 1.3465,
+      "step": 1531
+    },
+    {
+      "epoch": 0.13434379834180335,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0029826149777248305,
+      "loss": 1.2621,
+      "step": 1532
+    },
+    {
+      "epoch": 0.1344314901161779,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.002982549700372478,
+      "loss": 1.3645,
+      "step": 1533
+    },
+    {
+      "epoch": 0.13451918189055245,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0029824843014950326,
+      "loss": 1.2433,
+      "step": 1534
+    },
+    {
+      "epoch": 0.134606873664927,
+      "grad_norm": 0.07470703125,
+      "learning_rate": 0.002982418781098459,
+      "loss": 1.3254,
+      "step": 1535
+    },
+    {
+      "epoch": 0.13469456543930153,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0029823531391887322,
+      "loss": 1.3154,
+      "step": 1536
+    },
+    {
+      "epoch": 0.1347822572136761,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.0029822873757718387,
+      "loss": 1.3446,
+      "step": 1537
+    },
+    {
+      "epoch": 0.13486994898805063,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.0029822214908537758,
+      "loss": 1.2785,
+      "step": 1538
+    },
+    {
+      "epoch": 0.13495764076242517,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.0029821554844405517,
+      "loss": 1.3284,
+      "step": 1539
+    },
+    {
+      "epoch": 0.1350453325367997,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.002982089356538187,
+      "loss": 1.3399,
+      "step": 1540
+    },
+    {
+      "epoch": 0.13513302431117427,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 0.002982023107152711,
+      "loss": 1.3512,
+      "step": 1541
+    },
+    {
+      "epoch": 0.1352207160855488,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.0029819567362901664,
+      "loss": 1.3023,
+      "step": 1542
+    },
+    {
+      "epoch": 0.13530840785992335,
+      "grad_norm": 0.0888671875,
+      "learning_rate": 0.002981890243956606,
+      "loss": 1.3128,
+      "step": 1543
+    },
+    {
+      "epoch": 0.13539609963429788,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.002981823630158094,
+      "loss": 1.3158,
+      "step": 1544
+    },
+    {
+      "epoch": 0.13548379140867245,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.0029817568949007047,
+      "loss": 1.3173,
+      "step": 1545
+    },
+    {
+      "epoch": 0.135571483183047,
+      "grad_norm": 0.08935546875,
+      "learning_rate": 0.002981690038190524,
+      "loss": 1.302,
+      "step": 1546
+    },
+    {
+      "epoch": 0.13565917495742152,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.0029816230600336504,
+      "loss": 1.2854,
+      "step": 1547
+    },
+    {
+      "epoch": 0.1357468667317961,
+      "grad_norm": 0.08447265625,
+      "learning_rate": 0.0029815559604361905,
+      "loss": 1.2837,
+      "step": 1548
+    },
+    {
+      "epoch": 0.13583455850617063,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.002981488739404265,
+      "loss": 1.2879,
+      "step": 1549
+    },
+    {
+      "epoch": 0.13592225028054516,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.0029814213969440034,
+      "loss": 1.3034,
+      "step": 1550
+    },
+    {
+      "epoch": 0.1360099420549197,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0029813539330615477,
+      "loss": 1.3412,
+      "step": 1551
+    },
+    {
+      "epoch": 0.13609763382929427,
+      "grad_norm": 0.08203125,
+      "learning_rate": 0.0029812863477630502,
+      "loss": 1.2951,
+      "step": 1552
+    },
+    {
+      "epoch": 0.1361853256036688,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.002981218641054674,
+      "loss": 1.2204,
+      "step": 1553
+    },
+    {
+      "epoch": 0.13627301737804334,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.002981150812942595,
+      "loss": 1.2347,
+      "step": 1554
+    },
+    {
+      "epoch": 0.1363607091524179,
+      "grad_norm": 0.08935546875,
+      "learning_rate": 0.0029810828634329973,
+      "loss": 1.2983,
+      "step": 1555
+    },
+    {
+      "epoch": 0.13644840092679245,
+      "grad_norm": 0.06689453125,
+      "learning_rate": 0.002981014792532079,
+      "loss": 1.3449,
+      "step": 1556
+    },
+    {
+      "epoch": 0.13653609270116698,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0029809466002460477,
+      "loss": 1.3148,
+      "step": 1557
+    },
+    {
+      "epoch": 0.13662378447554152,
+      "grad_norm": 0.0654296875,
+      "learning_rate": 0.0029808782865811223,
+      "loss": 1.3352,
+      "step": 1558
+    },
+    {
+      "epoch": 0.1367114762499161,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 0.002980809851543533,
+      "loss": 1.4218,
+      "step": 1559
+    },
+    {
+      "epoch": 0.13679916802429062,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0029807412951395203,
+      "loss": 1.3049,
+      "step": 1560
+    },
+    {
+      "epoch": 0.13688685979866516,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.0029806726173753372,
+      "loss": 1.3568,
+      "step": 1561
+    },
+    {
+      "epoch": 0.1369745515730397,
+      "grad_norm": 0.0810546875,
+      "learning_rate": 0.0029806038182572463,
+      "loss": 1.3048,
+      "step": 1562
+    },
+    {
+      "epoch": 0.13706224334741426,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.0029805348977915216,
+      "loss": 1.3356,
+      "step": 1563
+    },
+    {
+      "epoch": 0.1371499351217888,
+      "grad_norm": 0.06396484375,
+      "learning_rate": 0.00298046585598445,
+      "loss": 1.296,
+      "step": 1564
+    },
+    {
+      "epoch": 0.13723762689616334,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.0029803966928423262,
+      "loss": 1.2985,
+      "step": 1565
+    },
+    {
+      "epoch": 0.1373253186705379,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.0029803274083714582,
+      "loss": 1.2839,
+      "step": 1566
+    },
+    {
+      "epoch": 0.13741301044491244,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.0029802580025781655,
+      "loss": 1.3112,
+      "step": 1567
+    },
+    {
+      "epoch": 0.13750070221928698,
+      "grad_norm": 0.07275390625,
+      "learning_rate": 0.0029801884754687767,
+      "loss": 1.3728,
+      "step": 1568
+    },
+    {
+      "epoch": 0.13758839399366152,
+      "grad_norm": 0.07177734375,
+      "learning_rate": 0.0029801188270496327,
+      "loss": 1.3091,
+      "step": 1569
+    },
+    {
+      "epoch": 0.13767608576803608,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.002980049057327086,
+      "loss": 1.2518,
+      "step": 1570
+    },
+    {
+      "epoch": 0.13776377754241062,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.002979979166307498,
+      "loss": 1.3197,
+      "step": 1571
+    },
+    {
+      "epoch": 0.13785146931678516,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 0.002979909153997243,
+      "loss": 1.3024,
+      "step": 1572
+    },
+    {
+      "epoch": 0.13793916109115972,
+      "grad_norm": 0.0859375,
+      "learning_rate": 0.002979839020402707,
+      "loss": 1.2366,
+      "step": 1573
+    },
+    {
+      "epoch": 0.13802685286553426,
+      "grad_norm": 0.08154296875,
+      "learning_rate": 0.0029797687655302853,
+      "loss": 1.2927,
+      "step": 1574
+    },
+    {
+      "epoch": 0.1381145446399088,
+      "grad_norm": 0.0791015625,
+      "learning_rate": 0.002979698389386385,
+      "loss": 1.3856,
+      "step": 1575
+    },
+    {
+      "epoch": 0.13820223641428334,
+      "grad_norm": 0.08251953125,
+      "learning_rate": 0.002979627891977424,
+      "loss": 1.3791,
+      "step": 1576
+    },
+    {
+      "epoch": 0.1382899281886579,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.0029795572733098317,
+      "loss": 1.2658,
+      "step": 1577
+    },
+    {
+      "epoch": 0.13837761996303244,
+      "grad_norm": 0.062255859375,
+      "learning_rate": 0.002979486533390048,
+      "loss": 1.3256,
+      "step": 1578
+    },
+    {
+      "epoch": 0.13846531173740698,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.002979415672224525,
+      "loss": 1.2792,
+      "step": 1579
+    },
+    {
+      "epoch": 0.13855300351178151,
+      "grad_norm": 0.0751953125,
+      "learning_rate": 0.0029793446898197244,
+      "loss": 1.2801,
+      "step": 1580
+    },
+    {
+      "epoch": 0.13864069528615608,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.00297927358618212,
+      "loss": 1.3045,
+      "step": 1581
+    },
+    {
+      "epoch": 0.13872838706053062,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.0029792023613181957,
+      "loss": 1.2642,
+      "step": 1582
+    },
+    {
+      "epoch": 0.13881607883490515,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0029791310152344473,
+      "loss": 1.2261,
+      "step": 1583
+    },
+    {
+      "epoch": 0.13890377060927972,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0029790595479373813,
+      "loss": 1.3044,
+      "step": 1584
+    },
+    {
+      "epoch": 0.13899146238365426,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0029789879594335164,
+      "loss": 1.3165,
+      "step": 1585
+    },
+    {
+      "epoch": 0.1390791541580288,
+      "grad_norm": 0.0859375,
+      "learning_rate": 0.0029789162497293794,
+      "loss": 1.2949,
+      "step": 1586
+    },
+    {
+      "epoch": 0.13916684593240333,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.002978844418831511,
+      "loss": 1.3131,
+      "step": 1587
+    },
+    {
+      "epoch": 0.1392545377067779,
+      "grad_norm": 0.23046875,
+      "learning_rate": 0.0029787724667464624,
+      "loss": 1.3262,
+      "step": 1588
+    },
+    {
+      "epoch": 0.13934222948115244,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.002978700393480795,
+      "loss": 1.217,
+      "step": 1589
+    },
+    {
+      "epoch": 0.13942992125552697,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.0029786281990410823,
