diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,16704 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 2382,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.505831176096957,
+      "learning_rate": 6.944444444444445e-07,
+      "loss": 0.3092,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.876311852539876,
+      "learning_rate": 1.388888888888889e-06,
+      "loss": 0.32,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.594724599465291,
+      "learning_rate": 2.0833333333333334e-06,
+      "loss": 0.3309,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.260379264703898,
+      "learning_rate": 2.777777777777778e-06,
+      "loss": 0.3186,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.555209220730907,
+      "learning_rate": 3.4722222222222224e-06,
+      "loss": 0.3016,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.9909267413876566,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 0.3187,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.2031315895631356,
+      "learning_rate": 4.861111111111111e-06,
+      "loss": 0.2838,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.4981226504069736,
+      "learning_rate": 5.555555555555556e-06,
+      "loss": 0.2825,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1195362477475772,
+      "learning_rate": 6.25e-06,
+      "loss": 0.2354,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.9734137498135732,
+      "learning_rate": 6.944444444444445e-06,
+      "loss": 0.2447,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2274650950192303,
+      "learning_rate": 7.63888888888889e-06,
+      "loss": 0.2253,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.2026559608951022,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.2175,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.2080276507017704,
+      "learning_rate": 9.027777777777777e-06,
+      "loss": 0.2354,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9959027109320846,
+      "learning_rate": 9.722222222222223e-06,
+      "loss": 0.2281,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8707625497310809,
+      "learning_rate": 1.0416666666666668e-05,
+      "loss": 0.2094,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8769208357919535,
+      "learning_rate": 1.1111111111111112e-05,
+      "loss": 0.2293,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0315288687490254,
+      "learning_rate": 1.1805555555555555e-05,
+      "loss": 0.2366,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0238163336729231,
+      "learning_rate": 1.25e-05,
+      "loss": 0.2165,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0473968754751415,
+      "learning_rate": 1.3194444444444446e-05,
+      "loss": 0.2125,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.8604519753746809,
+      "learning_rate": 1.388888888888889e-05,
+      "loss": 0.2145,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8014574427661126,
+      "learning_rate": 1.4583333333333335e-05,
+      "loss": 0.2114,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1491249042120122,
+      "learning_rate": 1.527777777777778e-05,
+      "loss": 0.201,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9004915098487288,
+      "learning_rate": 1.597222222222222e-05,
+      "loss": 0.2167,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7361390201368838,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.2134,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7906891893040704,
+      "learning_rate": 1.736111111111111e-05,
+      "loss": 0.2137,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8972544203573584,
+      "learning_rate": 1.8055555555555555e-05,
+      "loss": 0.2083,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5681258034688518,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 0.203,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7596246908439771,
+      "learning_rate": 1.9444444444444445e-05,
+      "loss": 0.2106,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.2328490481581964,
+      "learning_rate": 2.013888888888889e-05,
+      "loss": 0.2009,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8148921279263042,
+      "learning_rate": 2.0833333333333336e-05,
+      "loss": 0.2033,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0547394210199286,
+      "learning_rate": 2.152777777777778e-05,
+      "loss": 0.2133,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0268345879661689,
+      "learning_rate": 2.2222222222222223e-05,
+      "loss": 0.2239,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5724368050469452,
+      "learning_rate": 2.2916666666666667e-05,
+      "loss": 0.2056,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6057366058449886,
+      "learning_rate": 2.361111111111111e-05,
+      "loss": 0.2081,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0648168782742933,
+      "learning_rate": 2.4305555555555558e-05,
+      "loss": 0.2174,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6305691565849281,
+      "learning_rate": 2.5e-05,
+      "loss": 0.2087,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5838884004209798,
+      "learning_rate": 2.5694444444444445e-05,
+      "loss": 0.1875,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6442921527902131,
+      "learning_rate": 2.6388888888888892e-05,
+      "loss": 0.1974,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6049273440119587,
+      "learning_rate": 2.7083333333333332e-05,
+      "loss": 0.2155,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5924059277578564,
+      "learning_rate": 2.777777777777778e-05,
+      "loss": 0.2003,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6593474286795765,
+      "learning_rate": 2.8472222222222223e-05,
+      "loss": 0.1851,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6651479639066215,
+      "learning_rate": 2.916666666666667e-05,
+      "loss": 0.1979,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7332852815091829,
+      "learning_rate": 2.9861111111111113e-05,
+      "loss": 0.1978,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5147194573781568,
+      "learning_rate": 3.055555555555556e-05,
+      "loss": 0.1777,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.610645513479347,
+      "learning_rate": 3.125e-05,
+      "loss": 0.1929,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7116802014487239,
+      "learning_rate": 3.194444444444444e-05,
+      "loss": 0.207,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.708548853467484,
+      "learning_rate": 3.263888888888889e-05,
+      "loss": 0.1911,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6134114486928213,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.2074,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.315322339531238,
+      "learning_rate": 3.402777777777778e-05,
+      "loss": 0.1873,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.4581042037802392,
+      "learning_rate": 3.472222222222222e-05,
+      "loss": 0.2002,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6195434849882452,
+      "learning_rate": 3.541666666666667e-05,
+      "loss": 0.2004,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.4817036677991203,
+      "learning_rate": 3.611111111111111e-05,
+      "loss": 0.1862,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5478328906125305,
+      "learning_rate": 3.6805555555555556e-05,
+      "loss": 0.1901,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.4322189409404494,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.1929,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.1114890735313196,
+      "learning_rate": 3.8194444444444444e-05,
+      "loss": 0.1782,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5454600125252768,
+      "learning_rate": 3.888888888888889e-05,
+      "loss": 0.1807,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6399422452803984,
+      "learning_rate": 3.958333333333333e-05,
+      "loss": 0.1979,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5027887412129489,
+      "learning_rate": 4.027777777777778e-05,
+      "loss": 0.179,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5251316831073335,
+      "learning_rate": 4.0972222222222225e-05,
+      "loss": 0.1668,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4849312923986754,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.1711,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.47899058335856626,
+      "learning_rate": 4.236111111111111e-05,
+      "loss": 0.1852,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4414043905162519,
+      "learning_rate": 4.305555555555556e-05,
+      "loss": 0.1792,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4630173699164931,
+      "learning_rate": 4.375e-05,
+      "loss": 0.166,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4616548268686554,
+      "learning_rate": 4.4444444444444447e-05,
+      "loss": 0.1721,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6094437400832526,
+      "learning_rate": 4.5138888888888894e-05,
+      "loss": 0.191,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.49079205231875256,
+      "learning_rate": 4.5833333333333334e-05,
+      "loss": 0.187,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5581675029758306,
+      "learning_rate": 4.652777777777778e-05,
+      "loss": 0.1684,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6316072529048834,
+      "learning_rate": 4.722222222222222e-05,
+      "loss": 0.1945,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7259719108561538,
+      "learning_rate": 4.791666666666667e-05,
+      "loss": 0.187,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5200273012535201,
+      "learning_rate": 4.8611111111111115e-05,
+      "loss": 0.1927,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4533776427302155,
+      "learning_rate": 4.930555555555556e-05,
+      "loss": 0.1827,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5598220181804464,
+      "learning_rate": 5e-05,
+      "loss": 0.1996,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5129041711407194,
+      "learning_rate": 4.99999768801117e-05,
+      "loss": 0.1883,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4510537546923754,
+      "learning_rate": 4.999990752048953e-05,
+      "loss": 0.1639,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5005817541921228,
+      "learning_rate": 4.999979192126181e-05,
+      "loss": 0.1646,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.534949629640472,
+      "learning_rate": 4.9999630082642325e-05,
+      "loss": 0.1714,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.041346786916489,
+      "learning_rate": 4.999942200493043e-05,
+      "loss": 0.1656,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4874004927600869,
+      "learning_rate": 4.999916768851096e-05,
+      "loss": 0.1826,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4546406099417225,
+      "learning_rate": 4.999886713385432e-05,
+      "loss": 0.1941,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5192756878615733,
+      "learning_rate": 4.999852034151641e-05,
+      "loss": 0.1917,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6728876246629805,
+      "learning_rate": 4.999812731213864e-05,
+      "loss": 0.1802,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.42143086276844266,
+      "learning_rate": 4.999768804644796e-05,
+      "loss": 0.167,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5000324177487674,
+      "learning_rate": 4.999720254525684e-05,
+      "loss": 0.1805,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5011519511810362,
+      "learning_rate": 4.999667080946324e-05,
+      "loss": 0.1826,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.786575040394215,
+      "learning_rate": 4.999609284005068e-05,
+      "loss": 0.191,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4792285675577297,
+      "learning_rate": 4.999546863808815e-05,
+      "loss": 0.1787,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5868950496794698,
+      "learning_rate": 4.9994798204730166e-05,
+      "loss": 0.1871,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4267428848594329,
+      "learning_rate": 4.999408154121676e-05,
+      "loss": 0.1739,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5094266505072533,
+      "learning_rate": 4.999331864887347e-05,
+      "loss": 0.1934,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4086387722988919,
+      "learning_rate": 4.999250952911133e-05,
+      "loss": 0.1796,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4251236961685902,
+      "learning_rate": 4.999165418342689e-05,
+      "loss": 0.1646,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5505576181673397,
+      "learning_rate": 4.999075261340218e-05,
+      "loss": 0.1735,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6333089265924944,
+      "learning_rate": 4.9989804820704735e-05,
+      "loss": 0.1651,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.43039221519200616,
+      "learning_rate": 4.9988810807087584e-05,
+      "loss": 0.1832,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6220686862646253,
+      "learning_rate": 4.998777057438926e-05,
+      "loss": 0.1808,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6048163035204037,
+      "learning_rate": 4.998668412453374e-05,
+      "loss": 0.1744,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4619075681710722,
+      "learning_rate": 4.998555145953054e-05,
+      "loss": 0.1697,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5354444838760041,
+      "learning_rate": 4.998437258147462e-05,
+      "loss": 0.1815,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6085063403904147,
+      "learning_rate": 4.9983147492546414e-05,
+      "loss": 0.1875,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5414068673962986,
+      "learning_rate": 4.9981876195011844e-05,
+      "loss": 0.1666,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6356814576487988,
+      "learning_rate": 4.998055869122228e-05,
+      "loss": 0.1831,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7050794109252304,
+      "learning_rate": 4.997919498361457e-05,
+      "loss": 0.1734,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.53532783671682,
+      "learning_rate": 4.997778507471102e-05,
+      "loss": 0.1616,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.50110168295122,
+      "learning_rate": 4.997632896711939e-05,
+      "loss": 0.1689,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7773573908009797,
+      "learning_rate": 4.997482666353287e-05,
+      "loss": 0.1684,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4886281403919567,
+      "learning_rate": 4.9973278166730106e-05,
+      "loss": 0.1664,
+      "step": 106
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.517833215785634,
+      "learning_rate": 4.99716834795752e-05,
+      "loss": 0.1657,
+      "step": 107
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9958233837590985,
+      "learning_rate": 4.997004260501766e-05,
+      "loss": 0.1828,
+      "step": 108
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.45699676779632276,
+      "learning_rate": 4.996835554609244e-05,
+      "loss": 0.1505,
+      "step": 109
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.5265811586725444,
+      "learning_rate": 4.996662230591989e-05,
+      "loss": 0.1892,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6208722622075454,
+      "learning_rate": 4.996484288770582e-05,
+      "loss": 0.1644,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4362995935878586,
+      "learning_rate": 4.9963017294741407e-05,
+      "loss": 0.1754,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3994570642478068,
+      "learning_rate": 4.996114553040328e-05,
+      "loss": 0.1744,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5326062992220989,
+      "learning_rate": 4.995922759815339e-05,
+      "loss": 0.1663,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.549765393641323,
+      "learning_rate": 4.9957263501539174e-05,
+      "loss": 0.1692,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.48343215971188047,
+      "learning_rate": 4.9955253244193375e-05,
+      "loss": 0.1769,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6499922648276882,
+      "learning_rate": 4.995319682983418e-05,
+      "loss": 0.1906,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.44799251880779417,
+      "learning_rate": 4.995109426226508e-05,
+      "loss": 0.184,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4800215350260659,
+      "learning_rate": 4.994894554537498e-05,
+      "loss": 0.1749,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.44423409429150207,
+      "learning_rate": 4.9946750683138134e-05,
+      "loss": 0.1765,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.45702536866059634,
+      "learning_rate": 4.994450967961413e-05,
+      "loss": 0.1634,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6096461673682956,
+      "learning_rate": 4.994222253894791e-05,
+      "loss": 0.1686,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.42787153856926974,
+      "learning_rate": 4.993988926536975e-05,
+      "loss": 0.1479,
+      "step": 123
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5108659341644166,
+      "learning_rate": 4.9937509863195256e-05,
+      "loss": 0.1672,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.41972712194888373,
+      "learning_rate": 4.993508433682535e-05,
+      "loss": 0.1831,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3631124358611529,
+      "learning_rate": 4.993261269074625e-05,
+      "loss": 0.1642,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.40059516042409765,
+      "learning_rate": 4.9930094929529506e-05,
+      "loss": 0.1659,
+      "step": 127
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6148339993605405,
+      "learning_rate": 4.992753105783194e-05,
+      "loss": 0.1757,
+      "step": 128
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.39363762290385285,
+      "learning_rate": 4.992492108039566e-05,
+      "loss": 0.1585,
+      "step": 129
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.41435191858889847,
+      "learning_rate": 4.992226500204808e-05,
+      "loss": 0.1775,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3655543350875483,
+      "learning_rate": 4.9919562827701824e-05,
+      "loss": 0.144,
+      "step": 131
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.7210663686207939,
+      "learning_rate": 4.991681456235483e-05,
+      "loss": 0.1706,
+      "step": 132
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9054446181809322,
+      "learning_rate": 4.991402021109027e-05,
+      "loss": 0.1941,
+      "step": 133
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4833926613100019,
+      "learning_rate": 4.9911179779076544e-05,
+      "loss": 0.1743,
+      "step": 134
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.45552665316947527,
+      "learning_rate": 4.9908293271567286e-05,
+      "loss": 0.1799,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3605251483292452,
+      "learning_rate": 4.990536069390136e-05,
+      "loss": 0.1661,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3256527028656672,
+      "learning_rate": 4.990238205150284e-05,
+      "loss": 0.1515,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5494000429704912,
+      "learning_rate": 4.989935734988098e-05,
+      "loss": 0.1702,
+      "step": 138
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4088134234389422,
+      "learning_rate": 4.9896286594630255e-05,
+      "loss": 0.1549,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.7956246893190204,
+      "learning_rate": 4.98931697914303e-05,
+      "loss": 0.1597,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4286995729163948,
+      "learning_rate": 4.989000694604593e-05,
+      "loss": 0.1696,
+      "step": 141
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4363227798234796,
+      "learning_rate": 4.988679806432712e-05,
+      "loss": 0.1585,
+      "step": 142
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4464818665918074,
+      "learning_rate": 4.988354315220898e-05,
+      "loss": 0.1756,
+      "step": 143
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9131542256695875,
+      "learning_rate": 4.988024221571177e-05,
+      "loss": 0.161,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.394256044720401,
+      "learning_rate": 4.987689526094087e-05,
+      "loss": 0.185,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.510140577091025,
+      "learning_rate": 4.9873502294086785e-05,
+      "loss": 0.1617,
+      "step": 146
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.36616005306280286,
+      "learning_rate": 4.9870063321425105e-05,
+      "loss": 0.156,
+      "step": 147
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5757419758833113,
+      "learning_rate": 4.986657834931653e-05,
+      "loss": 0.167,
+      "step": 148
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.34139100244414955,
+      "learning_rate": 4.9863047384206835e-05,
+      "loss": 0.1934,
+      "step": 149
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3146406561419288,
+      "learning_rate": 4.985947043262686e-05,
+      "loss": 0.1583,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5267410347373885,
+      "learning_rate": 4.98558475011925e-05,
+      "loss": 0.1542,
+      "step": 151
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3491246210593201,
+      "learning_rate": 4.9852178596604705e-05,
+      "loss": 0.1695,
+      "step": 152
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.44159408096430236,
+      "learning_rate": 4.984846372564943e-05,
+      "loss": 0.1545,
+      "step": 153
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3580346062296655,
+      "learning_rate": 4.984470289519769e-05,
+      "loss": 0.1802,
+      "step": 154
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4015629416496644,
+      "learning_rate": 4.984089611220547e-05,
+      "loss": 0.1647,
+      "step": 155
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.37199258133600427,
+      "learning_rate": 4.9837043383713753e-05,
+      "loss": 0.1871,
+      "step": 156
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.28866757843796853,
+      "learning_rate": 4.983314471684853e-05,
+      "loss": 0.1512,
+      "step": 157
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.402880251825588,
+      "learning_rate": 4.982920011882074e-05,
+      "loss": 0.1562,
+      "step": 158
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.348275716049736,
+      "learning_rate": 4.982520959692626e-05,
+      "loss": 0.1667,
+      "step": 159
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.3347251213963409,
+      "learning_rate": 4.9821173158545936e-05,
+      "loss": 0.1582,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.3211879864412581,
+      "learning_rate": 4.9817090811145524e-05,
+      "loss": 0.1748,
+      "step": 161
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.257030108844957,
+      "learning_rate": 4.981296256227569e-05,
+      "loss": 0.1571,
+      "step": 162
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4087362812728143,
+      "learning_rate": 4.980878841957203e-05,
+      "loss": 0.1651,
+      "step": 163
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.517801391725505,
+      "learning_rate": 4.9804568390754974e-05,
+      "loss": 0.1709,
+      "step": 164
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4557759164490978,
+      "learning_rate": 4.980030248362987e-05,
+      "loss": 0.1623,
+      "step": 165
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.38298834176837165,
+      "learning_rate": 4.979599070608688e-05,
+      "loss": 0.15,
+      "step": 166
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9088564940000337,
+      "learning_rate": 4.979163306610105e-05,
+      "loss": 0.1667,
+      "step": 167
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.46548267734203874,
+      "learning_rate": 4.978722957173222e-05,
+      "loss": 0.1879,
+      "step": 168
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4141973126487384,
+      "learning_rate": 4.978278023112506e-05,
+      "loss": 0.1513,
+      "step": 169
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.3665636944160238,
+      "learning_rate": 4.977828505250903e-05,
+      "loss": 0.1754,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.49199965431061116,
+      "learning_rate": 4.977374404419837e-05,
+      "loss": 0.1685,
+      "step": 171
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.33039724817541666,
+      "learning_rate": 4.976915721459209e-05,
+      "loss": 0.1494,
+      "step": 172
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.39330650824967517,
+      "learning_rate": 4.976452457217394e-05,
+      "loss": 0.1682,
+      "step": 173
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.44071732634537353,
+      "learning_rate": 4.975984612551243e-05,
+      "loss": 0.1563,
+      "step": 174
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.7004646928317338,
+      "learning_rate": 4.975512188326077e-05,
+      "loss": 0.157,
+      "step": 175
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5801430139252254,
+      "learning_rate": 4.9750351854156864e-05,
+      "loss": 0.1626,
+      "step": 176
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.39720712164612026,
+      "learning_rate": 4.9745536047023324e-05,
+      "loss": 0.1546,
+      "step": 177
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.34700262445120433,
+      "learning_rate": 4.974067447076742e-05,
+      "loss": 0.1532,
+      "step": 178
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5411384486498156,
+      "learning_rate": 4.973576713438108e-05,
+      "loss": 0.1569,
+      "step": 179
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.28168394505914957,
+      "learning_rate": 4.973081404694088e-05,
+      "loss": 0.148,
+      "step": 180
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.44989435161975977,
+      "learning_rate": 4.9725815217607994e-05,
+      "loss": 0.1697,
+      "step": 181
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.31046541516968484,
+      "learning_rate": 4.972077065562821e-05,
+      "loss": 0.1577,
+      "step": 182
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.377734977307708,
+      "learning_rate": 4.9715680370331926e-05,
+      "loss": 0.1641,
+      "step": 183
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.260214762898074,
+      "learning_rate": 4.971054437113406e-05,
+      "loss": 0.1558,
+      "step": 184
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.31330789057624936,
+      "learning_rate": 4.9705362667534126e-05,
+      "loss": 0.1653,
+      "step": 185
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8510823967979427,
+      "learning_rate": 4.970013526911617e-05,
+      "loss": 0.1531,
+      "step": 186
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.32639160357979424,
+      "learning_rate": 4.969486218554871e-05,
+      "loss": 0.1509,
+      "step": 187
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.36144717651260766,
+      "learning_rate": 4.96895434265848e-05,
+      "loss": 0.175,
+      "step": 188
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.3673504376687341,
+      "learning_rate": 4.9684179002062e-05,
+      "loss": 0.1621,
+      "step": 189
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.31259050564224894,
+      "learning_rate": 4.967876892190227e-05,
+      "loss": 0.1621,
+      "step": 190
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.2939147104378102,
+      "learning_rate": 4.967331319611206e-05,
+      "loss": 0.1538,
+      "step": 191
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5280948073227805,
+      "learning_rate": 4.9667811834782224e-05,
+      "loss": 0.1698,
+      "step": 192
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.2834579432262214,
+      "learning_rate": 4.9662264848088034e-05,
+      "loss": 0.1549,
+      "step": 193
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.36801428775906864,
+      "learning_rate": 4.965667224628916e-05,
+      "loss": 0.1684,
+      "step": 194
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.42881667088774117,
+      "learning_rate": 4.96510340397296e-05,
+      "loss": 0.1675,
+      "step": 195
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.3881201042162419,
+      "learning_rate": 4.964535023883776e-05,
+      "loss": 0.1616,
+      "step": 196
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5271995829751434,
+      "learning_rate": 4.9639620854126326e-05,
+      "loss": 0.1531,
+      "step": 197
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.35893027926425874,
+      "learning_rate": 4.963384589619233e-05,
+      "loss": 0.1758,
+      "step": 198
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.3698122170987755,
+      "learning_rate": 4.962802537571707e-05,
+      "loss": 0.145,
+      "step": 199
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.30028023595625514,
+      "learning_rate": 4.9622159303466144e-05,
+      "loss": 0.1559,
+      "step": 200
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4112285471509381,
+      "learning_rate": 4.9616247690289375e-05,
+      "loss": 0.1464,
+      "step": 201
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.3546419745199511,
+      "learning_rate": 4.9610290547120835e-05,
+      "loss": 0.1661,
+      "step": 202
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3276429318206949,
+      "learning_rate": 4.9604287884978803e-05,
+      "loss": 0.1824,
+      "step": 203
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.855002827145104,
+      "learning_rate": 4.959823971496574e-05,
+      "loss": 0.163,
+      "step": 204
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3250528459427936,
+      "learning_rate": 4.959214604826831e-05,
+      "loss": 0.1613,
+      "step": 205
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3335693378150676,
+      "learning_rate": 4.958600689615728e-05,
+      "loss": 0.1482,
+      "step": 206
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.29953799370110085,
+      "learning_rate": 4.9579822269987574e-05,
+      "loss": 0.1589,
+      "step": 207
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3214603607385732,
+      "learning_rate": 4.957359218119824e-05,
+      "loss": 0.1556,
+      "step": 208
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.33802502005879204,
+      "learning_rate": 4.956731664131238e-05,
+      "loss": 0.1597,
+      "step": 209
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1924981326646698,
+      "learning_rate": 4.956099566193717e-05,
+      "loss": 0.1576,
+      "step": 210
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3293968536457082,
+      "learning_rate": 4.955462925476385e-05,
+      "loss": 0.18,
+      "step": 211
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3669510748066192,
+      "learning_rate": 4.9548217431567665e-05,
+      "loss": 0.1842,
+      "step": 212
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.37429970351430936,
+      "learning_rate": 4.954176020420788e-05,
+      "loss": 0.1521,
+      "step": 213
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3823744259454877,
+      "learning_rate": 4.953525758462769e-05,
+      "loss": 0.1547,
+      "step": 214
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3068887688727271,
+      "learning_rate": 4.952870958485432e-05,
+      "loss": 0.1539,
+      "step": 215
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.31978193044773395,
+      "learning_rate": 4.952211621699887e-05,
+      "loss": 0.1533,
+      "step": 216
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.38336966052905397,
+      "learning_rate": 4.951547749325638e-05,
+      "loss": 0.1568,
+      "step": 217
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.2930800740098956,
+      "learning_rate": 4.950879342590577e-05,
+      "loss": 0.1418,
+      "step": 218
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.2923913376073505,
+      "learning_rate": 4.9502064027309836e-05,
+      "loss": 0.1552,
+      "step": 219
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.2885974364956092,
+      "learning_rate": 4.949528930991522e-05,
+      "loss": 0.1618,
+      "step": 220
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3513504156126098,
+      "learning_rate": 4.948846928625236e-05,
+      "loss": 0.1559,
+      "step": 221
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3559880052894558,
+      "learning_rate": 4.948160396893553e-05,
+      "loss": 0.1381,
+      "step": 222
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6759249982115987,
+      "learning_rate": 4.947469337066275e-05,
+      "loss": 0.1501,
+      "step": 223
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3549169072992492,
+      "learning_rate": 4.9467737504215805e-05,
+      "loss": 0.1435,
+      "step": 224
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.30837832445675933,
+      "learning_rate": 4.9460736382460195e-05,
+      "loss": 0.1553,
+      "step": 225
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.426213693420787,
+      "learning_rate": 4.9453690018345144e-05,
+      "loss": 0.1652,
+      "step": 226
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.32266439898843996,
+      "learning_rate": 4.944659842490354e-05,
+      "loss": 0.1476,
+      "step": 227
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.2671355709680038,
+      "learning_rate": 4.943946161525192e-05,
+      "loss": 0.1541,
+      "step": 228
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.33129259550368234,
+      "learning_rate": 4.943227960259048e-05,
+      "loss": 0.1451,
+      "step": 229
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3620314297617122,
+      "learning_rate": 4.9425052400203e-05,
+      "loss": 0.1601,
+      "step": 230
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.30189056621698923,
+      "learning_rate": 4.941778002145684e-05,
+      "loss": 0.1459,
+      "step": 231
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.28246518039591356,
+      "learning_rate": 4.9410462479802945e-05,
+      "loss": 0.1359,
+      "step": 232
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3250606945184856,
+      "learning_rate": 4.9403099788775754e-05,
+      "loss": 0.159,
+      "step": 233
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3591176139641523,
+      "learning_rate": 4.939569196199325e-05,
+      "loss": 0.1522,
+      "step": 234
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.34160167133362246,
+      "learning_rate": 4.938823901315688e-05,
+      "loss": 0.1633,
+      "step": 235
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.2999074705218003,
+      "learning_rate": 4.9380740956051545e-05,
+      "loss": 0.1609,
+      "step": 236
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3041485631683432,
+      "learning_rate": 4.937319780454559e-05,
+      "loss": 0.1626,
+      "step": 237
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3384170081144906,
+      "learning_rate": 4.936560957259077e-05,
+      "loss": 0.1549,
+      "step": 238
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.36744662628512786,
+      "learning_rate": 4.9357976274222185e-05,
+      "loss": 0.1579,
+      "step": 239
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6654291540447684,
+      "learning_rate": 4.935029792355834e-05,
+      "loss": 0.1395,
+      "step": 240
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.28766735703449686,
+      "learning_rate": 4.934257453480103e-05,
+      "loss": 0.1577,
+      "step": 241
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.2507062409762846,
+      "learning_rate": 4.9334806122235376e-05,
+      "loss": 0.1534,
+      "step": 242
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.230297735660614,
+      "learning_rate": 4.932699270022976e-05,
+      "loss": 0.1356,
+      "step": 243
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.2763377750604221,
+      "learning_rate": 4.931913428323581e-05,
+      "loss": 0.1583,
+      "step": 244
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3068450771234095,
+      "learning_rate": 4.93112308857884e-05,
+      "loss": 0.1615,
+      "step": 245
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.28882177873166925,
+      "learning_rate": 4.9303282522505565e-05,
+      "loss": 0.1694,
+      "step": 246
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.29110546876671506,
+      "learning_rate": 4.929528920808854e-05,
+      "loss": 0.1505,
+      "step": 247
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.2922527507810248,
+      "learning_rate": 4.928725095732169e-05,
+      "loss": 0.1425,
+      "step": 248
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.703961755087846,
+      "learning_rate": 4.927916778507248e-05,
+      "loss": 0.1661,
+      "step": 249
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.2151818201654711,
+      "learning_rate": 4.927103970629148e-05,
+      "loss": 0.1339,
+      "step": 250
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2267034883239192,
+      "learning_rate": 4.9262866736012304e-05,
+      "loss": 0.1421,
+      "step": 251
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.27668578228999935,
+      "learning_rate": 4.925464888935162e-05,
+      "loss": 0.1502,
+      "step": 252
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.3144809264536009,
+      "learning_rate": 4.924638618150906e-05,
+      "loss": 0.1483,
+      "step": 253
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.249298059474405,
+      "learning_rate": 4.923807862776728e-05,
+      "loss": 0.1677,
+      "step": 254
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2891492701977114,
+      "learning_rate": 4.922972624349185e-05,
+      "loss": 0.1454,
+      "step": 255
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2833220488741825,
+      "learning_rate": 4.922132904413126e-05,
+      "loss": 0.1403,
+      "step": 256
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2601797899677536,
+      "learning_rate": 4.921288704521689e-05,
+      "loss": 0.1644,
+      "step": 257
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.3360778920835631,
+      "learning_rate": 4.920440026236301e-05,
+      "loss": 0.1501,
+      "step": 258
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2241749585352573,
+      "learning_rate": 4.919586871126667e-05,
+      "loss": 0.1307,
+      "step": 259
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.23803377008727716,
+      "learning_rate": 4.918729240770775e-05,
+      "loss": 0.1493,
+      "step": 260
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2541720336695473,
+      "learning_rate": 4.917867136754893e-05,
+      "loss": 0.1535,
+      "step": 261
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.251713953983326,
+      "learning_rate": 4.9170005606735594e-05,
+      "loss": 0.1506,
+      "step": 262
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.38103715090555207,
+      "learning_rate": 4.916129514129585e-05,
+      "loss": 0.1292,
+      "step": 263
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2512837972331519,
+      "learning_rate": 4.915253998734051e-05,
+      "loss": 0.135,
+      "step": 264
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2645672154213092,
+      "learning_rate": 4.9143740161063015e-05,
+      "loss": 0.1588,
+      "step": 265
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.3273141568932692,
+      "learning_rate": 4.9134895678739456e-05,
+      "loss": 0.1556,
+      "step": 266
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.24812276003743336,
+      "learning_rate": 4.91260065567285e-05,
+      "loss": 0.1572,
+      "step": 267
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.26170827037829125,
+      "learning_rate": 4.91170728114714e-05,
+      "loss": 0.1456,
+      "step": 268
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2782551653006925,
+      "learning_rate": 4.9108094459491916e-05,
+      "loss": 0.1447,
+      "step": 269
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.26501062892559696,
+      "learning_rate": 4.909907151739633e-05,
+      "loss": 0.1607,
+      "step": 270
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.26075538076191235,
+      "learning_rate": 4.909000400187341e-05,
+      "loss": 0.173,
+      "step": 271
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.22671357600241931,
+      "learning_rate": 4.908089192969434e-05,
+      "loss": 0.1489,
+      "step": 272
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2730878478928314,
+      "learning_rate": 4.907173531771273e-05,
+      "loss": 0.1544,
+      "step": 273
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.25590787581998936,
+      "learning_rate": 4.906253418286456e-05,
+      "loss": 0.1624,
+      "step": 274
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.37426492024617797,
+      "learning_rate": 4.9053288542168185e-05,
+      "loss": 0.1528,
+      "step": 275
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.26863800323800735,
+      "learning_rate": 4.904399841272423e-05,
+      "loss": 0.1479,
+      "step": 276
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.2518106015212866,
+      "learning_rate": 4.903466381171568e-05,
+      "loss": 0.1622,
+      "step": 277
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.22856182106611597,
+      "learning_rate": 4.902528475640768e-05,
+      "loss": 0.149,
+      "step": 278
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.31163203719068544,
+      "learning_rate": 4.901586126414768e-05,
+      "loss": 0.153,
+      "step": 279
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.2779280772691567,
+      "learning_rate": 4.900639335236527e-05,
+      "loss": 0.1605,
+      "step": 280
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.3656078440554968,
+      "learning_rate": 4.899688103857223e-05,
+      "loss": 0.1526,
+      "step": 281
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.2345471048092292,
+      "learning_rate": 4.898732434036244e-05,
+      "loss": 0.1524,
+      "step": 282
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.2710002227765519,
+      "learning_rate": 4.897772327541188e-05,
+      "loss": 0.1445,
+      "step": 283
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.25944352849666,
+      "learning_rate": 4.8968077861478606e-05,
+      "loss": 0.1522,
+      "step": 284
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.25035916985294826,
+      "learning_rate": 4.8958388116402685e-05,
+      "loss": 0.1503,
+      "step": 285
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.26387823237482777,
+      "learning_rate": 4.894865405810618e-05,
+      "loss": 0.1634,
+      "step": 286
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.23781727639760813,
+      "learning_rate": 4.893887570459312e-05,
+      "loss": 0.1325,
+      "step": 287
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.26473375012831907,
+      "learning_rate": 4.8929053073949456e-05,
+      "loss": 0.1601,
+      "step": 288
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.31717652489901665,
+      "learning_rate": 4.8919186184343046e-05,
+      "loss": 0.1605,
+      "step": 289
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.25561132780516277,
+      "learning_rate": 4.890927505402359e-05,
+      "loss": 0.1457,
+      "step": 290
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.2350806643291294,
+      "learning_rate": 4.8899319701322646e-05,
+      "loss": 0.1509,
+      "step": 291
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.25977500239193485,
+      "learning_rate": 4.888932014465352e-05,
+      "loss": 0.1491,
+      "step": 292
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.24330605229250188,
+      "learning_rate": 4.887927640251132e-05,
+      "loss": 0.1436,
+      "step": 293
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.21362267266988572,
+      "learning_rate": 4.8869188493472854e-05,
+      "loss": 0.1446,
+      "step": 294
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.2780394389207256,
+      "learning_rate": 4.885905643619664e-05,
+      "loss": 0.1568,
+      "step": 295
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.3282121881568022,
+      "learning_rate": 4.8848880249422815e-05,
+      "loss": 0.1595,
+      "step": 296
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4059743716125514,
+      "learning_rate": 4.883865995197319e-05,
+      "loss": 0.1535,
+      "step": 297
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.2534665902460271,
+      "learning_rate": 4.882839556275113e-05,
+      "loss": 0.152,
+      "step": 298
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.25147853244616303,
+      "learning_rate": 4.881808710074155e-05,
+      "loss": 0.1404,
+      "step": 299
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.2114740675937596,
+      "learning_rate": 4.880773458501089e-05,
+      "loss": 0.1384,
+      "step": 300
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.3978554640648006,
+      "learning_rate": 4.879733803470707e-05,
+      "loss": 0.1442,
+      "step": 301
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.3395078593094899,
+      "learning_rate": 4.878689746905946e-05,
+      "loss": 0.1475,
+      "step": 302
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.20995376558140022,
+      "learning_rate": 4.877641290737884e-05,
+      "loss": 0.1371,
+      "step": 303
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.25405144516236805,
+      "learning_rate": 4.876588436905736e-05,
+      "loss": 0.1558,
+      "step": 304
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.24345867938387258,
+      "learning_rate": 4.8755311873568505e-05,
+      "loss": 0.1566,
+      "step": 305
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.21854645826343833,
+      "learning_rate": 4.874469544046707e-05,
+      "loss": 0.1513,
+      "step": 306
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.2740976626547021,
+      "learning_rate": 4.8734035089389115e-05,
+      "loss": 0.1682,
+      "step": 307
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.23242987746059862,
+      "learning_rate": 4.872333084005194e-05,
+      "loss": 0.1475,
+      "step": 308
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.21405907503612592,
+      "learning_rate": 4.8712582712254016e-05,
+      "loss": 0.1554,
+      "step": 309
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.388095456055863,
+      "learning_rate": 4.870179072587499e-05,
+      "loss": 0.1376,
+      "step": 310
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.2247619935700204,
+      "learning_rate": 4.869095490087562e-05,
+      "loss": 0.1475,
+      "step": 311
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.21785148173278546,
+      "learning_rate": 4.868007525729775e-05,
+      "loss": 0.1421,
+      "step": 312
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.24396929422609678,
+      "learning_rate": 4.866915181526428e-05,
+      "loss": 0.1409,
+      "step": 313
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.20724863078048605,
+      "learning_rate": 4.865818459497911e-05,
+      "loss": 0.1277,
+      "step": 314
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.20793483040283653,
+      "learning_rate": 4.86471736167271e-05,
