diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5923 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 840,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0035714285714285713,
+      "grad_norm": 63.2909598676028,
+      "learning_rate": 0.0,
+      "loss": 10.9231,
+      "step": 1
+    },
+    {
+      "epoch": 0.007142857142857143,
+      "grad_norm": 64.32051917654996,
+      "learning_rate": 5.952380952380952e-07,
+      "loss": 10.9231,
+      "step": 2
+    },
+    {
+      "epoch": 0.010714285714285714,
+      "grad_norm": 62.05927927910702,
+      "learning_rate": 1.1904761904761904e-06,
+      "loss": 10.995,
+      "step": 3
+    },
+    {
+      "epoch": 0.014285714285714285,
+      "grad_norm": 62.06359449511473,
+      "learning_rate": 1.7857142857142857e-06,
+      "loss": 10.966,
+      "step": 4
+    },
+    {
+      "epoch": 0.017857142857142856,
+      "grad_norm": 64.55746045165326,
+      "learning_rate": 2.3809523809523808e-06,
+      "loss": 10.863,
+      "step": 5
+    },
+    {
+      "epoch": 0.02142857142857143,
+      "grad_norm": 70.50885850815543,
+      "learning_rate": 2.9761904761904763e-06,
+      "loss": 10.5207,
+      "step": 6
+    },
+    {
+      "epoch": 0.025,
+      "grad_norm": 111.77059065896604,
+      "learning_rate": 3.5714285714285714e-06,
+      "loss": 9.0843,
+      "step": 7
+    },
+    {
+      "epoch": 0.02857142857142857,
+      "grad_norm": 126.15250033943975,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 8.3813,
+      "step": 8
+    },
+    {
+      "epoch": 0.03214285714285714,
+      "grad_norm": 55.947093659486534,
+      "learning_rate": 4.7619047619047615e-06,
+      "loss": 3.1857,
+      "step": 9
+    },
+    {
+      "epoch": 0.03571428571428571,
+      "grad_norm": 46.96149532998265,
+      "learning_rate": 5.357142857142857e-06,
+      "loss": 2.7973,
+      "step": 10
+    },
+    {
+      "epoch": 0.039285714285714285,
+      "grad_norm": 29.415473154514356,
+      "learning_rate": 5.9523809523809525e-06,
+      "loss": 2.1328,
+      "step": 11
+    },
+    {
+      "epoch": 0.04285714285714286,
+      "grad_norm": 6.022923568028981,
+      "learning_rate": 6.547619047619048e-06,
+      "loss": 1.2311,
+      "step": 12
+    },
+    {
+      "epoch": 0.04642857142857143,
+      "grad_norm": 4.34487277273631,
+      "learning_rate": 7.142857142857143e-06,
+      "loss": 1.1888,
+      "step": 13
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.0394219197624555,
+      "learning_rate": 7.738095238095238e-06,
+      "loss": 1.0677,
+      "step": 14
+    },
+    {
+      "epoch": 0.05357142857142857,
+      "grad_norm": 2.2944335082596052,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.971,
+      "step": 15
+    },
+    {
+      "epoch": 0.05714285714285714,
+      "grad_norm": 1.522935870498678,
+      "learning_rate": 8.92857142857143e-06,
+      "loss": 0.9077,
+      "step": 16
+    },
+    {
+      "epoch": 0.060714285714285714,
+      "grad_norm": 124.78335039442808,
+      "learning_rate": 9.523809523809523e-06,
+      "loss": 1.0214,
+      "step": 17
+    },
+    {
+      "epoch": 0.06428571428571428,
+      "grad_norm": 14.928547529543073,
+      "learning_rate": 1.011904761904762e-05,
+      "loss": 0.7935,
+      "step": 18
+    },
+    {
+      "epoch": 0.06785714285714285,
+      "grad_norm": 1.496288141365076,
+      "learning_rate": 1.0714285714285714e-05,
+      "loss": 0.8129,
+      "step": 19
+    },
+    {
+      "epoch": 0.07142857142857142,
+      "grad_norm": 1.0089644605749692,
+      "learning_rate": 1.130952380952381e-05,
+      "loss": 0.7528,
+      "step": 20
+    },
+    {
+      "epoch": 0.075,
+      "grad_norm": 0.805145615284854,
+      "learning_rate": 1.1904761904761905e-05,
+      "loss": 0.735,
+      "step": 21
+    },
+    {
+      "epoch": 0.07857142857142857,
+      "grad_norm": 0.6866217260347119,
+      "learning_rate": 1.25e-05,
+      "loss": 0.7038,
+      "step": 22
+    },
+    {
+      "epoch": 0.08214285714285714,
+      "grad_norm": 0.8908378314344355,
+      "learning_rate": 1.3095238095238096e-05,
+      "loss": 0.7163,
+      "step": 23
+    },
+    {
+      "epoch": 0.08571428571428572,
+      "grad_norm": 0.8290961136864439,
+      "learning_rate": 1.3690476190476192e-05,
+      "loss": 0.6608,
+      "step": 24
+    },
+    {
+      "epoch": 0.08928571428571429,
+      "grad_norm": 0.5879196946319878,
+      "learning_rate": 1.4285714285714285e-05,
+      "loss": 0.6593,
+      "step": 25
+    },
+    {
+      "epoch": 0.09285714285714286,
+      "grad_norm": 0.5632843400750671,
+      "learning_rate": 1.4880952380952381e-05,
+      "loss": 0.6391,
+      "step": 26
+    },
+    {
+      "epoch": 0.09642857142857143,
+      "grad_norm": 0.6688568612307081,
+      "learning_rate": 1.5476190476190476e-05,
+      "loss": 0.6221,
+      "step": 27
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6027868455486277,
+      "learning_rate": 1.6071428571428572e-05,
+      "loss": 0.5923,
+      "step": 28
+    },
+    {
+      "epoch": 0.10357142857142858,
+      "grad_norm": 0.5200772854033066,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.615,
+      "step": 29
+    },
+    {
+      "epoch": 0.10714285714285714,
+      "grad_norm": 0.4926878901443814,
+      "learning_rate": 1.7261904761904763e-05,
+      "loss": 0.5765,
+      "step": 30
+    },
+    {
+      "epoch": 0.11071428571428571,
+      "grad_norm": 0.4352667593774574,
+      "learning_rate": 1.785714285714286e-05,
+      "loss": 0.5674,
+      "step": 31
+    },
+    {
+      "epoch": 0.11428571428571428,
+      "grad_norm": 0.4888702742851021,
+      "learning_rate": 1.8452380952380954e-05,
+      "loss": 0.5598,
+      "step": 32
+    },
+    {
+      "epoch": 0.11785714285714285,
+      "grad_norm": 0.46372857953630514,
+      "learning_rate": 1.9047619047619046e-05,
+      "loss": 0.5849,
+      "step": 33
+    },
+    {
+      "epoch": 0.12142857142857143,
+      "grad_norm": 0.32606791409340763,
+      "learning_rate": 1.9642857142857145e-05,
+      "loss": 0.5484,
+      "step": 34
+    },
+    {
+      "epoch": 0.125,
+      "grad_norm": 0.43097733037799724,
+      "learning_rate": 2.023809523809524e-05,
+      "loss": 0.5653,
+      "step": 35
+    },
+    {
+      "epoch": 0.12857142857142856,
+      "grad_norm": 0.44246236417916174,
+      "learning_rate": 2.0833333333333336e-05,
+      "loss": 0.5557,
+      "step": 36
+    },
+    {
+      "epoch": 0.13214285714285715,
+      "grad_norm": 0.40597817892093574,
+      "learning_rate": 2.1428571428571428e-05,
+      "loss": 0.5306,
+      "step": 37
+    },
+    {
+      "epoch": 0.1357142857142857,
+      "grad_norm": 0.3413531429640796,
+      "learning_rate": 2.2023809523809524e-05,
+      "loss": 0.552,
+      "step": 38
+    },
+    {
+      "epoch": 0.1392857142857143,
+      "grad_norm": 0.375090843516523,
+      "learning_rate": 2.261904761904762e-05,
+      "loss": 0.5362,
+      "step": 39
+    },
+    {
+      "epoch": 0.14285714285714285,
+      "grad_norm": 0.4370207719682257,
+      "learning_rate": 2.3214285714285715e-05,
+      "loss": 0.569,
+      "step": 40
+    },
+    {
+      "epoch": 0.14642857142857144,
+      "grad_norm": 0.3122725113951591,
+      "learning_rate": 2.380952380952381e-05,
+      "loss": 0.5239,
+      "step": 41
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.3000438325674209,
+      "learning_rate": 2.4404761904761906e-05,
+      "loss": 0.5184,
+      "step": 42
+    },
+    {
+      "epoch": 0.15357142857142858,
+      "grad_norm": 0.33646670669411705,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5013,
+      "step": 43
+    },
+    {
+      "epoch": 0.15714285714285714,
+      "grad_norm": 0.31948513886675145,
+      "learning_rate": 2.5595238095238093e-05,
+      "loss": 0.5286,
+      "step": 44
+    },
+    {
+      "epoch": 0.16071428571428573,
+      "grad_norm": 0.3029688423628134,
+      "learning_rate": 2.6190476190476192e-05,
+      "loss": 0.5028,
+      "step": 45
+    },
+    {
+      "epoch": 0.16428571428571428,
+      "grad_norm": 0.2924440937681386,
+      "learning_rate": 2.6785714285714288e-05,
+      "loss": 0.5129,
+      "step": 46
+    },
+    {
+      "epoch": 0.16785714285714284,
+      "grad_norm": 0.3059678924447749,
+      "learning_rate": 2.7380952380952383e-05,
+      "loss": 0.4943,
+      "step": 47
+    },
+    {
+      "epoch": 0.17142857142857143,
+      "grad_norm": 0.32671333512639517,
+      "learning_rate": 2.797619047619048e-05,
+      "loss": 0.5206,
+      "step": 48
+    },
+    {
+      "epoch": 0.175,
+      "grad_norm": 0.28713374508509426,
+      "learning_rate": 2.857142857142857e-05,
+      "loss": 0.5115,
+      "step": 49
+    },
+    {
+      "epoch": 0.17857142857142858,
+      "grad_norm": 0.33469017904235776,
+      "learning_rate": 2.916666666666667e-05,
+      "loss": 0.5114,
+      "step": 50
+    },
+    {
+      "epoch": 0.18214285714285713,
+      "grad_norm": 0.33026708212762995,
+      "learning_rate": 2.9761904761904762e-05,
+      "loss": 0.5236,
+      "step": 51
+    },
+    {
+      "epoch": 0.18571428571428572,
+      "grad_norm": 0.3013482634417232,
+      "learning_rate": 3.0357142857142857e-05,
+      "loss": 0.4919,
+      "step": 52
+    },
+    {
+      "epoch": 0.18928571428571428,
+      "grad_norm": 0.2793943955654439,
+      "learning_rate": 3.095238095238095e-05,
+      "loss": 0.5091,
+      "step": 53
+    },
+    {
+      "epoch": 0.19285714285714287,
+      "grad_norm": 0.24549969394854382,
+      "learning_rate": 3.154761904761905e-05,
+      "loss": 0.4884,
+      "step": 54
+    },
+    {
+      "epoch": 0.19642857142857142,
+      "grad_norm": 0.25660195357093535,
+      "learning_rate": 3.2142857142857144e-05,
+      "loss": 0.4731,
+      "step": 55
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.25453588138865724,
+      "learning_rate": 3.273809523809524e-05,
+      "loss": 0.4723,
+      "step": 56
+    },
+    {
+      "epoch": 0.20357142857142857,
+      "grad_norm": 0.2375164125200452,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.5101,
+      "step": 57
+    },
+    {
+      "epoch": 0.20714285714285716,
+      "grad_norm": 0.2503026075669353,
+      "learning_rate": 3.392857142857143e-05,
+      "loss": 0.4852,
+      "step": 58
+    },
+    {
+      "epoch": 0.21071428571428572,
+      "grad_norm": 0.249362662840246,
+      "learning_rate": 3.4523809523809526e-05,
+      "loss": 0.48,
+      "step": 59
+    },
+    {
+      "epoch": 0.21428571428571427,
+      "grad_norm": 0.2413801325167299,
+      "learning_rate": 3.511904761904762e-05,
+      "loss": 0.4739,
+      "step": 60
+    },
+    {
+      "epoch": 0.21785714285714286,
+      "grad_norm": 9.762479365163218,
+      "learning_rate": 3.571428571428572e-05,
+      "loss": 0.4639,
+      "step": 61
+    },
+    {
+      "epoch": 0.22142857142857142,
+      "grad_norm": 0.2818186289200878,
+      "learning_rate": 3.630952380952381e-05,
+      "loss": 0.4737,
+      "step": 62
+    },
+    {
+      "epoch": 0.225,
+      "grad_norm": 0.2275134312760537,
+      "learning_rate": 3.690476190476191e-05,
+      "loss": 0.4818,
+      "step": 63
+    },
+    {
+      "epoch": 0.22857142857142856,
+      "grad_norm": 0.2277916881964094,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.4883,
+      "step": 64
+    },
+    {
+      "epoch": 0.23214285714285715,
+      "grad_norm": 0.2748971223673055,
+      "learning_rate": 3.809523809523809e-05,
+      "loss": 0.4643,
+      "step": 65
+    },
+    {
+      "epoch": 0.2357142857142857,
+      "grad_norm": 0.2281545420927171,
+      "learning_rate": 3.8690476190476195e-05,
+      "loss": 0.4808,
+      "step": 66
+    },
+    {
+      "epoch": 0.2392857142857143,
+      "grad_norm": 0.2968996139365469,
+      "learning_rate": 3.928571428571429e-05,
+      "loss": 0.4965,
+      "step": 67
+    },
+    {
+      "epoch": 0.24285714285714285,
+      "grad_norm": 0.23799700331950313,
+      "learning_rate": 3.9880952380952386e-05,
+      "loss": 0.472,
+      "step": 68
+    },
+    {
+      "epoch": 0.24642857142857144,
+      "grad_norm": 0.27053098612062004,
+      "learning_rate": 4.047619047619048e-05,
+      "loss": 0.4559,
+      "step": 69
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.259041861754061,
+      "learning_rate": 4.107142857142857e-05,
+      "loss": 0.4681,