+      "loss": 1.3964,
+      "step": 1590
+    },
+    {
+      "epoch": 0.13951761302990154,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.0029785558834339067,
+      "loss": 1.3288,
+      "step": 1591
+    },
+    {
+      "epoch": 0.13960530480427608,
+      "grad_norm": 0.0791015625,
+      "learning_rate": 0.002978483446665865,
+      "loss": 1.2713,
+      "step": 1592
+    },
+    {
+      "epoch": 0.1396929965786506,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.0029784108887435617,
+      "loss": 1.3323,
+      "step": 1593
+    },
+    {
+      "epoch": 0.13978068835302515,
+      "grad_norm": 0.08154296875,
+      "learning_rate": 0.0029783382096736145,
+      "loss": 1.3132,
+      "step": 1594
+    },
+    {
+      "epoch": 0.13986838012739972,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0029782654094626525,
+      "loss": 1.3063,
+      "step": 1595
+    },
+    {
+      "epoch": 0.13995607190177425,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0029781924881173137,
+      "loss": 1.3055,
+      "step": 1596
+    },
+    {
+      "epoch": 0.1400437636761488,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.0029781194456442485,
+      "loss": 1.3568,
+      "step": 1597
+    },
+    {
+      "epoch": 0.14013145545052333,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0029780462820501188,
+      "loss": 1.2725,
+      "step": 1598
+    },
+    {
+      "epoch": 0.1402191472248979,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.0029779729973415958,
+      "loss": 1.2852,
+      "step": 1599
+    },
+    {
+      "epoch": 0.14030683899927243,
+      "grad_norm": 0.23046875,
+      "learning_rate": 0.0029778995915253643,
+      "loss": 1.3203,
+      "step": 1600
+    },
+    {
+      "epoch": 0.14039453077364697,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.002977826064608118,
+      "loss": 1.2179,
+      "step": 1601
+    },
+    {
+      "epoch": 0.14048222254802153,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.002977752416596562,
+      "loss": 1.238,
+      "step": 1602
+    },
+    {
+      "epoch": 0.14056991432239607,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.002977678647497413,
+      "loss": 1.3133,
+      "step": 1603
+    },
+    {
+      "epoch": 0.1406576060967706,
+      "grad_norm": 0.0791015625,
+      "learning_rate": 0.0029776047573173993,
+      "loss": 1.3257,
+      "step": 1604
+    },
+    {
+      "epoch": 0.14074529787114515,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.0029775307460632584,
+      "loss": 1.289,
+      "step": 1605
+    },
+    {
+      "epoch": 0.1408329896455197,
+      "grad_norm": 0.09423828125,
+      "learning_rate": 0.002977456613741741,
+      "loss": 1.3245,
+      "step": 1606
+    },
+    {
+      "epoch": 0.14092068141989425,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.002977382360359607,
+      "loss": 1.3248,
+      "step": 1607
+    },
+    {
+      "epoch": 0.1410083731942688,
+      "grad_norm": 0.078125,
+      "learning_rate": 0.002977307985923628,
+      "loss": 1.3047,
+      "step": 1608
+    },
+    {
+      "epoch": 0.14109606496864335,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.0029772334904405876,
+      "loss": 1.2525,
+      "step": 1609
+    },
+    {
+      "epoch": 0.1411837567430179,
+      "grad_norm": 0.07470703125,
+      "learning_rate": 0.002977158873917279,
+      "loss": 1.3219,
+      "step": 1610
+    },
+    {
+      "epoch": 0.14127144851739243,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0029770841363605067,
+      "loss": 1.2669,
+      "step": 1611
+    },
+    {
+      "epoch": 0.14135914029176697,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.0029770092777770874,
+      "loss": 1.3024,
+      "step": 1612
+    },
+    {
+      "epoch": 0.14144683206614153,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.002976934298173848,
+      "loss": 1.3109,
+      "step": 1613
+    },
+    {
+      "epoch": 0.14153452384051607,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.0029768591975576253,
+      "loss": 1.2855,
+      "step": 1614
+    },
+    {
+      "epoch": 0.1416222156148906,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 0.0029767839759352694,
+      "loss": 1.2878,
+      "step": 1615
+    },
+    {
+      "epoch": 0.14170990738926514,
+      "grad_norm": 0.06982421875,
+      "learning_rate": 0.0029767086333136405,
+      "loss": 1.3303,
+      "step": 1616
+    },
+    {
+      "epoch": 0.1417975991636397,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.0029766331696996083,
+      "loss": 1.3565,
+      "step": 1617
+    },
+    {
+      "epoch": 0.14188529093801425,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.002976557585100056,
+      "loss": 1.3557,
+      "step": 1618
+    },
+    {
+      "epoch": 0.14197298271238878,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.0029764818795218763,
+      "loss": 1.3797,
+      "step": 1619
+    },
+    {
+      "epoch": 0.14206067448676335,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.002976406052971974,
+      "loss": 1.3364,
+      "step": 1620
+    },
+    {
+      "epoch": 0.1421483662611379,
+      "grad_norm": 0.08154296875,
+      "learning_rate": 0.0029763301054572633,
+      "loss": 1.3215,
+      "step": 1621
+    },
+    {
+      "epoch": 0.14223605803551242,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0029762540369846708,
+      "loss": 1.2655,
+      "step": 1622
+    },
+    {
+      "epoch": 0.14232374980988696,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.002976177847561134,
+      "loss": 1.2315,
+      "step": 1623
+    },
+    {
+      "epoch": 0.14241144158426153,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0029761015371936013,
+      "loss": 1.315,
+      "step": 1624
+    },
+    {
+      "epoch": 0.14249913335863607,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.0029760251058890312,
+      "loss": 1.3761,
+      "step": 1625
+    },
+    {
+      "epoch": 0.1425868251330106,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.002975948553654395,
+      "loss": 1.3818,
+      "step": 1626
+    },
+    {
+      "epoch": 0.14267451690738517,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.002975871880496673,
+      "loss": 1.2746,
+      "step": 1627
+    },
+    {
+      "epoch": 0.1427622086817597,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.002975795086422859,
+      "loss": 1.2724,
+      "step": 1628
+    },
+    {
+      "epoch": 0.14284990045613424,
+      "grad_norm": 0.0634765625,
+      "learning_rate": 0.002975718171439955,
+      "loss": 1.2987,
+      "step": 1629
+    },
+    {
+      "epoch": 0.14293759223050878,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.002975641135554977,
+      "loss": 1.303,
+      "step": 1630
+    },
+    {
+      "epoch": 0.14302528400488335,
+      "grad_norm": 0.08642578125,
+      "learning_rate": 0.0029755639787749488,
+      "loss": 1.3062,
+      "step": 1631
+    },
+    {
+      "epoch": 0.14311297577925788,
+      "grad_norm": 0.09423828125,
+      "learning_rate": 0.0029754867011069076,
+      "loss": 1.2757,
+      "step": 1632
+    },
+    {
+      "epoch": 0.14320066755363242,
+      "grad_norm": 0.07275390625,
+      "learning_rate": 0.0029754093025579015,
+      "loss": 1.2948,
+      "step": 1633
+    },
+    {
+      "epoch": 0.14328835932800696,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.002975331783134988,
+      "loss": 1.2779,
+      "step": 1634
+    },
+    {
+      "epoch": 0.14337605110238152,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.0029752541428452375,
+      "loss": 1.254,
+      "step": 1635
+    },
+    {
+      "epoch": 0.14346374287675606,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.0029751763816957305,
+      "loss": 1.2744,
+      "step": 1636
+    },
+    {
+      "epoch": 0.1435514346511306,
+      "grad_norm": 0.0888671875,
+      "learning_rate": 0.002975098499693558,
+      "loss": 1.3138,
+      "step": 1637
+    },
+    {
+      "epoch": 0.14363912642550516,
+      "grad_norm": 0.0673828125,
+      "learning_rate": 0.0029750204968458233,
+      "loss": 1.3148,
+      "step": 1638
+    },
+    {
+      "epoch": 0.1437268181998797,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.0029749423731596394,
+      "loss": 1.2637,
+      "step": 1639
+    },
+    {
+      "epoch": 0.14381450997425424,
+      "grad_norm": 0.0732421875,
+      "learning_rate": 0.002974864128642132,
+      "loss": 1.273,
+      "step": 1640
+    },
+    {
+      "epoch": 0.14390220174862878,
+      "grad_norm": 0.072265625,
+      "learning_rate": 0.002974785763300436,
+      "loss": 1.3235,
+      "step": 1641
+    },
+    {
+      "epoch": 0.14398989352300334,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.002974707277141698,
+      "loss": 1.2963,
+      "step": 1642
+    },
+    {
+      "epoch": 0.14407758529737788,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.0029746286701730763,
+      "loss": 1.3017,
+      "step": 1643
+    },
+    {
+      "epoch": 0.14416527707175242,
+      "grad_norm": 0.0693359375,
+      "learning_rate": 0.0029745499424017395,
+      "loss": 1.3372,
+      "step": 1644
+    },
+    {
+      "epoch": 0.14425296884612698,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.002974471093834867,
+      "loss": 1.3172,
+      "step": 1645
+    },
+    {
+      "epoch": 0.14434066062050152,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.00297439212447965,
+      "loss": 1.3033,
+      "step": 1646
+    },
+    {
+      "epoch": 0.14442835239487606,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.00297431303434329,
+      "loss": 1.3191,
+      "step": 1647
+    },
+    {
+      "epoch": 0.1445160441692506,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.002974233823433,
+      "loss": 1.326,
+      "step": 1648
+    },
+    {
+      "epoch": 0.14460373594362516,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.002974154491756004,