+      "loss": 0.1371,
+      "step": 315
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.2925292398325892,
+      "learning_rate": 4.8636118900874064e-05,
+      "loss": 0.1541,
+      "step": 316
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.2489906584279097,
+      "learning_rate": 4.862502046786671e-05,
+      "loss": 0.1527,
+      "step": 317
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.29706918596109505,
+      "learning_rate": 4.86138783382326e-05,
+      "loss": 0.1424,
+      "step": 318
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.23710111099711076,
+      "learning_rate": 4.860269253258012e-05,
+      "loss": 0.1464,
+      "step": 319
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.3225357102725619,
+      "learning_rate": 4.859146307159842e-05,
+      "loss": 0.1664,
+      "step": 320
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4477584755167756,
+      "learning_rate": 4.858018997605742e-05,
+      "loss": 0.1441,
+      "step": 321
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.2655061297930301,
+      "learning_rate": 4.856887326680774e-05,
+      "loss": 0.151,
+      "step": 322
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.2510818240874215,
+      "learning_rate": 4.8557512964780674e-05,
+      "loss": 0.1513,
+      "step": 323
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.24649702119758882,
+      "learning_rate": 4.854610909098812e-05,
+      "loss": 0.1508,
+      "step": 324
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.21082825998560395,
+      "learning_rate": 4.8534661666522584e-05,
+      "loss": 0.1427,
+      "step": 325
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.2337443237371087,
+      "learning_rate": 4.852317071255712e-05,
+      "loss": 0.1372,
+      "step": 326
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.20623074261571012,
+      "learning_rate": 4.8511636250345294e-05,
+      "loss": 0.1384,
+      "step": 327
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.2574175176449261,
+      "learning_rate": 4.8500058301221144e-05,
+      "loss": 0.1359,
+      "step": 328
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.23733340731372732,
+      "learning_rate": 4.8488436886599144e-05,
+      "loss": 0.154,
+      "step": 329
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.2981594319721134,
+      "learning_rate": 4.847677202797415e-05,
+      "loss": 0.161,
+      "step": 330
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.24770709094196364,
+      "learning_rate": 4.8465063746921395e-05,
+      "loss": 0.139,
+      "step": 331
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.23477529083064017,
+      "learning_rate": 4.84533120650964e-05,
+      "loss": 0.147,
+      "step": 332
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.23724370082749097,
+      "learning_rate": 4.8441517004234975e-05,
+      "loss": 0.1616,
+      "step": 333
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.347769786562827,
+      "learning_rate": 4.842967858615316e-05,
+      "loss": 0.1639,
+      "step": 334
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.2593169609629534,
+      "learning_rate": 4.8417796832747186e-05,
+      "loss": 0.1471,
+      "step": 335
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.2401410194537142,
+      "learning_rate": 4.8405871765993433e-05,
+      "loss": 0.1543,
+      "step": 336
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.2649009523017183,
+      "learning_rate": 4.839390340794841e-05,
+      "loss": 0.1359,
+      "step": 337
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.2611331269400147,
+      "learning_rate": 4.838189178074867e-05,
+      "loss": 0.1435,
+      "step": 338
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.27081120625318617,
+      "learning_rate": 4.8369836906610816e-05,
+      "loss": 0.134,
+      "step": 339
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.2603011828452243,
+      "learning_rate": 4.8357738807831446e-05,
+      "loss": 0.1384,
+      "step": 340
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.2168400411702866,
+      "learning_rate": 4.8345597506787075e-05,
+      "loss": 0.1362,
+      "step": 341
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.26644881864133724,
+      "learning_rate": 4.833341302593417e-05,
+      "loss": 0.1351,
+      "step": 342
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.3751116094929113,
+      "learning_rate": 4.832118538780902e-05,
+      "loss": 0.1463,
+      "step": 343
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.038676452202723,
+      "learning_rate": 4.830891461502777e-05,
+      "loss": 0.1487,
+      "step": 344
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.23332921282055438,
+      "learning_rate": 4.829660073028631e-05,
+      "loss": 0.1433,
+      "step": 345
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.2314300357528207,
+      "learning_rate": 4.828424375636031e-05,
+      "loss": 0.136,
+      "step": 346
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.26876042677423917,
+      "learning_rate": 4.827184371610511e-05,
+      "loss": 0.1561,
+      "step": 347
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 7.331952532623471,
+      "learning_rate": 4.825940063245572e-05,
+      "loss": 0.1434,
+      "step": 348
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.2816094073108374,
+      "learning_rate": 4.824691452842675e-05,
+      "loss": 0.1379,
+      "step": 349
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.34816782132376123,
+      "learning_rate": 4.8234385427112385e-05,
+      "loss": 0.1485,
+      "step": 350
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.083450439663988,
+      "learning_rate": 4.822181335168634e-05,
+      "loss": 0.1518,
+      "step": 351
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.22753146734423657,
+      "learning_rate": 4.8209198325401815e-05,
+      "loss": 0.1283,
+      "step": 352
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.2629933336594103,
+      "learning_rate": 4.819654037159146e-05,
+      "loss": 0.1405,
+      "step": 353
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.2395757131014654,
+      "learning_rate": 4.818383951366729e-05,
+      "loss": 0.1238,
+      "step": 354
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.24531744825913318,
+      "learning_rate": 4.817109577512073e-05,
+      "loss": 0.1237,
+      "step": 355
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.543661482144856,
+      "learning_rate": 4.8158309179522454e-05,
+      "loss": 0.1393,
+      "step": 356
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.23701523584000475,
+      "learning_rate": 4.814547975052245e-05,
+      "loss": 0.1337,
+      "step": 357
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.23092241647129064,
+      "learning_rate": 4.813260751184992e-05,
+      "loss": 0.1433,
+      "step": 358
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.2878956119131095,
+      "learning_rate": 4.811969248731323e-05,
+      "loss": 0.1606,
+      "step": 359
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.39572867170223863,
+      "learning_rate": 4.81067347007999e-05,
+      "loss": 0.1548,
+      "step": 360
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.21749344070372317,
+      "learning_rate": 4.809373417627654e-05,
+      "loss": 0.1234,
+      "step": 361
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.27110294111597044,
+      "learning_rate": 4.808069093778879e-05,
+      "loss": 0.1358,
+      "step": 362
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.22173401897951495,
+      "learning_rate": 4.806760500946132e-05,
+      "loss": 0.1312,
+      "step": 363
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.2302818117359629,
+      "learning_rate": 4.805447641549774e-05,
+      "loss": 0.1313,
+      "step": 364
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.22735221534621344,
+      "learning_rate": 4.804130518018058e-05,
+      "loss": 0.1414,
+      "step": 365
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.2644514969734394,
+      "learning_rate": 4.802809132787125e-05,
+      "loss": 0.1475,
+      "step": 366
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.23638149918130982,
+      "learning_rate": 4.8014834883009966e-05,
+      "loss": 0.144,
+      "step": 367
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.40575519721298065,
+      "learning_rate": 4.800153587011573e-05,
+      "loss": 0.1619,
+      "step": 368
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.23253784888948237,
+      "learning_rate": 4.7988194313786275e-05,
+      "loss": 0.1345,
+      "step": 369
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.24898081754516282,
+      "learning_rate": 4.797481023869801e-05,
+      "loss": 0.1507,
+      "step": 370
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.25065502472084994,
+      "learning_rate": 4.796138366960603e-05,
+      "loss": 0.1405,
+      "step": 371
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.2653938855897259,
+      "learning_rate": 4.794791463134399e-05,
+      "loss": 0.1497,
+      "step": 372
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.2789552521528357,
+      "learning_rate": 4.793440314882408e-05,
+      "loss": 0.1252,
+      "step": 373
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.2693364206065251,
+      "learning_rate": 4.792084924703705e-05,
+      "loss": 0.1368,
+      "step": 374
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.25379040758571,
+      "learning_rate": 4.790725295105205e-05,
+      "loss": 0.1243,
+      "step": 375
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.26288177853190164,
+      "learning_rate": 4.7893614286016684e-05,
+      "loss": 0.1445,
+      "step": 376
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.26175623995134556,
+      "learning_rate": 4.7879933277156884e-05,
+      "loss": 0.1543,
+      "step": 377
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3107821982509933,
+      "learning_rate": 4.786620994977695e-05,
+      "loss": 0.1429,
+      "step": 378
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.30435490634185375,
+      "learning_rate": 4.78524443292594e-05,
+      "loss": 0.1562,
+      "step": 379
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.22517347828086134,
+      "learning_rate": 4.783863644106502e-05,
+      "loss": 0.1368,
+      "step": 380
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.217160212271337,
+      "learning_rate": 4.7824786310732754e-05,
+      "loss": 0.1405,
+      "step": 381
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.29874309034964575,
+      "learning_rate": 4.781089396387968e-05,
+      "loss": 0.1373,
+      "step": 382
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.21219537405667846,
+      "learning_rate": 4.779695942620094e-05,
+      "loss": 0.1387,
+      "step": 383
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.2383645403466058,
+      "learning_rate": 4.778298272346976e-05,
+      "loss": 0.1476,
+      "step": 384
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.23533917627567913,
+      "learning_rate": 4.77689638815373e-05,
+      "loss": 0.1357,
+      "step": 385
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3395749479278276,
+      "learning_rate": 4.775490292633269e-05,
+      "loss": 0.1544,
+      "step": 386
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.2969286740294667,
+      "learning_rate": 4.774079988386296e-05,
+      "loss": 0.1391,
+      "step": 387
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.23050102045884482,
+      "learning_rate": 4.772665478021296e-05,
+      "loss": 0.1373,
+      "step": 388
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3563315000157305,
+      "learning_rate": 4.7712467641545354e-05,
+      "loss": 0.1382,
+      "step": 389
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3414120993364458,
+      "learning_rate": 4.7698238494100536e-05,
+      "loss": 0.1522,
+      "step": 390
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.2817589114491083,
+      "learning_rate": 4.7683967364196624e-05,
+      "loss": 0.131,
+      "step": 391
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.29958936647532786,
+      "learning_rate": 4.766965427822936e-05,
+      "loss": 0.1463,
+      "step": 392
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.25158507145521003,
+      "learning_rate": 4.765529926267211e-05,
+      "loss": 0.1418,
+      "step": 393
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.32372763693024487,
+      "learning_rate": 4.764090234407577e-05,
+      "loss": 0.1443,
+      "step": 394
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.2825070089459905,
+      "learning_rate": 4.762646354906877e-05,
+      "loss": 0.1374,
+      "step": 395
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.24051120792520939,
+      "learning_rate": 4.7611982904356954e-05,
+      "loss": 0.1432,
+      "step": 396
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.3331042630247473,
+      "learning_rate": 4.759746043672362e-05,
+      "loss": 0.1348,
+      "step": 397
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.0417308778575092,
+      "learning_rate": 4.758289617302937e-05,
+      "loss": 0.1491,
+      "step": 398
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.2961273785726392,
+      "learning_rate": 4.7568290140212145e-05,
+      "loss": 0.1501,
+      "step": 399
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.38178640252357193,
+      "learning_rate": 4.755364236528713e-05,
+      "loss": 0.1425,
+      "step": 400
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.2526851103538564,
+      "learning_rate": 4.753895287534673e-05,
+      "loss": 0.1549,
+      "step": 401
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.3198087405488174,
+      "learning_rate": 4.752422169756048e-05,
+      "loss": 0.1495,
+      "step": 402
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.31017966818124154,
+      "learning_rate": 4.7509448859175043e-05,
+      "loss": 0.1374,
+      "step": 403
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.34372281413361067,
+      "learning_rate": 4.749463438751413e-05,
+      "loss": 0.136,
+      "step": 404
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.2568807743814327,
+      "learning_rate": 4.747977830997845e-05,
+      "loss": 0.1425,
+      "step": 405
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.27930164506931104,
+      "learning_rate": 4.746488065404567e-05,
+      "loss": 0.1464,
+      "step": 406
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.29934248086715903,
+      "learning_rate": 4.744994144727036e-05,
+      "loss": 0.1464,
+      "step": 407
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.2463620467282752,
+      "learning_rate": 4.743496071728396e-05,
+      "loss": 0.1477,
+      "step": 408
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.2718936275260185,
+      "learning_rate": 4.741993849179468e-05,
+      "loss": 0.134,
+      "step": 409
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.26623651947895743,
+      "learning_rate": 4.7404874798587494e-05,
+      "loss": 0.1534,
+      "step": 410
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.3168466927073915,
+      "learning_rate": 4.738976966552407e-05,
+      "loss": 0.138,
+      "step": 411
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.3075811664613153,
+      "learning_rate": 4.737462312054275e-05,
+      "loss": 0.1616,
+      "step": 412
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.3065361766340993,
+      "learning_rate": 4.7359435191658425e-05,
+      "loss": 0.1471,
+      "step": 413
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.3839757274514444,
+      "learning_rate": 4.7344205906962555e-05,
+      "loss": 0.1409,
+      "step": 414
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.2654921795680974,
+      "learning_rate": 4.73289352946231e-05,
+      "loss": 0.1329,
+      "step": 415
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.23867881174079836,
+      "learning_rate": 4.7313623382884435e-05,
+      "loss": 0.1434,
+      "step": 416
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.2526536711131106,
+      "learning_rate": 4.729827020006735e-05,
+      "loss": 0.1317,
+      "step": 417
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.2297920987011127,
+      "learning_rate": 4.728287577456894e-05,
+      "loss": 0.1467,
+      "step": 418
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3377186779829951,
+      "learning_rate": 4.7267440134862604e-05,
+      "loss": 0.1541,
+      "step": 419
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.28209200419629693,
+      "learning_rate": 4.725196330949797e-05,
+      "loss": 0.1457,
+      "step": 420
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3617528596974572,
+      "learning_rate": 4.723644532710082e-05,
+      "loss": 0.1329,
+      "step": 421
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.38012129521909727,
+      "learning_rate": 4.722088621637309e-05,
+      "loss": 0.1466,
+      "step": 422
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.27056264576859596,
+      "learning_rate": 4.7205286006092764e-05,
+      "loss": 0.1423,
+      "step": 423
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.25469954547537926,
+      "learning_rate": 4.718964472511386e-05,
+      "loss": 0.1399,
+      "step": 424
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.30559702251436316,
+      "learning_rate": 4.7173962402366334e-05,
+      "loss": 0.1469,
+      "step": 425
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3287814339800423,
+      "learning_rate": 4.715823906685609e-05,
+      "loss": 0.1367,
+      "step": 426
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3296032543932308,
+      "learning_rate": 4.7142474747664856e-05,
+      "loss": 0.145,
+      "step": 427
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.2843258689519891,
+      "learning_rate": 4.712666947395018e-05,
+      "loss": 0.1516,
+      "step": 428
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.44001827152155676,
+      "learning_rate": 4.711082327494536e-05,
+      "loss": 0.1246,
+      "step": 429
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.35542511109973773,
+      "learning_rate": 4.709493617995938e-05,
+      "loss": 0.1497,
+      "step": 430
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6441794149969033,
+      "learning_rate": 4.707900821837686e-05,
+      "loss": 0.1312,
+      "step": 431
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.2423152591904604,
+      "learning_rate": 4.7063039419658035e-05,
+      "loss": 0.1456,
+      "step": 432
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.2811792036638763,
+      "learning_rate": 4.704702981333864e-05,
+      "loss": 0.1408,
+      "step": 433
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3093479791446126,
+      "learning_rate": 4.703097942902989e-05,
+      "loss": 0.1373,
+      "step": 434
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.2951318820012811,
+      "learning_rate": 4.701488829641845e-05,
+      "loss": 0.1421,
+      "step": 435
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.42348208362141476,
+      "learning_rate": 4.6998756445266336e-05,
+      "loss": 0.1431,
+      "step": 436
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.2809362263136983,
+      "learning_rate": 4.698258390541086e-05,
+      "loss": 0.15,
+      "step": 437
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.24897832885492924,
+      "learning_rate": 4.696637070676462e-05,
+      "loss": 0.1356,
+      "step": 438
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.2655723957243506,
+      "learning_rate": 4.6950116879315385e-05,
+      "loss": 0.1475,
+      "step": 439
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.25479082323643903,
+      "learning_rate": 4.693382245312612e-05,
+      "loss": 0.1478,
+      "step": 440
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.22680456487137643,
+      "learning_rate": 4.6917487458334824e-05,
+      "loss": 0.1411,
+      "step": 441
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.21509754866249914,
+      "learning_rate": 4.690111192515457e-05,
+      "loss": 0.1384,
+      "step": 442
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.24038683806580535,
+      "learning_rate": 4.688469588387339e-05,
+      "loss": 0.1363,
+      "step": 443
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.30391987868319287,
+      "learning_rate": 4.686823936485426e-05,
+      "loss": 0.1281,
+      "step": 444
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3103440529756703,
+      "learning_rate": 4.685174239853499e-05,
+      "loss": 0.1345,
+      "step": 445
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3065679504987451,
+      "learning_rate": 4.6835205015428246e-05,
+      "loss": 0.1445,
+      "step": 446
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.5067982109774993,
+      "learning_rate": 4.681862724612141e-05,
+      "loss": 0.1406,
+      "step": 447
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3085531429849923,
+      "learning_rate": 4.6802009121276566e-05,
+      "loss": 0.147,
+      "step": 448
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.21001190429209293,
+      "learning_rate": 4.6785350671630467e-05,
+      "loss": 0.1242,
+      "step": 449
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.25152604359493425,
+      "learning_rate": 4.6768651927994434e-05,
+      "loss": 0.1418,
+      "step": 450
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.2991093954980607,
+      "learning_rate": 4.67519129212543e-05,
+      "loss": 0.1323,
+      "step": 451
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3310634444467528,
+      "learning_rate": 4.673513368237039e-05,
+      "loss": 0.1506,
+      "step": 452
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.2461620706340736,
+      "learning_rate": 4.671831424237743e-05,
+      "loss": 0.1416,
+      "step": 453
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3192805272117693,
+      "learning_rate": 4.670145463238451e-05,
+      "loss": 0.1457,
+      "step": 454
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.2702739351040745,
+      "learning_rate": 4.668455488357502e-05,
+      "loss": 0.144,
+      "step": 455
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.34905883412718697,
+      "learning_rate": 4.6667615027206564e-05,
+      "loss": 0.142,
+      "step": 456
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.28906794743612046,
+      "learning_rate": 4.665063509461097e-05,
+      "loss": 0.142,
+      "step": 457
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.31689436628615036,
+      "learning_rate": 4.6633615117194165e-05,
+      "loss": 0.1357,
+      "step": 458
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.33067954396889565,
+      "learning_rate": 4.6616555126436134e-05,
+      "loss": 0.1432,
+      "step": 459
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.2931086494388871,
+      "learning_rate": 4.65994551538909e-05,
+      "loss": 0.1352,
+      "step": 460
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.24850553046050688,
+      "learning_rate": 4.658231523118641e-05,
+      "loss": 0.1412,
+      "step": 461
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.27825965596414526,
+      "learning_rate": 4.6565135390024515e-05,
+      "loss": 0.1411,
+      "step": 462
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3046991282117659,
+      "learning_rate": 4.6547915662180905e-05,
+      "loss": 0.1379,
+      "step": 463
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3312763027955893,
+      "learning_rate": 4.653065607950502e-05,
+      "loss": 0.1471,
+      "step": 464
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.27217929274764113,
+      "learning_rate": 4.651335667392003e-05,
+      "loss": 0.1407,
+      "step": 465
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.27959438254564456,
+      "learning_rate": 4.649601747742277e-05,
+      "loss": 0.1369,
+      "step": 466
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.2823305468226839,
+      "learning_rate": 4.6478638522083654e-05,
+      "loss": 0.1494,
+      "step": 467
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3127901456505821,
+      "learning_rate": 4.6461219840046654e-05,
+      "loss": 0.1408,
+      "step": 468
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.36589784438664985,
+      "learning_rate": 4.64437614635292e-05,
+      "loss": 0.1302,
+      "step": 469
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3352322298859117,
+      "learning_rate": 4.642626342482215e-05,
+      "loss": 0.1501,
+      "step": 470
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.23888993612901877,
+      "learning_rate": 4.640872575628973e-05,
+      "loss": 0.1389,
+      "step": 471
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.26307671863607046,
+      "learning_rate": 4.639114849036944e-05,
+      "loss": 0.1294,
+      "step": 472
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.2513276287472555,
+      "learning_rate": 4.637353165957203e-05,
+      "loss": 0.1403,
+      "step": 473
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.19534136773380545,
+      "learning_rate": 4.635587529648146e-05,
+      "loss": 0.132,
+      "step": 474
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.29424097477701366,
+      "learning_rate": 4.6338179433754756e-05,
+      "loss": 0.1211,
+      "step": 475
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.23441407433255068,
+      "learning_rate": 4.632044410412204e-05,
+      "loss": 0.1538,
+      "step": 476
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.23210949677829354,
+      "learning_rate": 4.630266934038642e-05,
+      "loss": 0.1427,
+      "step": 477
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.214657873811937,
+      "learning_rate": 4.628485517542392e-05,
+      "loss": 0.1235,
+      "step": 478
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.2329289849056996,
+      "learning_rate": 4.6267001642183496e-05,
+      "loss": 0.1347,
+      "step": 479
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.23581060893897188,
+      "learning_rate": 4.6249108773686846e-05,
+      "loss": 0.1259,
+      "step": 480
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.19807188547735993,
+      "learning_rate": 4.6231176603028484e-05,
+      "loss": 0.1366,
+      "step": 481
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.22114230125305173,
+      "learning_rate": 4.6213205163375586e-05,
+      "loss": 0.1404,
+      "step": 482
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.290465628388217,
+      "learning_rate": 4.619519448796797e-05,
+      "loss": 0.1452,
+      "step": 483
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.22794781533990469,
+      "learning_rate": 4.617714461011802e-05,
+      "loss": 0.1369,
+      "step": 484
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.22828592620060587,
+      "learning_rate": 4.6159055563210604e-05,
+      "loss": 0.16,
+      "step": 485
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.25094864705550407,
+      "learning_rate": 4.61409273807031e-05,
+      "loss": 0.1398,
+      "step": 486
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.25282544043217764,
+      "learning_rate": 4.612276009612522e-05,
+      "loss": 0.1467,
+      "step": 487
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.18714858994946754,
+      "learning_rate": 4.6104553743078996e-05,
+      "loss": 0.1133,
+      "step": 488
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5495083177007971,
+      "learning_rate": 4.608630835523875e-05,
+      "loss": 0.1346,
+      "step": 489
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.20748745614800185,
+      "learning_rate": 4.606802396635098e-05,
+      "loss": 0.1389,
+      "step": 490
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2565975615768455,
+      "learning_rate": 4.604970061023434e-05,
+      "loss": 0.142,
+      "step": 491
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.24906584494993383,
+      "learning_rate": 4.6031338320779534e-05,
+      "loss": 0.1342,
+      "step": 492
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.26701224892368597,
+      "learning_rate": 4.601293713194929e-05,
+      "loss": 0.1507,
+      "step": 493
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.25167534045365747,
+      "learning_rate": 4.599449707777829e-05,
+      "loss": 0.1463,
+      "step": 494
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.22042947756625808,
+      "learning_rate": 4.5976018192373086e-05,
+      "loss": 0.1113,
+      "step": 495
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2095695083426717,
+      "learning_rate": 4.595750050991207e-05,
+      "loss": 0.1138,
+      "step": 496
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.24368956599597572,
+      "learning_rate": 4.593894406464537e-05,
+      "loss": 0.1297,
+      "step": 497
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2791395107708979,
+      "learning_rate": 4.592034889089482e-05,
+      "loss": 0.1365,
+      "step": 498
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.24429660436654924,
+      "learning_rate": 4.59017150230539e-05,
+      "loss": 0.1467,
+      "step": 499
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2437193458862282,
+      "learning_rate": 4.5883042495587637e-05,
+      "loss": 0.1345,
+      "step": 500
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.24441193580233098,
+      "learning_rate": 4.586433134303257e-05,
+      "loss": 0.1302,
+      "step": 501
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4267610051519048,
+      "learning_rate": 4.584558159999668e-05,
+      "loss": 0.1338,
+      "step": 502
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2372833148230934,
+      "learning_rate": 4.582679330115933e-05,
+      "loss": 0.1357,
+      "step": 503
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2841253414020195,
+      "learning_rate": 4.580796648127118e-05,
+      "loss": 0.1416,
+      "step": 504
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.24300530723473782,
+      "learning_rate": 4.578910117515416e-05,
+      "loss": 0.1527,
+      "step": 505
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.21547310188173485,
+      "learning_rate": 4.5770197417701365e-05,
+      "loss": 0.1438,
+      "step": 506
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.23957600197050985,
+      "learning_rate": 4.5751255243877015e-05,
+      "loss": 0.1437,
+      "step": 507
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.220436071686436,
+      "learning_rate": 4.573227468871639e-05,
+      "loss": 0.1396,
+      "step": 508
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.29925636242356635,
+      "learning_rate": 4.571325578732575e-05,
+      "loss": 0.1289,
+      "step": 509
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.32795525489128524,
+      "learning_rate": 4.569419857488228e-05,
+      "loss": 0.1523,
+      "step": 510
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.35708278179406977,
+      "learning_rate": 4.567510308663404e-05,
+      "loss": 0.1335,
+      "step": 511
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.31273919560083,
+      "learning_rate": 4.5655969357899874e-05,
+      "loss": 0.1364,
+      "step": 512
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.26936425352444593,
+      "learning_rate": 4.563679742406935e-05,
+      "loss": 0.136,
+      "step": 513
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.31382533534360335,
+      "learning_rate": 4.561758732060271e-05,
+      "loss": 0.1412,
+      "step": 514
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.27640214206090724,
+      "learning_rate": 4.559833908303079e-05,
+      "loss": 0.1353,
+      "step": 515
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.24063131611499286,
+      "learning_rate": 4.5579052746954955e-05,
+      "loss": 0.1374,
+      "step": 516
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.26030274174851176,
+      "learning_rate": 4.555972834804704e-05,
+      "loss": 0.1335,
+      "step": 517
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.28828500680678776,
+      "learning_rate": 4.5540365922049275e-05,
+      "loss": 0.1645,
+      "step": 518
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.29802589376307753,
+      "learning_rate": 4.5520965504774246e-05,
+      "loss": 0.133,
+      "step": 519
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.2814229129973559,
+      "learning_rate": 4.550152713210478e-05,
+      "loss": 0.1315,
+      "step": 520
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.33807192884201953,
+      "learning_rate": 4.548205083999392e-05,
+      "loss": 0.1589,
+      "step": 521
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.2445750986021557,
+      "learning_rate": 4.546253666446484e-05,
+      "loss": 0.1417,
+      "step": 522
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.34568149843169,
+      "learning_rate": 4.544298464161079e-05,
+      "loss": 0.1515,
+      "step": 523
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.27613283417321366,
+      "learning_rate": 4.5423394807595005e-05,
+      "loss": 0.1384,
+      "step": 524
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.27080191356009137,
+      "learning_rate": 4.5403767198650683e-05,
+      "loss": 0.1424,
+      "step": 525
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5269204473217596,
+      "learning_rate": 4.5384101851080864e-05,
+      "loss": 0.1424,
+      "step": 526
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.438210137458959,
+      "learning_rate": 4.5364398801258396e-05,
+      "loss": 0.1427,
+      "step": 527
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.44862225173452125,
+      "learning_rate": 4.534465808562587e-05,
+      "loss": 0.1332,
+      "step": 528
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.26681690929557084,
+      "learning_rate": 4.532487974069554e-05,
+      "loss": 0.1355,
+      "step": 529
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.2891258984745436,
+      "learning_rate": 4.530506380304925e-05,
+      "loss": 0.1479,
+      "step": 530
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4237505039069817,
+      "learning_rate": 4.528521030933839e-05,
+      "loss": 0.132,
+      "step": 531
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.24285789217926546,
+      "learning_rate": 4.526531929628379e-05,
+      "loss": 0.1538,
+      "step": 532
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5008120687269136,
+      "learning_rate": 4.52453908006757e-05,
+      "loss": 0.1395,
+      "step": 533
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.2959932438418103,
+      "learning_rate": 4.522542485937369e-05,
+      "loss": 0.1221,
+      "step": 534
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.29939461615886925,
+      "learning_rate": 4.5205421509306576e-05,
+      "loss": 0.1125,
+      "step": 535
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.2622200914314321,
+      "learning_rate": 4.5185380787472384e-05,
+      "loss": 0.1302,
+      "step": 536
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4081996154012246,
+      "learning_rate": 4.516530273093825e-05,
+      "loss": 0.1375,
+      "step": 537
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.33714950375494107,
+      "learning_rate": 4.514518737684038e-05,
+      "loss": 0.1401,
+      "step": 538
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.30296722789654745,
+      "learning_rate": 4.5125034762383936e-05,
+      "loss": 0.1392,
+      "step": 539
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.2787384851543654,
+      "learning_rate": 4.5104844924843016e-05,
+      "loss": 0.146,
+      "step": 540
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3093507532193742,
+      "learning_rate": 4.508461790156056e-05,
+      "loss": 0.1388,
+      "step": 541
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.30423437728862535,
+      "learning_rate": 4.5064353729948315e-05,
+      "loss": 0.1373,
+      "step": 542
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.377503880717078,
+      "learning_rate": 4.504405244748669e-05,
+      "loss": 0.1339,
+      "step": 543
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.29434330211341897,
+      "learning_rate": 4.5023714091724756e-05,
+      "loss": 0.1427,
+      "step": 544
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.2939613253591724,
+      "learning_rate": 4.500333870028016e-05,
+      "loss": 0.1305,
+      "step": 545
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.2945234594318532,
+      "learning_rate": 4.498292631083904e-05,
+      "loss": 0.1596,
+      "step": 546
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.43182768848002356,
+      "learning_rate": 4.4962476961155976e-05,
+      "loss": 0.1383,
+      "step": 547
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.39054811479379303,
+      "learning_rate": 4.4941990689053886e-05,
+      "loss": 0.1371,
+      "step": 548
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5157067720967706,
+      "learning_rate": 4.492146753242401e-05,
+      "loss": 0.1326,
+      "step": 549
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.28677867715599603,
+      "learning_rate": 4.49009075292258e-05,
+      "loss": 0.1329,
+      "step": 550
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.327185346873962,
+      "learning_rate": 4.488031071748684e-05,
+      "loss": 0.1399,
+      "step": 551
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6576166870463284,
+      "learning_rate": 4.485967713530281e-05,
+      "loss": 0.1319,
+      "step": 552
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.2793916483694725,
+      "learning_rate": 4.483900682083742e-05,
+      "loss": 0.1372,
+      "step": 553
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5666133325926903,
+      "learning_rate": 4.481829981232227e-05,
+      "loss": 0.1602,
+      "step": 554
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.25774633548559384,
+      "learning_rate": 4.479755614805688e-05,
+      "loss": 0.1349,
+      "step": 555
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.2879545012126699,
+      "learning_rate": 4.477677586640854e-05,
+      "loss": 0.1249,
+      "step": 556
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3922294150921172,
+      "learning_rate": 4.4755959005812256e-05,
+      "loss": 0.1364,
+      "step": 557
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.23910548158434056,
+      "learning_rate": 4.4735105604770735e-05,
+      "loss": 0.1365,
+      "step": 558
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.2474709620951068,
+      "learning_rate": 4.4714215701854225e-05,
+      "loss": 0.1304,
+      "step": 559
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.24331733584707,
+      "learning_rate": 4.469328933570051e-05,
+      "loss": 0.1326,
+      "step": 560
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.43159070677579225,
+      "learning_rate": 4.46723265450148e-05,
+      "loss": 0.1353,
+      "step": 561
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.36183397128456085,
+      "learning_rate": 4.465132736856969e-05,
+      "loss": 0.1479,
+      "step": 562
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.2827407095695767,
+      "learning_rate": 4.463029184520507e-05,
+      "loss": 0.1243,
+      "step": 563
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.24636234559414608,
+      "learning_rate": 4.4609220013828065e-05,
+      "loss": 0.1431,
+      "step": 564
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.28568894769151143,
+      "learning_rate": 4.4588111913412945e-05,
+      "loss": 0.1353,
+      "step": 565
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.22380738099505187,
+      "learning_rate": 4.4566967583001046e-05,
+      "loss": 0.1264,
+      "step": 566
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.25141595491872526,
+      "learning_rate": 4.454578706170075e-05,
+      "loss": 0.1404,
+      "step": 567
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3274083557669551,
+      "learning_rate": 4.452457038868735e-05,
+      "loss": 0.1389,
+      "step": 568
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3250774387620599,
+      "learning_rate": 4.450331760320302e-05,
+      "loss": 0.1456,
+      "step": 569
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.2541585973431589,
+      "learning_rate": 4.448202874455673e-05,
+      "loss": 0.1309,
+      "step": 570
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3086535154525258,
+      "learning_rate": 4.446070385212414e-05,
+      "loss": 0.1314,
+      "step": 571
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.2542854273133174,
+      "learning_rate": 4.4439342965347595e-05,
+      "loss": 0.1303,
+      "step": 572
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.24057649064089304,
+      "learning_rate": 4.4417946123736e-05,
+      "loss": 0.1363,
+      "step": 573
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4390456446431953,
+      "learning_rate": 4.4396513366864765e-05,
+      "loss": 0.1357,
+      "step": 574
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.2559714203788644,
+      "learning_rate": 4.4375044734375724e-05,
+      "loss": 0.119,
+      "step": 575
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.26085116708822836,
+      "learning_rate": 4.4353540265977064e-05,
+      "loss": 0.124,
+      "step": 576
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4521744530069156,
+      "learning_rate": 4.433200000144326e-05,
+      "loss": 0.1349,
+      "step": 577
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.23676985716701468,
+      "learning_rate": 4.431042398061499e-05,
+      "loss": 0.1409,
+      "step": 578
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.28663525701826703,
+      "learning_rate": 4.428881224339907e-05,
+      "loss": 0.1375,
+      "step": 579
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.2248225308751978,
+      "learning_rate": 4.426716482976838e-05,
+      "loss": 0.1257,
+      "step": 580
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.24313662628162414,
+      "learning_rate": 4.424548177976179e-05,
+      "loss": 0.1313,
+      "step": 581
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.28164718572578995,
+      "learning_rate": 4.422376313348405e-05,
+      "loss": 0.1289,
+      "step": 582
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.2374863018168902,
+      "learning_rate": 4.4202008931105795e-05,
+      "loss": 0.1353,
+      "step": 583
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.43183190419646,
+      "learning_rate": 4.4180219212863404e-05,
+      "loss": 0.1503,
+      "step": 584
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.26343673901481235,
+      "learning_rate": 4.4158394019058944e-05,
+      "loss": 0.1588,
+      "step": 585
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.34998014094645996,
+      "learning_rate": 4.413653339006008e-05,
+      "loss": 0.1412,
+      "step": 586
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.23070055818094928,
+      "learning_rate": 4.411463736630006e-05,
+      "loss": 0.1449,
+      "step": 587
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.2297267467267025,
+      "learning_rate": 4.409270598827756e-05,
+      "loss": 0.1294,
+      "step": 588
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.23657884733830212,
+      "learning_rate": 4.407073929655666e-05,
+      "loss": 0.136,
+      "step": 589
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.2361676679117266,
+      "learning_rate": 4.404873733176678e-05,
+      "loss": 0.1319,
+      "step": 590
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.23386420602754182,
+      "learning_rate": 4.402670013460252e-05,
+      "loss": 0.1419,
+      "step": 591
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.32362761787191663,
+      "learning_rate": 4.400462774582371e-05,
+      "loss": 0.1437,
+      "step": 592
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.25801400042068945,
+      "learning_rate": 4.398252020625523e-05,
+      "loss": 0.1444,
+      "step": 593
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.2586607783987809,
+      "learning_rate": 4.3960377556787e-05,
+      "loss": 0.1431,
+      "step": 594
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.2949725912604422,
+      "learning_rate": 4.393819983837385e-05,
+      "loss": 0.1315,
+      "step": 595
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.23660752032442448,
+      "learning_rate": 4.3915987092035505e-05,
+      "loss": 0.1372,
+      "step": 596
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.27338493186762136,
+      "learning_rate": 4.389373935885646e-05,
+      "loss": 0.1383,
+      "step": 597
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.2878419786166675,
+      "learning_rate": 4.387145667998591e-05,
+      "loss": 0.1296,
+      "step": 598
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.27099848461160336,
+      "learning_rate": 4.384913909663772e-05,
+      "loss": 0.1383,
+      "step": 599
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.2590618002328116,
+      "learning_rate": 4.382678665009028e-05,
+      "loss": 0.1327,
+      "step": 600
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.24599044010641058,
+      "learning_rate": 4.380439938168647e-05,
+      "loss": 0.1253,
+      "step": 601
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.30415909833019544,
+      "learning_rate": 4.378197733283359e-05,
+      "loss": 0.1479,
+      "step": 602
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3263883623850431,
+      "learning_rate": 4.375952054500326e-05,
+      "loss": 0.1469,
+      "step": 603
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.28643410311395895,
+      "learning_rate": 4.373702905973135e-05,
+      "loss": 0.1377,
+      "step": 604
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3399079216294787,
+      "learning_rate": 4.371450291861792e-05,
+      "loss": 0.1388,
+      "step": 605
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.25708643888411886,
+      "learning_rate": 4.369194216332712e-05,
+      "loss": 0.1311,
+      "step": 606
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.2638091298442061,
+      "learning_rate": 4.36693468355871e-05,
+      "loss": 0.1364,
+      "step": 607
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.2594893698569746,
+      "learning_rate": 4.3646716977189996e-05,
+      "loss": 0.1316,
+      "step": 608
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.2598150536936626,
+      "learning_rate": 4.362405262999178e-05,
+      "loss": 0.1248,
+      "step": 609
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.41640384557157,
+      "learning_rate": 4.360135383591224e-05,
+      "loss": 0.1373,
+      "step": 610
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3347693534268355,
+      "learning_rate": 4.357862063693486e-05,
+      "loss": 0.1422,
+      "step": 611
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.37244595681199966,