+      "step": 70
+    },
+    {
+      "epoch": 0.25357142857142856,
+      "grad_norm": 0.24404233122779534,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.4574,
+      "step": 71
+    },
+    {
+      "epoch": 0.2571428571428571,
+      "grad_norm": 0.2987213748361085,
+      "learning_rate": 4.226190476190476e-05,
+      "loss": 0.4844,
+      "step": 72
+    },
+    {
+      "epoch": 0.26071428571428573,
+      "grad_norm": 0.24066677179852564,
+      "learning_rate": 4.2857142857142856e-05,
+      "loss": 0.4465,
+      "step": 73
+    },
+    {
+      "epoch": 0.2642857142857143,
+      "grad_norm": 0.2621414084001365,
+      "learning_rate": 4.345238095238096e-05,
+      "loss": 0.458,
+      "step": 74
+    },
+    {
+      "epoch": 0.26785714285714285,
+      "grad_norm": 0.31069988039142715,
+      "learning_rate": 4.404761904761905e-05,
+      "loss": 0.4474,
+      "step": 75
+    },
+    {
+      "epoch": 0.2714285714285714,
+      "grad_norm": 0.22504750306783688,
+      "learning_rate": 4.464285714285715e-05,
+      "loss": 0.448,
+      "step": 76
+    },
+    {
+      "epoch": 0.275,
+      "grad_norm": 0.3147647092263142,
+      "learning_rate": 4.523809523809524e-05,
+      "loss": 0.4609,
+      "step": 77
+    },
+    {
+      "epoch": 0.2785714285714286,
+      "grad_norm": 0.22970521754176304,
+      "learning_rate": 4.5833333333333334e-05,
+      "loss": 0.4554,
+      "step": 78
+    },
+    {
+      "epoch": 0.28214285714285714,
+      "grad_norm": 0.27214869793472746,
+      "learning_rate": 4.642857142857143e-05,
+      "loss": 0.4555,
+      "step": 79
+    },
+    {
+      "epoch": 0.2857142857142857,
+      "grad_norm": 0.2551683442734359,
+      "learning_rate": 4.7023809523809525e-05,
+      "loss": 0.4571,
+      "step": 80
+    },
+    {
+      "epoch": 0.2892857142857143,
+      "grad_norm": 0.26926850053892176,
+      "learning_rate": 4.761904761904762e-05,
+      "loss": 0.4314,
+      "step": 81
+    },
+    {
+      "epoch": 0.29285714285714287,
+      "grad_norm": 0.2722600236485305,
+      "learning_rate": 4.8214285714285716e-05,
+      "loss": 0.4271,
+      "step": 82
+    },
+    {
+      "epoch": 0.29642857142857143,
+      "grad_norm": 0.27344393754444507,
+      "learning_rate": 4.880952380952381e-05,
+      "loss": 0.4559,
+      "step": 83
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.28468491697688036,
+      "learning_rate": 4.940476190476191e-05,
+      "loss": 0.4519,
+      "step": 84
+    },
+    {
+      "epoch": 0.30357142857142855,
+      "grad_norm": 0.23214295026064846,
+      "learning_rate": 5e-05,
+      "loss": 0.4398,
+      "step": 85
+    },
+    {
+      "epoch": 0.30714285714285716,
+      "grad_norm": 0.2588249550100154,
+      "learning_rate": 4.993386243386244e-05,
+      "loss": 0.4737,
+      "step": 86
+    },
+    {
+      "epoch": 0.3107142857142857,
+      "grad_norm": 0.2410386187260021,
+      "learning_rate": 4.986772486772487e-05,
+      "loss": 0.4294,
+      "step": 87
+    },
+    {
+      "epoch": 0.3142857142857143,
+      "grad_norm": 0.24950841529409806,
+      "learning_rate": 4.9801587301587306e-05,
+      "loss": 0.4412,
+      "step": 88
+    },
+    {
+      "epoch": 0.31785714285714284,
+      "grad_norm": 0.21826357893206527,
+      "learning_rate": 4.973544973544973e-05,
+      "loss": 0.4441,
+      "step": 89
+    },
+    {
+      "epoch": 0.32142857142857145,
+      "grad_norm": 0.30217242132977523,
+      "learning_rate": 4.966931216931217e-05,
+      "loss": 0.4513,
+      "step": 90
+    },
+    {
+      "epoch": 0.325,
+      "grad_norm": 0.239993738354605,
+      "learning_rate": 4.960317460317461e-05,
+      "loss": 0.4494,
+      "step": 91
+    },
+    {
+      "epoch": 0.32857142857142857,
+      "grad_norm": 0.26758264978328317,
+      "learning_rate": 4.9537037037037035e-05,
+      "loss": 0.4539,
+      "step": 92
+    },
+    {
+      "epoch": 0.33214285714285713,
+      "grad_norm": 0.24590915152595483,
+      "learning_rate": 4.9470899470899475e-05,
+      "loss": 0.4465,
+      "step": 93
+    },
+    {
+      "epoch": 0.3357142857142857,
+      "grad_norm": 0.24278870105122666,
+      "learning_rate": 4.940476190476191e-05,
+      "loss": 0.451,
+      "step": 94
+    },
+    {
+      "epoch": 0.3392857142857143,
+      "grad_norm": 0.2551125388393455,
+      "learning_rate": 4.933862433862434e-05,
+      "loss": 0.4451,
+      "step": 95
+    },
+    {
+      "epoch": 0.34285714285714286,
+      "grad_norm": 0.25412952446684417,
+      "learning_rate": 4.927248677248678e-05,
+      "loss": 0.431,
+      "step": 96
+    },
+    {
+      "epoch": 0.3464285714285714,
+      "grad_norm": 0.2961683574032674,
+      "learning_rate": 4.9206349206349204e-05,
+      "loss": 0.4523,
+      "step": 97
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.23008059190987482,
+      "learning_rate": 4.914021164021164e-05,
+      "loss": 0.4429,
+      "step": 98
+    },
+    {
+      "epoch": 0.3535714285714286,
+      "grad_norm": 0.28250989898936296,
+      "learning_rate": 4.9074074074074075e-05,
+      "loss": 0.4386,
+      "step": 99
+    },
+    {
+      "epoch": 0.35714285714285715,
+      "grad_norm": 0.269626173285728,
+      "learning_rate": 4.900793650793651e-05,
+      "loss": 0.4401,
+      "step": 100
+    },
+    {
+      "epoch": 0.3607142857142857,
+      "grad_norm": 0.2755301039891822,
+      "learning_rate": 4.894179894179895e-05,
+      "loss": 0.4355,
+      "step": 101
+    },
+    {
+      "epoch": 0.36428571428571427,
+      "grad_norm": 0.2760448520459721,
+      "learning_rate": 4.887566137566138e-05,
+      "loss": 0.4385,
+      "step": 102
+    },
+    {
+      "epoch": 0.3678571428571429,
+      "grad_norm": 0.2708128610937186,
+      "learning_rate": 4.880952380952381e-05,
+      "loss": 0.4543,
+      "step": 103
+    },
+    {
+      "epoch": 0.37142857142857144,
+      "grad_norm": 0.2547862854103764,
+      "learning_rate": 4.8743386243386244e-05,
+      "loss": 0.4263,
+      "step": 104
+    },
+    {
+      "epoch": 0.375,
+      "grad_norm": 0.29600449360827913,
+      "learning_rate": 4.8677248677248676e-05,
+      "loss": 0.4571,
+      "step": 105
+    },
+    {
+      "epoch": 0.37857142857142856,
+      "grad_norm": 0.32384641555851923,
+      "learning_rate": 4.8611111111111115e-05,
+      "loss": 0.4463,
+      "step": 106
+    },
+    {
+      "epoch": 0.3821428571428571,
+      "grad_norm": 0.2960474085530681,
+      "learning_rate": 4.854497354497355e-05,
+      "loss": 0.4149,
+      "step": 107
+    },
+    {
+      "epoch": 0.38571428571428573,
+      "grad_norm": 0.3266196897480769,
+      "learning_rate": 4.847883597883598e-05,
+      "loss": 0.422,
+      "step": 108
+    },
+    {
+      "epoch": 0.3892857142857143,
+      "grad_norm": 0.309904097486782,
+      "learning_rate": 4.841269841269841e-05,
+      "loss": 0.4308,
+      "step": 109
+    },
+    {
+      "epoch": 0.39285714285714285,
+      "grad_norm": 0.29460574315910837,
+      "learning_rate": 4.834656084656085e-05,
+      "loss": 0.4336,
+      "step": 110
+    },
+    {
+      "epoch": 0.3964285714285714,
+      "grad_norm": 0.26255967129404767,
+      "learning_rate": 4.8280423280423284e-05,
+      "loss": 0.4381,
+      "step": 111
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3052687329306859,
+      "learning_rate": 4.8214285714285716e-05,
+      "loss": 0.4339,
+      "step": 112
+    },
+    {
+      "epoch": 0.4035714285714286,
+      "grad_norm": 0.28366217638411834,
+      "learning_rate": 4.814814814814815e-05,
+      "loss": 0.4463,
+      "step": 113
+    },
+    {
+      "epoch": 0.40714285714285714,
+      "grad_norm": 0.282518291802335,
+      "learning_rate": 4.808201058201058e-05,
+      "loss": 0.4335,
+      "step": 114
+    },
+    {
+      "epoch": 0.4107142857142857,
+      "grad_norm": 0.2644295910909586,
+      "learning_rate": 4.801587301587302e-05,
+      "loss": 0.4288,
+      "step": 115
+    },
+    {
+      "epoch": 0.4142857142857143,
+      "grad_norm": 0.26488411410556284,
+      "learning_rate": 4.794973544973545e-05,
+      "loss": 0.4291,
+      "step": 116
+    },
+    {
+      "epoch": 0.41785714285714287,
+      "grad_norm": 0.30311144586789357,
+      "learning_rate": 4.7883597883597884e-05,
+      "loss": 0.4399,
+      "step": 117
+    },
+    {
+      "epoch": 0.42142857142857143,
+      "grad_norm": 0.2702893329908722,
+      "learning_rate": 4.781746031746032e-05,
+      "loss": 0.4325,
+      "step": 118
+    },
+    {
+      "epoch": 0.425,
+      "grad_norm": 0.243981921106788,
+      "learning_rate": 4.7751322751322756e-05,
+      "loss": 0.4279,
+      "step": 119
+    },
+    {
+      "epoch": 0.42857142857142855,
+      "grad_norm": 0.255682254878671,
+      "learning_rate": 4.768518518518519e-05,
+      "loss": 0.4188,
+      "step": 120
+    },
+    {
+      "epoch": 0.43214285714285716,
+      "grad_norm": 0.25097318197866175,
+      "learning_rate": 4.761904761904762e-05,
+      "loss": 0.4341,
+      "step": 121
+    },
+    {
+      "epoch": 0.4357142857142857,
+      "grad_norm": 0.2585935360830222,
+      "learning_rate": 4.755291005291005e-05,
+      "loss": 0.4361,
+      "step": 122
+    },
+    {
+      "epoch": 0.4392857142857143,
+      "grad_norm": 0.25508454797612123,
+      "learning_rate": 4.748677248677249e-05,
+      "loss": 0.4199,
+      "step": 123
+    },
+    {
+      "epoch": 0.44285714285714284,
+      "grad_norm": 0.2523301195494935,
+      "learning_rate": 4.7420634920634924e-05,
+      "loss": 0.418,
+      "step": 124
+    },
+    {
+      "epoch": 0.44642857142857145,
+      "grad_norm": 0.2700528176681201,
+      "learning_rate": 4.7354497354497356e-05,
+      "loss": 0.4339,
+      "step": 125
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.2451051756749464,
+      "learning_rate": 4.7288359788359796e-05,
+      "loss": 0.4427,
+      "step": 126
+    },
+    {
+      "epoch": 0.45357142857142857,
+      "grad_norm": 0.245562706757361,
+      "learning_rate": 4.722222222222222e-05,
+      "loss": 0.4236,
+      "step": 127
+    },
+    {
+      "epoch": 0.45714285714285713,
+      "grad_norm": 0.28423095630595446,
+      "learning_rate": 4.715608465608466e-05,
+      "loss": 0.4265,
+      "step": 128
+    },
+    {
+      "epoch": 0.4607142857142857,
+      "grad_norm": 0.23743304343118574,
+      "learning_rate": 4.708994708994709e-05,
+      "loss": 0.4167,
+      "step": 129
+    },
+    {
+      "epoch": 0.4642857142857143,
+      "grad_norm": 0.3238423796269778,
+      "learning_rate": 4.7023809523809525e-05,
+      "loss": 0.4601,
+      "step": 130
+    },
+    {
+      "epoch": 0.46785714285714286,
+      "grad_norm": 0.2311548261278586,
+      "learning_rate": 4.6957671957671964e-05,
+      "loss": 0.4244,
+      "step": 131
+    },
+    {
+      "epoch": 0.4714285714285714,
+      "grad_norm": 0.26809465843708497,
+      "learning_rate": 4.689153439153439e-05,
+      "loss": 0.4265,
+      "step": 132
+    },
+    {
+      "epoch": 0.475,
+      "grad_norm": 0.24887435791723833,
+      "learning_rate": 4.682539682539683e-05,
+      "loss": 0.4327,
+      "step": 133
+    },
+    {
+      "epoch": 0.4785714285714286,
+      "grad_norm": 0.23534998024581308,
+      "learning_rate": 4.675925925925926e-05,
+      "loss": 0.4297,
+      "step": 134
+    },
+    {
+      "epoch": 0.48214285714285715,
+      "grad_norm": 0.30099338539954895,
+      "learning_rate": 4.669312169312169e-05,
+      "loss": 0.4265,
+      "step": 135
+    },
+    {
+      "epoch": 0.4857142857142857,
+      "grad_norm": 0.22945918912139177,
+      "learning_rate": 4.662698412698413e-05,
+      "loss": 0.425,
+      "step": 136
+    },
+    {
+      "epoch": 0.48928571428571427,
+      "grad_norm": 0.2856343879201142,
+      "learning_rate": 4.656084656084656e-05,
+      "loss": 0.4197,
+      "step": 137
+    },
+    {
+      "epoch": 0.4928571428571429,
+      "grad_norm": 0.24038713435195247,
+      "learning_rate": 4.6494708994709e-05,
+      "loss": 0.407,
+      "step": 138
+    },
+    {
+      "epoch": 0.49642857142857144,
+      "grad_norm": 0.27952203501848993,
+      "learning_rate": 4.642857142857143e-05,
+      "loss": 0.4087,
+      "step": 139
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.2387761859939762,
+      "learning_rate": 4.636243386243386e-05,
+      "loss": 0.4007,