+      "loss": 1.2827,
+      "step": 1649
+    },
+    {
+      "epoch": 0.1446914277179997,
+      "grad_norm": 0.076171875,
+      "learning_rate": 0.002974075039319536,
+      "loss": 1.249,
+      "step": 1650
+    },
+    {
+      "epoch": 0.14477911949237424,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.0029739954661308437,
+      "loss": 1.2419,
+      "step": 1651
+    },
+    {
+      "epoch": 0.14486681126674877,
+      "grad_norm": 0.08251953125,
+      "learning_rate": 0.002973915772197182,
+      "loss": 1.2723,
+      "step": 1652
+    },
+    {
+      "epoch": 0.14495450304112334,
+      "grad_norm": 0.07763671875,
+      "learning_rate": 0.0029738359575258192,
+      "loss": 1.333,
+      "step": 1653
+    },
+    {
+      "epoch": 0.14504219481549788,
+      "grad_norm": 0.06689453125,
+      "learning_rate": 0.002973756022124035,
+      "loss": 1.308,
+      "step": 1654
+    },
+    {
+      "epoch": 0.14512988658987241,
+      "grad_norm": 0.062255859375,
+      "learning_rate": 0.0029736759659991186,
+      "loss": 1.3056,
+      "step": 1655
+    },
+    {
+      "epoch": 0.14521757836424698,
+      "grad_norm": 0.078125,
+      "learning_rate": 0.0029735957891583713,
+      "loss": 1.2666,
+      "step": 1656
+    },
+    {
+      "epoch": 0.14530527013862152,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.0029735154916091053,
+      "loss": 1.3374,
+      "step": 1657
+    },
+    {
+      "epoch": 0.14539296191299605,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.002973435073358643,
+      "loss": 1.335,
+      "step": 1658
+    },
+    {
+      "epoch": 0.1454806536873706,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0029733545344143176,
+      "loss": 1.3012,
+      "step": 1659
+    },
+    {
+      "epoch": 0.14556834546174516,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.002973273874783475,
+      "loss": 1.2851,
+      "step": 1660
+    },
+    {
+      "epoch": 0.1456560372361197,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.0029731930944734723,
+      "loss": 1.32,
+      "step": 1661
+    },
+    {
+      "epoch": 0.14574372901049423,
+      "grad_norm": 0.06787109375,
+      "learning_rate": 0.0029731121934916736,
+      "loss": 1.3265,
+      "step": 1662
+    },
+    {
+      "epoch": 0.14583142078486877,
+      "grad_norm": 0.07373046875,
+      "learning_rate": 0.0029730311718454594,
+      "loss": 1.4071,
+      "step": 1663
+    },
+    {
+      "epoch": 0.14591911255924334,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.002972950029542218,
+      "loss": 1.3511,
+      "step": 1664
+    },
+    {
+      "epoch": 0.14600680433361787,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.002972868766589348,
+      "loss": 1.3421,
+      "step": 1665
+    },
+    {
+      "epoch": 0.1460944961079924,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.0029727873829942623,
+      "loss": 1.343,
+      "step": 1666
+    },
+    {
+      "epoch": 0.14618218788236698,
+      "grad_norm": 0.07568359375,
+      "learning_rate": 0.002972705878764382,
+      "loss": 1.325,
+      "step": 1667
+    },
+    {
+      "epoch": 0.1462698796567415,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 0.0029726242539071398,
+      "loss": 1.2319,
+      "step": 1668
+    },
+    {
+      "epoch": 0.14635757143111605,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.0029725425084299803,
+      "loss": 1.2526,
+      "step": 1669
+    },
+    {
+      "epoch": 0.1464452632054906,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0029724606423403577,
+      "loss": 1.3712,
+      "step": 1670
+    },
+    {
+      "epoch": 0.14653295497986515,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0029723786556457386,
+      "loss": 1.2874,
+      "step": 1671
+    },
+    {
+      "epoch": 0.1466206467542397,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.0029722965483535996,
+      "loss": 1.2487,
+      "step": 1672
+    },
+    {
+      "epoch": 0.14670833852861423,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.0029722143204714293,
+      "loss": 1.3243,
+      "step": 1673
+    },
+    {
+      "epoch": 0.1467960303029888,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.0029721319720067262,
+      "loss": 1.314,
+      "step": 1674
+    },
+    {
+      "epoch": 0.14688372207736333,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.002972049502967,
+      "loss": 1.3093,
+      "step": 1675
+    },
+    {
+      "epoch": 0.14697141385173787,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0029719669133597723,
+      "loss": 1.3873,
+      "step": 1676
+    },
+    {
+      "epoch": 0.1470591056261124,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.002971884203192575,
+      "loss": 1.3275,
+      "step": 1677
+    },
+    {
+      "epoch": 0.14714679740048697,
+      "grad_norm": 0.07568359375,
+      "learning_rate": 0.00297180137247295,
+      "loss": 1.3174,
+      "step": 1678
+    },
+    {
+      "epoch": 0.1472344891748615,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.002971718421208453,
+      "loss": 1.295,
+      "step": 1679
+    },
+    {
+      "epoch": 0.14732218094923605,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.002971635349406647,
+      "loss": 1.2753,
+      "step": 1680
+    },
+    {
+      "epoch": 0.14740987272361059,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 0.0029715521570751096,
+      "loss": 1.3296,
+      "step": 1681
+    },
+    {
+      "epoch": 0.14749756449798515,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.002971468844221427,
+      "loss": 1.2714,
+      "step": 1682
+    },
+    {
+      "epoch": 0.1475852562723597,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.0029713854108531965,
+      "loss": 1.2962,
+      "step": 1683
+    },
+    {
+      "epoch": 0.14767294804673423,
+      "grad_norm": 0.08203125,
+      "learning_rate": 0.0029713018569780284,
+      "loss": 1.2805,
+      "step": 1684
+    },
+    {
+      "epoch": 0.1477606398211088,
+      "grad_norm": 0.08642578125,
+      "learning_rate": 0.0029712181826035415,
+      "loss": 1.2831,
+      "step": 1685
+    },
+    {
+      "epoch": 0.14784833159548333,
+      "grad_norm": 0.0751953125,
+      "learning_rate": 0.0029711343877373672,
+      "loss": 1.295,
+      "step": 1686
+    },
+    {
+      "epoch": 0.14793602336985787,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.0029710504723871474,
+      "loss": 1.2647,
+      "step": 1687
+    },
+    {
+      "epoch": 0.1480237151442324,
+      "grad_norm": 0.0732421875,
+      "learning_rate": 0.0029709664365605345,
+      "loss": 1.3441,
+      "step": 1688
+    },
+    {
+      "epoch": 0.14811140691860697,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.0029708822802651928,
+      "loss": 1.3064,
+      "step": 1689
+    },
+    {
+      "epoch": 0.1481990986929815,
+      "grad_norm": 0.08349609375,
+      "learning_rate": 0.0029707980035087967,
+      "loss": 1.2509,
+      "step": 1690
+    },
+    {
+      "epoch": 0.14828679046735604,
+      "grad_norm": 0.224609375,
+      "learning_rate": 0.002970713606299033,
+      "loss": 1.2616,
+      "step": 1691
+    },
+    {
+      "epoch": 0.1483744822417306,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.002970629088643597,
+      "loss": 1.3159,
+      "step": 1692
+    },
+    {
+      "epoch": 0.14846217401610515,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.0029705444505501977,
+      "loss": 1.3307,
+      "step": 1693
+    },
+    {
+      "epoch": 0.14854986579047968,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0029704596920265536,
+      "loss": 1.2977,
+      "step": 1694
+    },
+    {
+      "epoch": 0.14863755756485422,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.0029703748130803943,
+      "loss": 1.2808,
+      "step": 1695
+    },
+    {
+      "epoch": 0.1487252493392288,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.0029702898137194604,
+      "loss": 1.2882,
+      "step": 1696
+    },
+    {
+      "epoch": 0.14881294111360333,
+      "grad_norm": 0.09423828125,
+      "learning_rate": 0.0029702046939515036,
+      "loss": 1.3025,
+      "step": 1697
+    },
+    {
+      "epoch": 0.14890063288797786,
+      "grad_norm": 0.06640625,
+      "learning_rate": 0.002970119453784287,
+      "loss": 1.3931,
+      "step": 1698
+    },
+    {
+      "epoch": 0.1489883246623524,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.0029700340932255842,
+      "loss": 1.2934,
+      "step": 1699
+    },
+    {
+      "epoch": 0.14907601643672697,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.0029699486122831795,
+      "loss": 1.354,
+      "step": 1700
+    },
+    {
+      "epoch": 0.1491637082111015,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.002969863010964869,
+      "loss": 1.277,
+      "step": 1701
+    },
+    {
+      "epoch": 0.14925139998547604,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.002969777289278459,
+      "loss": 1.2748,
+      "step": 1702
+    },
+    {
+      "epoch": 0.1493390917598506,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.0029696914472317672,
+      "loss": 1.332,
+      "step": 1703
+    },
+    {
+      "epoch": 0.14942678353422514,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.0029696054848326226,
+      "loss": 1.3173,
+      "step": 1704
+    },
+    {
+      "epoch": 0.14951447530859968,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.002969519402088864,
+      "loss": 1.3306,
+      "step": 1705
+    },
+    {
+      "epoch": 0.14960216708297422,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.002969433199008342,
+      "loss": 1.2826,
+      "step": 1706
+    },
+    {
+      "epoch": 0.14968985885734878,
+      "grad_norm": 0.0888671875,
+      "learning_rate": 0.0029693468755989188,
+      "loss": 1.3347,
+      "step": 1707
+    },
+    {
+      "epoch": 0.14977755063172332,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.002969260431868466,