+      "learning_rate": 4.355585307510675e-05,
+      "loss": 0.12,
+      "step": 612
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.2855498613790247,
+      "learning_rate": 4.3533051192538596e-05,
+      "loss": 0.1508,
+      "step": 613
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.24344964913871922,
+      "learning_rate": 4.351021503140456e-05,
+      "loss": 0.1471,
+      "step": 614
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.34013641829305696,
+      "learning_rate": 4.348734463394219e-05,
+      "loss": 0.1432,
+      "step": 615
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3661009883123754,
+      "learning_rate": 4.346444004245239e-05,
+      "loss": 0.1326,
+      "step": 616
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.2443553732747609,
+      "learning_rate": 4.344150129929927e-05,
+      "loss": 0.1431,
+      "step": 617
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3182746426252024,
+      "learning_rate": 4.341852844691012e-05,
+      "loss": 0.1339,
+      "step": 618
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4374707610848554,
+      "learning_rate": 4.339552152777534e-05,
+      "loss": 0.1373,
+      "step": 619
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.2328681353698796,
+      "learning_rate": 4.337248058444832e-05,
+      "loss": 0.1387,
+      "step": 620
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.24035196918207313,
+      "learning_rate": 4.3349405659545365e-05,
+      "loss": 0.1489,
+      "step": 621
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.2282250383129986,
+      "learning_rate": 4.332629679574566e-05,
+      "loss": 0.1229,
+      "step": 622
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.2441622062483833,
+      "learning_rate": 4.3303154035791164e-05,
+      "loss": 0.1414,
+      "step": 623
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.2329663148035194,
+      "learning_rate": 4.32799774224865e-05,
+      "loss": 0.1373,
+      "step": 624
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.21504602932314495,
+      "learning_rate": 4.3256766998698936e-05,
+      "loss": 0.1313,
+      "step": 625
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.26689196500032525,
+      "learning_rate": 4.323352280735826e-05,
+      "loss": 0.1411,
+      "step": 626
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.2710828801575038,
+      "learning_rate": 4.321024489145673e-05,
+      "loss": 0.132,
+      "step": 627
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.2702774686534758,
+      "learning_rate": 4.318693329404896e-05,
+      "loss": 0.1231,
+      "step": 628
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.22364157070721635,
+      "learning_rate": 4.316358805825188e-05,
+      "loss": 0.1402,
+      "step": 629
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.2604000742236549,
+      "learning_rate": 4.3140209227244624e-05,
+      "loss": 0.1403,
+      "step": 630
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.5315519998045332,
+      "learning_rate": 4.311679684426847e-05,
+      "loss": 0.1413,
+      "step": 631
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6321734513970303,
+      "learning_rate": 4.309335095262676e-05,
+      "loss": 0.128,
+      "step": 632
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.2545518556592647,
+      "learning_rate": 4.306987159568479e-05,
+      "loss": 0.1265,
+      "step": 633
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.24879484556916578,
+      "learning_rate": 4.304635881686978e-05,
+      "loss": 0.1317,
+      "step": 634
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.26214651065926986,
+      "learning_rate": 4.302281265967076e-05,
+      "loss": 0.1303,
+      "step": 635
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.26459882159281034,
+      "learning_rate": 4.299923316763848e-05,
+      "loss": 0.1362,
+      "step": 636
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3087296094183048,
+      "learning_rate": 4.2975620384385364e-05,
+      "loss": 0.1452,
+      "step": 637
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.22508739766935096,
+      "learning_rate": 4.295197435358541e-05,
+      "loss": 0.132,
+      "step": 638
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.430289366423517,
+      "learning_rate": 4.292829511897409e-05,
+      "loss": 0.1389,
+      "step": 639
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5116660910053685,
+      "learning_rate": 4.2904582724348316e-05,
+      "loss": 0.1193,
+      "step": 640
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.28152167742818146,
+      "learning_rate": 4.2880837213566326e-05,
+      "loss": 0.1195,
+      "step": 641
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.2719918215437369,
+      "learning_rate": 4.2857058630547594e-05,
+      "loss": 0.1435,
+      "step": 642
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.2787390867016657,
+      "learning_rate": 4.2833247019272775e-05,
+      "loss": 0.1194,
+      "step": 643
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.2768651132328009,
+      "learning_rate": 4.2809402423783624e-05,
+      "loss": 0.1246,
+      "step": 644
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.2092286598502568,
+      "learning_rate": 4.278552488818288e-05,
+      "loss": 0.1346,
+      "step": 645
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.2821323386354148,
+      "learning_rate": 4.276161445663423e-05,
+      "loss": 0.1379,
+      "step": 646
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.20126962192912326,
+      "learning_rate": 4.273767117336217e-05,
+      "loss": 0.1236,
+      "step": 647
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.2400676798482724,
+      "learning_rate": 4.2713695082652015e-05,
+      "loss": 0.1198,
+      "step": 648
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.23949465411913112,
+      "learning_rate": 4.2689686228849716e-05,
+      "loss": 0.1329,
+      "step": 649
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3598992427152108,
+      "learning_rate": 4.266564465636183e-05,
+      "loss": 0.1195,
+      "step": 650
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3041119213162459,
+      "learning_rate": 4.264157040965543e-05,
+      "loss": 0.1108,
+      "step": 651
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.23050112500851822,
+      "learning_rate": 4.261746353325804e-05,
+      "loss": 0.1284,
+      "step": 652
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.27433883605918397,
+      "learning_rate": 4.259332407175751e-05,
+      "loss": 0.1234,
+      "step": 653
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.25815529510360985,
+      "learning_rate": 4.2569152069801994e-05,
+      "loss": 0.1267,
+      "step": 654
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.24818849154229652,
+      "learning_rate": 4.254494757209979e-05,
+      "loss": 0.1333,
+      "step": 655
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.30761090734954744,
+      "learning_rate": 4.252071062341933e-05,
+      "loss": 0.1268,
+      "step": 656
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.24746615725616938,
+      "learning_rate": 4.2496441268589046e-05,
+      "loss": 0.1264,
+      "step": 657
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3817194359669384,
+      "learning_rate": 4.2472139552497334e-05,
+      "loss": 0.1303,
+      "step": 658
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.25633629130260793,
+      "learning_rate": 4.244780552009243e-05,
+      "loss": 0.1252,
+      "step": 659
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3138167487033373,
+      "learning_rate": 4.242343921638234e-05,
+      "loss": 0.137,
+      "step": 660
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.45506910055450234,
+      "learning_rate": 4.2399040686434756e-05,
+      "loss": 0.148,
+      "step": 661
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.749260039776704,
+      "learning_rate": 4.237460997537699e-05,
+      "loss": 0.133,
+      "step": 662
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.2944335052772232,
+      "learning_rate": 4.235014712839586e-05,
+      "loss": 0.1134,
+      "step": 663
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3395863027130289,
+      "learning_rate": 4.232565219073763e-05,
+      "loss": 0.1483,
+      "step": 664
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.41462429838590503,
+      "learning_rate": 4.230112520770792e-05,
+      "loss": 0.1315,
+      "step": 665
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.37244716312412013,
+      "learning_rate": 4.227656622467162e-05,
+      "loss": 0.1366,
+      "step": 666
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.26760238527387986,
+      "learning_rate": 4.2251975287052804e-05,
+      "loss": 0.1106,
+      "step": 667
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.26124114762969153,
+      "learning_rate": 4.222735244033464e-05,
+      "loss": 0.1347,
+      "step": 668
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3315510643089393,
+      "learning_rate": 4.220269773005935e-05,
+      "loss": 0.1308,
+      "step": 669
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3636316416884437,
+      "learning_rate": 4.217801120182805e-05,
+      "loss": 0.1345,
+      "step": 670
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.28799225478584717,
+      "learning_rate": 4.2153292901300715e-05,
+      "loss": 0.152,
+      "step": 671
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.27659608481649917,
+      "learning_rate": 4.212854287419611e-05,
+      "loss": 0.13,
+      "step": 672
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.328756212582101,
+      "learning_rate": 4.210376116629165e-05,
+      "loss": 0.1364,
+      "step": 673
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.43714944014531854,
+      "learning_rate": 4.2078947823423364e-05,
+      "loss": 0.1322,
+      "step": 674
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.280297143831064,
+      "learning_rate": 4.205410289148579e-05,
+      "loss": 0.136,
+      "step": 675
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.47544598905331154,
+      "learning_rate": 4.20292264164319e-05,
+      "loss": 0.1231,
+      "step": 676
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.2878692376693704,
+      "learning_rate": 4.2004318444272985e-05,
+      "loss": 0.1167,
+      "step": 677
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3068815365001248,
+      "learning_rate": 4.197937902107863e-05,
+      "loss": 0.1218,
+      "step": 678
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3237309644422223,
+      "learning_rate": 4.1954408192976536e-05,
+      "loss": 0.1367,
+      "step": 679
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.531197794362997,
+      "learning_rate": 4.192940600615255e-05,
+      "loss": 0.1304,
+      "step": 680
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3187988946210869,
+      "learning_rate": 4.1904372506850484e-05,
+      "loss": 0.1341,
+      "step": 681
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.2596375713694015,
+      "learning_rate": 4.187930774137209e-05,
+      "loss": 0.1409,
+      "step": 682
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.27696623520580516,
+      "learning_rate": 4.18542117560769e-05,
+      "loss": 0.1263,
+      "step": 683
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.2362316441556187,
+      "learning_rate": 4.182908459738226e-05,
+      "loss": 0.1153,
+      "step": 684
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3337879175595662,
+      "learning_rate": 4.180392631176312e-05,
+      "loss": 0.1375,
+      "step": 685
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.2645777432257612,
+      "learning_rate": 4.177873694575202e-05,
+      "loss": 0.1321,
+      "step": 686
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3968745440772856,
+      "learning_rate": 4.175351654593899e-05,
+      "loss": 0.1271,
+      "step": 687
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.33708256737099496,
+      "learning_rate": 4.172826515897146e-05,
+      "loss": 0.1282,
+      "step": 688
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.30533306779234504,
+      "learning_rate": 4.170298283155416e-05,
+      "loss": 0.1497,
+      "step": 689
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3818797092899185,
+      "learning_rate": 4.167766961044907e-05,
+      "loss": 0.1342,
+      "step": 690
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.22427750039348165,
+      "learning_rate": 4.165232554247528e-05,
+      "loss": 0.13,
+      "step": 691
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.28198574530937215,
+      "learning_rate": 4.162695067450897e-05,
+      "loss": 0.134,
+      "step": 692
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.2705532753739328,
+      "learning_rate": 4.160154505348326e-05,
+      "loss": 0.1419,
+      "step": 693
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.21595679253148153,
+      "learning_rate": 4.157610872638815e-05,
+      "loss": 0.1192,
+      "step": 694
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.30560276355553595,
+      "learning_rate": 4.155064174027047e-05,
+      "loss": 0.1403,
+      "step": 695
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.22751769894493507,
+      "learning_rate": 4.1525144142233705e-05,
+      "loss": 0.1181,
+      "step": 696
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3156390591474674,
+      "learning_rate": 4.1499615979437986e-05,
+      "loss": 0.1366,
+      "step": 697
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.27368136635503965,
+      "learning_rate": 4.147405729909999e-05,
+      "loss": 0.1249,
+      "step": 698
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.2943082269051328,
+      "learning_rate": 4.144846814849282e-05,
+      "loss": 0.1263,
+      "step": 699
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.30669735404431925,
+      "learning_rate": 4.1422848574945924e-05,
+      "loss": 0.1378,
+      "step": 700
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.32478660861152736,
+      "learning_rate": 4.139719862584506e-05,
+      "loss": 0.1369,
+      "step": 701
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.26774744502607306,
+      "learning_rate": 4.137151834863213e-05,
+      "loss": 0.1307,
+      "step": 702
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.26603262844804304,
+      "learning_rate": 4.1345807790805144e-05,
+      "loss": 0.1348,
+      "step": 703
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.2602092571667875,
+      "learning_rate": 4.132006699991813e-05,
+      "loss": 0.1142,
+      "step": 704
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.2496996829825209,
+      "learning_rate": 4.1294296023581015e-05,
+      "loss": 0.1382,
+      "step": 705
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.24121131442548607,
+      "learning_rate": 4.126849490945958e-05,
+      "loss": 0.1319,
+      "step": 706
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3112494935410999,
+      "learning_rate": 4.124266370527531e-05,
+      "loss": 0.1237,
+      "step": 707
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.2712921111215056,
+      "learning_rate": 4.121680245880539e-05,
+      "loss": 0.1324,
+      "step": 708
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.30466658543269864,
+      "learning_rate": 4.119091121788256e-05,
+      "loss": 0.1394,
+      "step": 709
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.23297272225162585,
+      "learning_rate": 4.116499003039499e-05,
+      "loss": 0.1228,
+      "step": 710
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.22635459160489407,
+      "learning_rate": 4.113903894428632e-05,
+      "loss": 0.1262,
+      "step": 711
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.224286314374261,
+      "learning_rate": 4.1113058007555415e-05,
+      "loss": 0.1272,
+      "step": 712
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3112642574936697,
+      "learning_rate": 4.10870472682564e-05,
+      "loss": 0.1397,
+      "step": 713
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.21437734818679646,
+      "learning_rate": 4.106100677449851e-05,
+      "loss": 0.1088,
+      "step": 714
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.23847961300425505,
+      "learning_rate": 4.103493657444599e-05,
+      "loss": 0.1283,
+      "step": 715
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.2581129271875162,
+      "learning_rate": 4.100883671631806e-05,
+      "loss": 0.1447,
+      "step": 716
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4203311622149223,
+      "learning_rate": 4.098270724838879e-05,
+      "loss": 0.131,
+      "step": 717
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.23295929745451038,
+      "learning_rate": 4.0956548218987004e-05,
+      "loss": 0.1395,
+      "step": 718
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.225287326992596,
+      "learning_rate": 4.093035967649621e-05,
+      "loss": 0.1372,
+      "step": 719
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.2988854712720218,
+      "learning_rate": 4.0904141669354504e-05,
+      "loss": 0.1391,
+      "step": 720
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.2905751813087606,
+      "learning_rate": 4.087789424605447e-05,
+      "loss": 0.1413,
+      "step": 721
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.27833114988129015,
+      "learning_rate": 4.085161745514312e-05,
+      "loss": 0.1263,
+      "step": 722
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.26377262429629,
+      "learning_rate": 4.082531134522176e-05,
+      "loss": 0.1226,
+      "step": 723
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.28040987220705416,
+      "learning_rate": 4.079897596494594e-05,
+      "loss": 0.129,
+      "step": 724
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3089655225706308,
+      "learning_rate": 4.077261136302536e-05,
+      "loss": 0.1298,
+      "step": 725
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3883320392859413,
+      "learning_rate": 4.0746217588223724e-05,
+      "loss": 0.1168,
+      "step": 726
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.24255640016953053,
+      "learning_rate": 4.071979468935874e-05,
+      "loss": 0.1319,
+      "step": 727
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3034925468824987,
+      "learning_rate": 4.069334271530196e-05,
+      "loss": 0.1394,
+      "step": 728
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.2803882880212525,
+      "learning_rate": 4.0666861714978724e-05,
+      "loss": 0.1215,
+      "step": 729
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.21324538007860108,
+      "learning_rate": 4.064035173736804e-05,
+      "loss": 0.1233,
+      "step": 730
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.2378545524798451,
+      "learning_rate": 4.061381283150254e-05,
+      "loss": 0.1327,
+      "step": 731
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.24151453488283614,
+      "learning_rate": 4.058724504646834e-05,
+      "loss": 0.1206,
+      "step": 732
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.24452159924515776,
+      "learning_rate": 4.056064843140498e-05,
+      "loss": 0.1309,
+      "step": 733
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.2487323019152631,
+      "learning_rate": 4.053402303550533e-05,
+      "loss": 0.1222,
+      "step": 734
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.31847838893443786,
+      "learning_rate": 4.050736890801547e-05,
+      "loss": 0.1359,
+      "step": 735
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.24040468328261166,
+      "learning_rate": 4.048068609823464e-05,
+      "loss": 0.13,
+      "step": 736
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.49815906263813176,
+      "learning_rate": 4.045397465551513e-05,
+      "loss": 0.123,
+      "step": 737
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.31937013750209886,
+      "learning_rate": 4.042723462926219e-05,
+      "loss": 0.1182,
+      "step": 738
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.27071151534234916,
+      "learning_rate": 4.040046606893392e-05,
+      "loss": 0.1361,
+      "step": 739
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.2075969465684673,
+      "learning_rate": 4.0373669024041226e-05,
+      "loss": 0.13,
+      "step": 740
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.22102158168444247,
+      "learning_rate": 4.034684354414767e-05,
+      "loss": 0.1269,
+      "step": 741
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.2211964213007889,
+      "learning_rate": 4.0319989678869426e-05,
+      "loss": 0.1248,
+      "step": 742
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.21127407755509525,
+      "learning_rate": 4.029310747787516e-05,
+      "loss": 0.1208,
+      "step": 743
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.23497654659871173,
+      "learning_rate": 4.0266196990885955e-05,
+      "loss": 0.1281,
+      "step": 744
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.22100876673624092,
+      "learning_rate": 4.02392582676752e-05,
+      "loss": 0.1304,
+      "step": 745
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.35897601084420055,
+      "learning_rate": 4.021229135806853e-05,
+      "loss": 0.1312,
+      "step": 746
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.2937495151039835,
+      "learning_rate": 4.018529631194369e-05,
+      "loss": 0.1356,
+      "step": 747
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.25739158350220587,
+      "learning_rate": 4.0158273179230475e-05,
+      "loss": 0.1187,
+      "step": 748
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.23131673693774985,
+      "learning_rate": 4.013122200991064e-05,
+      "loss": 0.1192,
+      "step": 749
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.2839809129701794,
+      "learning_rate": 4.010414285401777e-05,
+      "loss": 0.1398,
+      "step": 750
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.32846832242842494,
+      "learning_rate": 4.007703576163724e-05,
+      "loss": 0.1381,
+      "step": 751
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20772075939760756,
+      "learning_rate": 4.0049900782906086e-05,
+      "loss": 0.1274,
+      "step": 752
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.23157029629681491,
+      "learning_rate": 4.002273796801292e-05,
+      "loss": 0.1345,
+      "step": 753
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.506600975797009,
+      "learning_rate": 3.9995547367197845e-05,
+      "loss": 0.1269,
+      "step": 754
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.23119522904719608,
+      "learning_rate": 3.996832903075235e-05,
+      "loss": 0.1447,
+      "step": 755
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.35252369519455856,
+      "learning_rate": 3.9941083009019223e-05,
+      "loss": 0.1335,
+      "step": 756
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.32737885072131534,
+      "learning_rate": 3.9913809352392474e-05,
+      "loss": 0.1318,
+      "step": 757
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.24190485772634274,
+      "learning_rate": 3.98865081113172e-05,
+      "loss": 0.1286,
+      "step": 758
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.243463973746657,
+      "learning_rate": 3.985917933628955e-05,
+      "loss": 0.1264,
+      "step": 759
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3201764652983342,
+      "learning_rate": 3.983182307785657e-05,
+      "loss": 0.1328,
+      "step": 760
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4676663800702981,
+      "learning_rate": 3.980443938661614e-05,
+      "loss": 0.1134,
+      "step": 761
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.2513360806090325,
+      "learning_rate": 3.977702831321692e-05,
+      "loss": 0.1153,
+      "step": 762
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3178485025347495,
+      "learning_rate": 3.974958990835816e-05,
+      "loss": 0.1226,
+      "step": 763
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20873807059514207,
+      "learning_rate": 3.9722124222789705e-05,
+      "loss": 0.1364,
+      "step": 764
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.2862983866460884,
+      "learning_rate": 3.969463130731183e-05,
+      "loss": 0.127,
+      "step": 765
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20831892224511372,
+      "learning_rate": 3.96671112127752e-05,
+      "loss": 0.1274,
+      "step": 766
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.25715427140042485,
+      "learning_rate": 3.9639563990080716e-05,
+      "loss": 0.1159,
+      "step": 767
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20340678818861685,
+      "learning_rate": 3.96119896901795e-05,
+      "loss": 0.1331,
+      "step": 768
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.22079230397665062,
+      "learning_rate": 3.95843883640727e-05,
+      "loss": 0.1404,
+      "step": 769
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.2202439355407089,
+      "learning_rate": 3.95567600628115e-05,
+      "loss": 0.1235,
+      "step": 770
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.27841969745712863,
+      "learning_rate": 3.9529104837496974e-05,
+      "loss": 0.1424,
+      "step": 771
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.2577283992458959,
+      "learning_rate": 3.9501422739279956e-05,
+      "loss": 0.1325,
+      "step": 772
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.23025827113985484,
+      "learning_rate": 3.9473713819361015e-05,
+      "loss": 0.1447,
+      "step": 773
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.24254206582090532,
+      "learning_rate": 3.9445978128990326e-05,
+      "loss": 0.1192,
+      "step": 774
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.24393619338816883,
+      "learning_rate": 3.9418215719467565e-05,
+      "loss": 0.11,
+      "step": 775
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.28846200291382884,
+      "learning_rate": 3.939042664214184e-05,
+      "loss": 0.123,
+      "step": 776
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.29781319997920996,
+      "learning_rate": 3.9362610948411585e-05,
+      "loss": 0.1248,
+      "step": 777
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3717012702841096,
+      "learning_rate": 3.9334768689724456e-05,
+      "loss": 0.136,
+      "step": 778
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3615679148120019,
+      "learning_rate": 3.9306899917577245e-05,
+      "loss": 0.1012,
+      "step": 779
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.24932085020273842,
+      "learning_rate": 3.9279004683515783e-05,
+      "loss": 0.1387,
+      "step": 780
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.2570940746585984,
+      "learning_rate": 3.925108303913485e-05,
+      "loss": 0.1338,
+      "step": 781
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.24465500968456286,
+      "learning_rate": 3.9223135036078064e-05,
+      "loss": 0.1132,
+      "step": 782
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.28060208480319876,
+      "learning_rate": 3.9195160726037805e-05,
+      "loss": 0.1244,
+      "step": 783
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.21373665782423806,
+      "learning_rate": 3.91671601607551e-05,
+      "loss": 0.1268,
+      "step": 784
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3983820364639343,
+      "learning_rate": 3.913913339201956e-05,
+      "loss": 0.1311,
+      "step": 785
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.42448557487327193,
+      "learning_rate": 3.911108047166924e-05,
+      "loss": 0.1402,
+      "step": 786
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.32583781888566715,
+      "learning_rate": 3.908300145159055e-05,
+      "loss": 0.1425,
+      "step": 787
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3105532942457085,
+      "learning_rate": 3.905489638371823e-05,
+      "loss": 0.1334,
+      "step": 788
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.314961148315096,
+      "learning_rate": 3.902676532003514e-05,
+      "loss": 0.1333,
+      "step": 789
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3051079873977507,
+      "learning_rate": 3.899860831257224e-05,
+      "loss": 0.1442,
+      "step": 790
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.2943406321372046,
+      "learning_rate": 3.8970425413408495e-05,
+      "loss": 0.1186,
+      "step": 791
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7811571502770833,
+      "learning_rate": 3.894221667467074e-05,
+      "loss": 0.1365,
+      "step": 792
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.2878895203487764,
+      "learning_rate": 3.89139821485336e-05,
+      "loss": 0.123,
+      "step": 793
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.22748950389140674,
+      "learning_rate": 3.8885721887219406e-05,
+      "loss": 0.1495,
+      "step": 794
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.35189873572183084,
+      "learning_rate": 3.8857435942998074e-05,
+      "loss": 0.1295,
+      "step": 795
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.20972517002818306,
+      "learning_rate": 3.882912436818705e-05,
+      "loss": 0.1139,
+      "step": 796
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.31037592116864976,
+      "learning_rate": 3.880078721515117e-05,
+      "loss": 0.1387,
+      "step": 797
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.25500923180099827,
+      "learning_rate": 3.8772424536302564e-05,
+      "loss": 0.1209,
+      "step": 798
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.22951042544756892,
+      "learning_rate": 3.87440363841006e-05,
+      "loss": 0.1223,
+      "step": 799
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.28529964832538296,
+      "learning_rate": 3.871562281105175e-05,
+      "loss": 0.125,
+      "step": 800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.2424340903908657,
+      "learning_rate": 3.86871838697095e-05,
+      "loss": 0.1384,
+      "step": 801
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.19133946186218648,
+      "learning_rate": 3.865871961267427e-05,
+      "loss": 0.1221,
+      "step": 802
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.27523306238392303,
+      "learning_rate": 3.863023009259329e-05,
+      "loss": 0.1312,
+      "step": 803
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.22961044254496116,
+      "learning_rate": 3.860171536216052e-05,
+      "loss": 0.1401,
+      "step": 804
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.19583892246225654,
+      "learning_rate": 3.8573175474116545e-05,
+      "loss": 0.1254,
+      "step": 805
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.24910110879416852,
+      "learning_rate": 3.85446104812485e-05,
+      "loss": 0.1348,
+      "step": 806
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.19116204965415973,
+      "learning_rate": 3.851602043638994e-05,
+      "loss": 0.1223,
+      "step": 807
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.21989001807336964,
+      "learning_rate": 3.848740539242075e-05,
+      "loss": 0.1238,
+      "step": 808
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.2302634036918583,
+      "learning_rate": 3.845876540226706e-05,
+      "loss": 0.1456,
+      "step": 809
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.2275614076587381,
+      "learning_rate": 3.843010051890114e-05,
+      "loss": 0.1203,
+      "step": 810
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.218866460110278,
+      "learning_rate": 3.840141079534131e-05,
+      "loss": 0.1387,
+      "step": 811
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.2586990336432796,
+      "learning_rate": 3.8372696284651814e-05,
+      "loss": 0.1389,
+      "step": 812
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.2115857260331374,
+      "learning_rate": 3.834395703994276e-05,
+      "loss": 0.1229,
+      "step": 813
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.2642730179187503,
+      "learning_rate": 3.8315193114369996e-05,
+      "loss": 0.1369,
+      "step": 814
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.22833547030569484,
+      "learning_rate": 3.828640456113504e-05,
+      "loss": 0.1364,
+      "step": 815
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.2578942507837022,
+      "learning_rate": 3.8257591433484906e-05,
+      "loss": 0.1391,
+      "step": 816
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.21276097749439515,
+      "learning_rate": 3.8228753784712126e-05,
+      "loss": 0.1326,
+      "step": 817
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.21246404200879496,
+      "learning_rate": 3.819989166815455e-05,
+      "loss": 0.1132,
+      "step": 818
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.20068054001029212,
+      "learning_rate": 3.817100513719529e-05,
+      "loss": 0.124,
+      "step": 819
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.22146773307513395,
+      "learning_rate": 3.814209424526262e-05,
+      "loss": 0.1316,
+      "step": 820
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.23951030258919137,
+      "learning_rate": 3.8113159045829864e-05,
+      "loss": 0.1076,
+      "step": 821
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.25531351660534524,
+      "learning_rate": 3.8084199592415305e-05,
+      "loss": 0.1381,
+      "step": 822
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.23227436685539996,
+      "learning_rate": 3.8055215938582086e-05,
+      "loss": 0.1439,
+      "step": 823
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.2027524930005183,
+      "learning_rate": 3.802620813793814e-05,
+      "loss": 0.1302,
+      "step": 824
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.2293928269204756,
+      "learning_rate": 3.799717624413599e-05,
+      "loss": 0.1437,
+      "step": 825
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.1743464057682883,
+      "learning_rate": 3.796812031087281e-05,
+      "loss": 0.1274,
+      "step": 826
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.23415991933526123,
+      "learning_rate": 3.793904039189018e-05,
+      "loss": 0.1258,
+      "step": 827
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.17838455005362286,
+      "learning_rate": 3.790993654097405e-05,
+      "loss": 0.1116,
+      "step": 828
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.27099488623623164,
+      "learning_rate": 3.788080881195465e-05,
+      "loss": 0.1277,
+      "step": 829
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.18974042030714686,
+      "learning_rate": 3.785165725870638e-05,
+      "loss": 0.1267,
+      "step": 830
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.20442981284455256,
+      "learning_rate": 3.782248193514766e-05,
+      "loss": 0.1272,
+      "step": 831
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.19425949021898983,
+      "learning_rate": 3.7793282895240926e-05,
+      "loss": 0.1276,
+      "step": 832
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.2405820058263144,
+      "learning_rate": 3.776406019299247e-05,
+      "loss": 0.1195,
+      "step": 833
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.25803395500421766,
+      "learning_rate": 3.7734813882452334e-05,
+      "loss": 0.1336,
+      "step": 834
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.20844847512869552,
+      "learning_rate": 3.770554401771423e-05,
+      "loss": 0.1277,
+      "step": 835
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.30635379335950963,
+      "learning_rate": 3.767625065291544e-05,
+      "loss": 0.1276,
+      "step": 836
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6254226341566934,
+      "learning_rate": 3.764693384223671e-05,
+      "loss": 0.1559,
+      "step": 837
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.24957889395265626,
+      "learning_rate": 3.761759363990215e-05,
+      "loss": 0.1366,
+      "step": 838
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.2436728381651345,
+      "learning_rate": 3.758823010017913e-05,
+      "loss": 0.127,
+      "step": 839
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.18141165362964315,
+      "learning_rate": 3.7558843277378206e-05,
+      "loss": 0.1243,
+      "step": 840
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.20914430546068327,
+      "learning_rate": 3.752943322585297e-05,
+      "loss": 0.123,
+      "step": 841
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3333610646691549,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.1241,
+      "step": 842
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.52541607985973,
+      "learning_rate": 3.747054365425872e-05,
+      "loss": 0.1245,
+      "step": 843
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.22762741050036892,
+      "learning_rate": 3.744106424311133e-05,
+      "loss": 0.1207,
+      "step": 844
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.368973573696696,
+      "learning_rate": 3.741156182108268e-05,
+      "loss": 0.1238,
+      "step": 845
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3187344651247262,
+      "learning_rate": 3.738203644274018e-05,
+      "loss": 0.1228,
+      "step": 846
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.23328547632678728,
+      "learning_rate": 3.735248816269372e-05,
+      "loss": 0.1371,
+      "step": 847
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.20519603397982816,
+      "learning_rate": 3.732291703559553e-05,
+      "loss": 0.1248,
+      "step": 848
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.23121621329974942,
+      "learning_rate": 3.72933231161401e-05,
+      "loss": 0.1299,
+      "step": 849
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3297866325562852,
+      "learning_rate": 3.726370645906407e-05,
+      "loss": 0.1274,
+      "step": 850
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.5750109159669348,
+      "learning_rate": 3.723406711914617e-05,
+      "loss": 0.1312,
+      "step": 851
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.19065266541421713,
+      "learning_rate": 3.7204405151207036e-05,
+      "loss": 0.1281,
+      "step": 852
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.16885514394568535,
+      "learning_rate": 3.717472061010918e-05,
+      "loss": 0.1095,
+      "step": 853
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.32359870148728154,
+      "learning_rate": 3.714501355075688e-05,
+      "loss": 0.1217,
+      "step": 854
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.18161944778459665,
+      "learning_rate": 3.711528402809603e-05,
+      "loss": 0.1133,
+      "step": 855
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.23522328144810725,
+      "learning_rate": 3.708553209711409e-05,
+      "loss": 0.1271,
+      "step": 856
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.24074764817224345,
+      "learning_rate": 3.705575781283999e-05,
+      "loss": 0.1344,
+      "step": 857
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.21775427745304415,
+      "learning_rate": 3.702596123034395e-05,
+      "loss": 0.1262,
+      "step": 858
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.33451731814482283,
+      "learning_rate": 3.699614240473748e-05,
+      "loss": 0.1207,
+      "step": 859
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.2426791005762291,
+      "learning_rate": 3.6966301391173206e-05,
+      "loss": 0.1382,
+      "step": 860
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8367936526725455,
+      "learning_rate": 3.69364382448448e-05,
+      "loss": 0.1326,
+      "step": 861
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.35327837641452137,
+      "learning_rate": 3.6906553020986876e-05,
+      "loss": 0.1457,
+      "step": 862
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.2600570188833623,
+      "learning_rate": 3.6876645774874876e-05,
+      "loss": 0.1411,
+      "step": 863
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.24138585582984232,
+      "learning_rate": 3.6846716561824965e-05,
+      "loss": 0.1316,
+      "step": 864
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.2099357619120145,
+      "learning_rate": 3.681676543719396e-05,
+      "loss": 0.1272,
+      "step": 865
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.17806326472947834,
+      "learning_rate": 3.67867924563792e-05,
+      "loss": 0.1219,
+      "step": 866
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.22425281562921937,
+      "learning_rate": 3.675679767481842e-05,
+      "loss": 0.1353,
+      "step": 867
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.32272467890706735,
+      "learning_rate": 3.672678114798972e-05,
+      "loss": 0.1377,
+      "step": 868
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.1744927538767948,
+      "learning_rate": 3.669674293141139e-05,
+      "loss": 0.1265,
+      "step": 869
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.21643139477333023,
+      "learning_rate": 3.6666683080641846e-05,
+      "loss": 0.1304,
+      "step": 870
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.26730125504234226,
+      "learning_rate": 3.6636601651279524e-05,
+      "loss": 0.1338,
+      "step": 871
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3587075567795276,
+      "learning_rate": 3.660649869896276e-05,
+      "loss": 0.1253,
+      "step": 872
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.26083481282511917,
+      "learning_rate": 3.657637427936972e-05,
+      "loss": 0.1326,
+      "step": 873
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.24036914855355815,
+      "learning_rate": 3.654622844821825e-05,
+      "loss": 0.1334,
+      "step": 874
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.2953541723718817,
+      "learning_rate": 3.651606126126581e-05,
+      "loss": 0.1221,
+      "step": 875
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3969524277328495,
+      "learning_rate": 3.648587277430936e-05,
+      "loss": 0.1223,
+      "step": 876
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.20521720015076136,
+      "learning_rate": 3.645566304318526e-05,
+      "loss": 0.1351,
+      "step": 877
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.22462205675652333,
+      "learning_rate": 3.642543212376916e-05,
+      "loss": 0.1198,
+      "step": 878
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.24892088210935426,
+      "learning_rate": 3.6395180071975885e-05,
+      "loss": 0.1288,
+      "step": 879
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.22816290686990046,
+      "learning_rate": 3.636490694375938e-05,
+      "loss": 0.1521,
+      "step": 880
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.20578233000278973,
+      "learning_rate": 3.6334612795112534e-05,
+      "loss": 0.1332,
+      "step": 881
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.2276598067557036,
+      "learning_rate": 3.6304297682067144e-05,
+      "loss": 0.1199,
+      "step": 882
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.18299234062975703,
+      "learning_rate": 3.627396166069377e-05,
+      "loss": 0.1114,
+      "step": 883
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4408210542687655,
+      "learning_rate": 3.624360478710165e-05,
+      "loss": 0.1157,
+      "step": 884
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3187112054046731,
+      "learning_rate": 3.621322711743858e-05,
+      "loss": 0.1455,
+      "step": 885
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.22035800569682767,
+      "learning_rate": 3.6182828707890816e-05,
+      "loss": 0.1317,
+      "step": 886
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.23292713387194636,
+      "learning_rate": 3.615240961468301e-05,
+      "loss": 0.1293,
+      "step": 887
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.22373495834401733,
+      "learning_rate": 3.6121969894078024e-05,
+      "loss": 0.1384,
+      "step": 888
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.21418111921674743,
+      "learning_rate": 3.60915096023769e-05,
+      "loss": 0.1143,
+      "step": 889
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.2700660695097094,
+      "learning_rate": 3.606102879591874e-05,
+      "loss": 0.1246,
+      "step": 890
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.24895985432505313,
+      "learning_rate": 3.603052753108053e-05,
+      "loss": 0.1192,
+      "step": 891
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.20567486565522328,
+      "learning_rate": 3.600000586427718e-05,
+      "loss": 0.1344,
+      "step": 892
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.20587734538526742,
+      "learning_rate": 3.596946385196126e-05,
+      "loss": 0.1392,
+      "step": 893
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.22934560913577634,
+      "learning_rate": 3.593890155062302e-05,
+      "loss": 0.1248,
+      "step": 894
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.27792256954759653,
+      "learning_rate": 3.5908319016790214e-05,
+      "loss": 0.1336,
+      "step": 895
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.2269833283982729,
+      "learning_rate": 3.587771630702803e-05,
+      "loss": 0.1205,
+      "step": 896
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.0515138232081827,
+      "learning_rate": 3.5847093477938956e-05,
+      "loss": 0.1372,
+      "step": 897
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.21767235568476812,
+      "learning_rate": 3.581645058616271e-05,
+      "loss": 0.1316,
+      "step": 898
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.23356348971141708,
+      "learning_rate": 3.5785787688376104e-05,
+      "loss": 0.1305,
+      "step": 899
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.32953262599556277,
+      "learning_rate": 3.5755104841292974e-05,
+      "loss": 0.1238,
+      "step": 900
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3655378641565382,
+      "learning_rate": 3.5724402101664023e-05,
+      "loss": 0.1181,
+      "step": 901
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.23707872005404748,
+      "learning_rate": 3.569367952627677e-05,
+      "loss": 0.1333,
+      "step": 902
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.24119922037147493,
+      "learning_rate": 3.566293717195543e-05,
+      "loss": 0.1419,
+      "step": 903
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.2204535421496677,
+      "learning_rate": 3.563217509556076e-05,
+      "loss": 0.1284,
+      "step": 904
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.20925067704926753,
+      "learning_rate": 3.5601393353990046e-05,
+      "loss": 0.118,
+      "step": 905
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6386538557848198,
+      "learning_rate": 3.557059200417691e-05,
+      "loss": 0.1049,
+      "step": 906
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.21554044008212292,
+      "learning_rate": 3.553977110309125e-05,
+      "loss": 0.1304,
+      "step": 907
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.2840332760372614,
+      "learning_rate": 3.550893070773914e-05,
+      "loss": 0.1282,
+      "step": 908
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.23885961256153607,