+      "step": 140
+    },
+    {
+      "epoch": 0.5035714285714286,
+      "grad_norm": 0.2990843773133364,
+      "learning_rate": 4.62962962962963e-05,
+      "loss": 0.4158,
+      "step": 141
+    },
+    {
+      "epoch": 0.5071428571428571,
+      "grad_norm": 0.2395051708147408,
+      "learning_rate": 4.623015873015873e-05,
+      "loss": 0.4195,
+      "step": 142
+    },
+    {
+      "epoch": 0.5107142857142857,
+      "grad_norm": 0.28415234831441183,
+      "learning_rate": 4.6164021164021165e-05,
+      "loss": 0.4322,
+      "step": 143
+    },
+    {
+      "epoch": 0.5142857142857142,
+      "grad_norm": 0.25542061027527535,
+      "learning_rate": 4.60978835978836e-05,
+      "loss": 0.4186,
+      "step": 144
+    },
+    {
+      "epoch": 0.5178571428571429,
+      "grad_norm": 0.25439554849335266,
+      "learning_rate": 4.603174603174603e-05,
+      "loss": 0.4327,
+      "step": 145
+    },
+    {
+      "epoch": 0.5214285714285715,
+      "grad_norm": 0.2275407464485459,
+      "learning_rate": 4.596560846560847e-05,
+      "loss": 0.4183,
+      "step": 146
+    },
+    {
+      "epoch": 0.525,
+      "grad_norm": 0.2437013225448749,
+      "learning_rate": 4.58994708994709e-05,
+      "loss": 0.4132,
+      "step": 147
+    },
+    {
+      "epoch": 0.5285714285714286,
+      "grad_norm": 0.24709156221978879,
+      "learning_rate": 4.5833333333333334e-05,
+      "loss": 0.4261,
+      "step": 148
+    },
+    {
+      "epoch": 0.5321428571428571,
+      "grad_norm": 0.2293353125764347,
+      "learning_rate": 4.576719576719577e-05,
+      "loss": 0.4118,
+      "step": 149
+    },
+    {
+      "epoch": 0.5357142857142857,
+      "grad_norm": 0.2336383233521205,
+      "learning_rate": 4.5701058201058205e-05,
+      "loss": 0.4303,
+      "step": 150
+    },
+    {
+      "epoch": 0.5392857142857143,
+      "grad_norm": 0.286992857160913,
+      "learning_rate": 4.563492063492064e-05,
+      "loss": 0.4012,
+      "step": 151
+    },
+    {
+      "epoch": 0.5428571428571428,
+      "grad_norm": 0.23683169398721024,
+      "learning_rate": 4.556878306878307e-05,
+      "loss": 0.4341,
+      "step": 152
+    },
+    {
+      "epoch": 0.5464285714285714,
+      "grad_norm": 0.26824990415868316,
+      "learning_rate": 4.55026455026455e-05,
+      "loss": 0.4054,
+      "step": 153
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.2654621301140528,
+      "learning_rate": 4.543650793650794e-05,
+      "loss": 0.4224,
+      "step": 154
+    },
+    {
+      "epoch": 0.5535714285714286,
+      "grad_norm": 0.25685386759711565,
+      "learning_rate": 4.5370370370370374e-05,
+      "loss": 0.4228,
+      "step": 155
+    },
+    {
+      "epoch": 0.5571428571428572,
+      "grad_norm": 0.27208392388218944,
+      "learning_rate": 4.5304232804232806e-05,
+      "loss": 0.4064,
+      "step": 156
+    },
+    {
+      "epoch": 0.5607142857142857,
+      "grad_norm": 0.2652994807674846,
+      "learning_rate": 4.523809523809524e-05,
+      "loss": 0.4118,
+      "step": 157
+    },
+    {
+      "epoch": 0.5642857142857143,
+      "grad_norm": 0.2737245453728529,
+      "learning_rate": 4.517195767195768e-05,
+      "loss": 0.4377,
+      "step": 158
+    },
+    {
+      "epoch": 0.5678571428571428,
+      "grad_norm": 0.24425615561533237,
+      "learning_rate": 4.510582010582011e-05,
+      "loss": 0.4166,
+      "step": 159
+    },
+    {
+      "epoch": 0.5714285714285714,
+      "grad_norm": 0.2851569690987044,
+      "learning_rate": 4.503968253968254e-05,
+      "loss": 0.4235,
+      "step": 160
+    },
+    {
+      "epoch": 0.575,
+      "grad_norm": 0.23761510252981802,
+      "learning_rate": 4.4973544973544974e-05,
+      "loss": 0.4037,
+      "step": 161
+    },
+    {
+      "epoch": 0.5785714285714286,
+      "grad_norm": 0.23964356821258676,
+      "learning_rate": 4.490740740740741e-05,
+      "loss": 0.4083,
+      "step": 162
+    },
+    {
+      "epoch": 0.5821428571428572,
+      "grad_norm": 0.29638190871296966,
+      "learning_rate": 4.4841269841269846e-05,
+      "loss": 0.4205,
+      "step": 163
+    },
+    {
+      "epoch": 0.5857142857142857,
+      "grad_norm": 0.26675058647143957,
+      "learning_rate": 4.477513227513228e-05,
+      "loss": 0.4335,
+      "step": 164
+    },
+    {
+      "epoch": 0.5892857142857143,
+      "grad_norm": 0.2814555800080778,
+      "learning_rate": 4.470899470899471e-05,
+      "loss": 0.4067,
+      "step": 165
+    },
+    {
+      "epoch": 0.5928571428571429,
+      "grad_norm": 0.27811545104460555,
+      "learning_rate": 4.464285714285715e-05,
+      "loss": 0.4181,
+      "step": 166
+    },
+    {
+      "epoch": 0.5964285714285714,
+      "grad_norm": 0.29532771522032414,
+      "learning_rate": 4.4576719576719575e-05,
+      "loss": 0.4275,
+      "step": 167
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.2783326735130141,
+      "learning_rate": 4.4510582010582014e-05,
+      "loss": 0.4099,
+      "step": 168
+    },
+    {
+      "epoch": 0.6035714285714285,
+      "grad_norm": 0.2658370303293187,
+      "learning_rate": 4.4444444444444447e-05,
+      "loss": 0.4203,
+      "step": 169
+    },
+    {
+      "epoch": 0.6071428571428571,
+      "grad_norm": 0.27122763103871084,
+      "learning_rate": 4.437830687830688e-05,
+      "loss": 0.4051,
+      "step": 170
+    },
+    {
+      "epoch": 0.6107142857142858,
+      "grad_norm": 0.2860101670180852,
+      "learning_rate": 4.431216931216932e-05,
+      "loss": 0.4198,
+      "step": 171
+    },
+    {
+      "epoch": 0.6142857142857143,
+      "grad_norm": 0.23735589793952236,
+      "learning_rate": 4.4246031746031744e-05,
+      "loss": 0.4135,
+      "step": 172
+    },
+    {
+      "epoch": 0.6178571428571429,
+      "grad_norm": 0.3151441586023076,
+      "learning_rate": 4.417989417989418e-05,
+      "loss": 0.4079,
+      "step": 173
+    },
+    {
+      "epoch": 0.6214285714285714,
+      "grad_norm": 0.26723730241064,
+      "learning_rate": 4.4113756613756615e-05,
+      "loss": 0.4241,
+      "step": 174
+    },
+    {
+      "epoch": 0.625,
+      "grad_norm": 0.2863799467230327,
+      "learning_rate": 4.404761904761905e-05,
+      "loss": 0.4141,
+      "step": 175
+    },
+    {
+      "epoch": 0.6285714285714286,
+      "grad_norm": 0.28274233897811024,
+      "learning_rate": 4.3981481481481486e-05,
+      "loss": 0.4131,
+      "step": 176
+    },
+    {
+      "epoch": 0.6321428571428571,
+      "grad_norm": 0.2351465563487619,
+      "learning_rate": 4.391534391534391e-05,
+      "loss": 0.4097,
+      "step": 177
+    },
+    {
+      "epoch": 0.6357142857142857,
+      "grad_norm": 0.2922749227878132,
+      "learning_rate": 4.384920634920635e-05,
+      "loss": 0.4014,
+      "step": 178
+    },
+    {
+      "epoch": 0.6392857142857142,
+      "grad_norm": 0.26136145292454754,
+      "learning_rate": 4.378306878306879e-05,
+      "loss": 0.4183,
+      "step": 179
+    },
+    {
+      "epoch": 0.6428571428571429,
+      "grad_norm": 0.227278661496956,
+      "learning_rate": 4.3716931216931216e-05,
+      "loss": 0.4202,
+      "step": 180
+    },
+    {
+      "epoch": 0.6464285714285715,
+      "grad_norm": 0.26076619910380017,
+      "learning_rate": 4.3650793650793655e-05,
+      "loss": 0.4406,
+      "step": 181
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.24674299698137575,
+      "learning_rate": 4.358465608465609e-05,
+      "loss": 0.4055,
+      "step": 182
+    },
+    {
+      "epoch": 0.6535714285714286,
+      "grad_norm": 0.2426322181385074,
+      "learning_rate": 4.351851851851852e-05,
+      "loss": 0.4042,
+      "step": 183
+    },
+    {
+      "epoch": 0.6571428571428571,
+      "grad_norm": 0.2415175143115033,
+      "learning_rate": 4.345238095238096e-05,
+      "loss": 0.4133,
+      "step": 184
+    },
+    {
+      "epoch": 0.6607142857142857,
+      "grad_norm": 0.22636870337636442,
+      "learning_rate": 4.3386243386243384e-05,
+      "loss": 0.411,
+      "step": 185
+    },
+    {
+      "epoch": 0.6642857142857143,
+      "grad_norm": 0.2744518855148423,
+      "learning_rate": 4.332010582010582e-05,
+      "loss": 0.4256,
+      "step": 186
+    },
+    {
+      "epoch": 0.6678571428571428,
+      "grad_norm": 0.21533148495850438,
+      "learning_rate": 4.3253968253968256e-05,
+      "loss": 0.3985,
+      "step": 187
+    },
+    {
+      "epoch": 0.6714285714285714,
+      "grad_norm": 0.2623043046032851,
+      "learning_rate": 4.318783068783069e-05,
+      "loss": 0.4215,
+      "step": 188
+    },
+    {
+      "epoch": 0.675,
+      "grad_norm": 0.2298234365468754,
+      "learning_rate": 4.312169312169313e-05,
+      "loss": 0.4042,
+      "step": 189
+    },
+    {
+      "epoch": 0.6785714285714286,
+      "grad_norm": 0.2697139147188533,
+      "learning_rate": 4.305555555555556e-05,
+      "loss": 0.4147,
+      "step": 190
+    },
+    {
+      "epoch": 0.6821428571428572,
+      "grad_norm": 0.22425975006511692,
+      "learning_rate": 4.298941798941799e-05,
+      "loss": 0.4162,
+      "step": 191
+    },
+    {
+      "epoch": 0.6857142857142857,
+      "grad_norm": 0.23539108045784032,
+      "learning_rate": 4.2923280423280424e-05,
+      "loss": 0.4125,
+      "step": 192
+    },
+    {
+      "epoch": 0.6892857142857143,
+      "grad_norm": 0.2440216598395288,
+      "learning_rate": 4.2857142857142856e-05,
+      "loss": 0.4128,
+      "step": 193
+    },
+    {
+      "epoch": 0.6928571428571428,
+      "grad_norm": 0.24859103904788782,
+      "learning_rate": 4.2791005291005295e-05,
+      "loss": 0.4008,
+      "step": 194
+    },
+    {
+      "epoch": 0.6964285714285714,
+      "grad_norm": 0.24315546314648262,
+      "learning_rate": 4.272486772486773e-05,
+      "loss": 0.4177,
+      "step": 195
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.22740583980875143,
+      "learning_rate": 4.265873015873016e-05,
+      "loss": 0.4046,
+      "step": 196
+    },
+    {
+      "epoch": 0.7035714285714286,
+      "grad_norm": 0.23966026665871448,
+      "learning_rate": 4.259259259259259e-05,
+      "loss": 0.4098,
+      "step": 197
+    },
+    {
+      "epoch": 0.7071428571428572,
+      "grad_norm": 0.22444747132306403,
+      "learning_rate": 4.252645502645503e-05,
+      "loss": 0.4056,
+      "step": 198
+    },
+    {
+      "epoch": 0.7107142857142857,
+      "grad_norm": 0.2534388690626414,
+      "learning_rate": 4.2460317460317464e-05,
+      "loss": 0.3999,
+      "step": 199
+    },
+    {
+      "epoch": 0.7142857142857143,
+      "grad_norm": 0.22570637557038012,
+      "learning_rate": 4.2394179894179896e-05,
+      "loss": 0.4021,
+      "step": 200
+    },
+    {
+      "epoch": 0.7178571428571429,
+      "grad_norm": 0.22914304254600884,
+      "learning_rate": 4.232804232804233e-05,
+      "loss": 0.397,
+      "step": 201
+    },
+    {
+      "epoch": 0.7214285714285714,
+      "grad_norm": 0.2580124480820263,
+      "learning_rate": 4.226190476190476e-05,
+      "loss": 0.4295,
+      "step": 202
+    },
+    {
+      "epoch": 0.725,
+      "grad_norm": 0.4378750602884544,
+      "learning_rate": 4.21957671957672e-05,
+      "loss": 0.425,
+      "step": 203
+    },
+    {
+      "epoch": 0.7285714285714285,
+      "grad_norm": 0.24126178240177254,
+      "learning_rate": 4.212962962962963e-05,
+      "loss": 0.3928,
+      "step": 204
+    },
+    {
+      "epoch": 0.7321428571428571,
+      "grad_norm": 0.23052024021932424,
+      "learning_rate": 4.2063492063492065e-05,
+      "loss": 0.4313,
+      "step": 205
+    },
+    {
+      "epoch": 0.7357142857142858,
+      "grad_norm": 0.2646011498598169,
+      "learning_rate": 4.1997354497354504e-05,
+      "loss": 0.4032,
+      "step": 206
+    },
+    {
+      "epoch": 0.7392857142857143,
+      "grad_norm": 0.23233463686180642,
+      "learning_rate": 4.193121693121693e-05,
+      "loss": 0.4157,
+      "step": 207
+    },
+    {
+      "epoch": 0.7428571428571429,
+      "grad_norm": 0.2659713743731251,
+      "learning_rate": 4.186507936507937e-05,
+      "loss": 0.4175,
+      "step": 208
+    },
+    {
+      "epoch": 0.7464285714285714,
+      "grad_norm": 0.26462104210938525,
+      "learning_rate": 4.17989417989418e-05,
+      "loss": 0.414,
+      "step": 209
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.26645218067931314,
+      "learning_rate": 4.173280423280423e-05,