+      "loss": 1.2668,
+      "step": 1708
+    },
+    {
+      "epoch": 0.14986524240609786,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.002969173867824868,
+      "loss": 1.2978,
+      "step": 1709
+    },
+    {
+      "epoch": 0.14995293418047242,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.002969087183476018,
+      "loss": 1.3336,
+      "step": 1710
+    },
+    {
+      "epoch": 0.15004062595484696,
+      "grad_norm": 0.0732421875,
+      "learning_rate": 0.0029690003788298224,
+      "loss": 1.3322,
+      "step": 1711
+    },
+    {
+      "epoch": 0.1501283177292215,
+      "grad_norm": 0.0625,
+      "learning_rate": 0.002968913453894197,
+      "loss": 1.2582,
+      "step": 1712
+    },
+    {
+      "epoch": 0.15021600950359604,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.00296882640867707,
+      "loss": 1.2509,
+      "step": 1713
+    },
+    {
+      "epoch": 0.1503037012779706,
+      "grad_norm": 0.0732421875,
+      "learning_rate": 0.0029687392431863788,
+      "loss": 1.3008,
+      "step": 1714
+    },
+    {
+      "epoch": 0.15039139305234514,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.0029686519574300724,
+      "loss": 1.4032,
+      "step": 1715
+    },
+    {
+      "epoch": 0.15047908482671968,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.0029685645514161123,
+      "loss": 1.2226,
+      "step": 1716
+    },
+    {
+      "epoch": 0.15056677660109422,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0029684770251524684,
+      "loss": 1.258,
+      "step": 1717
+    },
+    {
+      "epoch": 0.15065446837546878,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.002968389378647124,
+      "loss": 1.3325,
+      "step": 1718
+    },
+    {
+      "epoch": 0.15074216014984332,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.0029683016119080715,
+      "loss": 1.37,
+      "step": 1719
+    },
+    {
+      "epoch": 0.15082985192421786,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.002968213724943315,
+      "loss": 1.3357,
+      "step": 1720
+    },
+    {
+      "epoch": 0.15091754369859242,
+      "grad_norm": 0.25,
+      "learning_rate": 0.0029681257177608697,
+      "loss": 1.3055,
+      "step": 1721
+    },
+    {
+      "epoch": 0.15100523547296696,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.0029680375903687614,
+      "loss": 1.2907,
+      "step": 1722
+    },
+    {
+      "epoch": 0.1510929272473415,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0029679493427750277,
+      "loss": 1.3671,
+      "step": 1723
+    },
+    {
+      "epoch": 0.15118061902171603,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.002967860974987716,
+      "loss": 1.3317,
+      "step": 1724
+    },
+    {
+      "epoch": 0.1512683107960906,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.002967772487014886,
+      "loss": 1.3038,
+      "step": 1725
+    },
+    {
+      "epoch": 0.15135600257046514,
+      "grad_norm": 0.061279296875,
+      "learning_rate": 0.0029676838788646066,
+      "loss": 1.3724,
+      "step": 1726
+    },
+    {
+      "epoch": 0.15144369434483967,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0029675951505449593,
+      "loss": 1.3427,
+      "step": 1727
+    },
+    {
+      "epoch": 0.15153138611921424,
+      "grad_norm": 0.0791015625,
+      "learning_rate": 0.002967506302064035,
+      "loss": 1.3172,
+      "step": 1728
+    },
+    {
+      "epoch": 0.15161907789358878,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.0029674173334299377,
+      "loss": 1.266,
+      "step": 1729
+    },
+    {
+      "epoch": 0.15170676966796331,
+      "grad_norm": 0.07177734375,
+      "learning_rate": 0.00296732824465078,
+      "loss": 1.3791,
+      "step": 1730
+    },
+    {
+      "epoch": 0.15179446144233785,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.0029672390357346873,
+      "loss": 1.2942,
+      "step": 1731
+    },
+    {
+      "epoch": 0.15188215321671242,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.002967149706689795,
+      "loss": 1.3361,
+      "step": 1732
+    },
+    {
+      "epoch": 0.15196984499108696,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.00296706025752425,
+      "loss": 1.2497,
+      "step": 1733
+    },
+    {
+      "epoch": 0.1520575367654615,
+      "grad_norm": 0.06689453125,
+      "learning_rate": 0.002966970688246209,
+      "loss": 1.2908,
+      "step": 1734
+    },
+    {
+      "epoch": 0.15214522853983603,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.0029668809988638405,
+      "loss": 1.3164,
+      "step": 1735
+    },
+    {
+      "epoch": 0.1522329203142106,
+      "grad_norm": 0.0751953125,
+      "learning_rate": 0.002966791189385325,
+      "loss": 1.2638,
+      "step": 1736
+    },
+    {
+      "epoch": 0.15232061208858513,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.0029667012598188526,
+      "loss": 1.331,
+      "step": 1737
+    },
+    {
+      "epoch": 0.15240830386295967,
+      "grad_norm": 0.07373046875,
+      "learning_rate": 0.0029666112101726237,
+      "loss": 1.2418,
+      "step": 1738
+    },
+    {
+      "epoch": 0.15249599563733424,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.002966521040454852,
+      "loss": 1.31,
+      "step": 1739
+    },
+    {
+      "epoch": 0.15258368741170877,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0029664307506737596,
+      "loss": 1.3026,
+      "step": 1740
+    },
+    {
+      "epoch": 0.1526713791860833,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.0029663403408375813,
+      "loss": 1.2625,
+      "step": 1741
+    },
+    {
+      "epoch": 0.15275907096045785,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.002966249810954561,
+      "loss": 1.2715,
+      "step": 1742
+    },
+    {
+      "epoch": 0.15284676273483241,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.002966159161032957,
+      "loss": 1.2995,
+      "step": 1743
+    },
+    {
+      "epoch": 0.15293445450920695,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.002966068391081035,
+      "loss": 1.2396,
+      "step": 1744
+    },
+    {
+      "epoch": 0.1530221462835815,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.002965977501107073,
+      "loss": 1.2447,
+      "step": 1745
+    },
+    {
+      "epoch": 0.15310983805795605,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.0029658864911193596,
+      "loss": 1.3183,
+      "step": 1746
+    },
+    {
+      "epoch": 0.1531975298323306,
+      "grad_norm": 0.06201171875,
+      "learning_rate": 0.0029657953611261956,
+      "loss": 1.2936,
+      "step": 1747
+    },
+    {
+      "epoch": 0.15328522160670513,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0029657041111358917,
+      "loss": 1.2693,
+      "step": 1748
+    },
+    {
+      "epoch": 0.15337291338107967,
+      "grad_norm": 0.0771484375,
+      "learning_rate": 0.0029656127411567686,
+      "loss": 1.2768,
+      "step": 1749
+    },
+    {
+      "epoch": 0.15346060515545423,
+      "grad_norm": 0.08447265625,
+      "learning_rate": 0.0029655212511971603,
+      "loss": 1.2926,
+      "step": 1750
+    },
+    {
+      "epoch": 0.15354829692982877,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0029654296412654096,
+      "loss": 1.3025,
+      "step": 1751
+    },
+    {
+      "epoch": 0.1536359887042033,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0029653379113698717,
+      "loss": 1.3208,
+      "step": 1752
+    },
+    {
+      "epoch": 0.15372368047857785,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0029652460615189114,
+      "loss": 1.2135,
+      "step": 1753
+    },
+    {
+      "epoch": 0.1538113722529524,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.002965154091720906,
+      "loss": 1.3043,
+      "step": 1754
+    },
+    {
+      "epoch": 0.15389906402732695,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.002965062001984242,
+      "loss": 1.3056,
+      "step": 1755
+    },
+    {
+      "epoch": 0.15398675580170149,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.0029649697923173185,
+      "loss": 1.3145,
+      "step": 1756
+    },
+    {
+      "epoch": 0.15407444757607605,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0029648774627285446,
+      "loss": 1.2702,
+      "step": 1757
+    },
+    {
+      "epoch": 0.1541621393504506,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0029647850132263403,
+      "loss": 1.387,
+      "step": 1758
+    },
+    {
+      "epoch": 0.15424983112482513,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.002964692443819137,
+      "loss": 1.3019,
+      "step": 1759
+    },
+    {
+      "epoch": 0.15433752289919966,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.002964599754515377,
+      "loss": 1.2936,
+      "step": 1760
+    },
+    {
+      "epoch": 0.15442521467357423,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.002964506945323512,
+      "loss": 1.3148,
+      "step": 1761
+    },
+    {
+      "epoch": 0.15451290644794877,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.002964414016252008,
+      "loss": 1.271,
+      "step": 1762
+    },
+    {
+      "epoch": 0.1546005982223233,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.0029643209673093384,
+      "loss": 1.3235,
+      "step": 1763
+    },
+    {
+      "epoch": 0.15468828999669787,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0029642277985039893,
+      "loss": 1.3947,
+      "step": 1764
+    },
+    {
+      "epoch": 0.1547759817710724,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0029641345098444585,
+      "loss": 1.2814,
+      "step": 1765
+    },
+    {
+      "epoch": 0.15486367354544694,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.002964041101339252,
+      "loss": 1.3002,
+      "step": 1766
+    },
+    {
+      "epoch": 0.15495136531982148,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0029639475729968893,