+      "learning_rate": 3.5478070875162694e-05,
+      "loss": 0.1386,
+      "step": 909
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.20104861195132392,
+      "learning_rate": 3.544719166243998e-05,
+      "loss": 0.1369,
+      "step": 910
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.24636968917655522,
+      "learning_rate": 3.541629312668492e-05,
+      "loss": 0.1137,
+      "step": 911
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.2808434573795825,
+      "learning_rate": 3.5385375325047166e-05,
+      "loss": 0.1162,
+      "step": 912
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.25932112122752227,
+      "learning_rate": 3.535443831471201e-05,
+      "loss": 0.1245,
+      "step": 913
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.2018748085404597,
+      "learning_rate": 3.5323482152900254e-05,
+      "loss": 0.1159,
+      "step": 914
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.27640488396828167,
+      "learning_rate": 3.529250689686817e-05,
+      "loss": 0.124,
+      "step": 915
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.30593530902435806,
+      "learning_rate": 3.526151260390729e-05,
+      "loss": 0.1381,
+      "step": 916
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.22790767098568776,
+      "learning_rate": 3.523049933134439e-05,
+      "loss": 0.1361,
+      "step": 917
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.139082764893216,
+      "learning_rate": 3.519946713654134e-05,
+      "loss": 0.1196,
+      "step": 918
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.0170101652726802,
+      "learning_rate": 3.516841607689501e-05,
+      "loss": 0.1192,
+      "step": 919
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.27006067129644246,
+      "learning_rate": 3.5137346209837165e-05,
+      "loss": 0.1393,
+      "step": 920
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.25433070085170617,
+      "learning_rate": 3.5106257592834343e-05,
+      "loss": 0.122,
+      "step": 921
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.2511758284401212,
+      "learning_rate": 3.507515028338779e-05,
+      "loss": 0.1278,
+      "step": 922
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.30422155838460735,
+      "learning_rate": 3.5044024339033297e-05,
+      "loss": 0.1369,
+      "step": 923
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.29523515020098257,
+      "learning_rate": 3.501287981734113e-05,
+      "loss": 0.1256,
+      "step": 924
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5733610149203257,
+      "learning_rate": 3.498171677591593e-05,
+      "loss": 0.118,
+      "step": 925
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.31064524219567313,
+      "learning_rate": 3.495053527239656e-05,
+      "loss": 0.1127,
+      "step": 926
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.46956039906193087,
+      "learning_rate": 3.491933536445606e-05,
+      "loss": 0.1351,
+      "step": 927
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3798908395338638,
+      "learning_rate": 3.488811710980151e-05,
+      "loss": 0.1258,
+      "step": 928
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.27952032923722053,
+      "learning_rate": 3.485688056617391e-05,
+      "loss": 0.127,
+      "step": 929
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.2510246841604782,
+      "learning_rate": 3.4825625791348096e-05,
+      "loss": 0.1137,
+      "step": 930
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.24727656629898945,
+      "learning_rate": 3.479435284313261e-05,
+      "loss": 0.1258,
+      "step": 931
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.27083574692091555,
+      "learning_rate": 3.476306177936961e-05,
+      "loss": 0.1233,
+      "step": 932
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.23763117160735547,
+      "learning_rate": 3.4731752657934794e-05,
+      "loss": 0.1212,
+      "step": 933
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.34424744184765466,
+      "learning_rate": 3.470042553673721e-05,
+      "loss": 0.1271,
+      "step": 934
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3220840828047677,
+      "learning_rate": 3.466908047371923e-05,
+      "loss": 0.1338,
+      "step": 935
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.32428303454176455,
+      "learning_rate": 3.4637717526856406e-05,
+      "loss": 0.1357,
+      "step": 936
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.45050363480212524,
+      "learning_rate": 3.460633675415736e-05,
+      "loss": 0.1308,
+      "step": 937
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.2022410097432858,
+      "learning_rate": 3.457493821366369e-05,
+      "loss": 0.1279,
+      "step": 938
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3302788522550135,
+      "learning_rate": 3.4543521963449857e-05,
+      "loss": 0.1179,
+      "step": 939
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3277206665816927,
+      "learning_rate": 3.4512088061623075e-05,
+      "loss": 0.1326,
+      "step": 940
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.28317114102336804,
+      "learning_rate": 3.4480636566323215e-05,
+      "loss": 0.1269,
+      "step": 941
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3276342022555628,
+      "learning_rate": 3.444916753572266e-05,
+      "loss": 0.1227,
+      "step": 942
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.2394748136645073,
+      "learning_rate": 3.4417681028026276e-05,
+      "loss": 0.1231,
+      "step": 943
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4787391926667228,
+      "learning_rate": 3.4386177101471216e-05,
+      "loss": 0.1469,
+      "step": 944
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3238920952703891,
+      "learning_rate": 3.4354655814326845e-05,
+      "loss": 0.1359,
+      "step": 945
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.38670966148301944,
+      "learning_rate": 3.432311722489467e-05,
+      "loss": 0.1291,
+      "step": 946
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4077201041764966,
+      "learning_rate": 3.4291561391508185e-05,
+      "loss": 0.1246,
+      "step": 947
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.29736788807517045,
+      "learning_rate": 3.425998837253278e-05,
+      "loss": 0.1238,
+      "step": 948
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4002109130013521,
+      "learning_rate": 3.422839822636562e-05,
+      "loss": 0.1377,
+      "step": 949
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.41632723425144263,
+      "learning_rate": 3.4196791011435554e-05,
+      "loss": 0.1408,
+      "step": 950
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3012513758621095,
+      "learning_rate": 3.4165166786203015e-05,
+      "loss": 0.1258,
+      "step": 951
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3859780341438405,
+      "learning_rate": 3.413352560915988e-05,
+      "loss": 0.1241,
+      "step": 952
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.36959222645318845,
+      "learning_rate": 3.4101867538829394e-05,
+      "loss": 0.1319,
+      "step": 953
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5266685312253583,
+      "learning_rate": 3.4070192633766025e-05,
+      "loss": 0.1161,
+      "step": 954
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.311286645602151,
+      "learning_rate": 3.403850095255542e-05,
+      "loss": 0.1322,
+      "step": 955
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.37032790834698065,
+      "learning_rate": 3.400679255381421e-05,
+      "loss": 0.1255,
+      "step": 956
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.2915528008926978,
+      "learning_rate": 3.3975067496189965e-05,
+      "loss": 0.1206,
+      "step": 957
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.33670264459448385,
+      "learning_rate": 3.3943325838361084e-05,
+      "loss": 0.1152,
+      "step": 958
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.34561925317661707,
+      "learning_rate": 3.391156763903665e-05,
+      "loss": 0.1221,
+      "step": 959
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4852838448343069,
+      "learning_rate": 3.387979295695632e-05,
+      "loss": 0.1224,
+      "step": 960
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.24734666208927072,
+      "learning_rate": 3.3848001850890296e-05,
+      "loss": 0.1292,
+      "step": 961
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.21587745766536837,
+      "learning_rate": 3.381619437963911e-05,
+      "loss": 0.1108,
+      "step": 962
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.2746933531055994,
+      "learning_rate": 3.378437060203357e-05,
+      "loss": 0.1171,
+      "step": 963
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5171339181739222,
+      "learning_rate": 3.375253057693466e-05,
+      "loss": 0.1207,
+      "step": 964
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.32969985082750636,
+      "learning_rate": 3.3720674363233396e-05,
+      "loss": 0.1265,
+      "step": 965
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3404502517464877,
+      "learning_rate": 3.368880201985076e-05,
+      "loss": 0.1183,
+      "step": 966
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.2921119307772584,
+      "learning_rate": 3.365691360573754e-05,
+      "loss": 0.1293,
+      "step": 967
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3083998025274543,
+      "learning_rate": 3.3625009179874265e-05,
+      "loss": 0.1282,
+      "step": 968
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.36377939287045324,
+      "learning_rate": 3.359308880127108e-05,
+      "loss": 0.1299,
+      "step": 969
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3249181994316395,
+      "learning_rate": 3.3561152528967646e-05,
+      "loss": 0.135,
+      "step": 970
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.29131274771843335,
+      "learning_rate": 3.352920042203298e-05,
+      "loss": 0.1315,
+      "step": 971
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.25776812941850397,
+      "learning_rate": 3.349723253956542e-05,
+      "loss": 0.1201,
+      "step": 972
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.2284099683286508,
+      "learning_rate": 3.346524894069248e-05,
+      "loss": 0.111,
+      "step": 973
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.38932303538599766,
+      "learning_rate": 3.343324968457076e-05,
+      "loss": 0.1211,
+      "step": 974
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3228597517074561,
+      "learning_rate": 3.3401234830385756e-05,
+      "loss": 0.1259,
+      "step": 975
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.19877272629585194,
+      "learning_rate": 3.3369204437351886e-05,
+      "loss": 0.1131,
+      "step": 976
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.2829540397694802,
+      "learning_rate": 3.3337158564712267e-05,
+      "loss": 0.1247,
+      "step": 977
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4177281787875288,
+      "learning_rate": 3.3305097271738665e-05,
+      "loss": 0.1302,
+      "step": 978
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.31615970010391303,
+      "learning_rate": 3.327302061773136e-05,
+      "loss": 0.1215,
+      "step": 979
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3055993395352262,
+      "learning_rate": 3.324092866201904e-05,
+      "loss": 0.1188,
+      "step": 980
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.350147882772192,
+      "learning_rate": 3.320882146395871e-05,
+      "loss": 0.1205,
+      "step": 981
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.2171182234444422,
+      "learning_rate": 3.3176699082935545e-05,
+      "loss": 0.1244,
+      "step": 982
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.41745695114296005,
+      "learning_rate": 3.314456157836281e-05,
+      "loss": 0.13,
+      "step": 983
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3590736616307884,
+      "learning_rate": 3.3112409009681766e-05,
+      "loss": 0.1183,
+      "step": 984
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.237222842534061,
+      "learning_rate": 3.3080241436361506e-05,
+      "loss": 0.1248,
+      "step": 985
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.2649992915709618,
+      "learning_rate": 3.304805891789888e-05,
+      "loss": 0.1208,
+      "step": 986
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.339280228638389,
+      "learning_rate": 3.301586151381839e-05,
+      "loss": 0.1263,
+      "step": 987
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3121443744549699,
+      "learning_rate": 3.298364928367207e-05,
+      "loss": 0.1135,
+      "step": 988
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.2552108439814379,
+      "learning_rate": 3.295142228703938e-05,
+      "loss": 0.1231,
+      "step": 989
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.24765721270082117,
+      "learning_rate": 3.291918058352706e-05,
+      "loss": 0.1263,
+      "step": 990
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.9144501675314273,
+      "learning_rate": 3.288692423276911e-05,
+      "loss": 0.1346,
+      "step": 991
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.23730308880764447,
+      "learning_rate": 3.2854653294426566e-05,
+      "loss": 0.1279,
+      "step": 992
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.35694859288478226,
+      "learning_rate": 3.282236782818747e-05,
+      "loss": 0.1318,
+      "step": 993
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.24921931844731304,
+      "learning_rate": 3.279006789376674e-05,
+      "loss": 0.1197,
+      "step": 994
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.40490135696578833,
+      "learning_rate": 3.275775355090603e-05,
+      "loss": 0.1222,
+      "step": 995
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.2656555947746581,
+      "learning_rate": 3.272542485937369e-05,
+      "loss": 0.1323,
+      "step": 996
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4546682850923009,
+      "learning_rate": 3.2693081878964546e-05,
+      "loss": 0.1171,
+      "step": 997
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.5314542195009351,
+      "learning_rate": 3.2660724669499906e-05,
+      "loss": 0.1393,
+      "step": 998
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.2147693539180035,
+      "learning_rate": 3.2628353290827365e-05,
+      "loss": 0.1291,
+      "step": 999
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.23365688363080891,
+      "learning_rate": 3.259596780282074e-05,
+      "loss": 0.1242,
+      "step": 1000
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.29115760095098675,
+      "learning_rate": 3.256356826537994e-05,
+      "loss": 0.1266,
+      "step": 1001
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.2756480517653031,
+      "learning_rate": 3.253115473843086e-05,
+      "loss": 0.1287,
+      "step": 1002
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.20516983624124824,
+      "learning_rate": 3.249872728192527e-05,
+      "loss": 0.1213,
+      "step": 1003
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.2908548164848843,
+      "learning_rate": 3.24662859558407e-05,
+      "loss": 0.1299,
+      "step": 1004
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.2679818707484825,
+      "learning_rate": 3.2433830820180346e-05,
+      "loss": 0.132,
+      "step": 1005
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.39270491723770945,
+      "learning_rate": 3.240136193497293e-05,
+      "loss": 0.1244,
+      "step": 1006
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.25168857823334134,
+      "learning_rate": 3.2368879360272606e-05,
+      "loss": 0.128,
+      "step": 1007
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4665116093001654,
+      "learning_rate": 3.233638315615887e-05,
+      "loss": 0.1159,
+      "step": 1008
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3361657477909562,
+      "learning_rate": 3.23038733827364e-05,
+      "loss": 0.1245,
+      "step": 1009
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3428326580933736,
+      "learning_rate": 3.2271350100134975e-05,
+      "loss": 0.1236,
+      "step": 1010
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.308583186509405,
+      "learning_rate": 3.223881336850939e-05,
+      "loss": 0.128,
+      "step": 1011
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7642576421025281,
+      "learning_rate": 3.2206263248039276e-05,
+      "loss": 0.1413,
+      "step": 1012
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.322704419666017,
+      "learning_rate": 3.217369979892905e-05,
+      "loss": 0.1136,
+      "step": 1013
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.24307137374776414,
+      "learning_rate": 3.214112308140777e-05,
+      "loss": 0.1112,
+      "step": 1014
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3051565400215789,
+      "learning_rate": 3.210853315572906e-05,
+      "loss": 0.1205,
+      "step": 1015
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.23974603317074866,
+      "learning_rate": 3.207593008217092e-05,
+      "loss": 0.1186,
+      "step": 1016
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.2344022049624195,
+      "learning_rate": 3.2043313921035743e-05,
+      "loss": 0.1281,
+      "step": 1017
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3696651400970255,
+      "learning_rate": 3.201068473265007e-05,
+      "loss": 0.1304,
+      "step": 1018
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.2448256082612709,
+      "learning_rate": 3.197804257736456e-05,
+      "loss": 0.1179,
+      "step": 1019
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.32129084230691296,
+      "learning_rate": 3.1945387515553846e-05,
+      "loss": 0.1286,
+      "step": 1020
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6721711030192031,
+      "learning_rate": 3.191271960761645e-05,
+      "loss": 0.1217,
+      "step": 1021
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.26499313342468217,
+      "learning_rate": 3.188003891397463e-05,
+      "loss": 0.1347,
+      "step": 1022
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4385580709426176,
+      "learning_rate": 3.184734549507431e-05,
+      "loss": 0.1409,
+      "step": 1023
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.21161812231553753,
+      "learning_rate": 3.181463941138495e-05,
+      "loss": 0.1211,
+      "step": 1024
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.29180935894588156,
+      "learning_rate": 3.178192072339942e-05,
+      "loss": 0.12,
+      "step": 1025
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.2662623372843712,
+      "learning_rate": 3.174918949163392e-05,
+      "loss": 0.1233,
+      "step": 1026
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.23312421133969088,
+      "learning_rate": 3.171644577662785e-05,
+      "loss": 0.1117,
+      "step": 1027
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.22428667872205926,
+      "learning_rate": 3.1683689638943684e-05,
+      "loss": 0.1192,
+      "step": 1028
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.2661781610494002,
+      "learning_rate": 3.165092113916688e-05,
+      "loss": 0.1252,
+      "step": 1029
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.35720096084059394,
+      "learning_rate": 3.161814033790577e-05,
+      "loss": 0.1209,
+      "step": 1030
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3461089305045611,
+      "learning_rate": 3.158534729579142e-05,
+      "loss": 0.1268,
+      "step": 1031
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.27728074002634545,
+      "learning_rate": 3.1552542073477555e-05,
+      "loss": 0.1345,
+      "step": 1032
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8744007059497831,
+      "learning_rate": 3.1519724731640424e-05,
+      "loss": 0.1373,
+      "step": 1033
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3019719401470567,
+      "learning_rate": 3.1486895330978685e-05,
+      "loss": 0.1243,
+      "step": 1034
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.2620350740852829,
+      "learning_rate": 3.1454053932213304e-05,
+      "loss": 0.1326,
+      "step": 1035
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.21950628016804616,
+      "learning_rate": 3.142120059608744e-05,
+      "loss": 0.1228,
+      "step": 1036
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.2735004813229921,
+      "learning_rate": 3.138833538336633e-05,
+      "loss": 0.1176,
+      "step": 1037
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.23568263384307905,
+      "learning_rate": 3.135545835483718e-05,
+      "loss": 0.1234,
+      "step": 1038
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.34811683222096296,
+      "learning_rate": 3.132256957130904e-05,
+      "loss": 0.1452,
+      "step": 1039
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.28850883196171473,
+      "learning_rate": 3.1289669093612714e-05,
+      "loss": 0.1178,
+      "step": 1040
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3336710651109349,
+      "learning_rate": 3.125675698260065e-05,
+      "loss": 0.125,
+      "step": 1041
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.2584922836390666,
+      "learning_rate": 3.1223833299146756e-05,
+      "loss": 0.134,
+      "step": 1042
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.2051290881388176,
+      "learning_rate": 3.11908981041464e-05,
+      "loss": 0.1188,
+      "step": 1043
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.22815628367217383,
+      "learning_rate": 3.115795145851625e-05,
+      "loss": 0.1332,
+      "step": 1044
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.2344156512621993,
+      "learning_rate": 3.11249934231941e-05,
+      "loss": 0.1125,
+      "step": 1045
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.21330852679151735,
+      "learning_rate": 3.1092024059138836e-05,
+      "loss": 0.1275,
+      "step": 1046
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.26734480675913924,
+      "learning_rate": 3.105904342733032e-05,
+      "loss": 0.1146,
+      "step": 1047
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.23617585092406218,
+      "learning_rate": 3.1026051588769204e-05,
+      "loss": 0.1114,
+      "step": 1048
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.43807935329059966,
+      "learning_rate": 3.099304860447692e-05,
+      "loss": 0.1107,
+      "step": 1049
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.26973800314263324,
+      "learning_rate": 3.096003453549549e-05,
+      "loss": 0.1308,
+      "step": 1050
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.5629556714679972,
+      "learning_rate": 3.092700944288744e-05,
+      "loss": 0.1244,
+      "step": 1051
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.31965192121723157,
+      "learning_rate": 3.0893973387735687e-05,
+      "loss": 0.1246,
+      "step": 1052
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.2273119223652771,
+      "learning_rate": 3.0860926431143415e-05,
+      "loss": 0.1242,
+      "step": 1053
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.23783857587622562,
+      "learning_rate": 3.082786863423399e-05,
+      "loss": 0.124,
+      "step": 1054
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.19570815637568423,
+      "learning_rate": 3.079480005815081e-05,
+      "loss": 0.1129,
+      "step": 1055
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.22583482223282145,
+      "learning_rate": 3.076172076405722e-05,
+      "loss": 0.12,
+      "step": 1056
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3909255984965604,
+      "learning_rate": 3.072863081313639e-05,
+      "loss": 0.1111,
+      "step": 1057
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.24092438453258666,
+      "learning_rate": 3.069553026659119e-05,
+      "loss": 0.1244,
+      "step": 1058
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.5455146265805269,
+      "learning_rate": 3.0662419185644115e-05,
+      "loss": 0.1347,
+      "step": 1059
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.30612179721273214,
+      "learning_rate": 3.06292976315371e-05,
+      "loss": 0.1197,
+      "step": 1060
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3436265815580422,
+      "learning_rate": 3.059616566553149e-05,
+      "loss": 0.1194,
+      "step": 1061
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.21591926193148583,
+      "learning_rate": 3.056302334890786e-05,
+      "loss": 0.1146,
+      "step": 1062
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.26327944335577247,
+      "learning_rate": 3.052987074296596e-05,
+      "loss": 0.1237,
+      "step": 1063
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.2789496415443722,
+      "learning_rate": 3.0496707909024542e-05,
+      "loss": 0.1259,
+      "step": 1064
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.2859402089706897,
+      "learning_rate": 3.0463534908421298e-05,
+      "loss": 0.1103,
+      "step": 1065
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.28599898876688706,
+      "learning_rate": 3.0430351802512698e-05,
+      "loss": 0.1239,
+      "step": 1066
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.21709973608187036,
+      "learning_rate": 3.039715865267393e-05,
+      "loss": 0.1131,
+      "step": 1067
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.36484590367122105,
+      "learning_rate": 3.0363955520298742e-05,
+      "loss": 0.1175,
+      "step": 1068
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.2677757053296887,
+      "learning_rate": 3.033074246679935e-05,
+      "loss": 0.1274,
+      "step": 1069
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3321054682918382,
+      "learning_rate": 3.029751955360633e-05,
+      "loss": 0.1317,
+      "step": 1070
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.2735787405587081,
+      "learning_rate": 3.0264286842168477e-05,
+      "loss": 0.1453,
+      "step": 1071
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3377592622865575,
+      "learning_rate": 3.0231044393952712e-05,
+      "loss": 0.1161,
+      "step": 1072
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.2480888698730289,
+      "learning_rate": 3.0197792270443982e-05,
+      "loss": 0.1213,
+      "step": 1073
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.2970693989529065,
+      "learning_rate": 3.0164530533145123e-05,
+      "loss": 0.1207,
+      "step": 1074
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3020722253402678,
+      "learning_rate": 3.0131259243576726e-05,
+      "loss": 0.1112,
+      "step": 1075
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3199004682002101,
+      "learning_rate": 3.0097978463277076e-05,
+      "loss": 0.1335,
+      "step": 1076
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.25367230103958316,
+      "learning_rate": 3.0064688253802026e-05,
+      "loss": 0.1105,
+      "step": 1077
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.2995525849540768,
+      "learning_rate": 3.0031388676724836e-05,
+      "loss": 0.1178,
+      "step": 1078
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.2724705120009308,
+      "learning_rate": 2.9998079793636098e-05,
+      "loss": 0.1175,
+      "step": 1079
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.31645188538642155,
+      "learning_rate": 2.996476166614364e-05,
+      "loss": 0.1125,
+      "step": 1080
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.2851255622986633,
+      "learning_rate": 2.9931434355872367e-05,
+      "loss": 0.1247,
+      "step": 1081
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.22285159118444253,
+      "learning_rate": 2.989809792446417e-05,
+      "loss": 0.1284,
+      "step": 1082
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.232061534427358,
+      "learning_rate": 2.986475243357782e-05,
+      "loss": 0.0959,
+      "step": 1083
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.42560281537665967,
+      "learning_rate": 2.9831397944888833e-05,
+      "loss": 0.1239,
+      "step": 1084
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23139874600099342,
+      "learning_rate": 2.9798034520089385e-05,
+      "loss": 0.1184,
+      "step": 1085
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3535465364325666,
+      "learning_rate": 2.9764662220888157e-05,
+      "loss": 0.1207,
+      "step": 1086
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.21809165880011447,
+      "learning_rate": 2.9731281109010256e-05,
+      "loss": 0.1153,
+      "step": 1087
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23611763149701287,
+      "learning_rate": 2.9697891246197097e-05,
+      "loss": 0.1257,
+      "step": 1088
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3890412327836164,
+      "learning_rate": 2.966449269420627e-05,
+      "loss": 0.1224,
+      "step": 1089
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.2691356136931359,
+      "learning_rate": 2.9631085514811423e-05,
+      "loss": 0.1081,
+      "step": 1090
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.2522699730109621,
+      "learning_rate": 2.9597669769802196e-05,
+      "loss": 0.1279,
+      "step": 1091
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.2102601762941481,
+      "learning_rate": 2.9564245520984047e-05,
+      "loss": 0.1272,
+      "step": 1092
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.31385001172767035,
+      "learning_rate": 2.9530812830178162e-05,
+      "loss": 0.1206,
+      "step": 1093
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.22392026411544225,
+      "learning_rate": 2.9497371759221347e-05,
+      "loss": 0.1173,
+      "step": 1094
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.2115007017934818,
+      "learning_rate": 2.9463922369965917e-05,
+      "loss": 0.1285,
+      "step": 1095
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.2729071204894609,
+      "learning_rate": 2.943046472427956e-05,
+      "loss": 0.1354,
+      "step": 1096
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.21977956296958467,
+      "learning_rate": 2.9396998884045235e-05,
+      "loss": 0.1317,
+      "step": 1097
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23783260591333408,
+      "learning_rate": 2.936352491116106e-05,
+      "loss": 0.1172,
+      "step": 1098
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3612722099611281,
+      "learning_rate": 2.93300428675402e-05,
+      "loss": 0.1379,
+      "step": 1099
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.21289577312601732,
+      "learning_rate": 2.929655281511075e-05,
+      "loss": 0.1139,
+      "step": 1100
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.33316190267221624,
+      "learning_rate": 2.92630548158156e-05,
+      "loss": 0.1273,
+      "step": 1101
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.28082459846208485,
+      "learning_rate": 2.9229548931612348e-05,
+      "loss": 0.1228,
+      "step": 1102
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.22018460826158645,
+      "learning_rate": 2.9196035224473196e-05,
+      "loss": 0.1313,
+      "step": 1103
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.2378298760062803,
+      "learning_rate": 2.916251375638478e-05,
+      "loss": 0.1261,
+      "step": 1104
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.2035535489704069,
+      "learning_rate": 2.9128984589348114e-05,
+      "loss": 0.1091,
+      "step": 1105
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23802775583283856,
+      "learning_rate": 2.9095447785378443e-05,
+      "loss": 0.1218,
+      "step": 1106
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.2308604501914174,
+      "learning_rate": 2.9061903406505154e-05,
+      "loss": 0.1108,
+      "step": 1107
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.24785248735402737,
+      "learning_rate": 2.9028351514771606e-05,
+      "loss": 0.1139,
+      "step": 1108
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6336009026675509,
+      "learning_rate": 2.899479217223509e-05,
+      "loss": 0.1412,
+      "step": 1109
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.2964232846716362,
+      "learning_rate": 2.896122544096668e-05,
+      "loss": 0.1415,
+      "step": 1110
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.22803524776342365,
+      "learning_rate": 2.892765138305108e-05,
+      "loss": 0.1233,
+      "step": 1111
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.242480189897841,
+      "learning_rate": 2.8894070060586576e-05,
+      "loss": 0.1232,
+      "step": 1112
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.23264585414163133,
+      "learning_rate": 2.886048153568489e-05,
+      "loss": 0.1224,
+      "step": 1113
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.23010906556881042,
+      "learning_rate": 2.8826885870471043e-05,
+      "loss": 0.1154,
+      "step": 1114
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.36352973744401407,
+      "learning_rate": 2.8793283127083292e-05,
+      "loss": 0.1356,
+      "step": 1115
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.26544501978196355,
+      "learning_rate": 2.875967336767296e-05,
+      "loss": 0.1467,
+      "step": 1116
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.2502253400959257,
+      "learning_rate": 2.872605665440436e-05,
+      "loss": 0.1393,
+      "step": 1117
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.23729019175956934,
+      "learning_rate": 2.869243304945467e-05,
+      "loss": 0.1198,
+      "step": 1118
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.2959450905995426,
+      "learning_rate": 2.8658802615013807e-05,
+      "loss": 0.1304,
+      "step": 1119
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.24065213367264687,
+      "learning_rate": 2.862516541328431e-05,
+      "loss": 0.1201,
+      "step": 1120
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.31126433855388785,
+      "learning_rate": 2.859152150648126e-05,
+      "loss": 0.1401,
+      "step": 1121
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.227139335120596,
+      "learning_rate": 2.8557870956832132e-05,
+      "loss": 0.1175,
+      "step": 1122
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.2831801297922102,
+      "learning_rate": 2.8524213826576667e-05,
+      "loss": 0.1164,
+      "step": 1123
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.24670161780300998,
+      "learning_rate": 2.8490550177966797e-05,
+      "loss": 0.1185,
+      "step": 1124
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.2261263142662772,
+      "learning_rate": 2.845688007326651e-05,
+      "loss": 0.1261,
+      "step": 1125
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.23276170951341574,
+      "learning_rate": 2.8423203574751727e-05,
+      "loss": 0.1172,
+      "step": 1126
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.23398444884086755,
+      "learning_rate": 2.8389520744710196e-05,
+      "loss": 0.1255,
+      "step": 1127
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.20280662437168281,
+      "learning_rate": 2.8355831645441388e-05,
+      "loss": 0.1215,
+      "step": 1128
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.5496769946852627,
+      "learning_rate": 2.8322136339256356e-05,
+      "loss": 0.1228,
+      "step": 1129
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.18395133204823794,
+      "learning_rate": 2.8288434888477627e-05,
+      "loss": 0.1072,
+      "step": 1130
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.2129265656797621,
+      "learning_rate": 2.825472735543912e-05,
+      "loss": 0.1187,
+      "step": 1131
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2588266309406716,
+      "learning_rate": 2.8221013802485975e-05,
+      "loss": 0.1506,
+      "step": 1132
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.26004346021441604,
+      "learning_rate": 2.818729429197448e-05,
+      "loss": 0.1199,
+      "step": 1133
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2197495259903005,
+      "learning_rate": 2.815356888627195e-05,
+      "loss": 0.1218,
+      "step": 1134
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2515765349105225,
+      "learning_rate": 2.8119837647756574e-05,
+      "loss": 0.1204,
+      "step": 1135
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.24889106026741228,
+      "learning_rate": 2.808610063881737e-05,
+      "loss": 0.1079,
+      "step": 1136
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4672517181112952,
+      "learning_rate": 2.8052357921854e-05,
+      "loss": 0.1238,
+      "step": 1137
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4185387071787891,
+      "learning_rate": 2.8018609559276686e-05,
+      "loss": 0.1185,
+      "step": 1138
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2386780134672528,
+      "learning_rate": 2.7984855613506107e-05,
+      "loss": 0.121,
+      "step": 1139
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.46321444464317174,
+      "learning_rate": 2.795109614697326e-05,
+      "loss": 0.1181,
+      "step": 1140
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.23949571735138875,
+      "learning_rate": 2.7917331222119346e-05,
+      "loss": 0.1168,
+      "step": 1141
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3391417987756395,
+      "learning_rate": 2.7883560901395667e-05,
+      "loss": 0.1225,
+      "step": 1142
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.8281725562051542,
+      "learning_rate": 2.7849785247263515e-05,
+      "loss": 0.1169,
+      "step": 1143
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2628596105480252,
+      "learning_rate": 2.781600432219404e-05,
+      "loss": 0.1263,
+      "step": 1144
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.30935511217969053,
+      "learning_rate": 2.7782218188668118e-05,
+      "loss": 0.1177,
+      "step": 1145
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2271926028625351,
+      "learning_rate": 2.7748426909176307e-05,
+      "loss": 0.1172,
+      "step": 1146
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.25927098400437154,
+      "learning_rate": 2.7714630546218635e-05,
+      "loss": 0.1299,
+      "step": 1147
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.26879581061171065,
+      "learning_rate": 2.7680829162304567e-05,
+      "loss": 0.1176,
+      "step": 1148
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.19729799794674666,
+      "learning_rate": 2.7647022819952835e-05,
+      "loss": 0.1116,
+      "step": 1149
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.21491977287703362,
+      "learning_rate": 2.761321158169134e-05,
+      "loss": 0.1088,
+      "step": 1150
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.253316406813515,
+      "learning_rate": 2.757939551005706e-05,
+      "loss": 0.1218,
+      "step": 1151
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.18922158992116975,
+      "learning_rate": 2.754557466759589e-05,
+      "loss": 0.1276,
+      "step": 1152
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.20894741104288572,
+      "learning_rate": 2.7511749116862558e-05,
+      "loss": 0.1265,
+      "step": 1153
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.31832522472455277,
+      "learning_rate": 2.7477918920420504e-05,
+      "loss": 0.1146,
+      "step": 1154
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.20593564771529585,
+      "learning_rate": 2.7444084140841765e-05,
+      "loss": 0.1199,
+      "step": 1155
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.19596197884612868,
+      "learning_rate": 2.741024484070682e-05,
+      "loss": 0.0994,
+      "step": 1156
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.29371207722068565,
+      "learning_rate": 2.7376401082604564e-05,
+      "loss": 0.1101,
+      "step": 1157
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.2596858671237982,
+      "learning_rate": 2.7342552929132104e-05,
+      "loss": 0.124,
+      "step": 1158
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.22838988087201478,
+      "learning_rate": 2.7308700442894675e-05,
+      "loss": 0.1271,
+      "step": 1159
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3108620567759173,
+      "learning_rate": 2.7274843686505536e-05,
+      "loss": 0.1331,
+      "step": 1160
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.2689427839799072,
+      "learning_rate": 2.724098272258584e-05,
+      "loss": 0.1162,
+      "step": 1161
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.22670588619162219,
+      "learning_rate": 2.720711761376452e-05,
+      "loss": 0.1145,
+      "step": 1162
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.26253740157891037,
+      "learning_rate": 2.71732484226782e-05,
+      "loss": 0.1124,
+      "step": 1163
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.251392475490188,
+      "learning_rate": 2.7139375211970996e-05,
+      "loss": 0.1226,
+      "step": 1164
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6815378938852352,
+      "learning_rate": 2.710549804429452e-05,
+      "loss": 0.142,
+      "step": 1165
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.2564452407351033,
+      "learning_rate": 2.7071616982307684e-05,
+      "loss": 0.1119,
+      "step": 1166
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.2242703061682597,
+      "learning_rate": 2.7037732088676582e-05,
+      "loss": 0.1285,
+      "step": 1167
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.24974035192574656,
+      "learning_rate": 2.7003843426074416e-05,
+      "loss": 0.1247,
+      "step": 1168
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.1942412335191429,
+      "learning_rate": 2.6969951057181358e-05,
+      "loss": 0.1113,
+      "step": 1169
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.22754519352145877,
+      "learning_rate": 2.693605504468443e-05,
+      "loss": 0.1117,
+      "step": 1170
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.2367032849006637,
+      "learning_rate": 2.6902155451277377e-05,
+      "loss": 0.1253,
+      "step": 1171
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.257792837582943,
+      "learning_rate": 2.686825233966061e-05,
+      "loss": 0.1182,
+      "step": 1172
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.34125600233324455,
+      "learning_rate": 2.6834345772541002e-05,
+      "loss": 0.119,
+      "step": 1173
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.20067289920020595,
+      "learning_rate": 2.6800435812631854e-05,
+      "loss": 0.1198,
+      "step": 1174
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.40174072025149443,
+      "learning_rate": 2.6766522522652704e-05,
+      "loss": 0.1158,
+      "step": 1175
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.21033137724924414,
+      "learning_rate": 2.6732605965329283e-05,
+      "loss": 0.1139,
+      "step": 1176
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.24129314014216963,
+      "learning_rate": 2.6698686203393354e-05,
+      "loss": 0.1234,
+      "step": 1177
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.22786304185613926,
+      "learning_rate": 2.6664763299582602e-05,
+      "loss": 0.1181,
+      "step": 1178
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3905620004487617,
+      "learning_rate": 2.6630837316640523e-05,
+      "loss": 0.1306,
+      "step": 1179
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.25196341840091707,
+      "learning_rate": 2.659690831731631e-05,
+      "loss": 0.1194,
+      "step": 1180
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.22974888555002212,
+      "learning_rate": 2.656297636436475e-05,
+      "loss": 0.1262,
+      "step": 1181
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.22481129389897467,
+      "learning_rate": 2.652904152054607e-05,
+      "loss": 0.1275,
+      "step": 1182
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.20954901821647043,
+      "learning_rate": 2.649510384862586e-05,
+      "loss": 0.1212,
+      "step": 1183
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.28738396763254137,
+      "learning_rate": 2.6461163411374923e-05,
+      "loss": 0.1105,
+      "step": 1184
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.23116940410113643,
+      "learning_rate": 2.6427220271569203e-05,
+      "loss": 0.1329,
+      "step": 1185
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.0204446048737257,
+      "learning_rate": 2.6393274491989617e-05,
+      "loss": 0.123,
+      "step": 1186
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.24558302114980415,
+      "learning_rate": 2.6359326135421986e-05,
+      "loss": 0.1151,
+      "step": 1187
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.29339767073912726,
+      "learning_rate": 2.632537526465687e-05,
+      "loss": 0.132,
+      "step": 1188
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.23650992655485673,
+      "learning_rate": 2.629142194248952e-05,
+      "loss": 0.1041,
+      "step": 1189
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.2037466191646994,
+      "learning_rate": 2.625746623171968e-05,
+      "loss": 0.1063,
+      "step": 1190
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.26284881040281655,
+      "learning_rate": 2.622350819515153e-05,
+      "loss": 0.1147,
+      "step": 1191
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.2339115969126323,
+      "learning_rate": 2.6189547895593562e-05,
+      "loss": 0.1085,
+      "step": 1192
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.23723828638619437,
+      "learning_rate": 2.6155585395858435e-05,
+      "loss": 0.1089,
+      "step": 1193
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6357355864719927,
+      "learning_rate": 2.6121620758762877e-05,
+      "loss": 0.1205,
+      "step": 1194
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.26149531594011527,
+      "learning_rate": 2.6087654047127587e-05,
+      "loss": 0.1082,
+      "step": 1195
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.2754384154806975,
+      "learning_rate": 2.60536853237771e-05,
+      "loss": 0.121,
+      "step": 1196
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.25516192027882767,
+      "learning_rate": 2.6019714651539646e-05,
+      "loss": 0.1215,
+      "step": 1197
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.26336019660167964,
+      "learning_rate": 2.5985742093247078e-05,
+      "loss": 0.1131,
+      "step": 1198
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.2657872755443346,
+      "learning_rate": 2.5951767711734753e-05,
+      "loss": 0.11,
+      "step": 1199
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.2543363457963374,
+      "learning_rate": 2.591779156984137e-05,
+      "loss": 0.1135,
+      "step": 1200
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.19203660353338686,
+      "learning_rate": 2.5883813730408894e-05,
+      "loss": 0.114,
+      "step": 1201
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.197903285425131,
+      "learning_rate": 2.5849834256282447e-05,
+      "loss": 0.1103,
+      "step": 1202
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.21799251116688412,
+      "learning_rate": 2.5815853210310152e-05,
+      "loss": 0.1172,
+      "step": 1203
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.24951396330752126,
+      "learning_rate": 2.5781870655343045e-05,