+      "loss": 0.4105,
+      "step": 210
+    },
+    {
+      "epoch": 0.7535714285714286,
+      "grad_norm": 0.2559019869495972,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.4142,
+      "step": 211
+    },
+    {
+      "epoch": 0.7571428571428571,
+      "grad_norm": 0.23988304156040452,
+      "learning_rate": 4.16005291005291e-05,
+      "loss": 0.4017,
+      "step": 212
+    },
+    {
+      "epoch": 0.7607142857142857,
+      "grad_norm": 0.25680478833175463,
+      "learning_rate": 4.153439153439154e-05,
+      "loss": 0.3889,
+      "step": 213
+    },
+    {
+      "epoch": 0.7642857142857142,
+      "grad_norm": 0.2238786625080816,
+      "learning_rate": 4.1468253968253976e-05,
+      "loss": 0.4291,
+      "step": 214
+    },
+    {
+      "epoch": 0.7678571428571429,
+      "grad_norm": 0.2578493137182916,
+      "learning_rate": 4.14021164021164e-05,
+      "loss": 0.3985,
+      "step": 215
+    },
+    {
+      "epoch": 0.7714285714285715,
+      "grad_norm": 0.22509775707103208,
+      "learning_rate": 4.133597883597884e-05,
+      "loss": 0.4172,
+      "step": 216
+    },
+    {
+      "epoch": 0.775,
+      "grad_norm": 0.2440020259846848,
+      "learning_rate": 4.126984126984127e-05,
+      "loss": 0.4247,
+      "step": 217
+    },
+    {
+      "epoch": 0.7785714285714286,
+      "grad_norm": 0.23785580222825373,
+      "learning_rate": 4.1203703703703705e-05,
+      "loss": 0.3989,
+      "step": 218
+    },
+    {
+      "epoch": 0.7821428571428571,
+      "grad_norm": 0.20961022615181701,
+      "learning_rate": 4.1137566137566144e-05,
+      "loss": 0.4122,
+      "step": 219
+    },
+    {
+      "epoch": 0.7857142857142857,
+      "grad_norm": 0.2463653933760433,
+      "learning_rate": 4.107142857142857e-05,
+      "loss": 0.4001,
+      "step": 220
+    },
+    {
+      "epoch": 0.7892857142857143,
+      "grad_norm": 0.2216892411172763,
+      "learning_rate": 4.100529100529101e-05,
+      "loss": 0.4048,
+      "step": 221
+    },
+    {
+      "epoch": 0.7928571428571428,
+      "grad_norm": 0.23329971374849656,
+      "learning_rate": 4.093915343915344e-05,
+      "loss": 0.4165,
+      "step": 222
+    },
+    {
+      "epoch": 0.7964285714285714,
+      "grad_norm": 0.24306626848356686,
+      "learning_rate": 4.0873015873015874e-05,
+      "loss": 0.3853,
+      "step": 223
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.221940760926531,
+      "learning_rate": 4.080687830687831e-05,
+      "loss": 0.3878,
+      "step": 224
+    },
+    {
+      "epoch": 0.8035714285714286,
+      "grad_norm": 0.25785997406952393,
+      "learning_rate": 4.074074074074074e-05,
+      "loss": 0.3816,
+      "step": 225
+    },
+    {
+      "epoch": 0.8071428571428572,
+      "grad_norm": 0.2415040411158912,
+      "learning_rate": 4.067460317460318e-05,
+      "loss": 0.4029,
+      "step": 226
+    },
+    {
+      "epoch": 0.8107142857142857,
+      "grad_norm": 0.2213344065478689,
+      "learning_rate": 4.060846560846561e-05,
+      "loss": 0.3768,
+      "step": 227
+    },
+    {
+      "epoch": 0.8142857142857143,
+      "grad_norm": 0.24318423410982426,
+      "learning_rate": 4.054232804232804e-05,
+      "loss": 0.3709,
+      "step": 228
+    },
+    {
+      "epoch": 0.8178571428571428,
+      "grad_norm": 0.22460820181621888,
+      "learning_rate": 4.047619047619048e-05,
+      "loss": 0.3974,
+      "step": 229
+    },
+    {
+      "epoch": 0.8214285714285714,
+      "grad_norm": 0.2149127546515608,
+      "learning_rate": 4.041005291005291e-05,
+      "loss": 0.3753,
+      "step": 230
+    },
+    {
+      "epoch": 0.825,
+      "grad_norm": 0.2691563535325831,
+      "learning_rate": 4.0343915343915346e-05,
+      "loss": 0.3934,
+      "step": 231
+    },
+    {
+      "epoch": 0.8285714285714286,
+      "grad_norm": 0.23715026701441888,
+      "learning_rate": 4.027777777777778e-05,
+      "loss": 0.3982,
+      "step": 232
+    },
+    {
+      "epoch": 0.8321428571428572,
+      "grad_norm": 0.263142225931176,
+      "learning_rate": 4.021164021164021e-05,
+      "loss": 0.4035,
+      "step": 233
+    },
+    {
+      "epoch": 0.8357142857142857,
+      "grad_norm": 0.20257318430459098,
+      "learning_rate": 4.014550264550265e-05,
+      "loss": 0.4237,
+      "step": 234
+    },
+    {
+      "epoch": 0.8392857142857143,
+      "grad_norm": 0.20227425845081407,
+      "learning_rate": 4.007936507936508e-05,
+      "loss": 0.4063,
+      "step": 235
+    },
+    {
+      "epoch": 0.8428571428571429,
+      "grad_norm": 0.26087497658406905,
+      "learning_rate": 4.0013227513227514e-05,
+      "loss": 0.41,
+      "step": 236
+    },
+    {
+      "epoch": 0.8464285714285714,
+      "grad_norm": 0.22261831212667374,
+      "learning_rate": 3.9947089947089946e-05,
+      "loss": 0.4113,
+      "step": 237
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.23249668981731697,
+      "learning_rate": 3.9880952380952386e-05,
+      "loss": 0.3896,
+      "step": 238
+    },
+    {
+      "epoch": 0.8535714285714285,
+      "grad_norm": 0.1984011567598906,
+      "learning_rate": 3.981481481481482e-05,
+      "loss": 0.3864,
+      "step": 239
+    },
+    {
+      "epoch": 0.8571428571428571,
+      "grad_norm": 0.23109635530422953,
+      "learning_rate": 3.974867724867725e-05,
+      "loss": 0.4042,
+      "step": 240
+    },
+    {
+      "epoch": 0.8607142857142858,
+      "grad_norm": 0.23479821749222357,
+      "learning_rate": 3.968253968253968e-05,
+      "loss": 0.4175,
+      "step": 241
+    },
+    {
+      "epoch": 0.8642857142857143,
+      "grad_norm": 0.24783451053714153,
+      "learning_rate": 3.9616402116402115e-05,
+      "loss": 0.4085,
+      "step": 242
+    },
+    {
+      "epoch": 0.8678571428571429,
+      "grad_norm": 0.22718189125161153,
+      "learning_rate": 3.9550264550264554e-05,
+      "loss": 0.398,
+      "step": 243
+    },
+    {
+      "epoch": 0.8714285714285714,
+      "grad_norm": 0.23552368174031393,
+      "learning_rate": 3.9484126984126986e-05,
+      "loss": 0.3898,
+      "step": 244
+    },
+    {
+      "epoch": 0.875,
+      "grad_norm": 0.21456870439276007,
+      "learning_rate": 3.941798941798942e-05,
+      "loss": 0.3795,
+      "step": 245
+    },
+    {
+      "epoch": 0.8785714285714286,
+      "grad_norm": 0.25773881710456087,
+      "learning_rate": 3.935185185185186e-05,
+      "loss": 0.4066,
+      "step": 246
+    },
+    {
+      "epoch": 0.8821428571428571,
+      "grad_norm": 0.23908500272579614,
+      "learning_rate": 3.928571428571429e-05,
+      "loss": 0.4079,
+      "step": 247
+    },
+    {
+      "epoch": 0.8857142857142857,
+      "grad_norm": 0.24308979014035098,
+      "learning_rate": 3.921957671957672e-05,
+      "loss": 0.396,
+      "step": 248
+    },
+    {
+      "epoch": 0.8892857142857142,
+      "grad_norm": 0.22928610244720696,
+      "learning_rate": 3.9153439153439155e-05,
+      "loss": 0.4068,
+      "step": 249
+    },
+    {
+      "epoch": 0.8928571428571429,
+      "grad_norm": 0.21223778158028944,
+      "learning_rate": 3.908730158730159e-05,
+      "loss": 0.3984,
+      "step": 250
+    },
+    {
+      "epoch": 0.8964285714285715,
+      "grad_norm": 0.2920325563283691,
+      "learning_rate": 3.9021164021164026e-05,
+      "loss": 0.4196,
+      "step": 251
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.27038177338251934,
+      "learning_rate": 3.895502645502646e-05,
+      "loss": 0.3957,
+      "step": 252
+    },
+    {
+      "epoch": 0.9035714285714286,
+      "grad_norm": 0.2686736356156684,
+      "learning_rate": 3.888888888888889e-05,
+      "loss": 0.4004,
+      "step": 253
+    },
+    {
+      "epoch": 0.9071428571428571,
+      "grad_norm": 0.24034727038143308,
+      "learning_rate": 3.882275132275133e-05,
+      "loss": 0.4085,
+      "step": 254
+    },
+    {
+      "epoch": 0.9107142857142857,
+      "grad_norm": 0.26606840143746835,
+      "learning_rate": 3.8756613756613755e-05,
+      "loss": 0.4014,
+      "step": 255
+    },
+    {
+      "epoch": 0.9142857142857143,
+      "grad_norm": 0.26864675455968595,
+      "learning_rate": 3.8690476190476195e-05,
+      "loss": 0.3998,
+      "step": 256
+    },
+    {
+      "epoch": 0.9178571428571428,
+      "grad_norm": 0.22595913208009918,
+      "learning_rate": 3.862433862433863e-05,
+      "loss": 0.4034,
+      "step": 257
+    },
+    {
+      "epoch": 0.9214285714285714,
+      "grad_norm": 0.24576967114637668,
+      "learning_rate": 3.855820105820106e-05,
+      "loss": 0.3877,
+      "step": 258
+    },
+    {
+      "epoch": 0.925,
+      "grad_norm": 0.23903315513301104,
+      "learning_rate": 3.84920634920635e-05,
+      "loss": 0.3988,
+      "step": 259
+    },
+    {
+      "epoch": 0.9285714285714286,
+      "grad_norm": 0.24073663380258667,
+      "learning_rate": 3.8425925925925924e-05,
+      "loss": 0.3943,
+      "step": 260
+    },
+    {
+      "epoch": 0.9321428571428572,
+      "grad_norm": 0.24819557663261646,
+      "learning_rate": 3.835978835978836e-05,
+      "loss": 0.4013,
+      "step": 261
+    },
+    {
+      "epoch": 0.9357142857142857,
+      "grad_norm": 0.2443725602136866,
+      "learning_rate": 3.8293650793650795e-05,
+      "loss": 0.383,
+      "step": 262
+    },
+    {
+      "epoch": 0.9392857142857143,
+      "grad_norm": 0.20035392945671243,
+      "learning_rate": 3.822751322751323e-05,
+      "loss": 0.392,
+      "step": 263
+    },
+    {
+      "epoch": 0.9428571428571428,
+      "grad_norm": 0.22147050091882006,
+      "learning_rate": 3.816137566137567e-05,
+      "loss": 0.4067,
+      "step": 264
+    },
+    {
+      "epoch": 0.9464285714285714,
+      "grad_norm": 0.2276409307660139,
+      "learning_rate": 3.809523809523809e-05,
+      "loss": 0.3868,
+      "step": 265
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.22031756932029836,
+      "learning_rate": 3.802910052910053e-05,
+      "loss": 0.4129,
+      "step": 266
+    },
+    {
+      "epoch": 0.9535714285714286,
+      "grad_norm": 0.20229099540789539,
+      "learning_rate": 3.7962962962962964e-05,
+      "loss": 0.391,
+      "step": 267
+    },
+    {
+      "epoch": 0.9571428571428572,
+      "grad_norm": 0.24804874939197893,
+      "learning_rate": 3.7896825396825396e-05,
+      "loss": 0.3936,
+      "step": 268
+    },
+    {
+      "epoch": 0.9607142857142857,
+      "grad_norm": 0.230860486647842,
+      "learning_rate": 3.7830687830687835e-05,
+      "loss": 0.3899,
+      "step": 269
+    },
+    {
+      "epoch": 0.9642857142857143,
+      "grad_norm": 0.2540185502627217,
+      "learning_rate": 3.776455026455027e-05,
+      "loss": 0.4016,
+      "step": 270
+    },
+    {
+      "epoch": 0.9678571428571429,
+      "grad_norm": 0.2167633875601823,
+      "learning_rate": 3.76984126984127e-05,
+      "loss": 0.4122,
+      "step": 271
+    },
+    {
+      "epoch": 0.9714285714285714,
+      "grad_norm": 0.2732370762980548,
+      "learning_rate": 3.763227513227514e-05,
+      "loss": 0.4036,
+      "step": 272
+    },
+    {
+      "epoch": 0.975,
+      "grad_norm": 0.21168162389357492,
+      "learning_rate": 3.7566137566137564e-05,
+      "loss": 0.3873,
+      "step": 273
+    },
+    {
+      "epoch": 0.9785714285714285,
+      "grad_norm": 0.22875238906244957,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.3979,
+      "step": 274
+    },
+    {
+      "epoch": 0.9821428571428571,
+      "grad_norm": 0.2322711516974447,
+      "learning_rate": 3.7433862433862436e-05,
+      "loss": 0.4055,
+      "step": 275
+    },
+    {
+      "epoch": 0.9857142857142858,
+      "grad_norm": 0.2526956045023868,
+      "learning_rate": 3.736772486772487e-05,
+      "loss": 0.3955,
+      "step": 276
+    },
+    {
+      "epoch": 0.9892857142857143,
+      "grad_norm": 0.239905073763612,
+      "learning_rate": 3.730158730158731e-05,
+      "loss": 0.4091,
+      "step": 277
+    },
+    {
+      "epoch": 0.9928571428571429,
+      "grad_norm": 0.23579561663269719,
+      "learning_rate": 3.723544973544973e-05,
+      "loss": 0.3983,
+      "step": 278
+    },
+    {
+      "epoch": 0.9964285714285714,
+      "grad_norm": 0.2448385989114418,
+      "learning_rate": 3.716931216931217e-05,
+      "loss": 0.3895,
+      "step": 279
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.2488808028410814,
+      "learning_rate": 3.7103174603174604e-05,