+      "loss": 1.2534,
+      "step": 1767
+    },
+    {
+      "epoch": 0.15503905709419605,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0029638539248259,
+      "loss": 1.3228,
+      "step": 1768
+    },
+    {
+      "epoch": 0.15512674886857059,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.002963760156834825,
+      "loss": 1.2834,
+      "step": 1769
+    },
+    {
+      "epoch": 0.15521444064294512,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.0029636662690322143,
+      "loss": 1.3193,
+      "step": 1770
+    },
+    {
+      "epoch": 0.15530213241731966,
+      "grad_norm": 0.072265625,
+      "learning_rate": 0.0029635722614266315,
+      "loss": 1.3285,
+      "step": 1771
+    },
+    {
+      "epoch": 0.15538982419169423,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.002963478134026649,
+      "loss": 1.3342,
+      "step": 1772
+    },
+    {
+      "epoch": 0.15547751596606876,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.002963383886840852,
+      "loss": 1.2917,
+      "step": 1773
+    },
+    {
+      "epoch": 0.1555652077404433,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.002963289519877835,
+      "loss": 1.2716,
+      "step": 1774
+    },
+    {
+      "epoch": 0.15565289951481787,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.0029631950331462037,
+      "loss": 1.3703,
+      "step": 1775
+    },
+    {
+      "epoch": 0.1557405912891924,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.0029631004266545756,
+      "loss": 1.2962,
+      "step": 1776
+    },
+    {
+      "epoch": 0.15582828306356694,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.002963005700411578,
+      "loss": 1.2938,
+      "step": 1777
+    },
+    {
+      "epoch": 0.15591597483794148,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00296291085442585,
+      "loss": 1.3536,
+      "step": 1778
+    },
+    {
+      "epoch": 0.15600366661231604,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.0029628158887060416,
+      "loss": 1.2311,
+      "step": 1779
+    },
+    {
+      "epoch": 0.15609135838669058,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.002962720803260813,
+      "loss": 1.2521,
+      "step": 1780
+    },
+    {
+      "epoch": 0.15617905016106512,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.0029626255980988365,
+      "loss": 1.2764,
+      "step": 1781
+    },
+    {
+      "epoch": 0.15626674193543968,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.0029625302732287934,
+      "loss": 1.2357,
+      "step": 1782
+    },
+    {
+      "epoch": 0.15635443370981422,
+      "grad_norm": 0.076171875,
+      "learning_rate": 0.0029624348286593776,
+      "loss": 1.3649,
+      "step": 1783
+    },
+    {
+      "epoch": 0.15644212548418876,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.0029623392643992937,
+      "loss": 1.3346,
+      "step": 1784
+    },
+    {
+      "epoch": 0.1565298172585633,
+      "grad_norm": 0.07373046875,
+      "learning_rate": 0.0029622435804572563,
+      "loss": 1.3027,
+      "step": 1785
+    },
+    {
+      "epoch": 0.15661750903293786,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0029621477768419927,
+      "loss": 1.3065,
+      "step": 1786
+    },
+    {
+      "epoch": 0.1567052008073124,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.002962051853562238,
+      "loss": 1.2124,
+      "step": 1787
+    },
+    {
+      "epoch": 0.15679289258168694,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.0029619558106267424,
+      "loss": 1.3163,
+      "step": 1788
+    },
+    {
+      "epoch": 0.15688058435606148,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0029618596480442635,
+      "loss": 1.266,
+      "step": 1789
+    },
+    {
+      "epoch": 0.15696827613043604,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.0029617633658235707,
+      "loss": 1.3783,
+      "step": 1790
+    },
+    {
+      "epoch": 0.15705596790481058,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.0029616669639734457,
+      "loss": 1.3268,
+      "step": 1791
+    },
+    {
+      "epoch": 0.15714365967918512,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.002961570442502679,
+      "loss": 1.2861,
+      "step": 1792
+    },
+    {
+      "epoch": 0.15723135145355968,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.0029614738014200745,
+      "loss": 1.3748,
+      "step": 1793
+    },
+    {
+      "epoch": 0.15731904322793422,
+      "grad_norm": 0.0859375,
+      "learning_rate": 0.0029613770407344447,
+      "loss": 1.2581,
+      "step": 1794
+    },
+    {
+      "epoch": 0.15740673500230876,
+      "grad_norm": 0.07763671875,
+      "learning_rate": 0.002961280160454614,
+      "loss": 1.2899,
+      "step": 1795
+    },
+    {
+      "epoch": 0.1574944267766833,
+      "grad_norm": 0.07373046875,
+      "learning_rate": 0.0029611831605894172,
+      "loss": 1.2457,
+      "step": 1796
+    },
+    {
+      "epoch": 0.15758211855105786,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.0029610860411477015,
+      "loss": 1.3766,
+      "step": 1797
+    },
+    {
+      "epoch": 0.1576698103254324,
+      "grad_norm": 0.06494140625,
+      "learning_rate": 0.0029609888021383235,
+      "loss": 1.3176,
+      "step": 1798
+    },
+    {
+      "epoch": 0.15775750209980693,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.002960891443570151,
+      "loss": 1.2908,
+      "step": 1799
+    },
+    {
+      "epoch": 0.1578451938741815,
+      "grad_norm": 0.07470703125,
+      "learning_rate": 0.0029607939654520627,
+      "loss": 1.305,
+      "step": 1800
+    },
+    {
+      "epoch": 0.15793288564855604,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.0029606963677929485,
+      "loss": 1.2922,
+      "step": 1801
+    },
+    {
+      "epoch": 0.15802057742293057,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.0029605986506017093,
+      "loss": 1.2961,
+      "step": 1802
+    },
+    {
+      "epoch": 0.1581082691973051,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.0029605008138872562,
+      "loss": 1.2114,
+      "step": 1803
+    },
+    {
+      "epoch": 0.15819596097167968,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.0029604028576585124,
+      "loss": 1.2436,
+      "step": 1804
+    },
+    {
+      "epoch": 0.15828365274605422,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.00296030478192441,
+      "loss": 1.3183,
+      "step": 1805
+    },
+    {
+      "epoch": 0.15837134452042875,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.002960206586693895,
+      "loss": 1.2487,
+      "step": 1806
+    },
+    {
+      "epoch": 0.1584590362948033,
+      "grad_norm": 0.0703125,
+      "learning_rate": 0.002960108271975921,
+      "loss": 1.3022,
+      "step": 1807
+    },
+    {
+      "epoch": 0.15854672806917786,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0029600098377794543,
+      "loss": 1.3295,
+      "step": 1808
+    },
+    {
+      "epoch": 0.1586344198435524,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.0029599112841134723,
+      "loss": 1.297,
+      "step": 1809
+    },
+    {
+      "epoch": 0.15872211161792693,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.0029598126109869633,
+      "loss": 1.3147,
+      "step": 1810
+    },
+    {
+      "epoch": 0.1588098033923015,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.002959713818408925,
+      "loss": 1.2913,
+      "step": 1811
+    },
+    {
+      "epoch": 0.15889749516667603,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.0029596149063883677,
+      "loss": 1.233,
+      "step": 1812
+    },
+    {
+      "epoch": 0.15898518694105057,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.0029595158749343114,
+      "loss": 1.2982,
+      "step": 1813
+    },
+    {
+      "epoch": 0.1590728787154251,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.0029594167240557883,
+      "loss": 1.3088,
+      "step": 1814
+    },
+    {
+      "epoch": 0.15916057048979967,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.0029593174537618392,
+      "loss": 1.3154,
+      "step": 1815
+    },
+    {
+      "epoch": 0.1592482622641742,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0029592180640615195,
+      "loss": 1.2972,
+      "step": 1816
+    },
+    {
+      "epoch": 0.15933595403854875,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0029591185549638914,
+      "loss": 1.32,
+      "step": 1817
+    },
+    {
+      "epoch": 0.15942364581292331,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.002959018926478031,
+      "loss": 1.2938,
+      "step": 1818
+    },
+    {
+      "epoch": 0.15951133758729785,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.002958919178613023,
+      "loss": 1.2777,
+      "step": 1819
+    },
+    {
+      "epoch": 0.1595990293616724,
+      "grad_norm": 0.08251953125,
+      "learning_rate": 0.002958819311377965,
+      "loss": 1.3409,
+      "step": 1820
+    },
+    {
+      "epoch": 0.15968672113604693,
+      "grad_norm": 0.080078125,
+      "learning_rate": 0.002958719324781965,
+      "loss": 1.2738,
+      "step": 1821
+    },
+    {
+      "epoch": 0.1597744129104215,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.002958619218834141,
+      "loss": 1.2636,
+      "step": 1822
+    },
+    {
+      "epoch": 0.15986210468479603,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.002958518993543622,
+      "loss": 1.3311,
+      "step": 1823
+    },
+    {
+      "epoch": 0.15994979645917057,
+      "grad_norm": 0.06640625,
+      "learning_rate": 0.002958418648919549,
+      "loss": 1.3638,
+      "step": 1824
+    },
+    {
+      "epoch": 0.1600374882335451,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.0029583181849710733,
+      "loss": 1.3066,
+      "step": 1825
+    },
+    {
+      "epoch": 0.16012518000791967,
+      "grad_norm": 0.0634765625,
+      "learning_rate": 0.0029582176017073558,
+      "loss": 1.3149,
+      "step": 1826