+      "loss": 0.1108,
+      "step": 1204
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.2305220327597322,
+      "learning_rate": 2.5747886654234967e-05,
+      "loss": 0.1101,
+      "step": 1205
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.22702040539911192,
+      "learning_rate": 2.5713901269842404e-05,
+      "loss": 0.1283,
+      "step": 1206
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.23169522076468246,
+      "learning_rate": 2.5679914565024443e-05,
+      "loss": 0.1112,
+      "step": 1207
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3478380852183946,
+      "learning_rate": 2.564592660264258e-05,
+      "loss": 0.1073,
+      "step": 1208
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7326698823283713,
+      "learning_rate": 2.5611937445560634e-05,
+      "loss": 0.1049,
+      "step": 1209
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.24659716913927024,
+      "learning_rate": 2.5577947156644655e-05,
+      "loss": 0.1166,
+      "step": 1210
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.2639181299521172,
+      "learning_rate": 2.5543955798762798e-05,
+      "loss": 0.1129,
+      "step": 1211
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.2995614127138167,
+      "learning_rate": 2.550996343478514e-05,
+      "loss": 0.1248,
+      "step": 1212
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.2284565223979637,
+      "learning_rate": 2.5475970127583666e-05,
+      "loss": 0.1249,
+      "step": 1213
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.23187740418632194,
+      "learning_rate": 2.544197594003211e-05,
+      "loss": 0.1261,
+      "step": 1214
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.2027287542003588,
+      "learning_rate": 2.54079809350058e-05,
+      "loss": 0.1218,
+      "step": 1215
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.31015808480064766,
+      "learning_rate": 2.5373985175381594e-05,
+      "loss": 0.1251,
+      "step": 1216
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.26721317786704996,
+      "learning_rate": 2.533998872403775e-05,
+      "loss": 0.127,
+      "step": 1217
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.25782574904195965,
+      "learning_rate": 2.5305991643853806e-05,
+      "loss": 0.1135,
+      "step": 1218
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.2441616828164399,
+      "learning_rate": 2.527199399771044e-05,
+      "loss": 0.1253,
+      "step": 1219
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.23223410636136224,
+      "learning_rate": 2.523799584848942e-05,
+      "loss": 0.1336,
+      "step": 1220
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.1780605870853041,
+      "learning_rate": 2.52039972590734e-05,
+      "loss": 0.116,
+      "step": 1221
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.24116533413331767,
+      "learning_rate": 2.516999829234587e-05,
+      "loss": 0.1251,
+      "step": 1222
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3860254739551161,
+      "learning_rate": 2.5135999011191018e-05,
+      "loss": 0.1275,
+      "step": 1223
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.21371254284096586,
+      "learning_rate": 2.510199947849361e-05,
+      "loss": 0.1113,
+      "step": 1224
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.2106117806179564,
+      "learning_rate": 2.5067999757138884e-05,
+      "loss": 0.1116,
+      "step": 1225
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.2640446071140837,
+      "learning_rate": 2.5033999910012414e-05,
+      "loss": 0.1169,
+      "step": 1226
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.22585415113439986,
+      "learning_rate": 2.5e-05,
+      "loss": 0.122,
+      "step": 1227
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.17484842286097238,
+      "learning_rate": 2.4966000089987596e-05,
+      "loss": 0.1135,
+      "step": 1228
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2692517799444981,
+      "learning_rate": 2.4932000242861122e-05,
+      "loss": 0.1166,
+      "step": 1229
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.20392561800882372,
+      "learning_rate": 2.48980005215064e-05,
+      "loss": 0.1182,
+      "step": 1230
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.1914728488362021,
+      "learning_rate": 2.486400098880899e-05,
+      "loss": 0.1265,
+      "step": 1231
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.23107096485033862,
+      "learning_rate": 2.4830001707654134e-05,
+      "loss": 0.1327,
+      "step": 1232
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.20821535208191902,
+      "learning_rate": 2.479600274092661e-05,
+      "loss": 0.1259,
+      "step": 1233
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2638611915501704,
+      "learning_rate": 2.4762004151510584e-05,
+      "loss": 0.1139,
+      "step": 1234
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.27380124068072914,
+      "learning_rate": 2.4728006002289555e-05,
+      "loss": 0.1167,
+      "step": 1235
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2657854266571698,
+      "learning_rate": 2.4694008356146207e-05,
+      "loss": 0.1212,
+      "step": 1236
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.22579968372177545,
+      "learning_rate": 2.4660011275962258e-05,
+      "loss": 0.1202,
+      "step": 1237
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.21114114549231186,
+      "learning_rate": 2.4626014824618415e-05,
+      "loss": 0.1121,
+      "step": 1238
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2836672383661368,
+      "learning_rate": 2.4592019064994207e-05,
+      "loss": 0.1239,
+      "step": 1239
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3093955906778783,
+      "learning_rate": 2.45580240599679e-05,
+      "loss": 0.1106,
+      "step": 1240
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.25640124790145175,
+      "learning_rate": 2.4524029872416333e-05,
+      "loss": 0.1135,
+      "step": 1241
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2362338322797443,
+      "learning_rate": 2.4490036565214873e-05,
+      "loss": 0.1322,
+      "step": 1242
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.22699073590315796,
+      "learning_rate": 2.4456044201237215e-05,
+      "loss": 0.1085,
+      "step": 1243
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.17910866455095192,
+      "learning_rate": 2.4422052843355347e-05,
+      "loss": 0.1137,
+      "step": 1244
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.4138318740820604,
+      "learning_rate": 2.4388062554439372e-05,
+      "loss": 0.1435,
+      "step": 1245
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.22618093415878543,
+      "learning_rate": 2.4354073397357427e-05,
+      "loss": 0.1315,
+      "step": 1246
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.1962267488845416,
+      "learning_rate": 2.432008543497556e-05,
+      "loss": 0.1146,
+      "step": 1247
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.16747709429819374,
+      "learning_rate": 2.42860987301576e-05,
+      "loss": 0.1038,
+      "step": 1248
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.24776677216294182,
+      "learning_rate": 2.4252113345765046e-05,
+      "loss": 0.0979,
+      "step": 1249
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.1641724194669641,
+      "learning_rate": 2.4218129344656958e-05,
+      "loss": 0.1163,
+      "step": 1250
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.195783190135323,
+      "learning_rate": 2.418414678968985e-05,
+      "loss": 0.1222,
+      "step": 1251
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.1990004167488364,
+      "learning_rate": 2.4150165743717556e-05,
+      "loss": 0.1133,
+      "step": 1252
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.2226385766096943,
+      "learning_rate": 2.4116186269591105e-05,
+      "loss": 0.1145,
+      "step": 1253
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.19266384006947762,
+      "learning_rate": 2.408220843015864e-05,
+      "loss": 0.1128,
+      "step": 1254
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.20879389021303937,
+      "learning_rate": 2.4048232288265253e-05,
+      "loss": 0.1173,
+      "step": 1255
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7806212026199716,
+      "learning_rate": 2.4014257906752928e-05,
+      "loss": 0.1156,
+      "step": 1256
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.22327032163529717,
+      "learning_rate": 2.3980285348460363e-05,
+      "loss": 0.1267,
+      "step": 1257
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.2357064548606856,
+      "learning_rate": 2.3946314676222905e-05,
+      "loss": 0.1206,
+      "step": 1258
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.36675008163989425,
+      "learning_rate": 2.3912345952872416e-05,
+      "loss": 0.115,
+      "step": 1259
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.21966341651922291,
+      "learning_rate": 2.3878379241237136e-05,
+      "loss": 0.1279,
+      "step": 1260
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.23628054562445716,
+      "learning_rate": 2.384441460414158e-05,
+      "loss": 0.122,
+      "step": 1261
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.2278814531524817,
+      "learning_rate": 2.3810452104406444e-05,
+      "loss": 0.1319,
+      "step": 1262
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.1877171977466418,
+      "learning_rate": 2.3776491804848474e-05,
+      "loss": 0.1167,
+      "step": 1263
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.2049116855141644,
+      "learning_rate": 2.3742533768280322e-05,
+      "loss": 0.1121,
+      "step": 1264
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.2516311260780154,
+      "learning_rate": 2.370857805751048e-05,
+      "loss": 0.1153,
+      "step": 1265
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.6242123710616442,
+      "learning_rate": 2.3674624735343133e-05,
+      "loss": 0.1247,
+      "step": 1266
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.22828488299873198,
+      "learning_rate": 2.3640673864578023e-05,
+      "loss": 0.1132,
+      "step": 1267
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.24256852550127384,
+      "learning_rate": 2.360672550801039e-05,
+      "loss": 0.1211,
+      "step": 1268
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.2364278797547901,
+      "learning_rate": 2.35727797284308e-05,
+      "loss": 0.1062,
+      "step": 1269
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3839628819623954,
+      "learning_rate": 2.353883658862508e-05,
+      "loss": 0.1141,
+      "step": 1270
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.21694618556739156,
+      "learning_rate": 2.3504896151374144e-05,
+      "loss": 0.1013,
+      "step": 1271
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.29103637379455755,
+      "learning_rate": 2.3470958479453938e-05,
+      "loss": 0.1055,
+      "step": 1272
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.35752555413003967,
+      "learning_rate": 2.3437023635635254e-05,
+      "loss": 0.1091,
+      "step": 1273
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.2300945359916466,
+      "learning_rate": 2.34030916826837e-05,
+      "loss": 0.1039,
+      "step": 1274
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.1601955363389646,
+      "learning_rate": 2.3369162683359486e-05,
+      "loss": 0.1168,
+      "step": 1275
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5264723941870013,
+      "learning_rate": 2.3335236700417404e-05,
+      "loss": 0.1173,
+      "step": 1276
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.7540131975571011,
+      "learning_rate": 2.3301313796606652e-05,
+      "loss": 0.1035,
+      "step": 1277
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.2435458952610069,
+      "learning_rate": 2.326739403467073e-05,
+      "loss": 0.1204,
+      "step": 1278
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.27769275335670135,
+      "learning_rate": 2.3233477477347305e-05,
+      "loss": 0.1228,
+      "step": 1279
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.38849368407349677,
+      "learning_rate": 2.3199564187368156e-05,
+      "loss": 0.1234,
+      "step": 1280
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.2328844438795492,
+      "learning_rate": 2.3165654227459004e-05,
+      "loss": 0.1217,
+      "step": 1281
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3507158459482223,
+      "learning_rate": 2.3131747660339394e-05,
+      "loss": 0.1125,
+      "step": 1282
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.313981634121183,
+      "learning_rate": 2.309784454872262e-05,
+      "loss": 0.1394,
+      "step": 1283
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.23970977762448978,
+      "learning_rate": 2.3063944955315584e-05,
+      "loss": 0.1196,
+      "step": 1284
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.24858560296842813,
+      "learning_rate": 2.303004894281865e-05,
+      "loss": 0.1158,
+      "step": 1285
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.2744291356777743,
+      "learning_rate": 2.299615657392559e-05,
+      "loss": 0.1357,
+      "step": 1286
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.1954219223174465,
+      "learning_rate": 2.296226791132342e-05,
+      "loss": 0.1193,
+      "step": 1287
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.21197804140679902,
+      "learning_rate": 2.2928383017692322e-05,
+      "loss": 0.122,
+      "step": 1288
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.36756467755756184,
+      "learning_rate": 2.2894501955705477e-05,
+      "loss": 0.1165,
+      "step": 1289
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.39516688040039294,
+      "learning_rate": 2.2860624788029013e-05,
+      "loss": 0.1237,
+      "step": 1290
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3265240723739007,
+      "learning_rate": 2.2826751577321813e-05,
+      "loss": 0.123,
+      "step": 1291
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.2344457601311995,
+      "learning_rate": 2.2792882386235485e-05,
+      "loss": 0.1224,
+      "step": 1292
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.4593398271489293,
+      "learning_rate": 2.2759017277414166e-05,
+      "loss": 0.1157,
+      "step": 1293
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.33728694015725075,
+      "learning_rate": 2.2725156313494466e-05,
+      "loss": 0.1167,
+      "step": 1294
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.2582934736719085,
+      "learning_rate": 2.2691299557105328e-05,
+      "loss": 0.1085,
+      "step": 1295
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.2288517618231441,
+      "learning_rate": 2.2657447070867902e-05,
+      "loss": 0.1055,
+      "step": 1296
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.20236781567125628,
+      "learning_rate": 2.2623598917395438e-05,
+      "loss": 0.1087,
+      "step": 1297
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.33514236873808645,
+      "learning_rate": 2.258975515929318e-05,
+      "loss": 0.1289,
+      "step": 1298
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.25439243723594285,
+      "learning_rate": 2.2555915859158244e-05,
+      "loss": 0.1108,
+      "step": 1299
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.2298070509825204,
+      "learning_rate": 2.2522081079579498e-05,
+      "loss": 0.1162,
+      "step": 1300
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.2780026600630138,
+      "learning_rate": 2.2488250883137445e-05,
+      "loss": 0.1208,
+      "step": 1301
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.20845127125818602,
+      "learning_rate": 2.2454425332404122e-05,
+      "loss": 0.1129,
+      "step": 1302
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.21170635354361705,
+      "learning_rate": 2.2420604489942946e-05,
+      "loss": 0.1117,
+      "step": 1303
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.2119737621631178,
+      "learning_rate": 2.238678841830867e-05,
+      "loss": 0.1216,
+      "step": 1304
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.4803035105337743,
+      "learning_rate": 2.2352977180047175e-05,
+      "loss": 0.1164,
+      "step": 1305
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.4734221555931778,
+      "learning_rate": 2.2319170837695435e-05,
+      "loss": 0.1171,
+      "step": 1306
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.9003589201467294,
+      "learning_rate": 2.2285369453781364e-05,
+      "loss": 0.1267,
+      "step": 1307
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.33599594388447707,
+      "learning_rate": 2.2251573090823706e-05,
+      "loss": 0.118,
+      "step": 1308
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.32796918170779976,
+      "learning_rate": 2.2217781811331885e-05,
+      "loss": 0.124,
+      "step": 1309
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.261776328571547,
+      "learning_rate": 2.218399567780597e-05,
+      "loss": 0.1177,
+      "step": 1310
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.24399688699034153,
+      "learning_rate": 2.2150214752736488e-05,
+      "loss": 0.1193,
+      "step": 1311
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.21170123409465205,
+      "learning_rate": 2.211643909860433e-05,
+      "loss": 0.1154,
+      "step": 1312
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6101070680792533,
+      "learning_rate": 2.2082668777880653e-05,
+      "loss": 0.1351,
+      "step": 1313
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.24352423776241355,
+      "learning_rate": 2.2048903853026745e-05,
+      "loss": 0.1173,
+      "step": 1314
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.238218077534471,
+      "learning_rate": 2.2015144386493896e-05,
+      "loss": 0.1097,
+      "step": 1315
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.30911673360695435,
+      "learning_rate": 2.1981390440723316e-05,
+      "loss": 0.1085,
+      "step": 1316
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5635783628598078,
+      "learning_rate": 2.1947642078146004e-05,
+      "loss": 0.1276,
+      "step": 1317
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.24192927035783535,
+      "learning_rate": 2.1913899361182632e-05,
+      "loss": 0.1348,
+      "step": 1318
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3293731070214771,
+      "learning_rate": 2.1880162352243425e-05,
+      "loss": 0.1124,
+      "step": 1319
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.27459101342675146,
+      "learning_rate": 2.1846431113728064e-05,
+      "loss": 0.1113,
+      "step": 1320
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.294727151591201,
+      "learning_rate": 2.1812705708025526e-05,
+      "loss": 0.1046,
+      "step": 1321
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.19235854347708684,
+      "learning_rate": 2.1778986197514034e-05,
+      "loss": 0.1037,
+      "step": 1322
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.27789857704530874,
+      "learning_rate": 2.1745272644560885e-05,
+      "loss": 0.1231,
+      "step": 1323
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4501500849607945,
+      "learning_rate": 2.1711565111522372e-05,
+      "loss": 0.1302,
+      "step": 1324
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3197741981355104,
+      "learning_rate": 2.167786366074365e-05,
+      "loss": 0.1135,
+      "step": 1325
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.21075695345694428,
+      "learning_rate": 2.164416835455862e-05,
+      "loss": 0.1089,
+      "step": 1326
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.2569698899078435,
+      "learning_rate": 2.161047925528981e-05,
+      "loss": 0.1301,
+      "step": 1327
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.24702689478444836,
+      "learning_rate": 2.157679642524828e-05,
+      "loss": 0.114,
+      "step": 1328
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.23249162295619394,
+      "learning_rate": 2.1543119926733495e-05,
+      "loss": 0.1204,
+      "step": 1329
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.25243344374159715,
+      "learning_rate": 2.1509449822033205e-05,
+      "loss": 0.1255,
+      "step": 1330
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3827523044864369,
+      "learning_rate": 2.1475786173423335e-05,
+      "loss": 0.1225,
+      "step": 1331
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.33701889143239094,
+      "learning_rate": 2.1442129043167874e-05,
+      "loss": 0.1208,
+      "step": 1332
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.22150016075818965,
+      "learning_rate": 2.1408478493518742e-05,
+      "loss": 0.1128,
+      "step": 1333
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.2863883584433789,
+      "learning_rate": 2.137483458671569e-05,
+      "loss": 0.1107,
+      "step": 1334
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.24520176760550677,
+      "learning_rate": 2.13411973849862e-05,
+      "loss": 0.1212,
+      "step": 1335
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.25408594434307324,
+      "learning_rate": 2.1307566950545333e-05,
+      "loss": 0.1214,
+      "step": 1336
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.26532467470969856,
+      "learning_rate": 2.1273943345595637e-05,
+      "loss": 0.1224,
+      "step": 1337
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.26367072084636856,
+      "learning_rate": 2.1240326632327044e-05,
+      "loss": 0.111,
+      "step": 1338
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.504261018556106,
+      "learning_rate": 2.1206716872916714e-05,
+      "loss": 0.1094,
+      "step": 1339
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.27318136161607304,
+      "learning_rate": 2.117311412952896e-05,
+      "loss": 0.1212,
+      "step": 1340
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.26832866688945894,
+      "learning_rate": 2.1139518464315116e-05,
+      "loss": 0.1194,
+      "step": 1341
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.21857286089380346,
+      "learning_rate": 2.110592993941342e-05,
+      "loss": 0.1289,
+      "step": 1342
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.23368216029997166,
+      "learning_rate": 2.107234861694892e-05,
+      "loss": 0.1014,
+      "step": 1343
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.2433088874602415,
+      "learning_rate": 2.1038774559033327e-05,
+      "loss": 0.1166,
+      "step": 1344
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.33434995520395316,
+      "learning_rate": 2.1005207827764915e-05,
+      "loss": 0.126,
+      "step": 1345
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.6742346136949028,
+      "learning_rate": 2.09716484852284e-05,
+      "loss": 0.1191,
+      "step": 1346
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2421315651339868,
+      "learning_rate": 2.0938096593494855e-05,
+      "loss": 0.1166,
+      "step": 1347
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.24066391951905025,
+      "learning_rate": 2.090455221462156e-05,
+      "loss": 0.1188,
+      "step": 1348
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3361373504763993,
+      "learning_rate": 2.0871015410651888e-05,
+      "loss": 0.136,
+      "step": 1349
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.23518401560436872,
+      "learning_rate": 2.0837486243615227e-05,
+      "loss": 0.1082,
+      "step": 1350
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3218599139614129,
+      "learning_rate": 2.080396477552681e-05,
+      "loss": 0.1286,
+      "step": 1351
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.23953385594600327,
+      "learning_rate": 2.0770451068387654e-05,
+      "loss": 0.1232,
+      "step": 1352
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.21268717338313844,
+      "learning_rate": 2.0736945184184405e-05,
+      "loss": 0.1235,
+      "step": 1353
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.4194972804794982,
+      "learning_rate": 2.070344718488925e-05,
+      "loss": 0.1181,
+      "step": 1354
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2176897055645799,
+      "learning_rate": 2.06699571324598e-05,
+      "loss": 0.1105,
+      "step": 1355
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2734833499550044,
+      "learning_rate": 2.0636475088838946e-05,
+      "loss": 0.1231,
+      "step": 1356
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.20931772995921372,
+      "learning_rate": 2.0603001115954774e-05,
+      "loss": 0.116,
+      "step": 1357
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.28892029743876707,
+      "learning_rate": 2.0569535275720445e-05,
+      "loss": 0.1133,
+      "step": 1358
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3085648408723292,
+      "learning_rate": 2.0536077630034086e-05,
+      "loss": 0.1264,
+      "step": 1359
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.23128910398385255,
+      "learning_rate": 2.0502628240778655e-05,
+      "loss": 0.1149,
+      "step": 1360
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.34896854467639793,
+      "learning_rate": 2.046918716982185e-05,
+      "loss": 0.1079,
+      "step": 1361
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.24657482923503202,
+      "learning_rate": 2.0435754479015962e-05,
+      "loss": 0.1064,
+      "step": 1362
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.24611849938328598,
+      "learning_rate": 2.040233023019781e-05,
+      "loss": 0.1086,
+      "step": 1363
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3323342740190771,
+      "learning_rate": 2.0368914485188582e-05,
+      "loss": 0.1139,
+      "step": 1364
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.24242567036275717,
+      "learning_rate": 2.033550730579374e-05,
+      "loss": 0.1083,
+      "step": 1365
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.256493307250058,
+      "learning_rate": 2.0302108753802906e-05,
+      "loss": 0.1175,
+      "step": 1366
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.23995581661161194,
+      "learning_rate": 2.0268718890989753e-05,
+      "loss": 0.1048,
+      "step": 1367
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3232438378142975,
+      "learning_rate": 2.0235337779111856e-05,
+      "loss": 0.1185,
+      "step": 1368
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3917237815432992,
+      "learning_rate": 2.0201965479910624e-05,
+      "loss": 0.1311,
+      "step": 1369
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.31087357823389106,
+      "learning_rate": 2.0168602055111173e-05,
+      "loss": 0.1195,
+      "step": 1370
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.31070294618049893,
+      "learning_rate": 2.0135247566422186e-05,
+      "loss": 0.1312,
+      "step": 1371
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.2231789344247939,
+      "learning_rate": 2.0101902075535834e-05,
+      "loss": 0.1196,
+      "step": 1372
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.2592994344809001,
+      "learning_rate": 2.0068565644127645e-05,
+      "loss": 0.1165,
+      "step": 1373
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.26994989322308877,
+      "learning_rate": 2.003523833385637e-05,
+      "loss": 0.1211,
+      "step": 1374
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.0649960647563947,
+      "learning_rate": 2.0001920206363908e-05,
+      "loss": 0.1211,
+      "step": 1375
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5800679007086713,
+      "learning_rate": 1.996861132327517e-05,
+      "loss": 0.1296,
+      "step": 1376
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.38330632709233026,
+      "learning_rate": 1.993531174619798e-05,
+      "loss": 0.1272,
+      "step": 1377
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.508314422182805,
+      "learning_rate": 1.990202153672292e-05,
+      "loss": 0.1397,
+      "step": 1378
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.23625551409146456,
+      "learning_rate": 1.9868740756423286e-05,
+      "loss": 0.1048,
+      "step": 1379
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3135869155249492,
+      "learning_rate": 1.983546946685489e-05,
+      "loss": 0.1074,
+      "step": 1380
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.24856256342741986,
+      "learning_rate": 1.980220772955602e-05,
+      "loss": 0.1106,
+      "step": 1381
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.35533508881735604,
+      "learning_rate": 1.976895560604729e-05,
+      "loss": 0.1011,
+      "step": 1382
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.28526770630878684,
+      "learning_rate": 1.973571315783153e-05,
+      "loss": 0.1149,
+      "step": 1383
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3158456839589542,
+      "learning_rate": 1.9702480446393677e-05,
+      "loss": 0.116,
+      "step": 1384
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.2987754394837732,
+      "learning_rate": 1.9669257533200654e-05,
+      "loss": 0.1062,
+      "step": 1385
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.2661152720805441,
+      "learning_rate": 1.9636044479701267e-05,
+      "loss": 0.1252,
+      "step": 1386
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3017902191177092,
+      "learning_rate": 1.9602841347326077e-05,
+      "loss": 0.1266,
+      "step": 1387
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.24096786616874982,
+      "learning_rate": 1.956964819748731e-05,
+      "loss": 0.112,
+      "step": 1388
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.26953905301244624,
+      "learning_rate": 1.9536465091578708e-05,
+      "loss": 0.1114,
+      "step": 1389
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.36492962922364197,
+      "learning_rate": 1.9503292090975457e-05,
+      "loss": 0.1035,
+      "step": 1390
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.28119009248250576,
+      "learning_rate": 1.9470129257034046e-05,
+      "loss": 0.1197,
+      "step": 1391
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.39320370677407646,
+      "learning_rate": 1.9436976651092144e-05,
+      "loss": 0.1163,
+      "step": 1392
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.29831753923841525,
+      "learning_rate": 1.9403834334468522e-05,
+      "loss": 0.1251,
+      "step": 1393
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.22125498907780722,
+      "learning_rate": 1.9370702368462906e-05,
+      "loss": 0.1362,
+      "step": 1394
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.2161741839924953,
+      "learning_rate": 1.9337580814355888e-05,
+      "loss": 0.1134,
+      "step": 1395
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.2582875686634503,
+      "learning_rate": 1.930446973340881e-05,
+      "loss": 0.108,
+      "step": 1396
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.4642030019388871,
+      "learning_rate": 1.927136918686362e-05,
+      "loss": 0.1282,
+      "step": 1397
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.25564097378468215,
+      "learning_rate": 1.9238279235942787e-05,
+      "loss": 0.119,
+      "step": 1398
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.2605880171542009,
+      "learning_rate": 1.9205199941849196e-05,
+      "loss": 0.1137,
+      "step": 1399
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3558697382209097,
+      "learning_rate": 1.9172131365766022e-05,
+      "loss": 0.1221,
+      "step": 1400
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.28320529954702145,
+      "learning_rate": 1.9139073568856588e-05,
+      "loss": 0.133,
+      "step": 1401
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.26683706058252443,
+      "learning_rate": 1.9106026612264316e-05,
+      "loss": 0.1162,
+      "step": 1402
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.26646906475064813,
+      "learning_rate": 1.9072990557112564e-05,
+      "loss": 0.1036,
+      "step": 1403
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3723395958108291,
+      "learning_rate": 1.9039965464504518e-05,
+      "loss": 0.1276,
+      "step": 1404
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.28372859638578674,
+      "learning_rate": 1.9006951395523084e-05,
+      "loss": 0.1201,
+      "step": 1405
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3287355402982934,
+      "learning_rate": 1.89739484112308e-05,
+      "loss": 0.114,
+      "step": 1406
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.2576061774834209,
+      "learning_rate": 1.8940956572669692e-05,
+      "loss": 0.1188,
+      "step": 1407
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.27333979853854207,
+      "learning_rate": 1.8907975940861166e-05,
+      "loss": 0.1179,
+      "step": 1408
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3002994093448025,
+      "learning_rate": 1.8875006576805914e-05,
+      "loss": 0.112,
+      "step": 1409
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.28992434239590303,
+      "learning_rate": 1.884204854148376e-05,
+      "loss": 0.1055,
+      "step": 1410
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.2060223479101002,
+      "learning_rate": 1.88091018958536e-05,
+      "loss": 0.1215,
+      "step": 1411
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.24943482715400636,
+      "learning_rate": 1.8776166700853253e-05,
+      "loss": 0.1088,
+      "step": 1412
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.32105888934469096,
+      "learning_rate": 1.8743243017399366e-05,
+      "loss": 0.1053,
+      "step": 1413
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.30939522100226724,
+      "learning_rate": 1.871033090638729e-05,
+      "loss": 0.1239,
+      "step": 1414
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.2660505477294054,
+      "learning_rate": 1.867743042869097e-05,
+      "loss": 0.1156,
+      "step": 1415
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.2399422158582084,
+      "learning_rate": 1.8644541645162834e-05,
+      "loss": 0.1234,
+      "step": 1416
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3509267689749243,
+      "learning_rate": 1.8611664616633675e-05,
+      "loss": 0.1071,
+      "step": 1417
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.2708229874702106,
+      "learning_rate": 1.8578799403912567e-05,
+      "loss": 0.1115,
+      "step": 1418
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.35911131622667297,
+      "learning_rate": 1.85459460677867e-05,
+      "loss": 0.1267,
+      "step": 1419
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.4995333401672427,
+      "learning_rate": 1.8513104669021314e-05,
+      "loss": 0.1244,
+      "step": 1420
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3007660058697796,
+      "learning_rate": 1.8480275268359582e-05,
+      "loss": 0.1098,
+      "step": 1421
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.4064954319587962,
+      "learning_rate": 1.8447457926522454e-05,
+      "loss": 0.1373,
+      "step": 1422
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.23752316571476725,
+      "learning_rate": 1.8414652704208583e-05,
+      "loss": 0.1095,
+      "step": 1423
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.326295142778755,
+      "learning_rate": 1.8381859662094236e-05,
+      "loss": 0.1373,
+      "step": 1424
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.2769042398160702,
+      "learning_rate": 1.8349078860833123e-05,
+      "loss": 0.1262,
+      "step": 1425
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7288144759977627,
+      "learning_rate": 1.831631036105632e-05,
+      "loss": 0.1112,
+      "step": 1426
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3532834955441419,
+      "learning_rate": 1.8283554223372156e-05,
+      "loss": 0.1097,
+      "step": 1427
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.2936552454581024,
+      "learning_rate": 1.8250810508366077e-05,
+      "loss": 0.1102,
+      "step": 1428
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.35246011275777933,
+      "learning_rate": 1.8218079276600584e-05,
+      "loss": 0.1196,
+      "step": 1429
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3186278105852901,
+      "learning_rate": 1.8185360588615058e-05,
+      "loss": 0.0989,
+      "step": 1430
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.49598745541590106,
+      "learning_rate": 1.8152654504925692e-05,
+      "loss": 0.1198,
+      "step": 1431
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.37251272906564104,
+      "learning_rate": 1.8119961086025374e-05,
+      "loss": 0.1125,
+      "step": 1432
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.40207781014313854,
+      "learning_rate": 1.808728039238356e-05,
+      "loss": 0.1112,
+      "step": 1433
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.36331345704264006,
+      "learning_rate": 1.805461248444616e-05,
+      "loss": 0.1341,
+      "step": 1434
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.36798146333989895,
+      "learning_rate": 1.8021957422635444e-05,
+      "loss": 0.0998,
+      "step": 1435
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.48440205930427105,
+      "learning_rate": 1.7989315267349936e-05,
+      "loss": 0.1216,
+      "step": 1436
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6042560435766884,
+      "learning_rate": 1.795668607896426e-05,
+      "loss": 0.117,
+      "step": 1437
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.42891596655212383,
+      "learning_rate": 1.7924069917829074e-05,
+      "loss": 0.125,
+      "step": 1438
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.2610447587149842,
+      "learning_rate": 1.7891466844270956e-05,
+      "loss": 0.1171,
+      "step": 1439
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.650808721269614,
+      "learning_rate": 1.7858876918592236e-05,
+      "loss": 0.1352,
+      "step": 1440
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.9870462612828773,
+      "learning_rate": 1.782630020107096e-05,
+      "loss": 0.1111,
+      "step": 1441
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.45860101720536306,
+      "learning_rate": 1.779373675196073e-05,
+      "loss": 0.1279,
+      "step": 1442
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7152948768103583,
+      "learning_rate": 1.776118663149061e-05,
+      "loss": 0.0981,
+      "step": 1443
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.28506325082258954,
+      "learning_rate": 1.7728649899865024e-05,
+      "loss": 0.1066,
+      "step": 1444
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.2727761220971191,
+      "learning_rate": 1.7696126617263615e-05,
+      "loss": 0.1155,
+      "step": 1445
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.5450249421882601,
+      "learning_rate": 1.766361684384114e-05,
+      "loss": 0.1125,
+      "step": 1446
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.42343874090250117,
+      "learning_rate": 1.7631120639727393e-05,
+      "loss": 0.1109,
+      "step": 1447
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3028787576945943,
+      "learning_rate": 1.759863806502708e-05,
+      "loss": 0.115,
+      "step": 1448
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3907792883355111,
+      "learning_rate": 1.7566169179819657e-05,
+      "loss": 0.1236,
+      "step": 1449
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.4187948215033938,
+      "learning_rate": 1.75337140441593e-05,
+      "loss": 0.1057,
+      "step": 1450
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3242750579659611,
+      "learning_rate": 1.7501272718074737e-05,
+      "loss": 0.1218,
+      "step": 1451
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3393728250993398,
+      "learning_rate": 1.746884526156915e-05,
+      "loss": 0.1181,
+      "step": 1452
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3348290938332141,
+      "learning_rate": 1.7436431734620067e-05,
+      "loss": 0.1102,
+      "step": 1453
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.2669360376681576,
+      "learning_rate": 1.7404032197179264e-05,
+      "loss": 0.1073,
+      "step": 1454
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.2718588288723053,
+      "learning_rate": 1.737164670917264e-05,
+      "loss": 0.1084,
+      "step": 1455
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.2614260111093309,
+      "learning_rate": 1.73392753305001e-05,
+      "loss": 0.1121,
+      "step": 1456
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.46235822907393803,
+      "learning_rate": 1.730691812103546e-05,
+      "loss": 0.12,
+      "step": 1457
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.318202000564292,
+      "learning_rate": 1.7274575140626318e-05,
+      "loss": 0.1172,
+      "step": 1458
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.2317165046125462,
+      "learning_rate": 1.724224644909397e-05,
+      "loss": 0.1082,
+      "step": 1459
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.35258300361918216,
+      "learning_rate": 1.7209932106233263e-05,
+      "loss": 0.1082,
+      "step": 1460
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.25398676993334635,
+      "learning_rate": 1.717763217181253e-05,
+      "loss": 0.1176,
+      "step": 1461
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.4085383482525175,
+      "learning_rate": 1.714534670557344e-05,
+      "loss": 0.1097,
+      "step": 1462
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.2884573911976627,
+      "learning_rate": 1.7113075767230902e-05,
+      "loss": 0.106,
+      "step": 1463
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3757596009640975,
+      "learning_rate": 1.7080819416472944e-05,
+      "loss": 0.1302,
+      "step": 1464
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.24925307409023953,
+      "learning_rate": 1.7048577712960627e-05,
+      "loss": 0.1177,
+      "step": 1465
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.2758547940497246,
+      "learning_rate": 1.7016350716327932e-05,
+      "loss": 0.1276,
+      "step": 1466
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.2605936225414611,
+      "learning_rate": 1.698413848618161e-05,
+      "loss": 0.101,
+      "step": 1467
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3148685410810188,
+      "learning_rate": 1.6951941082101123e-05,
+      "loss": 0.1183,
+      "step": 1468
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7018712590147094,
+      "learning_rate": 1.6919758563638504e-05,
+      "loss": 0.1305,
+      "step": 1469
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.32528879539614963,
+      "learning_rate": 1.688759099031824e-05,
+      "loss": 0.1195,
+      "step": 1470
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.26931867259414616,
+      "learning_rate": 1.685543842163719e-05,
+      "loss": 0.1137,
+      "step": 1471
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.5594121279733659,
+      "learning_rate": 1.682330091706446e-05,
+      "loss": 0.1064,
+      "step": 1472
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.2801018045983819,
+      "learning_rate": 1.6791178536041297e-05,
+      "loss": 0.1182,
+      "step": 1473
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3107744028858906,
+      "learning_rate": 1.675907133798096e-05,
+      "loss": 0.1159,
+      "step": 1474
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.5066732923894974,
+      "learning_rate": 1.672697938226865e-05,
+      "loss": 0.1193,
+      "step": 1475
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.2668159942762188,
+      "learning_rate": 1.6694902728261337e-05,
+      "loss": 0.123,
+      "step": 1476
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3152176379161149,
+      "learning_rate": 1.666284143528774e-05,
+      "loss": 0.1202,
+      "step": 1477
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.4323266422726728,
+      "learning_rate": 1.663079556264812e-05,
+      "loss": 0.1097,
+      "step": 1478
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.4048641457935778,
+      "learning_rate": 1.6598765169614243e-05,
+      "loss": 0.1361,
+      "step": 1479
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.2542951610550642,
+      "learning_rate": 1.6566750315429254e-05,
+      "loss": 0.1087,
+      "step": 1480
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.23848024571388765,
+      "learning_rate": 1.6534751059307518e-05,
+      "loss": 0.1072,
+      "step": 1481
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.28357562881892073,
+      "learning_rate": 1.6502767460434588e-05,
+      "loss": 0.1134,
+      "step": 1482
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.42855366075200285,
+      "learning_rate": 1.6470799577967028e-05,
+      "loss": 0.1156,
+      "step": 1483
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3286483836878461,
+      "learning_rate": 1.643884747103236e-05,
+      "loss": 0.126,
+      "step": 1484
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3381825709614776,
+      "learning_rate": 1.6406911198728912e-05,
+      "loss": 0.1266,
+      "step": 1485
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3808753721331662,
+      "learning_rate": 1.637499082012574e-05,
+      "loss": 0.1217,
+      "step": 1486
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.35245487805539594,
+      "learning_rate": 1.6343086394262467e-05,
+      "loss": 0.1028,
+      "step": 1487
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3358890987621577,
+      "learning_rate": 1.6311197980149246e-05,
+      "loss": 0.0967,
+      "step": 1488
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3639619148922725,
+      "learning_rate": 1.6279325636766607e-05,
+      "loss": 0.1148,
+      "step": 1489
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7461685310656901,
+      "learning_rate": 1.6247469423065346e-05,
+      "loss": 0.1169,
+      "step": 1490
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.27545987334690974,
+      "learning_rate": 1.621562939796643e-05,
+      "loss": 0.1021,
+      "step": 1491
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.2899917216496325,
+      "learning_rate": 1.61838056203609e-05,
+      "loss": 0.1019,
+      "step": 1492
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.30531525126723985,
+      "learning_rate": 1.615199814910971e-05,
+      "loss": 0.1207,
+      "step": 1493
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.266628382326683,
+      "learning_rate": 1.6120207043043678e-05,
+      "loss": 0.1103,
+      "step": 1494
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.470428344309531,
+      "learning_rate": 1.6088432360963356e-05,
+      "loss": 0.1024,
+      "step": 1495
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.28196435986984164,
+      "learning_rate": 1.605667416163892e-05,
+      "loss": 0.1258,
+      "step": 1496
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.25313034876096574,
+      "learning_rate": 1.602493250381003e-05,
+      "loss": 0.1147,
+      "step": 1497
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.44968765386534,
+      "learning_rate": 1.59932074461858e-05,
+      "loss": 0.0965,
+      "step": 1498
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.27370883608047814,