+      "loss": 0.3971,
+      "step": 280
+    },
+    {
+      "epoch": 1.0035714285714286,
+      "grad_norm": 0.2698148053294866,
+      "learning_rate": 3.7037037037037037e-05,
+      "loss": 0.3304,
+      "step": 281
+    },
+    {
+      "epoch": 1.0071428571428571,
+      "grad_norm": 0.29787059318683307,
+      "learning_rate": 3.6970899470899476e-05,
+      "loss": 0.3311,
+      "step": 282
+    },
+    {
+      "epoch": 1.0107142857142857,
+      "grad_norm": 0.24043541486454786,
+      "learning_rate": 3.690476190476191e-05,
+      "loss": 0.3025,
+      "step": 283
+    },
+    {
+      "epoch": 1.0142857142857142,
+      "grad_norm": 0.2799724979964122,
+      "learning_rate": 3.683862433862434e-05,
+      "loss": 0.3193,
+      "step": 284
+    },
+    {
+      "epoch": 1.0178571428571428,
+      "grad_norm": 0.2813729843798218,
+      "learning_rate": 3.677248677248677e-05,
+      "loss": 0.3181,
+      "step": 285
+    },
+    {
+      "epoch": 1.0214285714285714,
+      "grad_norm": 0.2738275050905391,
+      "learning_rate": 3.6706349206349205e-05,
+      "loss": 0.3336,
+      "step": 286
+    },
+    {
+      "epoch": 1.025,
+      "grad_norm": 0.24807607300601758,
+      "learning_rate": 3.6640211640211644e-05,
+      "loss": 0.3112,
+      "step": 287
+    },
+    {
+      "epoch": 1.0285714285714285,
+      "grad_norm": 0.29586505992213535,
+      "learning_rate": 3.6574074074074076e-05,
+      "loss": 0.341,
+      "step": 288
+    },
+    {
+      "epoch": 1.032142857142857,
+      "grad_norm": 0.26756443828487003,
+      "learning_rate": 3.650793650793651e-05,
+      "loss": 0.321,
+      "step": 289
+    },
+    {
+      "epoch": 1.0357142857142858,
+      "grad_norm": 0.29387058464492444,
+      "learning_rate": 3.644179894179894e-05,
+      "loss": 0.3249,
+      "step": 290
+    },
+    {
+      "epoch": 1.0392857142857144,
+      "grad_norm": 0.26496962928385004,
+      "learning_rate": 3.637566137566138e-05,
+      "loss": 0.3451,
+      "step": 291
+    },
+    {
+      "epoch": 1.042857142857143,
+      "grad_norm": 0.2745844303127942,
+      "learning_rate": 3.630952380952381e-05,
+      "loss": 0.3361,
+      "step": 292
+    },
+    {
+      "epoch": 1.0464285714285715,
+      "grad_norm": 0.2751584435069459,
+      "learning_rate": 3.6243386243386245e-05,
+      "loss": 0.3242,
+      "step": 293
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.23303557949224518,
+      "learning_rate": 3.617724867724868e-05,
+      "loss": 0.3372,
+      "step": 294
+    },
+    {
+      "epoch": 1.0535714285714286,
+      "grad_norm": 0.28040670619840563,
+      "learning_rate": 3.611111111111111e-05,
+      "loss": 0.35,
+      "step": 295
+    },
+    {
+      "epoch": 1.0571428571428572,
+      "grad_norm": 0.29184317372238694,
+      "learning_rate": 3.604497354497355e-05,
+      "loss": 0.3239,
+      "step": 296
+    },
+    {
+      "epoch": 1.0607142857142857,
+      "grad_norm": 0.24229418125394608,
+      "learning_rate": 3.597883597883598e-05,
+      "loss": 0.3313,
+      "step": 297
+    },
+    {
+      "epoch": 1.0642857142857143,
+      "grad_norm": 0.24072324375762155,
+      "learning_rate": 3.591269841269841e-05,
+      "loss": 0.3237,
+      "step": 298
+    },
+    {
+      "epoch": 1.0678571428571428,
+      "grad_norm": 0.2833134707877109,
+      "learning_rate": 3.584656084656085e-05,
+      "loss": 0.3096,
+      "step": 299
+    },
+    {
+      "epoch": 1.0714285714285714,
+      "grad_norm": 0.24902535123305855,
+      "learning_rate": 3.578042328042328e-05,
+      "loss": 0.3364,
+      "step": 300
+    },
+    {
+      "epoch": 1.075,
+      "grad_norm": 0.23567472489176913,
+      "learning_rate": 3.571428571428572e-05,
+      "loss": 0.3251,
+      "step": 301
+    },
+    {
+      "epoch": 1.0785714285714285,
+      "grad_norm": 0.22714924651734877,
+      "learning_rate": 3.564814814814815e-05,
+      "loss": 0.3124,
+      "step": 302
+    },
+    {
+      "epoch": 1.082142857142857,
+      "grad_norm": 0.24562309005867838,
+      "learning_rate": 3.558201058201058e-05,
+      "loss": 0.3279,
+      "step": 303
+    },
+    {
+      "epoch": 1.0857142857142856,
+      "grad_norm": 0.22973724339159418,
+      "learning_rate": 3.551587301587302e-05,
+      "loss": 0.3168,
+      "step": 304
+    },
+    {
+      "epoch": 1.0892857142857142,
+      "grad_norm": 0.20137390322328863,
+      "learning_rate": 3.5449735449735446e-05,
+      "loss": 0.3174,
+      "step": 305
+    },
+    {
+      "epoch": 1.092857142857143,
+      "grad_norm": 0.2277877858022672,
+      "learning_rate": 3.5383597883597885e-05,
+      "loss": 0.3212,
+      "step": 306
+    },
+    {
+      "epoch": 1.0964285714285715,
+      "grad_norm": 0.23364121930336743,
+      "learning_rate": 3.5317460317460324e-05,
+      "loss": 0.2983,
+      "step": 307
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.23904953118952618,
+      "learning_rate": 3.525132275132275e-05,
+      "loss": 0.3116,
+      "step": 308
+    },
+    {
+      "epoch": 1.1035714285714286,
+      "grad_norm": 0.2019924615662405,
+      "learning_rate": 3.518518518518519e-05,
+      "loss": 0.2955,
+      "step": 309
+    },
+    {
+      "epoch": 1.1071428571428572,
+      "grad_norm": 0.25090290640968294,
+      "learning_rate": 3.511904761904762e-05,
+      "loss": 0.3193,
+      "step": 310
+    },
+    {
+      "epoch": 1.1107142857142858,
+      "grad_norm": 0.21445966828813937,
+      "learning_rate": 3.5052910052910054e-05,
+      "loss": 0.3156,
+      "step": 311
+    },
+    {
+      "epoch": 1.1142857142857143,
+      "grad_norm": 0.21489131772858427,
+      "learning_rate": 3.498677248677249e-05,
+      "loss": 0.312,
+      "step": 312
+    },
+    {
+      "epoch": 1.1178571428571429,
+      "grad_norm": 0.23514893605977252,
+      "learning_rate": 3.492063492063492e-05,
+      "loss": 0.3331,
+      "step": 313
+    },
+    {
+      "epoch": 1.1214285714285714,
+      "grad_norm": 0.2389724115268824,
+      "learning_rate": 3.485449735449736e-05,
+      "loss": 0.348,
+      "step": 314
+    },
+    {
+      "epoch": 1.125,
+      "grad_norm": 0.2001278134055622,
+      "learning_rate": 3.478835978835979e-05,
+      "loss": 0.3159,
+      "step": 315
+    },
+    {
+      "epoch": 1.1285714285714286,
+      "grad_norm": 0.19818461450565825,
+      "learning_rate": 3.472222222222222e-05,
+      "loss": 0.321,
+      "step": 316
+    },
+    {
+      "epoch": 1.1321428571428571,
+      "grad_norm": 0.20061721310501565,
+      "learning_rate": 3.465608465608466e-05,
+      "loss": 0.3152,
+      "step": 317
+    },
+    {
+      "epoch": 1.1357142857142857,
+      "grad_norm": 0.21155316484380368,
+      "learning_rate": 3.458994708994709e-05,
+      "loss": 0.3167,
+      "step": 318
+    },
+    {
+      "epoch": 1.1392857142857142,
+      "grad_norm": 0.18781590307206694,
+      "learning_rate": 3.4523809523809526e-05,
+      "loss": 0.3132,
+      "step": 319
+    },
+    {
+      "epoch": 1.1428571428571428,
+      "grad_norm": 0.21187239377014305,
+      "learning_rate": 3.445767195767196e-05,
+      "loss": 0.3437,
+      "step": 320
+    },
+    {
+      "epoch": 1.1464285714285714,
+      "grad_norm": 0.2165111470924026,
+      "learning_rate": 3.439153439153439e-05,
+      "loss": 0.3032,
+      "step": 321
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.2092350521500906,
+      "learning_rate": 3.432539682539683e-05,
+      "loss": 0.3344,
+      "step": 322
+    },
+    {
+      "epoch": 1.1535714285714285,
+      "grad_norm": 0.21119988368257825,
+      "learning_rate": 3.425925925925926e-05,
+      "loss": 0.3469,
+      "step": 323
+    },
+    {
+      "epoch": 1.157142857142857,
+      "grad_norm": 0.23852422869876627,
+      "learning_rate": 3.4193121693121694e-05,
+      "loss": 0.3271,
+      "step": 324
+    },
+    {
+      "epoch": 1.1607142857142858,
+      "grad_norm": 0.22250112661176832,
+      "learning_rate": 3.412698412698413e-05,
+      "loss": 0.3235,
+      "step": 325
+    },
+    {
+      "epoch": 1.1642857142857144,
+      "grad_norm": 0.22134395903358908,
+      "learning_rate": 3.406084656084656e-05,
+      "loss": 0.3139,
+      "step": 326
+    },
+    {
+      "epoch": 1.167857142857143,
+      "grad_norm": 0.2542356082706222,
+      "learning_rate": 3.3994708994709e-05,
+      "loss": 0.3282,
+      "step": 327
+    },
+    {
+      "epoch": 1.1714285714285715,
+      "grad_norm": 0.2054149682405068,
+      "learning_rate": 3.392857142857143e-05,
+      "loss": 0.3151,
+      "step": 328
+    },
+    {
+      "epoch": 1.175,
+      "grad_norm": 0.22979150072901058,
+      "learning_rate": 3.386243386243386e-05,
+      "loss": 0.3184,
+      "step": 329
+    },
+    {
+      "epoch": 1.1785714285714286,
+      "grad_norm": 0.22453404123938717,
+      "learning_rate": 3.3796296296296295e-05,
+      "loss": 0.3285,
+      "step": 330
+    },
+    {
+      "epoch": 1.1821428571428572,
+      "grad_norm": 0.23922789970929834,
+      "learning_rate": 3.3730158730158734e-05,
+      "loss": 0.316,
+      "step": 331
+    },
+    {
+      "epoch": 1.1857142857142857,
+      "grad_norm": 0.2221867028063624,
+      "learning_rate": 3.3664021164021167e-05,
+      "loss": 0.3312,
+      "step": 332
+    },
+    {
+      "epoch": 1.1892857142857143,
+      "grad_norm": 0.22023466333901925,
+      "learning_rate": 3.35978835978836e-05,
+      "loss": 0.3223,
+      "step": 333
+    },
+    {
+      "epoch": 1.1928571428571428,
+      "grad_norm": 0.22800515152150008,
+      "learning_rate": 3.353174603174603e-05,
+      "loss": 0.3389,
+      "step": 334
+    },
+    {
+      "epoch": 1.1964285714285714,
+      "grad_norm": 0.2213244017565832,
+      "learning_rate": 3.3465608465608464e-05,
+      "loss": 0.3099,
+      "step": 335
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.19686439444118947,
+      "learning_rate": 3.33994708994709e-05,
+      "loss": 0.3262,
+      "step": 336
+    },
+    {
+      "epoch": 1.2035714285714285,
+      "grad_norm": 0.2272820964685267,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.3149,
+      "step": 337
+    },
+    {
+      "epoch": 1.207142857142857,
+      "grad_norm": 0.23866396181627625,
+      "learning_rate": 3.326719576719577e-05,
+      "loss": 0.3315,
+      "step": 338
+    },
+    {
+      "epoch": 1.2107142857142856,
+      "grad_norm": 0.30018633838648967,
+      "learning_rate": 3.3201058201058206e-05,
+      "loss": 0.3531,
+      "step": 339
+    },
+    {
+      "epoch": 1.2142857142857142,
+      "grad_norm": 0.21629392974184772,
+      "learning_rate": 3.313492063492064e-05,
+      "loss": 0.3389,
+      "step": 340
+    },
+    {
+      "epoch": 1.217857142857143,
+      "grad_norm": 0.27501211744666204,
+      "learning_rate": 3.306878306878307e-05,
+      "loss": 0.3212,
+      "step": 341
+    },
+    {
+      "epoch": 1.2214285714285715,
+      "grad_norm": 0.24287612664362732,
+      "learning_rate": 3.30026455026455e-05,
+      "loss": 0.2969,
+      "step": 342
+    },
+    {
+      "epoch": 1.225,
+      "grad_norm": 0.200923745683883,
+      "learning_rate": 3.2936507936507936e-05,
+      "loss": 0.3102,
+      "step": 343
+    },
+    {
+      "epoch": 1.2285714285714286,
+      "grad_norm": 0.2541029653059413,
+      "learning_rate": 3.2870370370370375e-05,
+      "loss": 0.3101,
+      "step": 344
+    },
+    {
+      "epoch": 1.2321428571428572,
+      "grad_norm": 0.22152669833908464,
+      "learning_rate": 3.280423280423281e-05,
+      "loss": 0.3115,
+      "step": 345
+    },
+    {
+      "epoch": 1.2357142857142858,
+      "grad_norm": 0.2178591583397543,
+      "learning_rate": 3.273809523809524e-05,
+      "loss": 0.3149,
+      "step": 346
+    },
+    {
+      "epoch": 1.2392857142857143,
+      "grad_norm": 0.24126755975712302,
+      "learning_rate": 3.267195767195768e-05,
+      "loss": 0.3127,
+      "step": 347
+    },
+    {
+      "epoch": 1.2428571428571429,
+      "grad_norm": 0.24419259551386435,
+      "learning_rate": 3.2605820105820104e-05,
+      "loss": 0.3345,
+      "step": 348
+    },
+    {
+      "epoch": 1.2464285714285714,
+      "grad_norm": 0.23854345121889153,
+      "learning_rate": 3.253968253968254e-05,
+      "loss": 0.3264,
+      "step": 349
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.26829491514425446,
+      "learning_rate": 3.2473544973544976e-05,