+    },
+    {
+      "epoch": 0.1602128717822942,
+      "grad_norm": 0.07080078125,
+      "learning_rate": 0.00295811689913757,
+      "loss": 1.2287,
+      "step": 1827
+    },
+    {
+      "epoch": 0.16030056355666875,
+      "grad_norm": 0.0859375,
+      "learning_rate": 0.002958016077270901,
+      "loss": 1.2944,
+      "step": 1828
+    },
+    {
+      "epoch": 0.1603882553310433,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0029579151361165414,
+      "loss": 1.3327,
+      "step": 1829
+    },
+    {
+      "epoch": 0.16047594710541785,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.0029578140756836985,
+      "loss": 1.2067,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1605636388797924,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0029577128959815875,
+      "loss": 1.2926,
+      "step": 1831
+    },
+    {
+      "epoch": 0.16065133065416692,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.0029576115970194362,
+      "loss": 1.3344,
+      "step": 1832
+    },
+    {
+      "epoch": 0.1607390224285415,
+      "grad_norm": 0.07177734375,
+      "learning_rate": 0.0029575101788064826,
+      "loss": 1.2922,
+      "step": 1833
+    },
+    {
+      "epoch": 0.16082671420291603,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0029574086413519766,
+      "loss": 1.2902,
+      "step": 1834
+    },
+    {
+      "epoch": 0.16091440597729056,
+      "grad_norm": 0.068359375,
+      "learning_rate": 0.0029573069846651773,
+      "loss": 1.3136,
+      "step": 1835
+    },
+    {
+      "epoch": 0.16100209775166513,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.002957205208755356,
+      "loss": 1.1953,
+      "step": 1836
+    },
+    {
+      "epoch": 0.16108978952603967,
+      "grad_norm": 0.06982421875,
+      "learning_rate": 0.0029571033136317937,
+      "loss": 1.3136,
+      "step": 1837
+    },
+    {
+      "epoch": 0.1611774813004142,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 0.002957001299303784,
+      "loss": 1.2416,
+      "step": 1838
+    },
+    {
+      "epoch": 0.16126517307478874,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.0029568991657806295,
+      "loss": 1.2981,
+      "step": 1839
+    },
+    {
+      "epoch": 0.1613528648491633,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.002956796913071645,
+      "loss": 1.273,
+      "step": 1840
+    },
+    {
+      "epoch": 0.16144055662353785,
+      "grad_norm": 0.06298828125,
+      "learning_rate": 0.002956694541186155,
+      "loss": 1.2522,
+      "step": 1841
+    },
+    {
+      "epoch": 0.16152824839791238,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.002956592050133497,
+      "loss": 1.2785,
+      "step": 1842
+    },
+    {
+      "epoch": 0.16161594017228692,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.0029564894399230165,
+      "loss": 1.2929,
+      "step": 1843
+    },
+    {
+      "epoch": 0.16170363194666149,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.0029563867105640716,
+      "loss": 1.3277,
+      "step": 1844
+    },
+    {
+      "epoch": 0.16179132372103602,
+      "grad_norm": 0.08251953125,
+      "learning_rate": 0.002956283862066031,
+      "loss": 1.2834,
+      "step": 1845
+    },
+    {
+      "epoch": 0.16187901549541056,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0029561808944382744,
+      "loss": 1.24,
+      "step": 1846
+    },
+    {
+      "epoch": 0.16196670726978513,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.0029560778076901926,
+      "loss": 1.3224,
+      "step": 1847
+    },
+    {
+      "epoch": 0.16205439904415966,
+      "grad_norm": 0.07421875,
+      "learning_rate": 0.0029559746018311857,
+      "loss": 1.296,
+      "step": 1848
+    },
+    {
+      "epoch": 0.1621420908185342,
+      "grad_norm": 0.08154296875,
+      "learning_rate": 0.002955871276870667,
+      "loss": 1.3479,
+      "step": 1849
+    },
+    {
+      "epoch": 0.16222978259290874,
+      "grad_norm": 0.07177734375,
+      "learning_rate": 0.0029557678328180586,
+      "loss": 1.3016,
+      "step": 1850
+    },
+    {
+      "epoch": 0.1623174743672833,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 0.002955664269682795,
+      "loss": 1.3169,
+      "step": 1851
+    },
+    {
+      "epoch": 0.16240516614165784,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.0029555605874743204,
+      "loss": 1.2505,
+      "step": 1852
+    },
+    {
+      "epoch": 0.16249285791603238,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.0029554567862020904,
+      "loss": 1.3084,
+      "step": 1853
+    },
+    {
+      "epoch": 0.16258054969040694,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.002955352865875572,
+      "loss": 1.3385,
+      "step": 1854
+    },
+    {
+      "epoch": 0.16266824146478148,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.002955248826504241,
+      "loss": 1.3753,
+      "step": 1855
+    },
+    {
+      "epoch": 0.16275593323915602,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.002955144668097588,
+      "loss": 1.2402,
+      "step": 1856
+    },
+    {
+      "epoch": 0.16284362501353056,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.0029550403906651087,
+      "loss": 1.2476,
+      "step": 1857
+    },
+    {
+      "epoch": 0.16293131678790512,
+      "grad_norm": 0.0712890625,
+      "learning_rate": 0.0029549359942163157,
+      "loss": 1.2843,
+      "step": 1858
+    },
+    {
+      "epoch": 0.16301900856227966,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0029548314787607288,
+      "loss": 1.2993,
+      "step": 1859
+    },
+    {
+      "epoch": 0.1631067003366542,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.002954726844307879,
+      "loss": 1.3391,
+      "step": 1860
+    },
+    {
+      "epoch": 0.16319439211102874,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.0029546220908673094,
+      "loss": 1.319,
+      "step": 1861
+    },
+    {
+      "epoch": 0.1632820838854033,
+      "grad_norm": 0.07568359375,
+      "learning_rate": 0.0029545172184485733,
+      "loss": 1.2632,
+      "step": 1862
+    },
+    {
+      "epoch": 0.16336977565977784,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.002954412227061234,
+      "loss": 1.2358,
+      "step": 1863
+    },
+    {
+      "epoch": 0.16345746743415238,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.0029543071167148667,
+      "loss": 1.2597,
+      "step": 1864
+    },
+    {
+      "epoch": 0.16354515920852694,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.002954201887419058,
+      "loss": 1.2416,
+      "step": 1865
+    },
+    {
+      "epoch": 0.16363285098290148,
+      "grad_norm": 0.08544921875,
+      "learning_rate": 0.002954096539183404,
+      "loss": 1.2841,
+      "step": 1866
+    },
+    {
+      "epoch": 0.16372054275727602,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.0029539910720175113,
+      "loss": 1.2665,
+      "step": 1867
+    },
+    {
+      "epoch": 0.16380823453165055,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0029538854859309997,
+      "loss": 1.3402,
+      "step": 1868
+    },
+    {
+      "epoch": 0.16389592630602512,
+      "grad_norm": 0.0693359375,
+      "learning_rate": 0.0029537797809334977,
+      "loss": 1.2414,
+      "step": 1869
+    },
+    {
+      "epoch": 0.16398361808039966,
+      "grad_norm": 0.07275390625,
+      "learning_rate": 0.002953673957034645,
+      "loss": 1.2813,
+      "step": 1870
+    },
+    {
+      "epoch": 0.1640713098547742,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.0029535680142440935,
+      "loss": 1.2876,
+      "step": 1871
+    },
+    {
+      "epoch": 0.16415900162914873,
+      "grad_norm": 0.06298828125,
+      "learning_rate": 0.0029534619525715036,
+      "loss": 1.2672,
+      "step": 1872
+    },
+    {
+      "epoch": 0.1642466934035233,
+      "grad_norm": 0.080078125,
+      "learning_rate": 0.0029533557720265494,
+      "loss": 1.3085,
+      "step": 1873
+    },
+    {
+      "epoch": 0.16433438517789783,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.0029532494726189126,
+      "loss": 1.2638,
+      "step": 1874
+    },
+    {
+      "epoch": 0.16442207695227237,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.002953143054358288,
+      "loss": 1.2597,
+      "step": 1875
+    },
+    {
+      "epoch": 0.16450976872664694,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.0029530365172543817,
+      "loss": 1.2191,
+      "step": 1876
+    },
+    {
+      "epoch": 0.16459746050102148,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.002952929861316908,
+      "loss": 1.297,
+      "step": 1877
+    },
+    {
+      "epoch": 0.164685152275396,
+      "grad_norm": 0.0673828125,
+      "learning_rate": 0.002952823086555595,
+      "loss": 1.2995,
+      "step": 1878
+    },
+    {
+      "epoch": 0.16477284404977055,
+      "grad_norm": 0.0595703125,
+      "learning_rate": 0.0029527161929801798,
+      "loss": 1.2785,
+      "step": 1879
+    },
+    {
+      "epoch": 0.16486053582414512,
+      "grad_norm": 0.0703125,
+      "learning_rate": 0.00295260918060041,
+      "loss": 1.2669,
+      "step": 1880
+    },
+    {
+      "epoch": 0.16494822759851965,
+      "grad_norm": 0.0673828125,
+      "learning_rate": 0.0029525020494260462,
+      "loss": 1.2682,
+      "step": 1881
+    },
+    {
+      "epoch": 0.1650359193728942,
+      "grad_norm": 0.056396484375,
+      "learning_rate": 0.0029523947994668583,
+      "loss": 1.3312,
+      "step": 1882
+    },
+    {
+      "epoch": 0.16512361114726876,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0029522874307326263,
+      "loss": 1.3099,
+      "step": 1883
+    },
+    {
+      "epoch": 0.1652113029216433,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.002952179943233142,
+      "loss": 1.2679,
+      "step": 1884
+    },
+    {
+      "epoch": 0.16529899469601783,
+      "grad_norm": 0.07421875,
+      "learning_rate": 0.002952072336978209,