+      "learning_rate": 1.596149904744459e-05,
+      "loss": 0.11,
+      "step": 1499
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.22181583701146146,
+      "learning_rate": 1.5929807366233977e-05,
+      "loss": 0.1159,
+      "step": 1500
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6684319461186016,
+      "learning_rate": 1.589813246117062e-05,
+      "loss": 0.1172,
+      "step": 1501
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3862417382568489,
+      "learning_rate": 1.5866474390840125e-05,
+      "loss": 0.1215,
+      "step": 1502
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.5839555242019041,
+      "learning_rate": 1.583483321379699e-05,
+      "loss": 0.1015,
+      "step": 1503
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7690930382539971,
+      "learning_rate": 1.5803208988564458e-05,
+      "loss": 0.1226,
+      "step": 1504
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4261050500004941,
+      "learning_rate": 1.577160177363439e-05,
+      "loss": 0.1155,
+      "step": 1505
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3781174739071445,
+      "learning_rate": 1.5740011627467223e-05,
+      "loss": 0.1109,
+      "step": 1506
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.36777557731221977,
+      "learning_rate": 1.5708438608491814e-05,
+      "loss": 0.1119,
+      "step": 1507
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.2960466924210014,
+      "learning_rate": 1.5676882775105328e-05,
+      "loss": 0.1159,
+      "step": 1508
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.34568822439983804,
+      "learning_rate": 1.5645344185673154e-05,
+      "loss": 0.1249,
+      "step": 1509
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.31200978926673667,
+      "learning_rate": 1.5613822898528796e-05,
+      "loss": 0.1099,
+      "step": 1510
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3497605757081781,
+      "learning_rate": 1.558231897197373e-05,
+      "loss": 0.1179,
+      "step": 1511
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.22341273029809544,
+      "learning_rate": 1.555083246427734e-05,
+      "loss": 0.1008,
+      "step": 1512
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.34042592278913464,
+      "learning_rate": 1.551936343367679e-05,
+      "loss": 0.1169,
+      "step": 1513
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.28034574612664825,
+      "learning_rate": 1.5487911938376924e-05,
+      "loss": 0.105,
+      "step": 1514
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.6205322851578078,
+      "learning_rate": 1.5456478036550143e-05,
+      "loss": 0.122,
+      "step": 1515
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5048877259075881,
+      "learning_rate": 1.5425061786336314e-05,
+      "loss": 0.1163,
+      "step": 1516
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.25273033681143076,
+      "learning_rate": 1.5393663245842643e-05,
+      "loss": 0.1373,
+      "step": 1517
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.289604901706915,
+      "learning_rate": 1.5362282473143603e-05,
+      "loss": 0.1144,
+      "step": 1518
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3225805396350536,
+      "learning_rate": 1.5330919526280775e-05,
+      "loss": 0.1233,
+      "step": 1519
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.2815352422373032,
+      "learning_rate": 1.5299574463262794e-05,
+      "loss": 0.114,
+      "step": 1520
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3098572317617986,
+      "learning_rate": 1.5268247342065215e-05,
+      "loss": 0.1118,
+      "step": 1521
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.26130066361374793,
+      "learning_rate": 1.5236938220630392e-05,
+      "loss": 0.1091,
+      "step": 1522
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4144891543637434,
+      "learning_rate": 1.5205647156867403e-05,
+      "loss": 0.1207,
+      "step": 1523
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3220398903188043,
+      "learning_rate": 1.5174374208651912e-05,
+      "loss": 0.1082,
+      "step": 1524
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.2609253846192566,
+      "learning_rate": 1.5143119433826092e-05,
+      "loss": 0.1097,
+      "step": 1525
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.33153837275315967,
+      "learning_rate": 1.5111882890198486e-05,
+      "loss": 0.1115,
+      "step": 1526
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.2873230616475593,
+      "learning_rate": 1.5080664635543934e-05,
+      "loss": 0.1155,
+      "step": 1527
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.33166636976931957,
+      "learning_rate": 1.5049464727603452e-05,
+      "loss": 0.1208,
+      "step": 1528
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3624680140049228,
+      "learning_rate": 1.5018283224084084e-05,
+      "loss": 0.1203,
+      "step": 1529
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.39985335294767993,
+      "learning_rate": 1.4987120182658878e-05,
+      "loss": 0.1132,
+      "step": 1530
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.46679109440823774,
+      "learning_rate": 1.495597566096671e-05,
+      "loss": 0.1089,
+      "step": 1531
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.28013824258586406,
+      "learning_rate": 1.492484971661221e-05,
+      "loss": 0.1311,
+      "step": 1532
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.2956879076085355,
+      "learning_rate": 1.4893742407165656e-05,
+      "loss": 0.1113,
+      "step": 1533
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3142682332816632,
+      "learning_rate": 1.4862653790162851e-05,
+      "loss": 0.114,
+      "step": 1534
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.2658375984751835,
+      "learning_rate": 1.4831583923104999e-05,
+      "loss": 0.1072,
+      "step": 1535
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.34529455855747276,
+      "learning_rate": 1.4800532863458665e-05,
+      "loss": 0.1022,
+      "step": 1536
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3240890097821605,
+      "learning_rate": 1.4769500668655616e-05,
+      "loss": 0.1199,
+      "step": 1537
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3728783379610864,
+      "learning_rate": 1.4738487396092715e-05,
+      "loss": 0.1275,
+      "step": 1538
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3790192776366776,
+      "learning_rate": 1.4707493103131832e-05,
+      "loss": 0.1098,
+      "step": 1539
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.5175248269890591,
+      "learning_rate": 1.4676517847099747e-05,
+      "loss": 0.1272,
+      "step": 1540
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.2840099592680525,
+      "learning_rate": 1.4645561685288005e-05,
+      "loss": 0.1124,
+      "step": 1541
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3408418209263317,
+      "learning_rate": 1.4614624674952842e-05,
+      "loss": 0.1145,
+      "step": 1542
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.23883057591278112,
+      "learning_rate": 1.4583706873315089e-05,
+      "loss": 0.1082,
+      "step": 1543
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.2790257503328562,
+      "learning_rate": 1.4552808337560025e-05,
+      "loss": 0.1063,
+      "step": 1544
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.30158235465322647,
+      "learning_rate": 1.4521929124837308e-05,
+      "loss": 0.1095,
+      "step": 1545
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7099381427677307,
+      "learning_rate": 1.4491069292260868e-05,
+      "loss": 0.1063,
+      "step": 1546
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.32476065278876814,
+      "learning_rate": 1.4460228896908754e-05,
+      "loss": 0.125,
+      "step": 1547
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.23961866074856614,
+      "learning_rate": 1.4429407995823094e-05,
+      "loss": 0.105,
+      "step": 1548
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.4471971892685387,
+      "learning_rate": 1.4398606646009963e-05,
+      "loss": 0.1172,
+      "step": 1549
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.44107139534290263,
+      "learning_rate": 1.4367824904439242e-05,
+      "loss": 0.1218,
+      "step": 1550
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.2920502173875538,
+      "learning_rate": 1.4337062828044576e-05,
+      "loss": 0.0977,
+      "step": 1551
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3796488752060125,
+      "learning_rate": 1.4306320473723234e-05,
+      "loss": 0.108,
+      "step": 1552
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.4685843275505185,
+      "learning_rate": 1.4275597898335982e-05,
+      "loss": 0.1084,
+      "step": 1553
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.2940072743324732,
+      "learning_rate": 1.4244895158707031e-05,
+      "loss": 0.1272,
+      "step": 1554
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3196830824956346,
+      "learning_rate": 1.4214212311623892e-05,
+      "loss": 0.1073,
+      "step": 1555
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.25600173675610494,
+      "learning_rate": 1.4183549413837289e-05,
+      "loss": 0.1183,
+      "step": 1556
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.37425067447070626,
+      "learning_rate": 1.4152906522061048e-05,
+      "loss": 0.1162,
+      "step": 1557
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3742535116488289,
+      "learning_rate": 1.4122283692971983e-05,
+      "loss": 0.1108,
+      "step": 1558
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.22726399180268853,
+      "learning_rate": 1.4091680983209793e-05,
+      "loss": 0.1156,
+      "step": 1559
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3535873390956821,
+      "learning_rate": 1.4061098449376985e-05,
+      "loss": 0.0962,
+      "step": 1560
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3760375663926004,
+      "learning_rate": 1.4030536148038742e-05,
+      "loss": 0.1141,
+      "step": 1561
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.2462778204130549,
+      "learning_rate": 1.3999994135722822e-05,
+      "loss": 0.1112,
+      "step": 1562
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.2714735745863698,
+      "learning_rate": 1.3969472468919461e-05,
+      "loss": 0.1065,
+      "step": 1563
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.30179611020096986,
+      "learning_rate": 1.393897120408127e-05,
+      "loss": 0.1057,
+      "step": 1564
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.30837504671091,
+      "learning_rate": 1.3908490397623103e-05,
+      "loss": 0.1116,
+      "step": 1565
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.4236366199773709,
+      "learning_rate": 1.387803010592198e-05,
+      "loss": 0.1119,
+      "step": 1566
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.2688241739407226,
+      "learning_rate": 1.3847590385316997e-05,
+      "loss": 0.1078,
+      "step": 1567
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.25362282283408305,
+      "learning_rate": 1.3817171292109183e-05,
+      "loss": 0.1043,
+      "step": 1568
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.2565482515566807,
+      "learning_rate": 1.3786772882561427e-05,
+      "loss": 0.1207,
+      "step": 1569
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.8207138620291213,
+      "learning_rate": 1.3756395212898359e-05,
+      "loss": 0.095,
+      "step": 1570
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.31149630978764403,
+      "learning_rate": 1.372603833930623e-05,
+      "loss": 0.1119,
+      "step": 1571
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3120682669306964,
+      "learning_rate": 1.3695702317932862e-05,
+      "loss": 0.1111,
+      "step": 1572
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.6193188623058115,
+      "learning_rate": 1.366538720488747e-05,
+      "loss": 0.1058,
+      "step": 1573
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3446570097868727,
+      "learning_rate": 1.3635093056240625e-05,
+      "loss": 0.109,
+      "step": 1574
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3183302843744505,
+      "learning_rate": 1.3604819928024109e-05,
+      "loss": 0.1066,
+      "step": 1575
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.26173534840360985,
+      "learning_rate": 1.3574567876230847e-05,
+      "loss": 0.1112,
+      "step": 1576
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.27679621659036463,
+      "learning_rate": 1.354433695681474e-05,
+      "loss": 0.1216,
+      "step": 1577
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.2768638286286012,
+      "learning_rate": 1.3514127225690637e-05,
+      "loss": 0.1168,
+      "step": 1578
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.27122355619872535,
+      "learning_rate": 1.3483938738734198e-05,
+      "loss": 0.1209,
+      "step": 1579
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3689923402502186,
+      "learning_rate": 1.3453771551781757e-05,
+      "loss": 0.0991,
+      "step": 1580
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.32384656435650344,
+      "learning_rate": 1.3423625720630281e-05,
+      "loss": 0.1181,
+      "step": 1581
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3333698505119549,
+      "learning_rate": 1.3393501301037245e-05,
+      "loss": 0.1146,
+      "step": 1582
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3200414529554192,
+      "learning_rate": 1.3363398348720486e-05,
+      "loss": 0.0997,
+      "step": 1583
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.41375642240148414,
+      "learning_rate": 1.3333316919358157e-05,
+      "loss": 0.1176,
+      "step": 1584
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.37303498926309253,
+      "learning_rate": 1.3303257068588612e-05,
+      "loss": 0.1309,
+      "step": 1585
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3246750232535579,
+      "learning_rate": 1.3273218852010289e-05,
+      "loss": 0.1147,
+      "step": 1586
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.25362406401172355,
+      "learning_rate": 1.324320232518158e-05,
+      "loss": 0.1089,
+      "step": 1587
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.49856290773472456,
+      "learning_rate": 1.3213207543620812e-05,
+      "loss": 0.1065,
+      "step": 1588
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.30006429803119433,
+      "learning_rate": 1.318323456280604e-05,
+      "loss": 0.1085,
+      "step": 1589
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.22362694105007974,
+      "learning_rate": 1.3153283438175034e-05,
+      "loss": 0.1314,
+      "step": 1590
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.28473043568256545,
+      "learning_rate": 1.3123354225125128e-05,
+      "loss": 0.1122,
+      "step": 1591
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3118667125498441,
+      "learning_rate": 1.3093446979013122e-05,
+      "loss": 0.0902,
+      "step": 1592
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.29216157386006797,
+      "learning_rate": 1.3063561755155204e-05,
+      "loss": 0.0926,
+      "step": 1593
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3258638604875521,
+      "learning_rate": 1.30336986088268e-05,
+      "loss": 0.1034,
+      "step": 1594
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.28667112200554473,
+      "learning_rate": 1.300385759526253e-05,
+      "loss": 0.1107,
+      "step": 1595
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.35931957956008137,
+      "learning_rate": 1.2974038769656054e-05,
+      "loss": 0.112,
+      "step": 1596
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.2605124177638872,
+      "learning_rate": 1.2944242187160016e-05,
+      "loss": 0.1164,
+      "step": 1597
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.40687988802019354,
+      "learning_rate": 1.2914467902885901e-05,
+      "loss": 0.1294,
+      "step": 1598
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3062772491071066,
+      "learning_rate": 1.2884715971903966e-05,
+      "loss": 0.1233,
+      "step": 1599
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3035223681843266,
+      "learning_rate": 1.2854986449243126e-05,
+      "loss": 0.1165,
+      "step": 1600
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3439054955104647,
+      "learning_rate": 1.2825279389890817e-05,
+      "loss": 0.1266,
+      "step": 1601
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.29551334736313717,
+      "learning_rate": 1.2795594848792975e-05,
+      "loss": 0.1166,
+      "step": 1602
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.2715259686474053,
+      "learning_rate": 1.2765932880853836e-05,
+      "loss": 0.1092,
+      "step": 1603
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.5493572959589863,
+      "learning_rate": 1.273629354093593e-05,
+      "loss": 0.123,
+      "step": 1604
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.26843199602800544,
+      "learning_rate": 1.2706676883859903e-05,
+      "loss": 0.1206,
+      "step": 1605
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3079866311666048,
+      "learning_rate": 1.267708296440448e-05,
+      "loss": 0.0983,
+      "step": 1606
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.9292283334835236,
+      "learning_rate": 1.2647511837306284e-05,
+      "loss": 0.107,
+      "step": 1607
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3437264036585929,
+      "learning_rate": 1.261796355725982e-05,
+      "loss": 0.1073,
+      "step": 1608
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.32296485498334476,
+      "learning_rate": 1.2588438178917333e-05,
+      "loss": 0.1146,
+      "step": 1609
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.814504053000743,
+      "learning_rate": 1.2558935756888675e-05,
+      "loss": 0.1227,
+      "step": 1610
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.555679180137121,
+      "learning_rate": 1.252945634574128e-05,
+      "loss": 0.1067,
+      "step": 1611
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5733448559784373,
+      "learning_rate": 1.2500000000000006e-05,
+      "loss": 0.107,
+      "step": 1612
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.2382316543315384,
+      "learning_rate": 1.247056677414703e-05,
+      "loss": 0.1032,
+      "step": 1613
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.2300978726128643,
+      "learning_rate": 1.2441156722621793e-05,
+      "loss": 0.101,
+      "step": 1614
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5326605214275718,
+      "learning_rate": 1.2411769899820863e-05,
+      "loss": 0.107,
+      "step": 1615
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.37488619292792824,
+      "learning_rate": 1.2382406360097856e-05,
+      "loss": 0.1201,
+      "step": 1616
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.28939010417534505,
+      "learning_rate": 1.2353066157763304e-05,
+      "loss": 0.1076,
+      "step": 1617
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.31119138530749063,
+      "learning_rate": 1.232374934708457e-05,
+      "loss": 0.1159,
+      "step": 1618
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.32549281223729787,
+      "learning_rate": 1.2294455982285777e-05,
+      "loss": 0.1176,
+      "step": 1619
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.6651553299838262,
+      "learning_rate": 1.2265186117547669e-05,
+      "loss": 0.1125,
+      "step": 1620
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.9599536619510136,
+      "learning_rate": 1.223593980700753e-05,
+      "loss": 0.1219,
+      "step": 1621
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.23752515171806463,
+      "learning_rate": 1.220671710475907e-05,
+      "loss": 0.1113,
+      "step": 1622
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.25742850085803515,
+      "learning_rate": 1.217751806485235e-05,
+      "loss": 0.1155,
+      "step": 1623
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3470033382801706,
+      "learning_rate": 1.2148342741293642e-05,
+      "loss": 0.1114,
+      "step": 1624
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.2525464552105098,
+      "learning_rate": 1.2119191188045356e-05,
+      "loss": 0.107,
+      "step": 1625
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.2404776993574496,
+      "learning_rate": 1.2090063459025955e-05,
+      "loss": 0.1068,
+      "step": 1626
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.2323257585495817,
+      "learning_rate": 1.2060959608109823e-05,
+      "loss": 0.1048,
+      "step": 1627
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.30585278808055494,
+      "learning_rate": 1.2031879689127187e-05,
+      "loss": 0.1147,
+      "step": 1628
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.34655898954253544,
+      "learning_rate": 1.2002823755864012e-05,
+      "loss": 0.1208,
+      "step": 1629
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.27980520940627923,
+      "learning_rate": 1.1973791862061872e-05,
+      "loss": 0.1139,
+      "step": 1630
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4679805402590242,
+      "learning_rate": 1.194478406141792e-05,
+      "loss": 0.1139,
+      "step": 1631
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3997424752314078,
+      "learning_rate": 1.1915800407584704e-05,
+      "loss": 0.1248,
+      "step": 1632
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5119389127345466,
+      "learning_rate": 1.1886840954170142e-05,
+      "loss": 0.1203,
+      "step": 1633
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.45404486429691815,
+      "learning_rate": 1.185790575473738e-05,
+      "loss": 0.1005,
+      "step": 1634
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.31443642159777163,
+      "learning_rate": 1.1828994862804716e-05,
+      "loss": 0.1096,
+      "step": 1635
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.32784240190031183,
+      "learning_rate": 1.1800108331845455e-05,
+      "loss": 0.1225,
+      "step": 1636
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.31921912537838293,
+      "learning_rate": 1.1771246215287877e-05,
+      "loss": 0.1058,
+      "step": 1637
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4894247823337614,
+      "learning_rate": 1.1742408566515103e-05,
+      "loss": 0.103,
+      "step": 1638
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3364910964023013,
+      "learning_rate": 1.1713595438864974e-05,
+      "loss": 0.1166,
+      "step": 1639
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3431328749025995,
+      "learning_rate": 1.1684806885630004e-05,
+      "loss": 0.1312,
+      "step": 1640
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.271166130323169,
+      "learning_rate": 1.165604296005725e-05,
+      "loss": 0.1188,
+      "step": 1641
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3689151650024361,
+      "learning_rate": 1.1627303715348195e-05,
+      "loss": 0.1068,
+      "step": 1642
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3445268710971946,
+      "learning_rate": 1.1598589204658697e-05,
+      "loss": 0.1113,
+      "step": 1643
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.42108196121558983,
+      "learning_rate": 1.1569899481098859e-05,
+      "loss": 0.1146,
+      "step": 1644
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.2241917853893143,
+      "learning_rate": 1.1541234597732947e-05,
+      "loss": 0.1206,
+      "step": 1645
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.25043241958840295,
+      "learning_rate": 1.1512594607579255e-05,
+      "loss": 0.1024,
+      "step": 1646
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3030377988027378,
+      "learning_rate": 1.148397956361007e-05,
+      "loss": 0.109,
+      "step": 1647
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.2896105571084604,
+      "learning_rate": 1.1455389518751502e-05,
+      "loss": 0.1277,
+      "step": 1648
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4435855888118723,
+      "learning_rate": 1.1426824525883459e-05,
+      "loss": 0.1172,
+      "step": 1649
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.41094508426831006,
+      "learning_rate": 1.1398284637839487e-05,
+      "loss": 0.1175,
+      "step": 1650
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.2837278625607469,
+      "learning_rate": 1.1369769907406714e-05,
+      "loss": 0.1122,
+      "step": 1651
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.2810719132585285,
+      "learning_rate": 1.1341280387325728e-05,
+      "loss": 0.1047,
+      "step": 1652
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3583561379904284,
+      "learning_rate": 1.1312816130290501e-05,
+      "loss": 0.1171,
+      "step": 1653
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3444965468119708,
+      "learning_rate": 1.1284377188948258e-05,
+      "loss": 0.1152,
+      "step": 1654
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.25367997892808686,
+      "learning_rate": 1.1255963615899407e-05,
+      "loss": 0.098,
+      "step": 1655
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3091910045902887,
+      "learning_rate": 1.122757546369744e-05,
+      "loss": 0.1199,
+      "step": 1656
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.374628101486352,
+      "learning_rate": 1.1199212784848834e-05,
+      "loss": 0.1098,
+      "step": 1657
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.25261034176653735,
+      "learning_rate": 1.1170875631812944e-05,
+      "loss": 0.1206,
+      "step": 1658
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3062688235131243,
+      "learning_rate": 1.1142564057001928e-05,
+      "loss": 0.1086,
+      "step": 1659
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3014675246626487,
+      "learning_rate": 1.1114278112780602e-05,
+      "loss": 0.1029,
+      "step": 1660
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.27300326897474336,
+      "learning_rate": 1.1086017851466413e-05,
+      "loss": 0.1167,
+      "step": 1661
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.2926261174091477,
+      "learning_rate": 1.1057783325329268e-05,
+      "loss": 0.1191,
+      "step": 1662
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.4608644970858871,
+      "learning_rate": 1.1029574586591506e-05,
+      "loss": 0.0943,
+      "step": 1663
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.29374269648625817,
+      "learning_rate": 1.1001391687427759e-05,
+      "loss": 0.1086,
+      "step": 1664
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.28893400162584704,
+      "learning_rate": 1.0973234679964872e-05,
+      "loss": 0.1181,
+      "step": 1665
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.24866906257271476,
+      "learning_rate": 1.0945103616281777e-05,
+      "loss": 0.106,
+      "step": 1666
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.29823192514997116,
+      "learning_rate": 1.0916998548409449e-05,
+      "loss": 0.1052,
+      "step": 1667
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.27107000475768805,
+      "learning_rate": 1.0888919528330777e-05,
+      "loss": 0.1135,
+      "step": 1668
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.9301019591129689,
+      "learning_rate": 1.0860866607980446e-05,
+      "loss": 0.1112,
+      "step": 1669
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3055688819761507,
+      "learning_rate": 1.0832839839244902e-05,
+      "loss": 0.1206,
+      "step": 1670
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.23084015053475107,
+      "learning_rate": 1.0804839273962209e-05,
+      "loss": 0.1084,
+      "step": 1671
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.26815762245812447,
+      "learning_rate": 1.0776864963921946e-05,
+      "loss": 0.1199,
+      "step": 1672
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.4319439685887713,
+      "learning_rate": 1.0748916960865158e-05,
+      "loss": 0.112,
+      "step": 1673
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.28838035458689454,
+      "learning_rate": 1.0720995316484219e-05,
+      "loss": 0.1007,
+      "step": 1674
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.2983491018134642,
+      "learning_rate": 1.0693100082422763e-05,
+      "loss": 0.1099,
+      "step": 1675
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.33296639591549065,
+      "learning_rate": 1.0665231310275547e-05,
+      "loss": 0.1163,
+      "step": 1676
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3469061828535007,
+      "learning_rate": 1.0637389051588426e-05,
+      "loss": 0.1052,
+      "step": 1677
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3578065191005441,
+      "learning_rate": 1.0609573357858166e-05,
+      "loss": 0.1156,
+      "step": 1678
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3597127387776606,
+      "learning_rate": 1.0581784280532442e-05,
+      "loss": 0.1046,
+      "step": 1679
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3462867795828064,
+      "learning_rate": 1.0554021871009679e-05,
+      "loss": 0.1097,
+      "step": 1680
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.26820843239491743,
+      "learning_rate": 1.0526286180638983e-05,
+      "loss": 0.1196,
+      "step": 1681
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.33426162705773554,
+      "learning_rate": 1.049857726072005e-05,
+      "loss": 0.1075,
+      "step": 1682
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4089063141354508,
+      "learning_rate": 1.0470895162503037e-05,
+      "loss": 0.1102,
+      "step": 1683
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.2831572521266141,
+      "learning_rate": 1.04432399371885e-05,
+      "loss": 0.1154,
+      "step": 1684
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.274161012012429,
+      "learning_rate": 1.0415611635927306e-05,
+      "loss": 0.1071,
+      "step": 1685
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.2766670233094573,
+      "learning_rate": 1.0388010309820512e-05,
+      "loss": 0.1178,
+      "step": 1686
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.33799393206211936,
+      "learning_rate": 1.0360436009919283e-05,
+      "loss": 0.1338,
+      "step": 1687
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.28396677966855016,
+      "learning_rate": 1.0332888787224804e-05,
+      "loss": 0.1069,
+      "step": 1688
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.2938616038656009,
+      "learning_rate": 1.0305368692688174e-05,
+      "loss": 0.1234,
+      "step": 1689
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3400531931641028,
+      "learning_rate": 1.0277875777210299e-05,
+      "loss": 0.118,
+      "step": 1690
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.2921456096797843,
+      "learning_rate": 1.0250410091641848e-05,
+      "loss": 0.1154,
+      "step": 1691
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4387411626538218,
+      "learning_rate": 1.022297168678309e-05,
+      "loss": 0.1048,
+      "step": 1692
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.45349025064444815,
+      "learning_rate": 1.0195560613383862e-05,
+      "loss": 0.1142,
+      "step": 1693
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.2781201461301553,
+      "learning_rate": 1.0168176922143438e-05,
+      "loss": 0.1168,
+      "step": 1694
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.27214794270287945,
+      "learning_rate": 1.014082066371046e-05,
+      "loss": 0.1024,
+      "step": 1695
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.24953851956417544,
+      "learning_rate": 1.0113491888682802e-05,
+      "loss": 0.1103,
+      "step": 1696
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4925616277854429,
+      "learning_rate": 1.008619064760753e-05,
+      "loss": 0.1078,
+      "step": 1697
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3122473738377716,
+      "learning_rate": 1.0058916990980782e-05,
+      "loss": 0.1064,
+      "step": 1698
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.6897878293824378,
+      "learning_rate": 1.0031670969247656e-05,
+      "loss": 0.1201,
+      "step": 1699
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.2888198762246905,
+      "learning_rate": 1.0004452632802158e-05,
+      "loss": 0.1005,
+      "step": 1700
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.38436878535624786,
+      "learning_rate": 9.977262031987086e-06,
+      "loss": 0.1134,
+      "step": 1701
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.040576964982872,
+      "learning_rate": 9.95009921709392e-06,
+      "loss": 0.1096,
+      "step": 1702
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.45753183366226563,
+      "learning_rate": 9.922964238362762e-06,
+      "loss": 0.1303,
+      "step": 1703
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.2682252223485493,
+      "learning_rate": 9.895857145982232e-06,
+      "loss": 0.1207,
+      "step": 1704
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.31793001468411947,
+      "learning_rate": 9.86877799008937e-06,
+      "loss": 0.0987,
+      "step": 1705
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7856583661852863,
+      "learning_rate": 9.841726820769526e-06,
+      "loss": 0.1188,
+      "step": 1706
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.32328920546638534,
+      "learning_rate": 9.814703688056321e-06,
+      "loss": 0.1247,
+      "step": 1707
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3073811684300866,
+      "learning_rate": 9.787708641931476e-06,
+      "loss": 0.107,
+      "step": 1708
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3945169411987899,
+      "learning_rate": 9.7607417323248e-06,
+      "loss": 0.1243,
+      "step": 1709
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.2733269160312143,
+      "learning_rate": 9.733803009114045e-06,
+      "loss": 0.0963,
+      "step": 1710
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3520217159289507,
+      "learning_rate": 9.706892522124839e-06,
+      "loss": 0.1027,
+      "step": 1711
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.2581389843983647,
+      "learning_rate": 9.680010321130578e-06,
+      "loss": 0.1146,
+      "step": 1712
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3041054615229247,
+      "learning_rate": 9.653156455852342e-06,
+      "loss": 0.0967,
+      "step": 1713
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.38129904057292474,
+      "learning_rate": 9.626330975958784e-06,
+      "loss": 0.1007,
+      "step": 1714
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.8487694224870526,
+      "learning_rate": 9.599533931066082e-06,
+      "loss": 0.1099,
+      "step": 1715
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.26231475993742426,
+      "learning_rate": 9.572765370737815e-06,
+      "loss": 0.1125,
+      "step": 1716
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.30667436772021467,
+      "learning_rate": 9.546025344484869e-06,
+      "loss": 0.1172,
+      "step": 1717
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3832908246569038,
+      "learning_rate": 9.519313901765356e-06,
+      "loss": 0.0964,
+      "step": 1718
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3273658497042535,
+      "learning_rate": 9.492631091984535e-06,
+      "loss": 0.0971,
+      "step": 1719
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.2514862996350453,
+      "learning_rate": 9.465976964494682e-06,
+      "loss": 0.1123,
+      "step": 1720
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.30674178904613353,
+      "learning_rate": 9.439351568595022e-06,
+      "loss": 0.1134,
+      "step": 1721
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3495211475064707,
+      "learning_rate": 9.412754953531663e-06,
+      "loss": 0.1166,
+      "step": 1722
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.2500509992699186,
+      "learning_rate": 9.386187168497465e-06,
+      "loss": 0.1165,
+      "step": 1723
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7398073392263653,
+      "learning_rate": 9.359648262631961e-06,
+      "loss": 0.1077,
+      "step": 1724
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.344585138563564,
+      "learning_rate": 9.333138285021289e-06,
+      "loss": 0.1201,
+      "step": 1725
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.4099494426960492,
+      "learning_rate": 9.306657284698046e-06,
+      "loss": 0.1175,
+      "step": 1726
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.4679478116913521,
+      "learning_rate": 9.280205310641271e-06,
+      "loss": 0.1147,
+      "step": 1727
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.4385688168306654,
+      "learning_rate": 9.253782411776282e-06,
+      "loss": 0.1214,
+      "step": 1728
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.33076096789929677,
+      "learning_rate": 9.227388636974646e-06,
+      "loss": 0.1169,
+      "step": 1729
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5233615530948624,
+      "learning_rate": 9.201024035054053e-06,
+      "loss": 0.1143,
+      "step": 1730
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.34506616082970876,
+      "learning_rate": 9.174688654778243e-06,
+      "loss": 0.0995,
+      "step": 1731
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.38210044595111875,
+      "learning_rate": 9.148382544856884e-06,
+      "loss": 0.1138,
+      "step": 1732
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.4559344624892679,
+      "learning_rate": 9.122105753945531e-06,
+      "loss": 0.1203,
+      "step": 1733
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3413940451716188,
+      "learning_rate": 9.095858330645507e-06,
+      "loss": 0.1211,
+      "step": 1734
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.4187882036920371,
+      "learning_rate": 9.069640323503797e-06,
+      "loss": 0.1169,
+      "step": 1735
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3244480219967307,
+      "learning_rate": 9.043451781012999e-06,
+      "loss": 0.1096,
+      "step": 1736
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.33960291006055277,
+      "learning_rate": 9.01729275161122e-06,
+      "loss": 0.104,
+      "step": 1737
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.4291135253944239,
+      "learning_rate": 8.991163283681944e-06,
+      "loss": 0.1227,
+      "step": 1738
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.334619854735711,
+      "learning_rate": 8.965063425554015e-06,
+      "loss": 0.1055,
+      "step": 1739
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3135719618813803,
+      "learning_rate": 8.938993225501497e-06,
+      "loss": 0.1208,
+      "step": 1740
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.28189613122443397,
+      "learning_rate": 8.912952731743596e-06,
+      "loss": 0.1025,
+      "step": 1741
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.2964057539677339,
+      "learning_rate": 8.88694199244459e-06,
+      "loss": 0.1028,
+      "step": 1742
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3375154445734451,
+      "learning_rate": 8.860961055713692e-06,
+      "loss": 0.1016,
+      "step": 1743
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.2500581134180688,
+      "learning_rate": 8.835009969605012e-06,
+      "loss": 0.1057,
+      "step": 1744
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5249825092341744,
+      "learning_rate": 8.809088782117452e-06,
+      "loss": 0.1105,
+      "step": 1745
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.4985293957662395,
+      "learning_rate": 8.78319754119461e-06,
+      "loss": 0.1113,
+      "step": 1746
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.6594011148868202,
+      "learning_rate": 8.757336294724689e-06,
+      "loss": 0.1245,
+      "step": 1747
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.30444305096035296,
+      "learning_rate": 8.731505090540434e-06,
+      "loss": 0.1072,
+      "step": 1748
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3268861584986191,
+      "learning_rate": 8.705703976418989e-06,
+      "loss": 0.1064,
+      "step": 1749
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.0097772288920226,
+      "learning_rate": 8.679933000081881e-06,
+      "loss": 0.1039,
+      "step": 1750
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3138735929232096,
+      "learning_rate": 8.654192209194862e-06,
+      "loss": 0.1122,
+      "step": 1751
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.2716169367236768,
+      "learning_rate": 8.628481651367876e-06,
+      "loss": 0.1144,
+      "step": 1752
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4539811568723514,
+      "learning_rate": 8.60280137415494e-06,
+      "loss": 0.0966,
+      "step": 1753
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3483068089303449,
+      "learning_rate": 8.577151425054078e-06,
+      "loss": 0.1044,
+      "step": 1754
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3175999424955469,
+      "learning_rate": 8.551531851507186e-06,
+      "loss": 0.1141,
+      "step": 1755
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3374603295551562,
+      "learning_rate": 8.525942700900008e-06,
+      "loss": 0.1072,
+      "step": 1756
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.8658528101501237,
+      "learning_rate": 8.500384020562018e-06,
+      "loss": 0.1155,
+      "step": 1757
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3164089097276373,
+      "learning_rate": 8.474855857766304e-06,
+      "loss": 0.1101,
+      "step": 1758
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4503899128601106,
+      "learning_rate": 8.449358259729537e-06,
+      "loss": 0.1129,
+      "step": 1759
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3317269141253837,
+      "learning_rate": 8.423891273611855e-06,
+      "loss": 0.1097,
+      "step": 1760
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.2461160761272525,
+      "learning_rate": 8.398454946516752e-06,
+      "loss": 0.1103,
+      "step": 1761
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.41499543598225336,
+      "learning_rate": 8.373049325491036e-06,
+      "loss": 0.1368,
+      "step": 1762
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.31837404102543726,
+      "learning_rate": 8.347674457524721e-06,
+      "loss": 0.1134,
+      "step": 1763
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.30046894730266843,
+      "learning_rate": 8.322330389550942e-06,
+      "loss": 0.1152,
+      "step": 1764
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.5004149764364316,
+      "learning_rate": 8.297017168445844e-06,
+      "loss": 0.1214,
+      "step": 1765
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.509914238639227,
+      "learning_rate": 8.271734841028553e-06,
+      "loss": 0.1138,
+      "step": 1766
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.43869498091420905,
+      "learning_rate": 8.246483454061015e-06,
+      "loss": 0.098,
+      "step": 1767
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.316258650898693,
+      "learning_rate": 8.221263054247985e-06,
+      "loss": 0.109,
+      "step": 1768
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.43434150734700194,
+      "learning_rate": 8.196073688236886e-06,
+      "loss": 0.1182,
+      "step": 1769
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.6077013288219103,
+      "learning_rate": 8.170915402617741e-06,
+      "loss": 0.1137,
+      "step": 1770
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.30497031332517277,
+      "learning_rate": 8.145788243923103e-06,
+      "loss": 0.115,
+      "step": 1771
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4616952242540475,
+      "learning_rate": 8.120692258627927e-06,
+      "loss": 0.1138,
+      "step": 1772
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3830606530431249,
+      "learning_rate": 8.09562749314952e-06,
+      "loss": 0.1092,
+      "step": 1773
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.45751037667865796,
+      "learning_rate": 8.070593993847456e-06,
+      "loss": 0.1101,
+      "step": 1774
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4699559270780523,
+      "learning_rate": 8.045591807023466e-06,
+      "loss": 0.1069,
+      "step": 1775
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.44775197697497343,
+      "learning_rate": 8.020620978921382e-06,
+      "loss": 0.1147,
+      "step": 1776
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.39136230614297807,
+      "learning_rate": 7.99568155572701e-06,
+      "loss": 0.1258,
+      "step": 1777
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3798233956856507,
+      "learning_rate": 7.970773583568105e-06,
+      "loss": 0.103,
+      "step": 1778
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.35995746729113015,
+      "learning_rate": 7.94589710851421e-06,
+      "loss": 0.1067,
+      "step": 1779
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.41909508296558334,
+      "learning_rate": 7.921052176576644e-06,
+      "loss": 0.1192,
+      "step": 1780
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.48514397906530743,
+      "learning_rate": 7.896238833708358e-06,
+      "loss": 0.1103,
+      "step": 1781
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.27601390014509186,
+      "learning_rate": 7.871457125803896e-06,
+      "loss": 0.1118,
+      "step": 1782
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.40634372230451893,
+      "learning_rate": 7.846707098699282e-06,
+      "loss": 0.1046,
+      "step": 1783
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4009053262779818,
+      "learning_rate": 7.821988798171958e-06,
+      "loss": 0.1167,
+      "step": 1784
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.3109624606000105,
+      "learning_rate": 7.797302269940652e-06,
+      "loss": 0.1173,
+      "step": 1785
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3957767402429907,
+      "learning_rate": 7.772647559665356e-06,
+      "loss": 0.1127,
+      "step": 1786
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3237762497042359,
+      "learning_rate": 7.748024712947205e-06,
+      "loss": 0.1178,
+      "step": 1787
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.29080775212030235,
+      "learning_rate": 7.723433775328384e-06,
+      "loss": 0.1062,
+      "step": 1788
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.41774740503293967,
+      "learning_rate": 7.698874792292082e-06,
+      "loss": 0.0995,
+      "step": 1789
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.30915208814011597,
+      "learning_rate": 7.674347809262378e-06,
+      "loss": 0.1206,
+      "step": 1790
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3195410170660573,
+      "learning_rate": 7.649852871604146e-06,
+      "loss": 0.1029,
+      "step": 1791
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3810571565769786,
+      "learning_rate": 7.625390024623014e-06,
+      "loss": 0.091,
+      "step": 1792
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.36802357831586757,
+      "learning_rate": 7.600959313565245e-06,
+      "loss": 0.1226,
+      "step": 1793
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3177074722354379,