+      "loss": 0.3154,
+      "step": 350
+    },
+    {
+      "epoch": 1.2535714285714286,
+      "grad_norm": 0.24653312147713308,
+      "learning_rate": 3.240740740740741e-05,
+      "loss": 0.3168,
+      "step": 351
+    },
+    {
+      "epoch": 1.2571428571428571,
+      "grad_norm": 0.2479227335685136,
+      "learning_rate": 3.234126984126985e-05,
+      "loss": 0.3138,
+      "step": 352
+    },
+    {
+      "epoch": 1.2607142857142857,
+      "grad_norm": 0.28341469669919156,
+      "learning_rate": 3.227513227513227e-05,
+      "loss": 0.3302,
+      "step": 353
+    },
+    {
+      "epoch": 1.2642857142857142,
+      "grad_norm": 0.2342861766051803,
+      "learning_rate": 3.220899470899471e-05,
+      "loss": 0.3212,
+      "step": 354
+    },
+    {
+      "epoch": 1.2678571428571428,
+      "grad_norm": 0.2548844664588017,
+      "learning_rate": 3.2142857142857144e-05,
+      "loss": 0.3164,
+      "step": 355
+    },
+    {
+      "epoch": 1.2714285714285714,
+      "grad_norm": 0.21561104699960448,
+      "learning_rate": 3.2076719576719576e-05,
+      "loss": 0.3073,
+      "step": 356
+    },
+    {
+      "epoch": 1.275,
+      "grad_norm": 0.23613202295085758,
+      "learning_rate": 3.2010582010582015e-05,
+      "loss": 0.3365,
+      "step": 357
+    },
+    {
+      "epoch": 1.2785714285714285,
+      "grad_norm": 0.22827752938645943,
+      "learning_rate": 3.194444444444444e-05,
+      "loss": 0.318,
+      "step": 358
+    },
+    {
+      "epoch": 1.282142857142857,
+      "grad_norm": 0.2043181038028336,
+      "learning_rate": 3.187830687830688e-05,
+      "loss": 0.3165,
+      "step": 359
+    },
+    {
+      "epoch": 1.2857142857142856,
+      "grad_norm": 0.22529150141771823,
+      "learning_rate": 3.181216931216931e-05,
+      "loss": 0.3093,
+      "step": 360
+    },
+    {
+      "epoch": 1.2892857142857144,
+      "grad_norm": 0.19661747971704258,
+      "learning_rate": 3.1746031746031745e-05,
+      "loss": 0.314,
+      "step": 361
+    },
+    {
+      "epoch": 1.292857142857143,
+      "grad_norm": 0.2527114427346794,
+      "learning_rate": 3.1679894179894184e-05,
+      "loss": 0.3153,
+      "step": 362
+    },
+    {
+      "epoch": 1.2964285714285715,
+      "grad_norm": 0.2559317507920466,
+      "learning_rate": 3.1613756613756616e-05,
+      "loss": 0.3396,
+      "step": 363
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.21212107399992033,
+      "learning_rate": 3.154761904761905e-05,
+      "loss": 0.3308,
+      "step": 364
+    },
+    {
+      "epoch": 1.3035714285714286,
+      "grad_norm": 0.19996745238012004,
+      "learning_rate": 3.148148148148148e-05,
+      "loss": 0.3147,
+      "step": 365
+    },
+    {
+      "epoch": 1.3071428571428572,
+      "grad_norm": 0.25108404182287347,
+      "learning_rate": 3.141534391534391e-05,
+      "loss": 0.3278,
+      "step": 366
+    },
+    {
+      "epoch": 1.3107142857142857,
+      "grad_norm": 0.2408955388306113,
+      "learning_rate": 3.134920634920635e-05,
+      "loss": 0.2989,
+      "step": 367
+    },
+    {
+      "epoch": 1.3142857142857143,
+      "grad_norm": 0.1985698918120325,
+      "learning_rate": 3.1283068783068784e-05,
+      "loss": 0.3175,
+      "step": 368
+    },
+    {
+      "epoch": 1.3178571428571428,
+      "grad_norm": 0.2133709106770074,
+      "learning_rate": 3.121693121693122e-05,
+      "loss": 0.3226,
+      "step": 369
+    },
+    {
+      "epoch": 1.3214285714285714,
+      "grad_norm": 0.24062496857349938,
+      "learning_rate": 3.1150793650793656e-05,
+      "loss": 0.3298,
+      "step": 370
+    },
+    {
+      "epoch": 1.325,
+      "grad_norm": 0.22570905262563243,
+      "learning_rate": 3.108465608465609e-05,
+      "loss": 0.3164,
+      "step": 371
+    },
+    {
+      "epoch": 1.3285714285714285,
+      "grad_norm": 0.2216946004644928,
+      "learning_rate": 3.101851851851852e-05,
+      "loss": 0.3094,
+      "step": 372
+    },
+    {
+      "epoch": 1.332142857142857,
+      "grad_norm": 0.24283219042356244,
+      "learning_rate": 3.095238095238095e-05,
+      "loss": 0.3228,
+      "step": 373
+    },
+    {
+      "epoch": 1.3357142857142856,
+      "grad_norm": 0.20515002332219523,
+      "learning_rate": 3.0886243386243385e-05,
+      "loss": 0.3144,
+      "step": 374
+    },
+    {
+      "epoch": 1.3392857142857144,
+      "grad_norm": 0.260843446000358,
+      "learning_rate": 3.0820105820105824e-05,
+      "loss": 0.329,
+      "step": 375
+    },
+    {
+      "epoch": 1.342857142857143,
+      "grad_norm": 0.22590319491488972,
+      "learning_rate": 3.075396825396826e-05,
+      "loss": 0.3362,
+      "step": 376
+    },
+    {
+      "epoch": 1.3464285714285715,
+      "grad_norm": 0.20350969651674122,
+      "learning_rate": 3.068783068783069e-05,
+      "loss": 0.3059,
+      "step": 377
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.20611847354125643,
+      "learning_rate": 3.062169312169312e-05,
+      "loss": 0.2939,
+      "step": 378
+    },
+    {
+      "epoch": 1.3535714285714286,
+      "grad_norm": 0.21221324628906074,
+      "learning_rate": 3.055555555555556e-05,
+      "loss": 0.3288,
+      "step": 379
+    },
+    {
+      "epoch": 1.3571428571428572,
+      "grad_norm": 0.22140549617387273,
+      "learning_rate": 3.048941798941799e-05,
+      "loss": 0.3074,
+      "step": 380
+    },
+    {
+      "epoch": 1.3607142857142858,
+      "grad_norm": 0.22600940520728127,
+      "learning_rate": 3.0423280423280425e-05,
+      "loss": 0.3201,
+      "step": 381
+    },
+    {
+      "epoch": 1.3642857142857143,
+      "grad_norm": 0.2234212970881212,
+      "learning_rate": 3.0357142857142857e-05,
+      "loss": 0.3154,
+      "step": 382
+    },
+    {
+      "epoch": 1.3678571428571429,
+      "grad_norm": 0.23099207984503414,
+      "learning_rate": 3.0291005291005293e-05,
+      "loss": 0.3333,
+      "step": 383
+    },
+    {
+      "epoch": 1.3714285714285714,
+      "grad_norm": 0.23672897421810646,
+      "learning_rate": 3.022486772486773e-05,
+      "loss": 0.3306,
+      "step": 384
+    },
+    {
+      "epoch": 1.375,
+      "grad_norm": 0.21729277773509562,
+      "learning_rate": 3.0158730158730158e-05,
+      "loss": 0.3181,
+      "step": 385
+    },
+    {
+      "epoch": 1.3785714285714286,
+      "grad_norm": 0.22425038550065843,
+      "learning_rate": 3.0092592592592593e-05,
+      "loss": 0.3026,
+      "step": 386
+    },
+    {
+      "epoch": 1.3821428571428571,
+      "grad_norm": 0.25228506635706743,
+      "learning_rate": 3.002645502645503e-05,
+      "loss": 0.3201,
+      "step": 387
+    },
+    {
+      "epoch": 1.3857142857142857,
+      "grad_norm": 0.23343869756517963,
+      "learning_rate": 2.996031746031746e-05,
+      "loss": 0.3217,
+      "step": 388
+    },
+    {
+      "epoch": 1.3892857142857142,
+      "grad_norm": 0.8598302341191663,
+      "learning_rate": 2.9894179894179897e-05,
+      "loss": 0.3653,
+      "step": 389
+    },
+    {
+      "epoch": 1.3928571428571428,
+      "grad_norm": 0.19532195529440863,
+      "learning_rate": 2.9828042328042326e-05,
+      "loss": 0.3119,
+      "step": 390
+    },
+    {
+      "epoch": 1.3964285714285714,
+      "grad_norm": 0.20471206531930983,
+      "learning_rate": 2.9761904761904762e-05,
+      "loss": 0.3226,
+      "step": 391
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.20413127461432676,
+      "learning_rate": 2.9695767195767198e-05,
+      "loss": 0.3284,
+      "step": 392
+    },
+    {
+      "epoch": 1.4035714285714285,
+      "grad_norm": 0.19623116493458143,
+      "learning_rate": 2.962962962962963e-05,
+      "loss": 0.3183,
+      "step": 393
+    },
+    {
+      "epoch": 1.407142857142857,
+      "grad_norm": 0.21206567728526776,
+      "learning_rate": 2.9563492063492066e-05,
+      "loss": 0.3093,
+      "step": 394
+    },
+    {
+      "epoch": 1.4107142857142856,
+      "grad_norm": 0.20617095503392882,
+      "learning_rate": 2.94973544973545e-05,
+      "loss": 0.3187,
+      "step": 395
+    },
+    {
+      "epoch": 1.4142857142857144,
+      "grad_norm": 0.23685195138619927,
+      "learning_rate": 2.943121693121693e-05,
+      "loss": 0.3342,
+      "step": 396
+    },
+    {
+      "epoch": 1.417857142857143,
+      "grad_norm": 0.21346685351721095,
+      "learning_rate": 2.9365079365079366e-05,
+      "loss": 0.3266,
+      "step": 397
+    },
+    {
+      "epoch": 1.4214285714285715,
+      "grad_norm": 0.23472126939283025,
+      "learning_rate": 2.92989417989418e-05,
+      "loss": 0.3078,
+      "step": 398
+    },
+    {
+      "epoch": 1.425,
+      "grad_norm": 0.23985227072574442,
+      "learning_rate": 2.9232804232804234e-05,
+      "loss": 0.3292,
+      "step": 399
+    },
+    {
+      "epoch": 1.4285714285714286,
+      "grad_norm": 0.2300004385676989,
+      "learning_rate": 2.916666666666667e-05,
+      "loss": 0.3142,
+      "step": 400
+    },
+    {
+      "epoch": 1.4321428571428572,
+      "grad_norm": 0.21126502782167725,
+      "learning_rate": 2.91005291005291e-05,
+      "loss": 0.3131,
+      "step": 401
+    },
+    {
+      "epoch": 1.4357142857142857,
+      "grad_norm": 0.2538536591176184,
+      "learning_rate": 2.9034391534391538e-05,
+      "loss": 0.3176,
+      "step": 402
+    },
+    {
+      "epoch": 1.4392857142857143,
+      "grad_norm": 0.23384353289727688,
+      "learning_rate": 2.8968253968253974e-05,
+      "loss": 0.3365,
+      "step": 403
+    },
+    {
+      "epoch": 1.4428571428571428,
+      "grad_norm": 0.2427166057898595,
+      "learning_rate": 2.8902116402116402e-05,
+      "loss": 0.3203,
+      "step": 404
+    },
+    {
+      "epoch": 1.4464285714285714,
+      "grad_norm": 0.23992698913276472,
+      "learning_rate": 2.8835978835978838e-05,
+      "loss": 0.3219,
+      "step": 405
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.5861576988467794,
+      "learning_rate": 2.876984126984127e-05,
+      "loss": 0.3537,
+      "step": 406
+    },
+    {
+      "epoch": 1.4535714285714285,
+      "grad_norm": 0.31328887308387066,
+      "learning_rate": 2.8703703703703706e-05,
+      "loss": 0.3386,
+      "step": 407
+    },
+    {
+      "epoch": 1.457142857142857,
+      "grad_norm": 0.23240544010689496,
+      "learning_rate": 2.8637566137566142e-05,
+      "loss": 0.3174,
+      "step": 408
+    },
+    {
+      "epoch": 1.4607142857142856,
+      "grad_norm": 0.2776270309370733,
+      "learning_rate": 2.857142857142857e-05,
+      "loss": 0.321,
+      "step": 409
+    },
+    {
+      "epoch": 1.4642857142857144,
+      "grad_norm": 0.23938979812686723,
+      "learning_rate": 2.8505291005291007e-05,
+      "loss": 0.3162,
+      "step": 410
+    },
+    {
+      "epoch": 1.467857142857143,
+      "grad_norm": 0.2000828647462872,
+      "learning_rate": 2.8439153439153442e-05,
+      "loss": 0.3251,
+      "step": 411
+    },
+    {
+      "epoch": 1.4714285714285715,
+      "grad_norm": 0.29320943218223916,
+      "learning_rate": 2.8373015873015875e-05,
+      "loss": 0.3342,
+      "step": 412
+    },
+    {
+      "epoch": 1.475,
+      "grad_norm": 0.23319584435226867,
+      "learning_rate": 2.830687830687831e-05,
+      "loss": 0.3118,
+      "step": 413
+    },
+    {
+      "epoch": 1.4785714285714286,
+      "grad_norm": 0.20265214177059132,
+      "learning_rate": 2.824074074074074e-05,
+      "loss": 0.3143,
+      "step": 414
+    },
+    {
+      "epoch": 1.4821428571428572,
+      "grad_norm": 0.29261456441843703,
+      "learning_rate": 2.8174603174603175e-05,
+      "loss": 0.3234,
+      "step": 415
+    },
+    {
+      "epoch": 1.4857142857142858,
+      "grad_norm": 0.21902541055251015,
+      "learning_rate": 2.810846560846561e-05,
+      "loss": 0.3252,
+      "step": 416
+    },
+    {
+      "epoch": 1.4892857142857143,
+      "grad_norm": 0.20725428596574114,
+      "learning_rate": 2.8042328042328043e-05,
+      "loss": 0.3209,
+      "step": 417
+    },
+    {
+      "epoch": 1.4928571428571429,
+      "grad_norm": 0.2037255763102093,
+      "learning_rate": 2.797619047619048e-05,
+      "loss": 0.3008,
+      "step": 418
+    },
+    {
+      "epoch": 1.4964285714285714,
+      "grad_norm": 0.2247878306895995,
+      "learning_rate": 2.7910052910052914e-05,
+      "loss": 0.3237,
+      "step": 419
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.20265912038997164,