+      "loss": 1.2699,
+      "step": 1885
+    },
+    {
+      "epoch": 0.16538668647039237,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.00295196461197764,
+      "loss": 1.2717,
+      "step": 1886
+    },
+    {
+      "epoch": 0.16547437824476693,
+      "grad_norm": 0.076171875,
+      "learning_rate": 0.0029518567682412593,
+      "loss": 1.2344,
+      "step": 1887
+    },
+    {
+      "epoch": 0.16556207001914147,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0029517488057789025,
+      "loss": 1.2496,
+      "step": 1888
+    },
+    {
+      "epoch": 0.165649761793516,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.002951640724600414,
+      "loss": 1.32,
+      "step": 1889
+    },
+    {
+      "epoch": 0.16573745356789055,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0029515325247156526,
+      "loss": 1.3256,
+      "step": 1890
+    },
+    {
+      "epoch": 0.1658251453422651,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.0029514242061344847,
+      "loss": 1.3123,
+      "step": 1891
+    },
+    {
+      "epoch": 0.16591283711663965,
+      "grad_norm": 0.0751953125,
+      "learning_rate": 0.0029513157688667883,
+      "loss": 1.317,
+      "step": 1892
+    },
+    {
+      "epoch": 0.1660005288910142,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.002951207212922454,
+      "loss": 1.2483,
+      "step": 1893
+    },
+    {
+      "epoch": 0.16608822066538875,
+      "grad_norm": 0.07373046875,
+      "learning_rate": 0.00295109853831138,
+      "loss": 1.3007,
+      "step": 1894
+    },
+    {
+      "epoch": 0.1661759124397633,
+      "grad_norm": 0.0625,
+      "learning_rate": 0.0029509897450434785,
+      "loss": 1.2602,
+      "step": 1895
+    },
+    {
+      "epoch": 0.16626360421413783,
+      "grad_norm": 0.0751953125,
+      "learning_rate": 0.002950880833128671,
+      "loss": 1.3378,
+      "step": 1896
+    },
+    {
+      "epoch": 0.16635129598851237,
+      "grad_norm": 0.06982421875,
+      "learning_rate": 0.0029507718025768895,
+      "loss": 1.3414,
+      "step": 1897
+    },
+    {
+      "epoch": 0.16643898776288693,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.0029506626533980776,
+      "loss": 1.268,
+      "step": 1898
+    },
+    {
+      "epoch": 0.16652667953726147,
+      "grad_norm": 0.08544921875,
+      "learning_rate": 0.002950553385602189,
+      "loss": 1.2873,
+      "step": 1899
+    },
+    {
+      "epoch": 0.166614371311636,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.002950443999199189,
+      "loss": 1.2687,
+      "step": 1900
+    },
+    {
+      "epoch": 0.16670206308601057,
+      "grad_norm": 0.06494140625,
+      "learning_rate": 0.002950334494199054,
+      "loss": 1.307,
+      "step": 1901
+    },
+    {
+      "epoch": 0.1667897548603851,
+      "grad_norm": 0.0888671875,
+      "learning_rate": 0.0029502248706117693,
+      "loss": 1.3191,
+      "step": 1902
+    },
+    {
+      "epoch": 0.16687744663475965,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 0.002950115128447333,
+      "loss": 1.3759,
+      "step": 1903
+    },
+    {
+      "epoch": 0.16696513840913418,
+      "grad_norm": 0.0751953125,
+      "learning_rate": 0.0029500052677157535,
+      "loss": 1.2491,
+      "step": 1904
+    },
+    {
+      "epoch": 0.16705283018350875,
+      "grad_norm": 0.06982421875,
+      "learning_rate": 0.0029498952884270493,
+      "loss": 1.2773,
+      "step": 1905
+    },
+    {
+      "epoch": 0.1671405219578833,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.00294978519059125,
+      "loss": 1.2975,
+      "step": 1906
+    },
+    {
+      "epoch": 0.16722821373225782,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.0029496749742183976,
+      "loss": 1.3059,
+      "step": 1907
+    },
+    {
+      "epoch": 0.16731590550663236,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.002949564639318542,
+      "loss": 1.3331,
+      "step": 1908
+    },
+    {
+      "epoch": 0.16740359728100693,
+      "grad_norm": 0.0791015625,
+      "learning_rate": 0.0029494541859017465,
+      "loss": 1.2586,
+      "step": 1909
+    },
+    {
+      "epoch": 0.16749128905538146,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.0029493436139780838,
+      "loss": 1.3039,
+      "step": 1910
+    },
+    {
+      "epoch": 0.167578980829756,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0029492329235576375,
+      "loss": 1.281,
+      "step": 1911
+    },
+    {
+      "epoch": 0.16766667260413057,
+      "grad_norm": 0.0859375,
+      "learning_rate": 0.0029491221146505024,
+      "loss": 1.2381,
+      "step": 1912
+    },
+    {
+      "epoch": 0.1677543643785051,
+      "grad_norm": 0.0712890625,
+      "learning_rate": 0.0029490111872667846,
+      "loss": 1.2411,
+      "step": 1913
+    },
+    {
+      "epoch": 0.16784205615287964,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0029489001414165997,
+      "loss": 1.2881,
+      "step": 1914
+    },
+    {
+      "epoch": 0.16792974792725418,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.0029487889771100746,
+      "loss": 1.1976,
+      "step": 1915
+    },
+    {
+      "epoch": 0.16801743970162875,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.002948677694357348,
+      "loss": 1.2403,
+      "step": 1916
+    },
+    {
+      "epoch": 0.16810513147600328,
+      "grad_norm": 0.07861328125,
+      "learning_rate": 0.002948566293168568,
+      "loss": 1.2295,
+      "step": 1917
+    },
+    {
+      "epoch": 0.16819282325037782,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0029484547735538946,
+      "loss": 1.3096,
+      "step": 1918
+    },
+    {
+      "epoch": 0.1682805150247524,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.0029483431355234973,
+      "loss": 1.2894,
+      "step": 1919
+    },
+    {
+      "epoch": 0.16836820679912692,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.002948231379087558,
+      "loss": 1.3074,
+      "step": 1920
+    },
+    {
+      "epoch": 0.16845589857350146,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0029481195042562686,
+      "loss": 1.2618,
+      "step": 1921
+    },
+    {
+      "epoch": 0.168543590347876,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.002948007511039831,
+      "loss": 1.3031,
+      "step": 1922
+    },
+    {
+      "epoch": 0.16863128212225056,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.002947895399448459,
+      "loss": 1.317,
+      "step": 1923
+    },
+    {
+      "epoch": 0.1687189738966251,
+      "grad_norm": 0.125,
+      "learning_rate": 0.0029477831694923774,
+      "loss": 1.2723,
+      "step": 1924
+    },
+    {
+      "epoch": 0.16880666567099964,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0029476708211818212,
+      "loss": 1.3451,
+      "step": 1925
+    },
+    {
+      "epoch": 0.16889435744537418,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.002947558354527036,
+      "loss": 1.2635,
+      "step": 1926
+    },
+    {
+      "epoch": 0.16898204921974874,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.002947445769538278,
+      "loss": 1.276,
+      "step": 1927
+    },
+    {
+      "epoch": 0.16906974099412328,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0029473330662258155,
+      "loss": 1.3344,
+      "step": 1928
+    },
+    {
+      "epoch": 0.16915743276849782,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.002947220244599926,
+      "loss": 1.2571,
+      "step": 1929
+    },
+    {
+      "epoch": 0.16924512454287238,
+      "grad_norm": 0.06884765625,
+      "learning_rate": 0.0029471073046709,
+      "loss": 1.2501,
+      "step": 1930
+    },
+    {
+      "epoch": 0.16933281631724692,
+      "grad_norm": 0.076171875,
+      "learning_rate": 0.0029469942464490353,
+      "loss": 1.2875,
+      "step": 1931
+    },
+    {
+      "epoch": 0.16942050809162146,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.002946881069944644,
+      "loss": 1.2909,
+      "step": 1932
+    },
+    {
+      "epoch": 0.169508199865996,
+      "grad_norm": 0.072265625,
+      "learning_rate": 0.002946767775168047,
+      "loss": 1.3028,
+      "step": 1933
+    },
+    {
+      "epoch": 0.16959589164037056,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.002946654362129576,
+      "loss": 1.2602,
+      "step": 1934
+    },
+    {
+      "epoch": 0.1696835834147451,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.0029465408308395748,
+      "loss": 1.319,
+      "step": 1935
+    },
+    {
+      "epoch": 0.16977127518911964,
+      "grad_norm": 0.078125,
+      "learning_rate": 0.002946427181308397,
+      "loss": 1.2907,
+      "step": 1936
+    },
+    {
+      "epoch": 0.1698589669634942,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.0029463134135464066,
+      "loss": 1.2207,
+      "step": 1937
+    },
+    {
+      "epoch": 0.16994665873786874,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.0029461995275639795,
+      "loss": 1.2865,
+      "step": 1938
+    },
+    {
+      "epoch": 0.17003435051224328,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0029460855233715013,
+      "loss": 1.2687,
+      "step": 1939
+    },
+    {
+      "epoch": 0.17012204228661781,
+      "grad_norm": 0.0771484375,
+      "learning_rate": 0.0029459714009793694,
+      "loss": 1.2635,
+      "step": 1940
+    },
+    {
+      "epoch": 0.17020973406099238,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.0029458571603979914,
+      "loss": 1.3113,
+      "step": 1941
+    },
+    {
+      "epoch": 0.17029742583536692,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.0029457428016377856,
+      "loss": 1.2634,
+      "step": 1942
+    },
+    {
+      "epoch": 0.17038511760974145,
+      "grad_norm": 0.07275390625,
+      "learning_rate": 0.002945628324709181,
+      "loss": 1.3609,
+      "step": 1943
+    },
+    {
+      "epoch": 0.170472809384116,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.0029455137296226183,