+      "learning_rate": 7.576560783617668e-06,
+      "loss": 0.1185,
+      "step": 1794
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3642030520327992,
+      "learning_rate": 7.552194479907571e-06,
+      "loss": 0.1035,
+      "step": 1795
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3286402086284856,
+      "learning_rate": 7.5278604475026714e-06,
+      "loss": 0.12,
+      "step": 1796
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.28661011210665316,
+      "learning_rate": 7.503558731410959e-06,
+      "loss": 0.1068,
+      "step": 1797
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.45102484723455577,
+      "learning_rate": 7.479289376580681e-06,
+      "loss": 0.0953,
+      "step": 1798
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5503061030951557,
+      "learning_rate": 7.455052427900213e-06,
+      "loss": 0.1053,
+      "step": 1799
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.33442335984517135,
+      "learning_rate": 7.430847930198009e-06,
+      "loss": 0.1134,
+      "step": 1800
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5358921893971026,
+      "learning_rate": 7.40667592824249e-06,
+      "loss": 0.1061,
+      "step": 1801
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.436246559431753,
+      "learning_rate": 7.382536466741974e-06,
+      "loss": 0.1042,
+      "step": 1802
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.43850654284228924,
+      "learning_rate": 7.358429590344581e-06,
+      "loss": 0.1151,
+      "step": 1803
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.39425406374404715,
+      "learning_rate": 7.3343553436381825e-06,
+      "loss": 0.1052,
+      "step": 1804
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6588123919887777,
+      "learning_rate": 7.310313771150293e-06,
+      "loss": 0.1165,
+      "step": 1805
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3490991650327682,
+      "learning_rate": 7.286304917347986e-06,
+      "loss": 0.1115,
+      "step": 1806
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3297130133709311,
+      "learning_rate": 7.262328826637826e-06,
+      "loss": 0.1118,
+      "step": 1807
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.45721020761380593,
+      "learning_rate": 7.238385543365783e-06,
+      "loss": 0.1113,
+      "step": 1808
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.29792727877831193,
+      "learning_rate": 7.214475111817132e-06,
+      "loss": 0.1013,
+      "step": 1809
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.450280570207537,
+      "learning_rate": 7.190597576216385e-06,
+      "loss": 0.1119,
+      "step": 1810
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.47640562830893307,
+      "learning_rate": 7.166752980727226e-06,
+      "loss": 0.1063,
+      "step": 1811
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.36543497473995357,
+      "learning_rate": 7.142941369452411e-06,
+      "loss": 0.1076,
+      "step": 1812
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5871081617417724,
+      "learning_rate": 7.119162786433675e-06,
+      "loss": 0.1017,
+      "step": 1813
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.43952526031264233,
+      "learning_rate": 7.095417275651686e-06,
+      "loss": 0.1186,
+      "step": 1814
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.3302016722632333,
+      "learning_rate": 7.071704881025915e-06,
+      "loss": 0.1176,
+      "step": 1815
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5017500056657505,
+      "learning_rate": 7.0480256464146025e-06,
+      "loss": 0.1097,
+      "step": 1816
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.36381753242446263,
+      "learning_rate": 7.0243796156146404e-06,
+      "loss": 0.1054,
+      "step": 1817
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.39606402406064867,
+      "learning_rate": 7.000766832361525e-06,
+      "loss": 0.126,
+      "step": 1818
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.897169974773895,
+      "learning_rate": 6.977187340329242e-06,
+      "loss": 0.104,
+      "step": 1819
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.4851175654317836,
+      "learning_rate": 6.953641183130224e-06,
+      "loss": 0.1162,
+      "step": 1820
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.4156460106809062,
+      "learning_rate": 6.930128404315214e-06,
+      "loss": 0.1102,
+      "step": 1821
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.4382741483961508,
+      "learning_rate": 6.906649047373246e-06,
+      "loss": 0.1153,
+      "step": 1822
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.2572835051472882,
+      "learning_rate": 6.883203155731535e-06,
+      "loss": 0.1107,
+      "step": 1823
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4854888579196191,
+      "learning_rate": 6.859790772755381e-06,
+      "loss": 0.1207,
+      "step": 1824
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.35233303738324373,
+      "learning_rate": 6.836411941748122e-06,
+      "loss": 0.0998,
+      "step": 1825
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.35403261381138196,
+      "learning_rate": 6.813066705951046e-06,
+      "loss": 0.1055,
+      "step": 1826
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3080711167592087,
+      "learning_rate": 6.789755108543275e-06,
+      "loss": 0.1112,
+      "step": 1827
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4336942019481867,
+      "learning_rate": 6.76647719264174e-06,
+      "loss": 0.1343,
+      "step": 1828
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.46952136546868384,
+      "learning_rate": 6.743233001301067e-06,
+      "loss": 0.0976,
+      "step": 1829
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.336481345965579,
+      "learning_rate": 6.720022577513507e-06,
+      "loss": 0.1145,
+      "step": 1830
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.30707356156655646,
+      "learning_rate": 6.696845964208845e-06,
+      "loss": 0.101,
+      "step": 1831
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.6774521464288685,
+      "learning_rate": 6.673703204254347e-06,
+      "loss": 0.1143,
+      "step": 1832
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.41480456816265565,
+      "learning_rate": 6.650594340454644e-06,
+      "loss": 0.1076,
+      "step": 1833
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.2569649776196121,
+      "learning_rate": 6.6275194155516915e-06,
+      "loss": 0.1086,
+      "step": 1834
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.2747411734710381,
+      "learning_rate": 6.604478472224659e-06,
+      "loss": 0.0996,
+      "step": 1835
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4137330396746929,
+      "learning_rate": 6.5814715530898745e-06,
+      "loss": 0.109,
+      "step": 1836
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.30090921118734626,
+      "learning_rate": 6.55849870070073e-06,
+      "loss": 0.1112,
+      "step": 1837
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3567456047364232,
+      "learning_rate": 6.535559957547613e-06,
+      "loss": 0.105,
+      "step": 1838
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.3128454861772778,
+      "learning_rate": 6.51265536605781e-06,
+      "loss": 0.121,
+      "step": 1839
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4736424845524447,
+      "learning_rate": 6.4897849685954445e-06,
+      "loss": 0.1143,
+      "step": 1840
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.2781552942289591,
+      "learning_rate": 6.466948807461406e-06,
+      "loss": 0.1078,
+      "step": 1841
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.380193161847238,
+      "learning_rate": 6.4441469248932515e-06,
+      "loss": 0.094,
+      "step": 1842
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.35291796309818596,
+      "learning_rate": 6.421379363065142e-06,
+      "loss": 0.1268,
+      "step": 1843
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4063992860541648,
+      "learning_rate": 6.398646164087763e-06,
+      "loss": 0.1174,
+      "step": 1844
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.2999701482940684,
+      "learning_rate": 6.375947370008217e-06,
+      "loss": 0.1117,
+      "step": 1845
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4243343582859199,
+      "learning_rate": 6.353283022810014e-06,
+      "loss": 0.0999,
+      "step": 1846
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7935423483136271,
+      "learning_rate": 6.330653164412908e-06,
+      "loss": 0.1037,
+      "step": 1847
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3913796995156574,
+      "learning_rate": 6.308057836672893e-06,
+      "loss": 0.1201,
+      "step": 1848
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.34154487531954575,
+      "learning_rate": 6.2854970813820806e-06,
+      "loss": 0.1204,
+      "step": 1849
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.049411784234066,
+      "learning_rate": 6.2629709402686535e-06,
+      "loss": 0.1001,
+      "step": 1850
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.4470974704924731,
+      "learning_rate": 6.240479454996745e-06,
+      "loss": 0.1141,
+      "step": 1851
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3528702832162143,
+      "learning_rate": 6.218022667166412e-06,
+      "loss": 0.0951,
+      "step": 1852
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.8517336720893761,
+      "learning_rate": 6.195600618313538e-06,
+      "loss": 0.0986,
+      "step": 1853
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.42510955226693403,
+      "learning_rate": 6.173213349909729e-06,
+      "loss": 0.1097,
+      "step": 1854
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.2836046880216134,
+      "learning_rate": 6.150860903362282e-06,
+      "loss": 0.1089,
+      "step": 1855
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.41727289683608787,
+      "learning_rate": 6.128543320014091e-06,
+      "loss": 0.1176,
+      "step": 1856
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7025219302298807,
+      "learning_rate": 6.106260641143546e-06,
+      "loss": 0.1195,
+      "step": 1857
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.27604585081268485,
+      "learning_rate": 6.084012907964495e-06,
+      "loss": 0.0924,
+      "step": 1858
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3574947815465219,
+      "learning_rate": 6.0618001616261464e-06,
+      "loss": 0.1098,
+      "step": 1859
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.46250844057339163,
+      "learning_rate": 6.039622443213008e-06,
+      "loss": 0.1155,
+      "step": 1860
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3577601751211578,
+      "learning_rate": 6.017479793744771e-06,
+      "loss": 0.1194,
+      "step": 1861
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3460541418316472,
+      "learning_rate": 5.995372254176301e-06,
+      "loss": 0.1227,
+      "step": 1862
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.33462277729384166,
+      "learning_rate": 5.973299865397486e-06,
+      "loss": 0.1103,
+      "step": 1863
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.4400133581475585,
+      "learning_rate": 5.951262668233232e-06,
+      "loss": 0.1169,
+      "step": 1864
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3227836979997747,
+      "learning_rate": 5.9292607034433375e-06,
+      "loss": 0.1078,
+      "step": 1865
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3548308031594565,
+      "learning_rate": 5.907294011722439e-06,
+      "loss": 0.1094,
+      "step": 1866
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.2666965176673935,
+      "learning_rate": 5.885362633699945e-06,
+      "loss": 0.1279,
+      "step": 1867
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3996537118127708,
+      "learning_rate": 5.863466609939927e-06,
+      "loss": 0.0964,
+      "step": 1868
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6203641928522153,
+      "learning_rate": 5.841605980941067e-06,
+      "loss": 0.1065,
+      "step": 1869
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.30924425261170974,
+      "learning_rate": 5.8197807871366e-06,
+      "loss": 0.11,
+      "step": 1870
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3141267276859039,
+      "learning_rate": 5.797991068894201e-06,
+      "loss": 0.1203,
+      "step": 1871
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3574476222880627,
+      "learning_rate": 5.776236866515947e-06,
+      "loss": 0.1204,
+      "step": 1872
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3924796672209146,
+      "learning_rate": 5.75451822023822e-06,
+      "loss": 0.1008,
+      "step": 1873
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.2782741553270613,
+      "learning_rate": 5.732835170231621e-06,
+      "loss": 0.0927,
+      "step": 1874
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.38883096296038694,
+      "learning_rate": 5.711187756600927e-06,
+      "loss": 0.101,
+      "step": 1875
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.25694913338905007,
+      "learning_rate": 5.689576019385015e-06,
+      "loss": 0.0982,
+      "step": 1876
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.2881627820456557,
+      "learning_rate": 5.6679999985567446e-06,
+      "loss": 0.109,
+      "step": 1877
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.29075039922108425,
+      "learning_rate": 5.646459734022938e-06,
+      "loss": 0.1342,
+      "step": 1878
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.33743495585367705,
+      "learning_rate": 5.624955265624282e-06,
+      "loss": 0.1108,
+      "step": 1879
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4685836894692767,
+      "learning_rate": 5.603486633135238e-06,
+      "loss": 0.1164,
+      "step": 1880
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3654321095927535,
+      "learning_rate": 5.582053876264001e-06,
+      "loss": 0.1142,
+      "step": 1881
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.31940222725252776,
+      "learning_rate": 5.560657034652406e-06,
+      "loss": 0.1172,
+      "step": 1882
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.369484291334686,
+      "learning_rate": 5.5392961478758695e-06,
+      "loss": 0.1214,
+      "step": 1883
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3588832487307221,
+      "learning_rate": 5.517971255443283e-06,
+      "loss": 0.1222,
+      "step": 1884
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.2802641096999261,
+      "learning_rate": 5.49668239679699e-06,
+      "loss": 0.1167,
+      "step": 1885
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3904295633167014,
+      "learning_rate": 5.475429611312658e-06,
+      "loss": 0.1124,
+      "step": 1886
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.28117870677196366,
+      "learning_rate": 5.454212938299255e-06,
+      "loss": 0.1222,
+      "step": 1887
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4139611746435104,
+      "learning_rate": 5.433032416998957e-06,
+      "loss": 0.0981,
+      "step": 1888
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3072948847669026,
+      "learning_rate": 5.411888086587058e-06,
+      "loss": 0.1166,
+      "step": 1889
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.409764915539942,
+      "learning_rate": 5.390779986171934e-06,
+      "loss": 0.1162,
+      "step": 1890
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.31676357450534426,
+      "learning_rate": 5.369708154794934e-06,
+      "loss": 0.1142,
+      "step": 1891
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3670483976413192,
+      "learning_rate": 5.348672631430318e-06,
+      "loss": 0.1158,
+      "step": 1892
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.2567679148136321,
+      "learning_rate": 5.327673454985211e-06,
+      "loss": 0.1071,
+      "step": 1893
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2686688297092955,
+      "learning_rate": 5.3067106642995025e-06,
+      "loss": 0.1045,
+      "step": 1894
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.33652109375068234,
+      "learning_rate": 5.285784298145779e-06,
+      "loss": 0.1122,
+      "step": 1895
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.335670436997812,
+      "learning_rate": 5.264894395229267e-06,
+      "loss": 0.1088,
+      "step": 1896
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5563820616830429,
+      "learning_rate": 5.244040994187746e-06,
+      "loss": 0.1178,
+      "step": 1897
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.32671110179172463,
+      "learning_rate": 5.223224133591476e-06,
+      "loss": 0.1049,
+      "step": 1898
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3285148647729933,
+      "learning_rate": 5.202443851943126e-06,
+      "loss": 0.0999,
+      "step": 1899
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.36425645951040414,
+      "learning_rate": 5.181700187677732e-06,
+      "loss": 0.1053,
+      "step": 1900
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.34300040917134395,
+      "learning_rate": 5.160993179162585e-06,
+      "loss": 0.109,
+      "step": 1901
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4208066314516151,
+      "learning_rate": 5.140322864697183e-06,
+      "loss": 0.1134,
+      "step": 1902
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3526795712507116,
+      "learning_rate": 5.119689282513166e-06,
+      "loss": 0.1131,
+      "step": 1903
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3294837198961809,
+      "learning_rate": 5.0990924707742025e-06,
+      "loss": 0.1126,
+      "step": 1904
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.31285379419302556,
+      "learning_rate": 5.078532467575994e-06,
+      "loss": 0.1233,
+      "step": 1905
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.38041071976803714,
+      "learning_rate": 5.058009310946119e-06,
+      "loss": 0.1148,
+      "step": 1906
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3204976942938043,
+      "learning_rate": 5.037523038844033e-06,
+      "loss": 0.1075,
+      "step": 1907
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4451255803404825,
+      "learning_rate": 5.017073689160961e-06,
+      "loss": 0.1069,
+      "step": 1908
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4003362781243539,
+      "learning_rate": 4.996661299719846e-06,
+      "loss": 0.1094,
+      "step": 1909
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.38669237817751384,
+      "learning_rate": 4.976285908275247e-06,
+      "loss": 0.1053,
+      "step": 1910
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.2702136818583738,
+      "learning_rate": 4.955947552513313e-06,
+      "loss": 0.1,
+      "step": 1911
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3649307379360587,
+      "learning_rate": 4.935646270051689e-06,
+      "loss": 0.1147,
+      "step": 1912
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7069948304917711,
+      "learning_rate": 4.915382098439436e-06,
+      "loss": 0.0926,
+      "step": 1913
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3307633353292947,
+      "learning_rate": 4.895155075156987e-06,
+      "loss": 0.1101,
+      "step": 1914
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.31015441374410535,
+      "learning_rate": 4.8749652376160775e-06,
+      "loss": 0.111,
+      "step": 1915
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.29881413619928776,
+      "learning_rate": 4.8548126231596295e-06,
+      "loss": 0.1028,
+      "step": 1916
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.38603344074487383,
+      "learning_rate": 4.83469726906175e-06,
+      "loss": 0.1146,
+      "step": 1917
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3923055899400095,
+      "learning_rate": 4.814619212527615e-06,
+      "loss": 0.1124,
+      "step": 1918
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.32085368425520183,
+      "learning_rate": 4.79457849069343e-06,
+      "loss": 0.1012,
+      "step": 1919
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.28518484140357625,
+      "learning_rate": 4.7745751406263165e-06,
+      "loss": 0.1162,
+      "step": 1920
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3484728408375457,
+      "learning_rate": 4.754609199324309e-06,
+      "loss": 0.1031,
+      "step": 1921
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3238659446671702,
+      "learning_rate": 4.734680703716215e-06,
+      "loss": 0.11,
+      "step": 1922
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3120410100432846,
+      "learning_rate": 4.714789690661616e-06,
+      "loss": 0.1037,
+      "step": 1923
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3656855296483595,
+      "learning_rate": 4.69493619695075e-06,
+      "loss": 0.1109,
+      "step": 1924
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3824922439143014,
+      "learning_rate": 4.67512025930446e-06,
+      "loss": 0.1106,
+      "step": 1925
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.44754014456732405,
+      "learning_rate": 4.6553419143741255e-06,
+      "loss": 0.1062,
+      "step": 1926
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.2754065163595192,
+      "learning_rate": 4.6356011987416075e-06,
+      "loss": 0.1067,
+      "step": 1927
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5256839746001637,
+      "learning_rate": 4.615898148919148e-06,
+      "loss": 0.1165,
+      "step": 1928
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.46671833965047,
+      "learning_rate": 4.596232801349326e-06,
+      "loss": 0.1062,
+      "step": 1929
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5947739759840375,
+      "learning_rate": 4.576605192404998e-06,
+      "loss": 0.1128,
+      "step": 1930
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3824587669328823,
+      "learning_rate": 4.5570153583892165e-06,
+      "loss": 0.1091,
+      "step": 1931
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.6271019948562396,
+      "learning_rate": 4.537463335535161e-06,
+      "loss": 0.1049,
+      "step": 1932
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.3215558128856078,
+      "learning_rate": 4.517949160006085e-06,
+      "loss": 0.101,
+      "step": 1933
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.33090932727160527,
+      "learning_rate": 4.498472867895223e-06,
+      "loss": 0.1102,
+      "step": 1934
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.9253963240736406,
+      "learning_rate": 4.4790344952257615e-06,
+      "loss": 0.1056,
+      "step": 1935
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3362193607032271,
+      "learning_rate": 4.459634077950728e-06,
+      "loss": 0.1067,
+      "step": 1936
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.4546968242575051,
+      "learning_rate": 4.440271651952965e-06,
+      "loss": 0.1143,
+      "step": 1937
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.38905078823738104,
+      "learning_rate": 4.420947253045049e-06,
+      "loss": 0.1071,
+      "step": 1938
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.49521017026177777,
+      "learning_rate": 4.401660916969216e-06,
+      "loss": 0.1039,
+      "step": 1939
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.2946575915416108,
+      "learning_rate": 4.382412679397294e-06,
+      "loss": 0.0988,
+      "step": 1940
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.35574092831946175,
+      "learning_rate": 4.36320257593065e-06,
+      "loss": 0.1117,
+      "step": 1941
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.29778767510514387,
+      "learning_rate": 4.344030642100133e-06,
+      "loss": 0.1025,
+      "step": 1942
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.409893009914383,
+      "learning_rate": 4.3248969133659614e-06,
+      "loss": 0.1101,
+      "step": 1943
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3398061464560722,
+      "learning_rate": 4.305801425117722e-06,
+      "loss": 0.0973,
+      "step": 1944
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5974911518889044,
+      "learning_rate": 4.286744212674259e-06,
+      "loss": 0.1051,
+      "step": 1945
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3231706748719576,
+      "learning_rate": 4.267725311283616e-06,
+      "loss": 0.1252,
+      "step": 1946
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.29172628304000703,
+      "learning_rate": 4.248744756122986e-06,
+      "loss": 0.1016,
+      "step": 1947
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8729251628170546,
+      "learning_rate": 4.229802582298633e-06,
+      "loss": 0.1051,
+      "step": 1948
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.237165840915252,
+      "learning_rate": 4.210898824845841e-06,
+      "loss": 0.1252,
+      "step": 1949
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3702377270577044,
+      "learning_rate": 4.192033518728819e-06,
+      "loss": 0.0883,
+      "step": 1950
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.4496931489616092,
+      "learning_rate": 4.173206698840676e-06,
+      "loss": 0.105,
+      "step": 1951
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.37306421131847406,
+      "learning_rate": 4.154418400003324e-06,
+      "loss": 0.0926,
+      "step": 1952
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4373296533839181,
+      "learning_rate": 4.135668656967434e-06,
+      "loss": 0.1139,
+      "step": 1953
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2719540417461219,
+      "learning_rate": 4.116957504412367e-06,
+      "loss": 0.1041,
+      "step": 1954
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3664516644078235,
+      "learning_rate": 4.098284976946101e-06,
+      "loss": 0.1068,
+      "step": 1955
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4018357068847783,
+      "learning_rate": 4.079651109105184e-06,
+      "loss": 0.1174,
+      "step": 1956
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.37959707447211727,
+      "learning_rate": 4.061055935354643e-06,
+      "loss": 0.1147,
+      "step": 1957
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.43657941937537564,
+      "learning_rate": 4.042499490087939e-06,
+      "loss": 0.122,
+      "step": 1958
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2252324465107311,
+      "learning_rate": 4.023981807626917e-06,
+      "loss": 0.1099,
+      "step": 1959
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.34653893812236963,
+      "learning_rate": 4.005502922221713e-06,
+      "loss": 0.1158,
+      "step": 1960
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3633251619304228,
+      "learning_rate": 3.987062868050709e-06,
+      "loss": 0.1229,
+      "step": 1961
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.30124029502024635,
+      "learning_rate": 3.968661679220468e-06,
+      "loss": 0.1027,
+      "step": 1962
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3195267259850453,
+      "learning_rate": 3.950299389765666e-06,
+      "loss": 0.1133,
+      "step": 1963
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.510251987903077,
+      "learning_rate": 3.931976033649021e-06,
+      "loss": 0.1079,
+      "step": 1964
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.47882109437823844,
+      "learning_rate": 3.913691644761258e-06,
+      "loss": 0.1094,
+      "step": 1965
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.22854017439849356,
+      "learning_rate": 3.8954462569210095e-06,
+      "loss": 0.1015,
+      "step": 1966
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.26399182640112573,
+      "learning_rate": 3.8772399038747855e-06,
+      "loss": 0.1024,
+      "step": 1967
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.24949136320319837,
+      "learning_rate": 3.859072619296897e-06,
+      "loss": 0.1097,
+      "step": 1968
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.4571528002936169,
+      "learning_rate": 3.840944436789393e-06,
+      "loss": 0.1137,
+      "step": 1969
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.40004140246186076,
+      "learning_rate": 3.822855389881991e-06,
+      "loss": 0.1112,
+      "step": 1970
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.2923386307561353,
+      "learning_rate": 3.804805512032031e-06,
+      "loss": 0.0991,
+      "step": 1971
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.27163788482011997,
+      "learning_rate": 3.786794836624416e-06,
+      "loss": 0.0985,
+      "step": 1972
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3371915255245527,
+      "learning_rate": 3.768823396971516e-06,
+      "loss": 0.1128,
+      "step": 1973
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3559608834587082,
+      "learning_rate": 3.7508912263131524e-06,
+      "loss": 0.1022,
+      "step": 1974
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.21448869152654035,
+      "learning_rate": 3.732998357816514e-06,
+      "loss": 0.11,
+      "step": 1975
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.26518418882638745,
+      "learning_rate": 3.715144824576078e-06,
+      "loss": 0.1013,
+      "step": 1976
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.4316647786389924,
+      "learning_rate": 3.697330659613588e-06,
+      "loss": 0.1189,
+      "step": 1977
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.23850646802650402,
+      "learning_rate": 3.679555895877959e-06,
+      "loss": 0.11,
+      "step": 1978
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.2664740515977864,
+      "learning_rate": 3.661820566245247e-06,
+      "loss": 0.1151,
+      "step": 1979
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.29829540591281584,
+      "learning_rate": 3.644124703518542e-06,
+      "loss": 0.1182,
+      "step": 1980
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.28873289408915437,
+      "learning_rate": 3.6264683404279713e-06,
+      "loss": 0.0914,
+      "step": 1981
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.36560720629227217,
+      "learning_rate": 3.6088515096305674e-06,
+      "loss": 0.1322,
+      "step": 1982
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.62554270714657,
+      "learning_rate": 3.591274243710277e-06,
+      "loss": 0.1036,
+      "step": 1983
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3034650767394291,
+      "learning_rate": 3.573736575177847e-06,
+      "loss": 0.1307,
+      "step": 1984
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.34598683166175964,
+      "learning_rate": 3.556238536470799e-06,
+      "loss": 0.0936,
+      "step": 1985
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.27947863158793373,
+      "learning_rate": 3.5387801599533475e-06,
+      "loss": 0.1049,
+      "step": 1986
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.4618549571248451,
+      "learning_rate": 3.5213614779163518e-06,
+      "loss": 0.1243,
+      "step": 1987
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.49454637810592794,
+      "learning_rate": 3.503982522577237e-06,
+      "loss": 0.1166,
+      "step": 1988
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.30024929583018867,
+      "learning_rate": 3.4866433260799748e-06,
+      "loss": 0.1191,
+      "step": 1989
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.28164033876951666,
+      "learning_rate": 3.4693439204949858e-06,
+      "loss": 0.1006,
+      "step": 1990
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.8317762836489961,
+      "learning_rate": 3.4520843378190976e-06,
+      "loss": 0.1046,
+      "step": 1991
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.23694452437048885,
+      "learning_rate": 3.43486460997548e-06,
+      "loss": 0.1012,
+      "step": 1992
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3400786927571775,
+      "learning_rate": 3.4176847688135893e-06,
+      "loss": 0.109,
+      "step": 1993
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.32298584164739613,
+      "learning_rate": 3.4005448461091077e-06,
+      "loss": 0.1053,
+      "step": 1994
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.0154087170155794,
+      "learning_rate": 3.383444873563871e-06,
+      "loss": 0.0961,
+      "step": 1995
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.29705142912508664,
+      "learning_rate": 3.366384882805845e-06,
+      "loss": 0.1019,
+      "step": 1996
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4641617036349675,
+      "learning_rate": 3.3493649053890326e-06,
+      "loss": 0.1106,
+      "step": 1997
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3165961380004892,
+      "learning_rate": 3.3323849727934426e-06,
+      "loss": 0.0931,
+      "step": 1998
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.2865321195751479,
+      "learning_rate": 3.3154451164249922e-06,
+      "loss": 0.0973,
+      "step": 1999
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3242346430917135,
+      "learning_rate": 3.2985453676154937e-06,
+      "loss": 0.0976,
+      "step": 2000
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.44277702858752965,
+      "learning_rate": 3.2816857576225766e-06,
+      "loss": 0.1102,
+      "step": 2001
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.49800776510908323,
+      "learning_rate": 3.264866317629617e-06,
+      "loss": 0.1077,
+      "step": 2002
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3623649426462525,
+      "learning_rate": 3.248087078745704e-06,
+      "loss": 0.1086,
+      "step": 2003
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.28582018395640457,
+      "learning_rate": 3.2313480720055745e-06,
+      "loss": 0.1139,
+      "step": 2004
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.38602598570312235,
+      "learning_rate": 3.2146493283695345e-06,
+      "loss": 0.104,
+      "step": 2005
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.33350338620559494,
+      "learning_rate": 3.1979908787234338e-06,
+      "loss": 0.116,
+      "step": 2006
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4265041290983197,
+      "learning_rate": 3.181372753878595e-06,
+      "loss": 0.109,
+      "step": 2007
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3864341931475218,
+      "learning_rate": 3.164794984571759e-06,
+      "loss": 0.0928,
+      "step": 2008
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.37560207353027125,
+      "learning_rate": 3.1482576014650074e-06,
+      "loss": 0.114,
+      "step": 2009
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.42243451425287765,
+      "learning_rate": 3.13176063514575e-06,
+      "loss": 0.1169,
+      "step": 2010
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.44183358393881067,
+      "learning_rate": 3.1153041161266134e-06,
+      "loss": 0.1209,
+      "step": 2011
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.456504209462481,
+      "learning_rate": 3.0988880748454353e-06,
+      "loss": 0.115,
+      "step": 2012
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3036643394566577,
+      "learning_rate": 3.0825125416651777e-06,
+      "loss": 0.1038,
+      "step": 2013
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.38343425650473373,
+      "learning_rate": 3.066177546873883e-06,
+      "loss": 0.1089,
+      "step": 2014
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.34192702477658216,
+      "learning_rate": 3.0498831206846156e-06,
+      "loss": 0.1123,
+      "step": 2015
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3414429440118312,
+      "learning_rate": 3.0336292932353904e-06,
+      "loss": 0.1101,
+      "step": 2016
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.360714155617802,
+      "learning_rate": 3.017416094589151e-06,
+      "loss": 0.1214,
+      "step": 2017
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.33743044091581714,
+      "learning_rate": 3.0012435547336737e-06,
+      "loss": 0.1032,
+      "step": 2018
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.5102813146974998,
+      "learning_rate": 2.98511170358155e-06,
+      "loss": 0.1264,
+      "step": 2019
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3448853592036761,
+      "learning_rate": 2.9690205709701104e-06,
+      "loss": 0.105,
+      "step": 2020
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.46351521010740643,
+      "learning_rate": 2.9529701866613663e-06,
+      "loss": 0.1195,
+      "step": 2021
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.22775300268040602,
+      "learning_rate": 2.9369605803419715e-06,
+      "loss": 0.1134,
+      "step": 2022
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.40369772353699257,
+      "learning_rate": 2.920991781623139e-06,
+      "loss": 0.1208,
+      "step": 2023
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.32410101920819806,
+      "learning_rate": 2.905063820040629e-06,
+      "loss": 0.1222,
+      "step": 2024
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.24157150945483016,
+      "learning_rate": 2.889176725054643e-06,
+      "loss": 0.1151,
+      "step": 2025
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3625043415604677,
+      "learning_rate": 2.8733305260498196e-06,
+      "loss": 0.0988,
+      "step": 2026
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.39572488143523915,
+      "learning_rate": 2.857525252335144e-06,
+      "loss": 0.109,
+      "step": 2027
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.30803097585855443,
+      "learning_rate": 2.8417609331439165e-06,
+      "loss": 0.1312,
+      "step": 2028
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.33574988787679744,
+      "learning_rate": 2.8260375976336683e-06,
+      "loss": 0.1136,
+      "step": 2029
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3106460571598377,
+      "learning_rate": 2.8103552748861476e-06,
+      "loss": 0.1138,
+      "step": 2030
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.45525123688069175,
+      "learning_rate": 2.7947139939072413e-06,
+      "loss": 0.1167,
+      "step": 2031
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.26718163752774005,
+      "learning_rate": 2.779113783626916e-06,
+      "loss": 0.1058,
+      "step": 2032
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3312426645253653,
+      "learning_rate": 2.7635546728991822e-06,
+      "loss": 0.1233,
+      "step": 2033
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.6263654974914741,
+      "learning_rate": 2.74803669050204e-06,
+      "loss": 0.1038,
+      "step": 2034
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.251511244012359,
+      "learning_rate": 2.7325598651373992e-06,
+      "loss": 0.1103,
+      "step": 2035
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.5467978716621458,
+      "learning_rate": 2.7171242254310613e-06,
+      "loss": 0.1125,
+      "step": 2036
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3166344622241757,
+      "learning_rate": 2.7017297999326537e-06,
+      "loss": 0.1085,
+      "step": 2037
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.28076797624154604,
+      "learning_rate": 2.6863766171155683e-06,
+      "loss": 0.1202,
+      "step": 2038
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.0077031754945027,
+      "learning_rate": 2.671064705376905e-06,
+      "loss": 0.1005,
+      "step": 2039
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.39515206421250404,
+      "learning_rate": 2.6557940930374504e-06,
+      "loss": 0.102,
+      "step": 2040
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.315029510049465,
+      "learning_rate": 2.6405648083415834e-06,
+      "loss": 0.1036,
+      "step": 2041
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3499805750141062,
+      "learning_rate": 2.625376879457253e-06,
+      "loss": 0.1128,
+      "step": 2042
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.28949950009613773,
+      "learning_rate": 2.6102303344759242e-06,
+      "loss": 0.1151,
+      "step": 2043
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4059439861128851,
+      "learning_rate": 2.5951252014125074e-06,
+      "loss": 0.1145,
+      "step": 2044
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3084410202307766,
+      "learning_rate": 2.5800615082053227e-06,
+      "loss": 0.1175,
+      "step": 2045
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3111538004681346,
+      "learning_rate": 2.565039282716045e-06,
+      "loss": 0.1114,
+      "step": 2046
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3039026315771797,
+      "learning_rate": 2.5500585527296396e-06,
+      "loss": 0.1051,
+      "step": 2047
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3777109241510023,
+      "learning_rate": 2.5351193459543345e-06,
+      "loss": 0.1102,
+      "step": 2048
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3078887036195819,
+      "learning_rate": 2.520221690021557e-06,
+      "loss": 0.1248,
+      "step": 2049
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3293178957358001,
+      "learning_rate": 2.505365612485874e-06,
+      "loss": 0.1149,
+      "step": 2050
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.28144260192696,
+      "learning_rate": 2.4905511408249564e-06,
+      "loss": 0.1091,
+      "step": 2051
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.45926634716259507,
+      "learning_rate": 2.475778302439524e-06,
+      "loss": 0.1247,
+      "step": 2052
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3851049989756811,
+      "learning_rate": 2.461047124653279e-06,
+      "loss": 0.0999,
+      "step": 2053
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.26051189959942145,
+      "learning_rate": 2.446357634712873e-06,
+      "loss": 0.1226,
+      "step": 2054
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.0265568700907814,
+      "learning_rate": 2.431709859787862e-06,
+      "loss": 0.1007,
+      "step": 2055
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.5868471996195491,
+      "learning_rate": 2.4171038269706348e-06,
+      "loss": 0.1134,
+      "step": 2056
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.31019031320440454,
+      "learning_rate": 2.4025395632763846e-06,
+      "loss": 0.1077,
+      "step": 2057
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.2664801842128446,
+      "learning_rate": 2.3880170956430468e-06,
+      "loss": 0.0981,
+      "step": 2058
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.6388783639444267,
+      "learning_rate": 2.3735364509312353e-06,
+      "loss": 0.1009,
+      "step": 2059
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.27058420848719283,
+      "learning_rate": 2.3590976559242278e-06,
+      "loss": 0.121,
+      "step": 2060
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.2870197205313154,
+      "learning_rate": 2.3447007373278956e-06,
+      "loss": 0.1209,
+      "step": 2061
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4312914561172497,
+      "learning_rate": 2.330345721770641e-06,
+      "loss": 0.1041,
+      "step": 2062
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.517918518827338,
+      "learning_rate": 2.316032635803378e-06,
+      "loss": 0.1066,
+      "step": 2063
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3847095240744173,
+      "learning_rate": 2.301761505899466e-06,
+      "loss": 0.1126,
+      "step": 2064
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.2928097232730518,
+      "learning_rate": 2.28753235845465e-06,
+      "loss": 0.1067,
+      "step": 2065
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.2754233444390498,
+      "learning_rate": 2.2733452197870403e-06,
+      "loss": 0.1125,
+      "step": 2066
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.31809826433305316,
+      "learning_rate": 2.2592001161370392e-06,
+      "loss": 0.1037,
+      "step": 2067
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3115943381126067,
+      "learning_rate": 2.24509707366731e-06,
+      "loss": 0.1025,
+      "step": 2068
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.28925656172394,
+      "learning_rate": 2.2310361184627064e-06,
+      "loss": 0.1029,
+      "step": 2069
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.48326440901415163,
+      "learning_rate": 2.2170172765302504e-06,
+      "loss": 0.1132,
+      "step": 2070
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.5826288485127242,
+      "learning_rate": 2.2030405737990617e-06,
+      "loss": 0.1052,
+      "step": 2071
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.42091236069771826,
+      "learning_rate": 2.189106036120328e-06,
+      "loss": 0.118,
+      "step": 2072
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.40023945031820324,
+      "learning_rate": 2.175213689267247e-06,
+      "loss": 0.1143,
+      "step": 2073
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.29429448965062194,
+      "learning_rate": 2.1613635589349756e-06,
+      "loss": 0.1249,
+      "step": 2074
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.2610596244342566,
+      "learning_rate": 2.1475556707406e-06,
+      "loss": 0.1027,
+      "step": 2075
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.2887135688310749,
+      "learning_rate": 2.1337900502230584e-06,
+      "loss": 0.1033,
+      "step": 2076
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3950858961815221,
+      "learning_rate": 2.120066722843117e-06,
+      "loss": 0.1075,
+      "step": 2077
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.43181936522454545,
+      "learning_rate": 2.106385713983325e-06,
+      "loss": 0.1117,
+      "step": 2078
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.2580134156240275,
+      "learning_rate": 2.0927470489479517e-06,
+      "loss": 0.0935,
+      "step": 2079
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.2891599699489009,
+      "learning_rate": 2.0791507529629524e-06,
+      "loss": 0.1065,
+      "step": 2080
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3615406288746834,
+      "learning_rate": 2.0655968511759134e-06,
+      "loss": 0.1053,
+      "step": 2081
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.28307752730671515,
+      "learning_rate": 2.0520853686560178e-06,
+      "loss": 0.1236,
+      "step": 2082
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.22505576663629795,
+      "learning_rate": 2.038616330393972e-06,
+      "loss": 0.0985,
+      "step": 2083
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.31059690675157,
+      "learning_rate": 2.0251897613019902e-06,
+      "loss": 0.0936,
+      "step": 2084
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.36827287586414037,
+      "learning_rate": 2.0118056862137357e-06,
+      "loss": 0.1095,
+      "step": 2085
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4072869909947939,
+      "learning_rate": 1.9984641298842755e-06,
+      "loss": 0.0933,
+      "step": 2086
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4847055337461746,
+      "learning_rate": 1.9851651169900343e-06,
+      "loss": 0.1315,
+      "step": 2087
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4438381718973021,
+      "learning_rate": 1.97190867212875e-06,
+      "loss": 0.1027,
+      "step": 2088
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.2630856968272414,
+      "learning_rate": 1.958694819819415e-06,
+      "loss": 0.1056,
+      "step": 2089
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7469556265699129,