+      "learning_rate": 2.7843915343915343e-05,
+      "loss": 0.3185,
+      "step": 420
+    },
+    {
+      "epoch": 1.5035714285714286,
+      "grad_norm": 0.20958471843211285,
+      "learning_rate": 2.777777777777778e-05,
+      "loss": 0.3216,
+      "step": 421
+    },
+    {
+      "epoch": 1.5071428571428571,
+      "grad_norm": 0.1990650121427334,
+      "learning_rate": 2.771164021164021e-05,
+      "loss": 0.3209,
+      "step": 422
+    },
+    {
+      "epoch": 1.5107142857142857,
+      "grad_norm": 0.19458844775945988,
+      "learning_rate": 2.7645502645502647e-05,
+      "loss": 0.3104,
+      "step": 423
+    },
+    {
+      "epoch": 1.5142857142857142,
+      "grad_norm": 0.20350657720039508,
+      "learning_rate": 2.7579365079365083e-05,
+      "loss": 0.3184,
+      "step": 424
+    },
+    {
+      "epoch": 1.5178571428571428,
+      "grad_norm": 0.2007877704693728,
+      "learning_rate": 2.7513227513227512e-05,
+      "loss": 0.3294,
+      "step": 425
+    },
+    {
+      "epoch": 1.5214285714285714,
+      "grad_norm": 0.2060081466193361,
+      "learning_rate": 2.7447089947089948e-05,
+      "loss": 0.3117,
+      "step": 426
+    },
+    {
+      "epoch": 1.525,
+      "grad_norm": 0.1994491380622727,
+      "learning_rate": 2.7380952380952383e-05,
+      "loss": 0.3275,
+      "step": 427
+    },
+    {
+      "epoch": 1.5285714285714285,
+      "grad_norm": 0.217161191053233,
+      "learning_rate": 2.7314814814814816e-05,
+      "loss": 0.333,
+      "step": 428
+    },
+    {
+      "epoch": 1.532142857142857,
+      "grad_norm": 0.2142065113708073,
+      "learning_rate": 2.724867724867725e-05,
+      "loss": 0.3326,
+      "step": 429
+    },
+    {
+      "epoch": 1.5357142857142856,
+      "grad_norm": 0.2222056169146798,
+      "learning_rate": 2.718253968253968e-05,
+      "loss": 0.3152,
+      "step": 430
+    },
+    {
+      "epoch": 1.5392857142857141,
+      "grad_norm": 0.19991300270010534,
+      "learning_rate": 2.7116402116402116e-05,
+      "loss": 0.3159,
+      "step": 431
+    },
+    {
+      "epoch": 1.5428571428571427,
+      "grad_norm": 0.227016895055148,
+      "learning_rate": 2.7050264550264555e-05,
+      "loss": 0.3292,
+      "step": 432
+    },
+    {
+      "epoch": 1.5464285714285713,
+      "grad_norm": 0.21633576708437588,
+      "learning_rate": 2.6984126984126984e-05,
+      "loss": 0.3226,
+      "step": 433
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.21008517987023625,
+      "learning_rate": 2.691798941798942e-05,
+      "loss": 0.3352,
+      "step": 434
+    },
+    {
+      "epoch": 1.5535714285714286,
+      "grad_norm": 0.22215488980314949,
+      "learning_rate": 2.6851851851851855e-05,
+      "loss": 0.3166,
+      "step": 435
+    },
+    {
+      "epoch": 1.5571428571428572,
+      "grad_norm": 0.21289955891792955,
+      "learning_rate": 2.6785714285714288e-05,
+      "loss": 0.3238,
+      "step": 436
+    },
+    {
+      "epoch": 1.5607142857142857,
+      "grad_norm": 0.1939401079014445,
+      "learning_rate": 2.6719576719576723e-05,
+      "loss": 0.3379,
+      "step": 437
+    },
+    {
+      "epoch": 1.5642857142857143,
+      "grad_norm": 0.20162658366119965,
+      "learning_rate": 2.6653439153439152e-05,
+      "loss": 0.3126,
+      "step": 438
+    },
+    {
+      "epoch": 1.5678571428571428,
+      "grad_norm": 0.1924644358372577,
+      "learning_rate": 2.6587301587301588e-05,
+      "loss": 0.3073,
+      "step": 439
+    },
+    {
+      "epoch": 1.5714285714285714,
+      "grad_norm": 0.17972308541497478,
+      "learning_rate": 2.6521164021164024e-05,
+      "loss": 0.3037,
+      "step": 440
+    },
+    {
+      "epoch": 1.575,
+      "grad_norm": 0.21359235703650079,
+      "learning_rate": 2.6455026455026456e-05,
+      "loss": 0.3141,
+      "step": 441
+    },
+    {
+      "epoch": 1.5785714285714287,
+      "grad_norm": 0.1955585930360594,
+      "learning_rate": 2.6388888888888892e-05,
+      "loss": 0.3064,
+      "step": 442
+    },
+    {
+      "epoch": 1.5821428571428573,
+      "grad_norm": 0.21090455325075624,
+      "learning_rate": 2.6322751322751328e-05,
+      "loss": 0.3114,
+      "step": 443
+    },
+    {
+      "epoch": 1.5857142857142859,
+      "grad_norm": 0.20711435484343846,
+      "learning_rate": 2.6256613756613757e-05,
+      "loss": 0.3162,
+      "step": 444
+    },
+    {
+      "epoch": 1.5892857142857144,
+      "grad_norm": 0.21194449354663925,
+      "learning_rate": 2.6190476190476192e-05,
+      "loss": 0.3117,
+      "step": 445
+    },
+    {
+      "epoch": 1.592857142857143,
+      "grad_norm": 0.18482680962266437,
+      "learning_rate": 2.6124338624338625e-05,
+      "loss": 0.3212,
+      "step": 446
+    },
+    {
+      "epoch": 1.5964285714285715,
+      "grad_norm": 0.21950983031164695,
+      "learning_rate": 2.605820105820106e-05,
+      "loss": 0.3332,
+      "step": 447
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.21212890641722967,
+      "learning_rate": 2.5992063492063496e-05,
+      "loss": 0.3079,
+      "step": 448
+    },
+    {
+      "epoch": 1.6035714285714286,
+      "grad_norm": 0.191054730836077,
+      "learning_rate": 2.5925925925925925e-05,
+      "loss": 0.3313,
+      "step": 449
+    },
+    {
+      "epoch": 1.6071428571428572,
+      "grad_norm": 0.2005372337118027,
+      "learning_rate": 2.585978835978836e-05,
+      "loss": 0.2997,
+      "step": 450
+    },
+    {
+      "epoch": 1.6107142857142858,
+      "grad_norm": 0.21322581888379724,
+      "learning_rate": 2.5793650793650796e-05,
+      "loss": 0.3194,
+      "step": 451
+    },
+    {
+      "epoch": 1.6142857142857143,
+      "grad_norm": 0.19768718741922656,
+      "learning_rate": 2.572751322751323e-05,
+      "loss": 0.3125,
+      "step": 452
+    },
+    {
+      "epoch": 1.6178571428571429,
+      "grad_norm": 0.20067951909069295,
+      "learning_rate": 2.5661375661375664e-05,
+      "loss": 0.3071,
+      "step": 453
+    },
+    {
+      "epoch": 1.6214285714285714,
+      "grad_norm": 0.20945964894018265,
+      "learning_rate": 2.5595238095238093e-05,
+      "loss": 0.3416,
+      "step": 454
+    },
+    {
+      "epoch": 1.625,
+      "grad_norm": 0.1973239619373787,
+      "learning_rate": 2.552910052910053e-05,
+      "loss": 0.3305,
+      "step": 455
+    },
+    {
+      "epoch": 1.6285714285714286,
+      "grad_norm": 0.17720609931249395,
+      "learning_rate": 2.5462962962962965e-05,
+      "loss": 0.3169,
+      "step": 456
+    },
+    {
+      "epoch": 1.6321428571428571,
+      "grad_norm": 0.22060569625951437,
+      "learning_rate": 2.5396825396825397e-05,
+      "loss": 0.3162,
+      "step": 457
+    },
+    {
+      "epoch": 1.6357142857142857,
+      "grad_norm": 0.20083477106080036,
+      "learning_rate": 2.5330687830687833e-05,
+      "loss": 0.312,
+      "step": 458
+    },
+    {
+      "epoch": 1.6392857142857142,
+      "grad_norm": 0.19956890421189233,
+      "learning_rate": 2.526455026455027e-05,
+      "loss": 0.3187,
+      "step": 459
+    },
+    {
+      "epoch": 1.6428571428571428,
+      "grad_norm": 0.19641521986301247,
+      "learning_rate": 2.5198412698412697e-05,
+      "loss": 0.3157,
+      "step": 460
+    },
+    {
+      "epoch": 1.6464285714285714,
+      "grad_norm": 0.2032607078246991,
+      "learning_rate": 2.5132275132275137e-05,
+      "loss": 0.3117,
+      "step": 461
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.1940346788122237,
+      "learning_rate": 2.5066137566137565e-05,
+      "loss": 0.3218,
+      "step": 462
+    },
+    {
+      "epoch": 1.6535714285714285,
+      "grad_norm": 0.19593388982972978,
+      "learning_rate": 2.5e-05,
+      "loss": 0.3245,
+      "step": 463
+    },
+    {
+      "epoch": 1.657142857142857,
+      "grad_norm": 0.20410176467531405,
+      "learning_rate": 2.4933862433862434e-05,
+      "loss": 0.3038,
+      "step": 464
+    },
+    {
+      "epoch": 1.6607142857142856,
+      "grad_norm": 0.1996744015277083,
+      "learning_rate": 2.4867724867724866e-05,
+      "loss": 0.3173,
+      "step": 465
+    },
+    {
+      "epoch": 1.6642857142857141,
+      "grad_norm": 0.2523244496845766,
+      "learning_rate": 2.4801587301587305e-05,
+      "loss": 0.324,
+      "step": 466
+    },
+    {
+      "epoch": 1.6678571428571427,
+      "grad_norm": 0.2017399657248579,
+      "learning_rate": 2.4735449735449737e-05,
+      "loss": 0.3111,
+      "step": 467
+    },
+    {
+      "epoch": 1.6714285714285713,
+      "grad_norm": 0.1993767338330293,
+      "learning_rate": 2.466931216931217e-05,
+      "loss": 0.3229,
+      "step": 468
+    },
+    {
+      "epoch": 1.675,
+      "grad_norm": 0.19308949653853058,
+      "learning_rate": 2.4603174603174602e-05,
+      "loss": 0.3181,
+      "step": 469
+    },
+    {
+      "epoch": 1.6785714285714286,
+      "grad_norm": 0.19181680433038142,
+      "learning_rate": 2.4537037037037038e-05,
+      "loss": 0.3079,
+      "step": 470
+    },
+    {
+      "epoch": 1.6821428571428572,
+      "grad_norm": 0.1956953415023735,
+      "learning_rate": 2.4470899470899473e-05,
+      "loss": 0.3174,
+      "step": 471
+    },
+    {
+      "epoch": 1.6857142857142857,
+      "grad_norm": 0.1892930603320641,
+      "learning_rate": 2.4404761904761906e-05,
+      "loss": 0.3215,
+      "step": 472
+    },
+    {
+      "epoch": 1.6892857142857143,
+      "grad_norm": 0.19472880714766616,
+      "learning_rate": 2.4338624338624338e-05,
+      "loss": 0.3167,
+      "step": 473
+    },
+    {
+      "epoch": 1.6928571428571428,
+      "grad_norm": 0.185898106721318,
+      "learning_rate": 2.4272486772486774e-05,
+      "loss": 0.31,
+      "step": 474
+    },
+    {
+      "epoch": 1.6964285714285714,
+      "grad_norm": 0.18970369746221966,
+      "learning_rate": 2.4206349206349206e-05,
+      "loss": 0.3172,
+      "step": 475
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.22922372319881004,
+      "learning_rate": 2.4140211640211642e-05,
+      "loss": 0.3344,
+      "step": 476
+    },
+    {
+      "epoch": 1.7035714285714287,
+      "grad_norm": 0.20186000081464095,
+      "learning_rate": 2.4074074074074074e-05,
+      "loss": 0.3196,
+      "step": 477
+    },
+    {
+      "epoch": 1.7071428571428573,
+      "grad_norm": 0.2209695735064375,
+      "learning_rate": 2.400793650793651e-05,
+      "loss": 0.3072,
+      "step": 478
+    },
+    {
+      "epoch": 1.7107142857142859,
+      "grad_norm": 0.18234840916526965,
+      "learning_rate": 2.3941798941798942e-05,
+      "loss": 0.3326,
+      "step": 479
+    },
+    {
+      "epoch": 1.7142857142857144,
+      "grad_norm": 0.19574395469048686,
+      "learning_rate": 2.3875661375661378e-05,
+      "loss": 0.2941,
+      "step": 480
+    },
+    {
+      "epoch": 1.717857142857143,
+      "grad_norm": 0.20375440568534917,
+      "learning_rate": 2.380952380952381e-05,
+      "loss": 0.3195,
+      "step": 481
+    },
+    {
+      "epoch": 1.7214285714285715,
+      "grad_norm": 0.19552782319787138,
+      "learning_rate": 2.3743386243386246e-05,
+      "loss": 0.3271,
+      "step": 482
+    },
+    {
+      "epoch": 1.725,
+      "grad_norm": 0.1882297810602148,
+      "learning_rate": 2.3677248677248678e-05,
+      "loss": 0.3122,
+      "step": 483
+    },
+    {
+      "epoch": 1.7285714285714286,
+      "grad_norm": 0.223446608383744,
+      "learning_rate": 2.361111111111111e-05,
+      "loss": 0.346,
+      "step": 484
+    },
+    {
+      "epoch": 1.7321428571428572,
+      "grad_norm": 0.2088153798369964,
+      "learning_rate": 2.3544973544973546e-05,
+      "loss": 0.3496,
+      "step": 485
+    },
+    {
+      "epoch": 1.7357142857142858,
+      "grad_norm": 0.19931941090636426,
+      "learning_rate": 2.3478835978835982e-05,
+      "loss": 0.3276,
+      "step": 486
+    },
+    {
+      "epoch": 1.7392857142857143,
+      "grad_norm": 0.2065067477608148,
+      "learning_rate": 2.3412698412698414e-05,
+      "loss": 0.3077,
+      "step": 487
+    },
+    {
+      "epoch": 1.7428571428571429,
+      "grad_norm": 0.1867609537629706,
+      "learning_rate": 2.3346560846560847e-05,
+      "loss": 0.3037,
+      "step": 488
+    },
+    {
+      "epoch": 1.7464285714285714,
+      "grad_norm": 0.20534561340093374,
+      "learning_rate": 2.328042328042328e-05,
+      "loss": 0.3161,
+      "step": 489