+      "loss": 1.2946,
+      "step": 1944
+    },
+    {
+      "epoch": 0.17056050115849056,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.0029453990163885472,
+      "loss": 1.3569,
+      "step": 1945
+    },
+    {
+      "epoch": 0.1706481929328651,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.00294528418501743,
+      "loss": 1.2823,
+      "step": 1946
+    },
+    {
+      "epoch": 0.17073588470723963,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.0029451692355197396,
+      "loss": 1.2662,
+      "step": 1947
+    },
+    {
+      "epoch": 0.1708235764816142,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0029450541679059573,
+      "loss": 1.2931,
+      "step": 1948
+    },
+    {
+      "epoch": 0.17091126825598874,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0029449389821865793,
+      "loss": 1.1969,
+      "step": 1949
+    },
+    {
+      "epoch": 0.17099896003036327,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.0029448236783721077,
+      "loss": 1.2791,
+      "step": 1950
+    },
+    {
+      "epoch": 0.1710866518047378,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.00294470825647306,
+      "loss": 1.3965,
+      "step": 1951
+    },
+    {
+      "epoch": 0.17117434357911238,
+      "grad_norm": 0.072265625,
+      "learning_rate": 0.0029445927164999616,
+      "loss": 1.3565,
+      "step": 1952
+    },
+    {
+      "epoch": 0.1712620353534869,
+      "grad_norm": 0.08154296875,
+      "learning_rate": 0.0029444770584633488,
+      "loss": 1.2791,
+      "step": 1953
+    },
+    {
+      "epoch": 0.17134972712786145,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.0029443612823737706,
+      "loss": 1.2292,
+      "step": 1954
+    },
+    {
+      "epoch": 0.17143741890223602,
+      "grad_norm": 0.0732421875,
+      "learning_rate": 0.0029442453882417845,
+      "loss": 1.2841,
+      "step": 1955
+    },
+    {
+      "epoch": 0.17152511067661055,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.00294412937607796,
+      "loss": 1.2832,
+      "step": 1956
+    },
+    {
+      "epoch": 0.1716128024509851,
+      "grad_norm": 0.0712890625,
+      "learning_rate": 0.002944013245892877,
+      "loss": 1.3156,
+      "step": 1957
+    },
+    {
+      "epoch": 0.17170049422535963,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.0029438969976971273,
+      "loss": 1.291,
+      "step": 1958
+    },
+    {
+      "epoch": 0.1717881859997342,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.002943780631501311,
+      "loss": 1.2569,
+      "step": 1959
+    },
+    {
+      "epoch": 0.17187587777410873,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.002943664147316041,
+      "loss": 1.3101,
+      "step": 1960
+    },
+    {
+      "epoch": 0.17196356954848327,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0029435475451519404,
+      "loss": 1.2592,
+      "step": 1961
+    },
+    {
+      "epoch": 0.1720512613228578,
+      "grad_norm": 0.08642578125,
+      "learning_rate": 0.002943430825019643,
+      "loss": 1.3996,
+      "step": 1962
+    },
+    {
+      "epoch": 0.17213895309723237,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.002943313986929793,
+      "loss": 1.2734,
+      "step": 1963
+    },
+    {
+      "epoch": 0.1722266448716069,
+      "grad_norm": 0.07861328125,
+      "learning_rate": 0.0029431970308930466,
+      "loss": 1.3204,
+      "step": 1964
+    },
+    {
+      "epoch": 0.17231433664598145,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.0029430799569200687,
+      "loss": 1.27,
+      "step": 1965
+    },
+    {
+      "epoch": 0.172402028420356,
+      "grad_norm": 0.07177734375,
+      "learning_rate": 0.0029429627650215374,
+      "loss": 1.2319,
+      "step": 1966
+    },
+    {
+      "epoch": 0.17248972019473055,
+      "grad_norm": 0.076171875,
+      "learning_rate": 0.0029428454552081393,
+      "loss": 1.3078,
+      "step": 1967
+    },
+    {
+      "epoch": 0.1725774119691051,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.002942728027490574,
+      "loss": 1.2591,
+      "step": 1968
+    },
+    {
+      "epoch": 0.17266510374347963,
+      "grad_norm": 0.08642578125,
+      "learning_rate": 0.002942610481879549,
+      "loss": 1.3353,
+      "step": 1969
+    },
+    {
+      "epoch": 0.1727527955178542,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0029424928183857846,
+      "loss": 1.3396,
+      "step": 1970
+    },
+    {
+      "epoch": 0.17284048729222873,
+      "grad_norm": 0.080078125,
+      "learning_rate": 0.002942375037020012,
+      "loss": 1.3243,
+      "step": 1971
+    },
+    {
+      "epoch": 0.17292817906660327,
+      "grad_norm": 0.06884765625,
+      "learning_rate": 0.0029422571377929726,
+      "loss": 1.3106,
+      "step": 1972
+    },
+    {
+      "epoch": 0.17301587084097783,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.0029421391207154184,
+      "loss": 1.2602,
+      "step": 1973
+    },
+    {
+      "epoch": 0.17310356261535237,
+      "grad_norm": 0.064453125,
+      "learning_rate": 0.0029420209857981116,
+      "loss": 1.2806,
+      "step": 1974
+    },
+    {
+      "epoch": 0.1731912543897269,
+      "grad_norm": 0.08935546875,
+      "learning_rate": 0.0029419027330518255,
+      "loss": 1.2117,
+      "step": 1975
+    },
+    {
+      "epoch": 0.17327894616410144,
+      "grad_norm": 0.0751953125,
+      "learning_rate": 0.002941784362487346,
+      "loss": 1.3052,
+      "step": 1976
+    },
+    {
+      "epoch": 0.173366637938476,
+      "grad_norm": 0.0771484375,
+      "learning_rate": 0.0029416658741154675,
+      "loss": 1.2397,
+      "step": 1977
+    },
+    {
+      "epoch": 0.17345432971285055,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.0029415472679469954,
+      "loss": 1.2818,
+      "step": 1978
+    },
+    {
+      "epoch": 0.17354202148722508,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.0029414285439927465,
+      "loss": 1.2803,
+      "step": 1979
+    },
+    {
+      "epoch": 0.17362971326159962,
+      "grad_norm": 0.0654296875,
+      "learning_rate": 0.0029413097022635486,
+      "loss": 1.2976,
+      "step": 1980
+    },
+    {
+      "epoch": 0.1737174050359742,
+      "grad_norm": 0.07080078125,
+      "learning_rate": 0.002941190742770239,
+      "loss": 1.3047,
+      "step": 1981
+    },
+    {
+      "epoch": 0.17380509681034872,
+      "grad_norm": 0.07421875,
+      "learning_rate": 0.002941071665523667,
+      "loss": 1.2546,
+      "step": 1982
+    },
+    {
+      "epoch": 0.17389278858472326,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.0029409524705346917,
+      "loss": 1.2918,
+      "step": 1983
+    },
+    {
+      "epoch": 0.17398048035909783,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.002940833157814184,
+      "loss": 1.3862,
+      "step": 1984
+    },
+    {
+      "epoch": 0.17406817213347237,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.0029407137273730244,
+      "loss": 1.3218,
+      "step": 1985
+    },
+    {
+      "epoch": 0.1741558639078469,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.002940594179222105,
+      "loss": 1.3668,
+      "step": 1986
+    },
+    {
+      "epoch": 0.17424355568222144,
+      "grad_norm": 0.23046875,
+      "learning_rate": 0.002940474513372328,
+      "loss": 1.2302,
+      "step": 1987
+    },
+    {
+      "epoch": 0.174331247456596,
+      "grad_norm": 0.07421875,
+      "learning_rate": 0.0029403547298346064,
+      "loss": 1.3277,
+      "step": 1988
+    },
+    {
+      "epoch": 0.17441893923097054,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0029402348286198653,
+      "loss": 1.2634,
+      "step": 1989
+    },
+    {
+      "epoch": 0.17450663100534508,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.002940114809739038,
+      "loss": 1.2542,
+      "step": 1990
+    },
+    {
+      "epoch": 0.17459432277971965,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.0029399946732030706,
+      "loss": 1.3228,
+      "step": 1991
+    },
+    {
+      "epoch": 0.17468201455409418,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0029398744190229187,
+      "loss": 1.2353,
+      "step": 1992
+    },
+    {
+      "epoch": 0.17476970632846872,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.00293975404720955,
+      "loss": 1.3104,
+      "step": 1993
+    },
+    {
+      "epoch": 0.17485739810284326,
+      "grad_norm": 0.07861328125,
+      "learning_rate": 0.0029396335577739413,
+      "loss": 1.2444,
+      "step": 1994
+    },
+    {
+      "epoch": 0.17494508987721782,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.0029395129507270817,
+      "loss": 1.3507,
+      "step": 1995
+    },
+    {
+      "epoch": 0.17503278165159236,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.00293939222607997,
+      "loss": 1.3499,
+      "step": 1996
+    },
+    {
+      "epoch": 0.1751204734259669,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.002939271383843615,
+      "loss": 1.2426,
+      "step": 1997
+    },
+    {
+      "epoch": 0.17520816520034144,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.002939150424029039,
+      "loss": 1.2648,
+      "step": 1998
+    },
+    {
+      "epoch": 0.175295856974716,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.002939029346647272,
+      "loss": 1.3051,
+      "step": 1999
+    },
+    {
+      "epoch": 0.17538354874909054,
+      "grad_norm": 0.07568359375,
+      "learning_rate": 0.0029389081517093556,
+      "loss": 1.3173,
+      "step": 2000
+    },
+    {
+      "epoch": 0.17538354874909054,
+      "eval_loss": 1.2995423078536987,
+      "eval_runtime": 429.2876,
+      "eval_samples_per_second": 33.653,
+      "eval_steps_per_second": 8.414,
+      "step": 2000
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 11403,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 9.897567584256e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}