+      "learning_rate": 1.9455235845022624e-06,
+      "loss": 0.107,
+      "step": 2090
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.6724165647212293,
+      "learning_rate": 1.9323949905386856e-06,
+      "loss": 0.1189,
+      "step": 2091
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.36639339995379755,
+      "learning_rate": 1.9193090622112136e-06,
+      "loss": 0.1156,
+      "step": 2092
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.24203359422829668,
+      "learning_rate": 1.906265823723466e-06,
+      "loss": 0.0999,
+      "step": 2093
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.39810461221203997,
+      "learning_rate": 1.8932652992001032e-06,
+      "loss": 0.1215,
+      "step": 2094
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.5541391796830281,
+      "learning_rate": 1.8803075126867715e-06,
+      "loss": 0.1012,
+      "step": 2095
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.2736327180926601,
+      "learning_rate": 1.8673924881500826e-06,
+      "loss": 0.11,
+      "step": 2096
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.33774904660382593,
+      "learning_rate": 1.8545202494775509e-06,
+      "loss": 0.0983,
+      "step": 2097
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7142181030976656,
+      "learning_rate": 1.8416908204775485e-06,
+      "loss": 0.1111,
+      "step": 2098
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.2759353834262672,
+      "learning_rate": 1.8289042248792742e-06,
+      "loss": 0.1072,
+      "step": 2099
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.2987978129337262,
+      "learning_rate": 1.8161604863327074e-06,
+      "loss": 0.0951,
+      "step": 2100
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.46218055379521905,
+      "learning_rate": 1.8034596284085465e-06,
+      "loss": 0.1023,
+      "step": 2101
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.34760853828792687,
+      "learning_rate": 1.790801674598186e-06,
+      "loss": 0.1084,
+      "step": 2102
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.29396961598193744,
+      "learning_rate": 1.7781866483136623e-06,
+      "loss": 0.113,
+      "step": 2103
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.249356349147582,
+      "learning_rate": 1.7656145728876249e-06,
+      "loss": 0.0978,
+      "step": 2104
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4509181313412786,
+      "learning_rate": 1.7530854715732597e-06,
+      "loss": 0.1079,
+      "step": 2105
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.39886494089694835,
+      "learning_rate": 1.7405993675442905e-06,
+      "loss": 0.1037,
+      "step": 2106
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3002166550325826,
+      "learning_rate": 1.7281562838948966e-06,
+      "loss": 0.109,
+      "step": 2107
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.37755136490796326,
+      "learning_rate": 1.7157562436396962e-06,
+      "loss": 0.1022,
+      "step": 2108
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.23132762459237188,
+      "learning_rate": 1.703399269713693e-06,
+      "loss": 0.1088,
+      "step": 2109
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3787079366157527,
+      "learning_rate": 1.6910853849722351e-06,
+      "loss": 0.1068,
+      "step": 2110
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.506504292510756,
+      "learning_rate": 1.678814612190982e-06,
+      "loss": 0.1255,
+      "step": 2111
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.25252389341069986,
+      "learning_rate": 1.6665869740658312e-06,
+      "loss": 0.1041,
+      "step": 2112
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.46908631618137303,
+      "learning_rate": 1.6544024932129255e-06,
+      "loss": 0.0934,
+      "step": 2113
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3168798506669845,
+      "learning_rate": 1.642261192168562e-06,
+      "loss": 0.113,
+      "step": 2114
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.2935410432213498,
+      "learning_rate": 1.6301630933891866e-06,
+      "loss": 0.1097,
+      "step": 2115
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.4158163008937928,
+      "learning_rate": 1.6181082192513353e-06,
+      "loss": 0.1208,
+      "step": 2116
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.37989564502664885,
+      "learning_rate": 1.606096592051598e-06,
+      "loss": 0.1063,
+      "step": 2117
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.2901943777021299,
+      "learning_rate": 1.59412823400657e-06,
+      "loss": 0.0947,
+      "step": 2118
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3487895858042682,
+      "learning_rate": 1.5822031672528186e-06,
+      "loss": 0.1009,
+      "step": 2119
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3086903417213815,
+      "learning_rate": 1.5703214138468452e-06,
+      "loss": 0.1134,
+      "step": 2120
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.303316281186802,
+      "learning_rate": 1.558482995765026e-06,
+      "loss": 0.1093,
+      "step": 2121
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.36884740117762643,
+      "learning_rate": 1.5466879349036012e-06,
+      "loss": 0.0949,
+      "step": 2122
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.4496713998885864,
+      "learning_rate": 1.534936253078606e-06,
+      "loss": 0.123,
+      "step": 2123
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3324224693863234,
+      "learning_rate": 1.5232279720258507e-06,
+      "loss": 0.1105,
+      "step": 2124
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.28148306126377237,
+      "learning_rate": 1.5115631134008628e-06,
+      "loss": 0.1119,
+      "step": 2125
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.24783461626708775,
+      "learning_rate": 1.4999416987788617e-06,
+      "loss": 0.1228,
+      "step": 2126
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.32503609834682523,
+      "learning_rate": 1.4883637496547142e-06,
+      "loss": 0.087,
+      "step": 2127
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.2527727285629537,
+      "learning_rate": 1.476829287442888e-06,
+      "loss": 0.1263,
+      "step": 2128
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3676023570232795,
+      "learning_rate": 1.465338333477423e-06,
+      "loss": 0.1179,
+      "step": 2129
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.35337760584548245,
+      "learning_rate": 1.4538909090118846e-06,
+      "loss": 0.1184,
+      "step": 2130
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3630149273024841,
+      "learning_rate": 1.4424870352193282e-06,
+      "loss": 0.099,
+      "step": 2131
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.40094891890888124,
+      "learning_rate": 1.4311267331922534e-06,
+      "loss": 0.0958,
+      "step": 2132
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.289861424692419,
+      "learning_rate": 1.4198100239425749e-06,
+      "loss": 0.0988,
+      "step": 2133
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3319824863041316,
+      "learning_rate": 1.4085369284015854e-06,
+      "loss": 0.1001,
+      "step": 2134
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5187665458456897,
+      "learning_rate": 1.3973074674198893e-06,
+      "loss": 0.1231,
+      "step": 2135
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.6434517363084737,
+      "learning_rate": 1.3861216617674032e-06,
+      "loss": 0.1324,
+      "step": 2136
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.29586923180060304,
+      "learning_rate": 1.3749795321332887e-06,
+      "loss": 0.1059,
+      "step": 2137
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3418750162890369,
+      "learning_rate": 1.3638810991259355e-06,
+      "loss": 0.1267,
+      "step": 2138
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3738393954680121,
+      "learning_rate": 1.352826383272901e-06,
+      "loss": 0.1289,
+      "step": 2139
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4867314934766,
+      "learning_rate": 1.3418154050208936e-06,
+      "loss": 0.1143,
+      "step": 2140
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.19380172052120762,
+      "learning_rate": 1.3308481847357196e-06,
+      "loss": 0.0941,
+      "step": 2141
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.31212495710401944,
+      "learning_rate": 1.3199247427022526e-06,
+      "loss": 0.1167,
+      "step": 2142
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.26730374493853154,
+      "learning_rate": 1.3090450991243841e-06,
+      "loss": 0.104,
+      "step": 2143
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3127964949963654,
+      "learning_rate": 1.2982092741250145e-06,
+      "loss": 0.1163,
+      "step": 2144
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.33016093913486494,
+      "learning_rate": 1.2874172877459845e-06,
+      "loss": 0.1054,
+      "step": 2145
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3784551313248559,
+      "learning_rate": 1.27666915994806e-06,
+      "loss": 0.1187,
+      "step": 2146
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.272230977085158,
+      "learning_rate": 1.265964910610884e-06,
+      "loss": 0.1044,
+      "step": 2147
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.448598115311016,
+      "learning_rate": 1.2553045595329298e-06,
+      "loss": 0.111,
+      "step": 2148
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.26492495713874353,
+      "learning_rate": 1.2446881264315009e-06,
+      "loss": 0.0993,
+      "step": 2149
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.3486255458215914,
+      "learning_rate": 1.2341156309426449e-06,
+      "loss": 0.1158,
+      "step": 2150
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.44885577315448166,
+      "learning_rate": 1.2235870926211619e-06,
+      "loss": 0.1041,
+      "step": 2151
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.29218398297841386,
+      "learning_rate": 1.2131025309405409e-06,
+      "loss": 0.1206,
+      "step": 2152
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.28183461967553064,
+      "learning_rate": 1.2026619652929338e-06,
+      "loss": 0.1142,
+      "step": 2153
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.2535323848950225,
+      "learning_rate": 1.192265414989116e-06,
+      "loss": 0.0958,
+      "step": 2154
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.2955801340302899,
+      "learning_rate": 1.1819128992584555e-06,
+      "loss": 0.0999,
+      "step": 2155
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2833506093850544,
+      "learning_rate": 1.1716044372488744e-06,
+      "loss": 0.1056,
+      "step": 2156
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2535679255504902,
+      "learning_rate": 1.1613400480268099e-06,
+      "loss": 0.1041,
+      "step": 2157
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.39998015911444357,
+      "learning_rate": 1.1511197505771843e-06,
+      "loss": 0.1014,
+      "step": 2158
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.8584403330790167,
+      "learning_rate": 1.1409435638033706e-06,
+      "loss": 0.124,
+      "step": 2159
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.32561033011959784,
+      "learning_rate": 1.1308115065271491e-06,
+      "loss": 0.0998,
+      "step": 2160
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.35485596282130166,
+      "learning_rate": 1.120723597488682e-06,
+      "loss": 0.1044,
+      "step": 2161
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3102570021716889,
+      "learning_rate": 1.1106798553464804e-06,
+      "loss": 0.1147,
+      "step": 2162
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.563465955252973,
+      "learning_rate": 1.100680298677359e-06,
+      "loss": 0.1,
+      "step": 2163
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.266603821310788,
+      "learning_rate": 1.0907249459764096e-06,
+      "loss": 0.1027,
+      "step": 2164
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3246039306151965,
+      "learning_rate": 1.0808138156569614e-06,
+      "loss": 0.094,
+      "step": 2165
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.5415782960579164,
+      "learning_rate": 1.0709469260505477e-06,
+      "loss": 0.0984,
+      "step": 2166
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.34396499616288506,
+      "learning_rate": 1.0611242954068873e-06,
+      "loss": 0.1022,
+      "step": 2167
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.5198199852967915,
+      "learning_rate": 1.0513459418938248e-06,
+      "loss": 0.1022,
+      "step": 2168
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3559923390122219,
+      "learning_rate": 1.0416118835973182e-06,
+      "loss": 0.1152,
+      "step": 2169
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4865549510641347,
+      "learning_rate": 1.0319221385213933e-06,
+      "loss": 0.108,
+      "step": 2170
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2756573999572265,
+      "learning_rate": 1.0222767245881197e-06,
+      "loss": 0.1022,
+      "step": 2171
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.33979925737068606,
+      "learning_rate": 1.0126756596375686e-06,
+      "loss": 0.1147,
+      "step": 2172
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.41879600592108773,
+      "learning_rate": 1.0031189614277765e-06,
+      "loss": 0.1174,
+      "step": 2173
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.6534382185304043,
+      "learning_rate": 9.936066476347317e-07,
+      "loss": 0.1143,
+      "step": 2174
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2903701704142886,
+      "learning_rate": 9.841387358523246e-07,
+      "loss": 0.1242,
+      "step": 2175
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.38532537209530215,
+      "learning_rate": 9.747152435923224e-07,
+      "loss": 0.1262,
+      "step": 2176
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3272860042375846,
+      "learning_rate": 9.653361882843298e-07,
+      "loss": 0.1166,
+      "step": 2177
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2923879205878471,
+      "learning_rate": 9.560015872757649e-07,
+      "loss": 0.1004,
+      "step": 2178
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.26049768709197696,
+      "learning_rate": 9.467114578318226e-07,
+      "loss": 0.1056,
+      "step": 2179
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.2461068615397645,
+      "learning_rate": 9.374658171354411e-07,
+      "loss": 0.1078,
+      "step": 2180
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3676896183768612,
+      "learning_rate": 9.282646822872748e-07,
+      "loss": 0.1121,
+      "step": 2181
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5357818811091826,
+      "learning_rate": 9.191080703056604e-07,
+      "loss": 0.1006,
+      "step": 2182
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.38846226131247846,
+      "learning_rate": 9.099959981265921e-07,
+      "loss": 0.1023,
+      "step": 2183
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.46043465238156533,
+      "learning_rate": 9.009284826036691e-07,
+      "loss": 0.1029,
+      "step": 2184
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.273659446202888,
+      "learning_rate": 8.919055405080895e-07,
+      "loss": 0.1005,
+      "step": 2185
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.34202209547945434,
+      "learning_rate": 8.829271885286094e-07,
+      "loss": 0.1081,
+      "step": 2186
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.2104596422948685,
+      "learning_rate": 8.739934432715035e-07,
+      "loss": 0.1068,
+      "step": 2187
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4769154545613333,
+      "learning_rate": 8.651043212605486e-07,
+      "loss": 0.1094,
+      "step": 2188
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.2799469222766154,
+      "learning_rate": 8.562598389369902e-07,
+      "loss": 0.1038,
+      "step": 2189
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.45928585791735976,
+      "learning_rate": 8.474600126594984e-07,
+      "loss": 0.1062,
+      "step": 2190
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.617184611364542,
+      "learning_rate": 8.387048587041535e-07,
+      "loss": 0.1231,
+      "step": 2191
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3135713669359545,
+      "learning_rate": 8.299943932644105e-07,
+      "loss": 0.105,
+      "step": 2192
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 5.719539897664096,
+      "learning_rate": 8.213286324510738e-07,
+      "loss": 0.1111,
+      "step": 2193
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4403734676238107,
+      "learning_rate": 8.127075922922472e-07,
+      "loss": 0.1016,
+      "step": 2194
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.8908390427326282,
+      "learning_rate": 8.041312887333397e-07,
+      "loss": 0.1406,
+      "step": 2195
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5101285707610961,
+      "learning_rate": 7.955997376369984e-07,
+      "loss": 0.1194,
+      "step": 2196
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3084888106423402,
+      "learning_rate": 7.871129547831063e-07,
+      "loss": 0.0991,
+      "step": 2197
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.40922384799087175,
+      "learning_rate": 7.786709558687405e-07,
+      "loss": 0.1049,
+      "step": 2198
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5607398487290948,
+      "learning_rate": 7.702737565081497e-07,
+      "loss": 0.1071,
+      "step": 2199
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.37586172709977994,
+      "learning_rate": 7.619213722327185e-07,
+      "loss": 0.1153,
+      "step": 2200
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5586062857873637,
+      "learning_rate": 7.536138184909391e-07,
+      "loss": 0.1094,
+      "step": 2201
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3105225376497351,
+      "learning_rate": 7.453511106483902e-07,
+      "loss": 0.1034,
+      "step": 2202
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.0599845758794955,
+      "learning_rate": 7.371332639876994e-07,
+      "loss": 0.1189,
+      "step": 2203
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.33998170377448894,
+      "learning_rate": 7.289602937085277e-07,
+      "loss": 0.103,
+      "step": 2204
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3879257848757765,
+      "learning_rate": 7.208322149275248e-07,
+      "loss": 0.1133,
+      "step": 2205
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.302089931035153,
+      "learning_rate": 7.127490426783123e-07,
+      "loss": 0.1042,
+      "step": 2206
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.24920441590017486,
+      "learning_rate": 7.047107919114588e-07,
+      "loss": 0.1217,
+      "step": 2207
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3597352742968475,
+      "learning_rate": 6.967174774944324e-07,
+      "loss": 0.1194,
+      "step": 2208
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.35378448898103304,
+      "learning_rate": 6.887691142116042e-07,
+      "loss": 0.1126,
+      "step": 2209
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.26616865278288454,
+      "learning_rate": 6.808657167641897e-07,
+      "loss": 0.1141,
+      "step": 2210
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3299868740814219,
+      "learning_rate": 6.730072997702425e-07,
+      "loss": 0.1071,
+      "step": 2211
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.29896398978899336,
+      "learning_rate": 6.651938777646222e-07,
+      "loss": 0.1095,
+      "step": 2212
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.27053770328157795,
+      "learning_rate": 6.574254651989687e-07,
+      "loss": 0.1092,
+      "step": 2213
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3980322527172842,
+      "learning_rate": 6.497020764416633e-07,
+      "loss": 0.1135,
+      "step": 2214
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.30136797121968273,
+      "learning_rate": 6.420237257778156e-07,
+      "loss": 0.1072,
+      "step": 2215
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3942004311556152,
+      "learning_rate": 6.3439042740924e-07,
+      "loss": 0.0954,
+      "step": 2216
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4565102031505753,
+      "learning_rate": 6.268021954544096e-07,
+      "loss": 0.1227,
+      "step": 2217
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4497123867557345,
+      "learning_rate": 6.192590439484558e-07,
+      "loss": 0.1105,
+      "step": 2218
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.2848133562969757,
+      "learning_rate": 6.117609868431268e-07,
+      "loss": 0.1097,
+      "step": 2219
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.37028585869793945,
+      "learning_rate": 6.043080380067539e-07,
+      "loss": 0.1135,
+      "step": 2220
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.26470965497704907,
+      "learning_rate": 5.969002112242494e-07,
+      "loss": 0.0991,
+      "step": 2221
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5305254369187237,
+      "learning_rate": 5.895375201970616e-07,
+      "loss": 0.1102,
+      "step": 2222
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.2940121637539013,
+      "learning_rate": 5.822199785431636e-07,
+      "loss": 0.1111,
+      "step": 2223
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.38261112680236986,
+      "learning_rate": 5.749475997970072e-07,
+      "loss": 0.1107,
+      "step": 2224
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5519129170990189,
+      "learning_rate": 5.677203974095268e-07,
+      "loss": 0.121,
+      "step": 2225
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.27355589798400665,
+      "learning_rate": 5.605383847480827e-07,
+      "loss": 0.1073,
+      "step": 2226
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.26810356486763576,
+      "learning_rate": 5.534015750964683e-07,
+      "loss": 0.1035,
+      "step": 2227
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3396423095931366,
+      "learning_rate": 5.463099816548579e-07,
+      "loss": 0.1051,
+      "step": 2228
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.33266135988995077,
+      "learning_rate": 5.392636175398064e-07,
+      "loss": 0.1089,
+      "step": 2229
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.2976215141396653,
+      "learning_rate": 5.322624957841998e-07,
+      "loss": 0.1142,
+      "step": 2230
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.33842817896513644,
+      "learning_rate": 5.253066293372549e-07,
+      "loss": 0.114,
+      "step": 2231
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3626627276086838,
+      "learning_rate": 5.183960310644748e-07,
+      "loss": 0.1156,
+      "step": 2232
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.2975017224245907,
+      "learning_rate": 5.115307137476405e-07,
+      "loss": 0.1187,
+      "step": 2233
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.2784964426226879,
+      "learning_rate": 5.047106900847864e-07,
+      "loss": 0.1042,
+      "step": 2234
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.261019695267096,
+      "learning_rate": 4.979359726901639e-07,
+      "loss": 0.1088,
+      "step": 2235
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.34886045977610813,
+      "learning_rate": 4.912065740942329e-07,
+      "loss": 0.1122,
+      "step": 2236
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.41292202228588776,
+      "learning_rate": 4.84522506743626e-07,
+      "loss": 0.1042,
+      "step": 2237
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3828805114848851,
+      "learning_rate": 4.778837830011373e-07,
+      "loss": 0.1071,
+      "step": 2238
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.26961615694909463,
+      "learning_rate": 4.712904151456865e-07,
+      "loss": 0.106,
+      "step": 2239
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.39636461152040914,
+      "learning_rate": 4.647424153723101e-07,
+      "loss": 0.1113,
+      "step": 2240
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.299649861338779,
+      "learning_rate": 4.582397957921286e-07,
+      "loss": 0.1037,
+      "step": 2241
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3178438710488053,
+      "learning_rate": 4.517825684323324e-07,
+      "loss": 0.1002,
+      "step": 2242
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.4699388774673982,
+      "learning_rate": 4.4537074523615106e-07,
+      "loss": 0.1037,
+      "step": 2243
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3190159532206101,
+      "learning_rate": 4.390043380628317e-07,
+      "loss": 0.1108,
+      "step": 2244
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3277649318453508,
+      "learning_rate": 4.326833586876272e-07,
+      "loss": 0.0975,
+      "step": 2245
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.38417207820613397,
+      "learning_rate": 4.264078188017662e-07,
+      "loss": 0.097,
+      "step": 2246
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.5347688992935326,
+      "learning_rate": 4.201777300124249e-07,
+      "loss": 0.1066,
+      "step": 2247
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.35390178856598153,
+      "learning_rate": 4.139931038427275e-07,
+      "loss": 0.1092,
+      "step": 2248
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.170470007855778,
+      "learning_rate": 4.078539517316987e-07,
+      "loss": 0.1033,
+      "step": 2249
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.4288041550451131,
+      "learning_rate": 4.0176028503425835e-07,
+      "loss": 0.1084,
+      "step": 2250
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.49504577038902314,
+      "learning_rate": 3.9571211502120175e-07,
+      "loss": 0.1231,
+      "step": 2251
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3815303898181039,
+      "learning_rate": 3.897094528791667e-07,
+      "loss": 0.1022,
+      "step": 2252
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4309187114571592,
+      "learning_rate": 3.837523097106277e-07,
+      "loss": 0.1063,
+      "step": 2253
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.304433599060903,
+      "learning_rate": 3.778406965338599e-07,
+      "loss": 0.0961,
+      "step": 2254
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.27788033120956923,
+      "learning_rate": 3.719746242829308e-07,
+      "loss": 0.0999,
+      "step": 2255
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.344656564903897,
+      "learning_rate": 3.6615410380767544e-07,
+      "loss": 0.1139,
+      "step": 2256
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4707452434153392,
+      "learning_rate": 3.603791458736766e-07,
+      "loss": 0.1108,
+      "step": 2257
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.27291688279811527,
+      "learning_rate": 3.546497611622429e-07,
+      "loss": 0.0926,
+      "step": 2258
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4267163623155552,
+      "learning_rate": 3.4896596027039763e-07,
+      "loss": 0.1155,
+      "step": 2259
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.31036696952617915,
+      "learning_rate": 3.4332775371084815e-07,
+      "loss": 0.1111,
+      "step": 2260
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.27387646986144937,
+      "learning_rate": 3.377351519119665e-07,
+      "loss": 0.106,
+      "step": 2261
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.29828217442644833,
+      "learning_rate": 3.321881652177783e-07,
+      "loss": 0.1252,
+      "step": 2262
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4248203323258589,
+      "learning_rate": 3.266868038879434e-07,
+      "loss": 0.1012,
+      "step": 2263
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.515097727377051,
+      "learning_rate": 3.212310780977307e-07,
+      "loss": 0.1006,
+      "step": 2264
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.26103547570584973,
+      "learning_rate": 3.1582099793800444e-07,
+      "loss": 0.1008,
+      "step": 2265
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.30431767238144697,
+      "learning_rate": 3.104565734151965e-07,
+      "loss": 0.1069,
+      "step": 2266
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3819708364494844,
+      "learning_rate": 3.0513781445129783e-07,
+      "loss": 0.1079,
+      "step": 2267
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.29786488737842814,
+      "learning_rate": 2.9986473088384203e-07,
+      "loss": 0.107,
+      "step": 2268
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3732149482124049,
+      "learning_rate": 2.946373324658719e-07,
+      "loss": 0.1027,
+      "step": 2269
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.5070817476646035,
+      "learning_rate": 2.894556288659395e-07,
+      "loss": 0.1097,
+      "step": 2270
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3203906652244629,
+      "learning_rate": 2.8431962966807836e-07,
+      "loss": 0.1025,
+      "step": 2271
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.2853851344008999,
+      "learning_rate": 2.7922934437178695e-07,
+      "loss": 0.1193,
+      "step": 2272
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.32385843442148715,
+      "learning_rate": 2.741847823920091e-07,
+      "loss": 0.1063,
+      "step": 2273
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.2901565022123352,
+      "learning_rate": 2.691859530591229e-07,
+      "loss": 0.1038,
+      "step": 2274
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3500133328801598,
+      "learning_rate": 2.6423286561892136e-07,
+      "loss": 0.1041,
+      "step": 2275
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.2747530940352288,
+      "learning_rate": 2.593255292325847e-07,
+      "loss": 0.1204,
+      "step": 2276
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3528957752844534,
+      "learning_rate": 2.544639529766829e-07,
+      "loss": 0.1125,
+      "step": 2277
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.380475874588395,
+      "learning_rate": 2.4964814584314264e-07,
+      "loss": 0.1131,
+      "step": 2278
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.2974767283861541,
+      "learning_rate": 2.448781167392361e-07,
+      "loss": 0.1065,
+      "step": 2279
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.9770526851876286,
+      "learning_rate": 2.4015387448756975e-07,
+      "loss": 0.0999,
+      "step": 2280
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3045915982323744,
+      "learning_rate": 2.354754278260568e-07,
+      "loss": 0.1188,
+      "step": 2281
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.32653345671873985,
+      "learning_rate": 2.3084278540791427e-07,
+      "loss": 0.109,
+      "step": 2282
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.36060061415984684,
+      "learning_rate": 2.262559558016325e-07,
+      "loss": 0.1073,
+      "step": 2283
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3944621023292748,
+      "learning_rate": 2.2171494749097245e-07,
+      "loss": 0.0951,
+      "step": 2284
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3235663445833311,
+      "learning_rate": 2.1721976887494055e-07,
+      "loss": 0.1088,
+      "step": 2285
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.27751375139700685,
+      "learning_rate": 2.127704282677806e-07,
+      "loss": 0.1018,
+      "step": 2286
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.8460954606767217,
+      "learning_rate": 2.0836693389895135e-07,
+      "loss": 0.1014,
+      "step": 2287
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.35284582497404704,
+      "learning_rate": 2.0400929391312107e-07,
+      "loss": 0.1154,
+      "step": 2288
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.30824839587612146,
+      "learning_rate": 1.99697516370137e-07,
+      "loss": 0.1034,
+      "step": 2289
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.8160486568923421,
+      "learning_rate": 1.954316092450281e-07,
+      "loss": 0.1244,
+      "step": 2290
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.24318000632622622,
+      "learning_rate": 1.912115804279746e-07,
+      "loss": 0.1072,
+      "step": 2291
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.471166531961367,
+      "learning_rate": 1.870374377243078e-07,
+      "loss": 0.0976,
+      "step": 2292
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3416562185729039,
+      "learning_rate": 1.8290918885447983e-07,
+      "loss": 0.0986,
+      "step": 2293
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.22988491818540524,
+      "learning_rate": 1.7882684145406614e-07,
+      "loss": 0.1022,
+      "step": 2294
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.51596414809335,
+      "learning_rate": 1.7479040307374073e-07,
+      "loss": 0.1008,
+      "step": 2295
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.31196666393364186,
+      "learning_rate": 1.7079988117926493e-07,
+      "loss": 0.1094,
+      "step": 2296
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.4191208314175795,
+      "learning_rate": 1.6685528315146803e-07,
+      "loss": 0.1023,
+      "step": 2297
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.2773513000942579,
+      "learning_rate": 1.6295661628624447e-07,
+      "loss": 0.102,
+      "step": 2298
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.39334069183781983,
+      "learning_rate": 1.5910388779453721e-07,
+      "loss": 0.1175,
+      "step": 2299
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.29533005497805154,
+      "learning_rate": 1.5529710480231276e-07,
+      "loss": 0.093,
+      "step": 2300
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.31480965541398565,
+      "learning_rate": 1.515362743505694e-07,
+      "loss": 0.1057,
+      "step": 2301
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.6314455989349365,
+      "learning_rate": 1.4782140339530127e-07,
+      "loss": 0.1028,
+      "step": 2302
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4713640762857292,
+      "learning_rate": 1.4415249880750103e-07,
+      "loss": 0.1169,
+      "step": 2303
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.30541731952886925,
+      "learning_rate": 1.4052956737314326e-07,
+      "loss": 0.1052,
+      "step": 2304
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.2846539052274588,
+      "learning_rate": 1.3695261579316777e-07,
+      "loss": 0.1186,
+      "step": 2305
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.42831314360098244,
+      "learning_rate": 1.3342165068347125e-07,
+      "loss": 0.1197,
+      "step": 2306
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.25488635328054055,
+      "learning_rate": 1.29936678574899e-07,
+      "loss": 0.1115,
+      "step": 2307
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.5033554438526455,
+      "learning_rate": 1.2649770591322274e-07,
+      "loss": 0.1044,
+      "step": 2308
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.6737192119277112,
+      "learning_rate": 1.23104739059135e-07,
+      "loss": 0.1078,
+      "step": 2309
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.34888616608889783,
+      "learning_rate": 1.1975778428823524e-07,
+      "loss": 0.1038,
+      "step": 2310
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.2790875246978235,
+      "learning_rate": 1.164568477910244e-07,
+      "loss": 0.1057,
+      "step": 2311
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.24174404598308313,
+      "learning_rate": 1.1320193567288529e-07,
+      "loss": 0.0921,
+      "step": 2312
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.5802166115672507,
+      "learning_rate": 1.0999305395407167e-07,
+      "loss": 0.1195,
+      "step": 2313
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3649966822664672,
+      "learning_rate": 1.068302085697026e-07,
+      "loss": 0.116,
+      "step": 2314
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3791584313337832,
+      "learning_rate": 1.0371340536975139e-07,
+      "loss": 0.1158,
+      "step": 2315
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.28890187440023896,
+      "learning_rate": 1.006426501190233e-07,
+      "loss": 0.113,
+      "step": 2316
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.342412437537247,
+      "learning_rate": 9.761794849716676e-08,
+      "loss": 0.1126,
+      "step": 2317
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3160835731079688,
+      "learning_rate": 9.463930609864002e-08,
+      "loss": 0.1116,
+      "step": 2318
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.37104948438436997,
+      "learning_rate": 9.170672843271666e-08,
+      "loss": 0.0976,
+      "step": 2319
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.2998772264249333,
+      "learning_rate": 8.882022092346065e-08,
+      "loss": 0.0974,
+      "step": 2320
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3308332183773306,
+      "learning_rate": 8.59797889097319e-08,
+      "loss": 0.1026,
+      "step": 2321
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3040238656888801,
+      "learning_rate": 8.318543764516961e-08,
+      "loss": 0.1166,
+      "step": 2322
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5371142214265257,
+      "learning_rate": 8.043717229817837e-08,
+      "loss": 0.1044,
+      "step": 2323
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.24539667785126662,
+      "learning_rate": 7.773499795193095e-08,
+      "loss": 0.1049,
+      "step": 2324
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3841007071620765,
+      "learning_rate": 7.507891960433778e-08,
+      "loss": 0.1205,
+      "step": 2325
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.2902215224007677,
+      "learning_rate": 7.246894216806355e-08,
+      "loss": 0.1195,
+      "step": 2326
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.2958634005988022,
+      "learning_rate": 6.990507047049676e-08,
+      "loss": 0.1219,
+      "step": 2327
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.2539677667228061,
+      "learning_rate": 6.73873092537497e-08,
+      "loss": 0.1072,
+      "step": 2328
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.2905214366568891,
+      "learning_rate": 6.491566317465558e-08,
+      "loss": 0.1021,
+      "step": 2329
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.28895941409118997,
+      "learning_rate": 6.249013680474369e-08,
+      "loss": 0.1133,
+      "step": 2330
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.283080581796697,
+      "learning_rate": 6.011073463024764e-08,
+      "loss": 0.1133,
+      "step": 2331
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4223359697778957,
+      "learning_rate": 5.7777461052091474e-08,
+      "loss": 0.0971,
+      "step": 2332
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3688472910349817,
+      "learning_rate": 5.54903203858731e-08,
+      "loss": 0.119,
+      "step": 2333
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.2961213411104044,
+      "learning_rate": 5.324931686186973e-08,
+      "loss": 0.1005,
+      "step": 2334
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4020515382243391,
+      "learning_rate": 5.1054454625021344e-08,
+      "loss": 0.1019,
+      "step": 2335
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3597523388323017,
+      "learning_rate": 4.890573773492502e-08,
+      "loss": 0.1061,
+      "step": 2336
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.33827224727901256,
+      "learning_rate": 4.680317016582669e-08,
+      "loss": 0.1027,
+      "step": 2337
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.38978834012889246,
+      "learning_rate": 4.474675580662113e-08,
+      "loss": 0.1057,
+      "step": 2338
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3601720977366954,
+      "learning_rate": 4.2736498460829724e-08,
+      "loss": 0.1145,
+      "step": 2339
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.33949388377186007,
+      "learning_rate": 4.07724018466088e-08,
+      "loss": 0.0939,
+      "step": 2340
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.30957134505600525,
+      "learning_rate": 3.8854469596730224e-08,
+      "loss": 0.1017,
+      "step": 2341
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4227721827288734,
+      "learning_rate": 3.698270525858971e-08,
+      "loss": 0.104,
+      "step": 2342
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.33467066618632146,
+      "learning_rate": 3.515711229418184e-08,
+      "loss": 0.1089,
+      "step": 2343
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.24022679886116027,
+      "learning_rate": 3.3377694080111156e-08,
+      "loss": 0.0984,
+      "step": 2344
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.36654272289301515,
+      "learning_rate": 3.1644453907567206e-08,
+      "loss": 0.1156,
+      "step": 2345
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.29594325589052567,
+      "learning_rate": 2.995739498234118e-08,
+      "loss": 0.1099,
+      "step": 2346
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.2817953919467619,
+      "learning_rate": 2.831652042480093e-08,
+      "loss": 0.1013,
+      "step": 2347
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.4243943846735716,
+      "learning_rate": 2.6721833269890974e-08,
+      "loss": 0.1056,
+      "step": 2348
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.30796883938731046,
+      "learning_rate": 2.5173336467135267e-08,
+      "loss": 0.1003,
+      "step": 2349
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.30608159009970465,
+      "learning_rate": 2.3671032880612233e-08,
+      "loss": 0.1003,
+      "step": 2350
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.39046228945358846,
+      "learning_rate": 2.221492528897695e-08,
+      "loss": 0.1158,
+      "step": 2351
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.38276889928695884,
+      "learning_rate": 2.0805016385427865e-08,
+      "loss": 0.1071,
+      "step": 2352
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3973853840976732,
+      "learning_rate": 1.944130877772343e-08,
+      "loss": 0.1138,
+      "step": 2353
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.9274701359092353,
+      "learning_rate": 1.812380498815991e-08,
+      "loss": 0.1077,
+      "step": 2354
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3594139121207124,
+      "learning_rate": 1.6852507453588016e-08,
+      "loss": 0.1275,
+      "step": 2355
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.27358884633472946,
+      "learning_rate": 1.562741852538241e-08,
+      "loss": 0.113,
+      "step": 2356
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.365858309304738,
+      "learning_rate": 1.4448540469458316e-08,
+      "loss": 0.109,
+      "step": 2357
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3473543029853427,
+      "learning_rate": 1.3315875466257677e-08,
+      "loss": 0.107,
+      "step": 2358
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.37357500897740803,
+      "learning_rate": 1.2229425610746358e-08,
+      "loss": 0.1184,
+      "step": 2359
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3475633104397232,
+      "learning_rate": 1.1189192912416934e-08,
+      "loss": 0.1154,
+      "step": 2360
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.35381879267208705,
+      "learning_rate": 1.0195179295269252e-08,
+      "loss": 0.1015,
+      "step": 2361
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.25816893069274427,
+      "learning_rate": 9.247386597824314e-09,
+      "loss": 0.1131,
+      "step": 2362
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3767343107161633,
+      "learning_rate": 8.345816573113174e-09,
+      "loss": 0.1072,
+      "step": 2363
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.4190534290573895,
+      "learning_rate": 7.49047088866861e-09,
+      "loss": 0.1202,
+      "step": 2364
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.2808099061620016,
+      "learning_rate": 6.681351126530677e-09,
+      "loss": 0.1088,
+      "step": 2365
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.36437383513736926,
+      "learning_rate": 5.918458783241154e-09,
+      "loss": 0.119,
+      "step": 2366
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.27427886071548857,
+      "learning_rate": 5.201795269837995e-09,
+      "loss": 0.0995,
+      "step": 2367
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.2969786786715016,
+      "learning_rate": 4.531361911855325e-09,
+      "loss": 0.1311,
+      "step": 2368
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.42512237509723305,
+      "learning_rate": 3.907159949323447e-09,
+      "loss": 0.1323,
+      "step": 2369
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8205134814957196,
+      "learning_rate": 3.3291905367577313e-09,
+      "loss": 0.1164,
+      "step": 2370
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.2916569434652092,
+      "learning_rate": 2.797454743164174e-09,
+      "loss": 0.1143,
+      "step": 2371
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.1092499684602326,
+      "learning_rate": 2.3119535520421674e-09,
+      "loss": 0.1155,
+      "step": 2372
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.4743728311318732,
+      "learning_rate": 1.8726878613622987e-09,
+      "loss": 0.1032,
+      "step": 2373
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.37276208979170766,
+      "learning_rate": 1.479658483594104e-09,
+      "loss": 0.1096,
+      "step": 2374
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3028700294082978,
+      "learning_rate": 1.132866145678313e-09,
+      "loss": 0.1004,
+      "step": 2375
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.2740656896211313,
+      "learning_rate": 8.323114890379513e-10,
+      "loss": 0.1094,
+      "step": 2376
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.34694539769130955,
+      "learning_rate": 5.779950695755654e-10,
+      "loss": 0.1088,
+      "step": 2377
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3034373720944762,
+      "learning_rate": 3.699173576759973e-10,
+      "loss": 0.1096,
+      "step": 2378
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.096758928484376,
+      "learning_rate": 2.0807873819528313e-10,
+      "loss": 0.0912,
+      "step": 2379
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6572999913086256,
+      "learning_rate": 9.247951046897907e-11,
+      "loss": 0.11,
+      "step": 2380
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3431866278198902,
+      "learning_rate": 2.3119888306610828e-11,
+      "loss": 0.1223,
+      "step": 2381
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.2626322006776617,
+      "learning_rate": 0.0,
+      "loss": 0.1028,
+      "step": 2382
+    },
+    {
+      "epoch": 1.0,
+      "step": 2382,
+      "total_flos": 951480335171584.0,
+      "train_loss": 0.1271405643513731,
+      "train_runtime": 17313.6163,
+      "train_samples_per_second": 11.009,
+      "train_steps_per_second": 0.138
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2382,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 30000,
+  "total_flos": 951480335171584.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}