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.20397006275556817,
+      "learning_rate": 2.3214285714285715e-05,
+      "loss": 0.3183,
+      "step": 490
+    },
+    {
+      "epoch": 1.7535714285714286,
+      "grad_norm": 0.18550740271690497,
+      "learning_rate": 2.314814814814815e-05,
+      "loss": 0.3145,
+      "step": 491
+    },
+    {
+      "epoch": 1.7571428571428571,
+      "grad_norm": 0.2164734820955605,
+      "learning_rate": 2.3082010582010583e-05,
+      "loss": 0.329,
+      "step": 492
+    },
+    {
+      "epoch": 1.7607142857142857,
+      "grad_norm": 0.1849960524787468,
+      "learning_rate": 2.3015873015873015e-05,
+      "loss": 0.3055,
+      "step": 493
+    },
+    {
+      "epoch": 1.7642857142857142,
+      "grad_norm": 0.20223940369546317,
+      "learning_rate": 2.294973544973545e-05,
+      "loss": 0.3137,
+      "step": 494
+    },
+    {
+      "epoch": 1.7678571428571428,
+      "grad_norm": 0.2140087236470544,
+      "learning_rate": 2.2883597883597886e-05,
+      "loss": 0.3089,
+      "step": 495
+    },
+    {
+      "epoch": 1.7714285714285714,
+      "grad_norm": 0.1815126520211804,
+      "learning_rate": 2.281746031746032e-05,
+      "loss": 0.2993,
+      "step": 496
+    },
+    {
+      "epoch": 1.775,
+      "grad_norm": 0.2041694054419762,
+      "learning_rate": 2.275132275132275e-05,
+      "loss": 0.3208,
+      "step": 497
+    },
+    {
+      "epoch": 1.7785714285714285,
+      "grad_norm": 0.210298194454685,
+      "learning_rate": 2.2685185185185187e-05,
+      "loss": 0.33,
+      "step": 498
+    },
+    {
+      "epoch": 1.782142857142857,
+      "grad_norm": 0.19106104857071166,
+      "learning_rate": 2.261904761904762e-05,
+      "loss": 0.3154,
+      "step": 499
+    },
+    {
+      "epoch": 1.7857142857142856,
+      "grad_norm": 0.20668125243479243,
+      "learning_rate": 2.2552910052910055e-05,
+      "loss": 0.3298,
+      "step": 500
+    },
+    {
+      "epoch": 1.7892857142857141,
+      "grad_norm": 0.1883941632461236,
+      "learning_rate": 2.2486772486772487e-05,
+      "loss": 0.3236,
+      "step": 501
+    },
+    {
+      "epoch": 1.7928571428571427,
+      "grad_norm": 0.19354564473395905,
+      "learning_rate": 2.2420634920634923e-05,
+      "loss": 0.3079,
+      "step": 502
+    },
+    {
+      "epoch": 1.7964285714285713,
+      "grad_norm": 0.201284596806385,
+      "learning_rate": 2.2354497354497355e-05,
+      "loss": 0.3253,
+      "step": 503
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.19462419963121394,
+      "learning_rate": 2.2288359788359788e-05,
+      "loss": 0.3096,
+      "step": 504
+    },
+    {
+      "epoch": 1.8035714285714286,
+      "grad_norm": 0.18545828954299734,
+      "learning_rate": 2.2222222222222223e-05,
+      "loss": 0.3144,
+      "step": 505
+    },
+    {
+      "epoch": 1.8071428571428572,
+      "grad_norm": 0.18545753850141772,
+      "learning_rate": 2.215608465608466e-05,
+      "loss": 0.2985,
+      "step": 506
+    },
+    {
+      "epoch": 1.8107142857142857,
+      "grad_norm": 0.19859488170476142,
+      "learning_rate": 2.208994708994709e-05,
+      "loss": 0.3205,
+      "step": 507
+    },
+    {
+      "epoch": 1.8142857142857143,
+      "grad_norm": 0.18466331139884634,
+      "learning_rate": 2.2023809523809524e-05,
+      "loss": 0.3145,
+      "step": 508
+    },
+    {
+      "epoch": 1.8178571428571428,
+      "grad_norm": 0.22390415922563478,
+      "learning_rate": 2.1957671957671956e-05,
+      "loss": 0.315,
+      "step": 509
+    },
+    {
+      "epoch": 1.8214285714285714,
+      "grad_norm": 0.21124575576084206,
+      "learning_rate": 2.1891534391534395e-05,
+      "loss": 0.3237,
+      "step": 510
+    },
+    {
+      "epoch": 1.825,
+      "grad_norm": 0.20693747665063159,
+      "learning_rate": 2.1825396825396827e-05,
+      "loss": 0.3198,
+      "step": 511
+    },
+    {
+      "epoch": 1.8285714285714287,
+      "grad_norm": 0.19495576946062607,
+      "learning_rate": 2.175925925925926e-05,
+      "loss": 0.3081,
+      "step": 512
+    },
+    {
+      "epoch": 1.8321428571428573,
+      "grad_norm": 0.23868906868809317,
+      "learning_rate": 2.1693121693121692e-05,
+      "loss": 0.3232,
+      "step": 513
+    },
+    {
+      "epoch": 1.8357142857142859,
+      "grad_norm": 0.17252911380952607,
+      "learning_rate": 2.1626984126984128e-05,
+      "loss": 0.3059,
+      "step": 514
+    },
+    {
+      "epoch": 1.8392857142857144,
+      "grad_norm": 0.23501424825446657,
+      "learning_rate": 2.1560846560846563e-05,
+      "loss": 0.3194,
+      "step": 515
+    },
+    {
+      "epoch": 1.842857142857143,
+      "grad_norm": 0.21493985457796252,
+      "learning_rate": 2.1494708994708996e-05,
+      "loss": 0.3157,
+      "step": 516
+    },
+    {
+      "epoch": 1.8464285714285715,
+      "grad_norm": 0.19642554922820646,
+      "learning_rate": 2.1428571428571428e-05,
+      "loss": 0.3176,
+      "step": 517
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.1845929990522953,
+      "learning_rate": 2.1362433862433864e-05,
+      "loss": 0.3012,
+      "step": 518
+    },
+    {
+      "epoch": 1.8535714285714286,
+      "grad_norm": 0.18043634947714388,
+      "learning_rate": 2.1296296296296296e-05,
+      "loss": 0.3025,
+      "step": 519
+    },
+    {
+      "epoch": 1.8571428571428572,
+      "grad_norm": 0.1958846811143409,
+      "learning_rate": 2.1230158730158732e-05,
+      "loss": 0.2963,
+      "step": 520
+    },
+    {
+      "epoch": 1.8607142857142858,
+      "grad_norm": 0.23274522759162825,
+      "learning_rate": 2.1164021164021164e-05,
+      "loss": 0.3184,
+      "step": 521
+    },
+    {
+      "epoch": 1.8642857142857143,
+      "grad_norm": 0.20138408058426882,
+      "learning_rate": 2.10978835978836e-05,
+      "loss": 0.3215,
+      "step": 522
+    },
+    {
+      "epoch": 1.8678571428571429,
+      "grad_norm": 0.2241286592937016,
+      "learning_rate": 2.1031746031746032e-05,
+      "loss": 0.3077,
+      "step": 523
+    },
+    {
+      "epoch": 1.8714285714285714,
+      "grad_norm": 0.21406632761867334,
+      "learning_rate": 2.0965608465608465e-05,
+      "loss": 0.3058,
+      "step": 524
+    },
+    {
+      "epoch": 1.875,
+      "grad_norm": 0.21684431977850274,
+      "learning_rate": 2.08994708994709e-05,
+      "loss": 0.3153,
+      "step": 525
+    },
+    {
+      "epoch": 1.8785714285714286,
+      "grad_norm": 0.20371182399019253,
+      "learning_rate": 2.0833333333333336e-05,
+      "loss": 0.3236,
+      "step": 526
+    },
+    {
+      "epoch": 1.8821428571428571,
+      "grad_norm": 0.2137963270776002,
+      "learning_rate": 2.076719576719577e-05,
+      "loss": 0.3189,
+      "step": 527
+    },
+    {
+      "epoch": 1.8857142857142857,
+      "grad_norm": 0.20878852304240703,
+      "learning_rate": 2.07010582010582e-05,
+      "loss": 0.3106,
+      "step": 528
+    },
+    {
+      "epoch": 1.8892857142857142,
+      "grad_norm": 0.18538731070963968,
+      "learning_rate": 2.0634920634920636e-05,
+      "loss": 0.3172,
+      "step": 529
+    },
+    {
+      "epoch": 1.8928571428571428,
+      "grad_norm": 0.212624038689977,
+      "learning_rate": 2.0568783068783072e-05,
+      "loss": 0.3348,
+      "step": 530
+    },
+    {
+      "epoch": 1.8964285714285714,
+      "grad_norm": 0.19471907972129623,
+      "learning_rate": 2.0502645502645504e-05,
+      "loss": 0.3018,
+      "step": 531
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.202045861415553,
+      "learning_rate": 2.0436507936507937e-05,
+      "loss": 0.312,
+      "step": 532
+    },
+    {
+      "epoch": 1.9035714285714285,
+      "grad_norm": 0.20035244007982278,
+      "learning_rate": 2.037037037037037e-05,
+      "loss": 0.3185,
+      "step": 533
+    },
+    {
+      "epoch": 1.907142857142857,
+      "grad_norm": 0.20160058835353356,
+      "learning_rate": 2.0304232804232805e-05,
+      "loss": 0.3033,
+      "step": 534
+    },
+    {
+      "epoch": 1.9107142857142856,
+      "grad_norm": 0.17917779819107657,
+      "learning_rate": 2.023809523809524e-05,
+      "loss": 0.3084,
+      "step": 535
+    },
+    {
+      "epoch": 1.9142857142857141,
+      "grad_norm": 0.21286970360237184,
+      "learning_rate": 2.0171957671957673e-05,
+      "loss": 0.3229,
+      "step": 536
+    },
+    {
+      "epoch": 1.9178571428571427,
+      "grad_norm": 0.2080750529361991,
+      "learning_rate": 2.0105820105820105e-05,
+      "loss": 0.3192,
+      "step": 537
+    },
+    {
+      "epoch": 1.9214285714285713,
+      "grad_norm": 0.17649995719455544,
+      "learning_rate": 2.003968253968254e-05,
+      "loss": 0.313,
+      "step": 538
+    },
+    {
+      "epoch": 1.925,
+      "grad_norm": 0.20130395889298808,
+      "learning_rate": 1.9973544973544973e-05,
+      "loss": 0.3083,
+      "step": 539
+    },
+    {
+      "epoch": 1.9285714285714286,
+      "grad_norm": 0.19412311029563648,
+      "learning_rate": 1.990740740740741e-05,
+      "loss": 0.3038,
+      "step": 540
+    },
+    {
+      "epoch": 1.9321428571428572,
+      "grad_norm": 0.19071547590212706,
+      "learning_rate": 1.984126984126984e-05,
+      "loss": 0.3194,
+      "step": 541
+    },
+    {
+      "epoch": 1.9357142857142857,
+      "grad_norm": 0.19400026778264837,
+      "learning_rate": 1.9775132275132277e-05,
+      "loss": 0.3276,
+      "step": 542
+    },
+    {
+      "epoch": 1.9392857142857143,
+      "grad_norm": 0.20955171659005756,
+      "learning_rate": 1.970899470899471e-05,
+      "loss": 0.3024,
+      "step": 543
+    },
+    {
+      "epoch": 1.9428571428571428,
+      "grad_norm": 0.20219056251040063,
+      "learning_rate": 1.9642857142857145e-05,
+      "loss": 0.3241,
+      "step": 544
+    },
+    {
+      "epoch": 1.9464285714285714,
+      "grad_norm": 0.20143101487930265,
+      "learning_rate": 1.9576719576719577e-05,
+      "loss": 0.3127,
+      "step": 545
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.2131592134130349,
+      "learning_rate": 1.9510582010582013e-05,
+      "loss": 0.3048,
+      "step": 546
+    },
+    {
+      "epoch": 1.9535714285714287,
+      "grad_norm": 0.19695806604354119,
+      "learning_rate": 1.9444444444444445e-05,
+      "loss": 0.3226,
+      "step": 547
+    },
+    {
+      "epoch": 1.9571428571428573,
+      "grad_norm": 0.18131867081030673,
+      "learning_rate": 1.9378306878306878e-05,
+      "loss": 0.3119,
+      "step": 548
+    },
+    {
+      "epoch": 1.9607142857142859,
+      "grad_norm": 0.1841474316240587,
+      "learning_rate": 1.9312169312169313e-05,
+      "loss": 0.3132,
+      "step": 549
+    },
+    {
+      "epoch": 1.9642857142857144,
+      "grad_norm": 0.2041255598593875,
+      "learning_rate": 1.924603174603175e-05,
+      "loss": 0.3169,
+      "step": 550
+    },
+    {
+      "epoch": 1.967857142857143,
+      "grad_norm": 0.19664678721217965,
+      "learning_rate": 1.917989417989418e-05,
+      "loss": 0.324,
+      "step": 551
+    },
+    {
+      "epoch": 1.9714285714285715,
+      "grad_norm": 0.18677665911523328,
+      "learning_rate": 1.9113756613756614e-05,
+      "loss": 0.3106,
+      "step": 552
+    },
+    {
+      "epoch": 1.975,
+      "grad_norm": 0.19979315943157583,
+      "learning_rate": 1.9047619047619046e-05,
+      "loss": 0.322,
+      "step": 553
+    },
+    {
+      "epoch": 1.9785714285714286,
+      "grad_norm": 0.18920886270236184,
+      "learning_rate": 1.8981481481481482e-05,
+      "loss": 0.3037,
+      "step": 554
+    },
+    {
+      "epoch": 1.9821428571428572,
+      "grad_norm": 0.1884207693611572,
+      "learning_rate": 1.8915343915343918e-05,
+      "loss": 0.3037,
+      "step": 555
+    },
+    {
+      "epoch": 1.9857142857142858,
+      "grad_norm": 0.19161767629954962,
+      "learning_rate": 1.884920634920635e-05,
+      "loss": 0.3046,
+      "step": 556
+    },
+    {
+      "epoch": 1.9892857142857143,
+      "grad_norm": 0.18691004744787804,
+      "learning_rate": 1.8783068783068782e-05,
+      "loss": 0.3164,
+      "step": 557
+    },
+    {
+      "epoch": 1.9928571428571429,
+      "grad_norm": 0.21237318063393876,
+      "learning_rate": 1.8716931216931218e-05,
+      "loss": 0.3256,
+      "step": 558
+    },
+    {
+      "epoch": 1.9964285714285714,
+      "grad_norm": 0.18508104972415634,
+      "learning_rate": 1.8650793650793654e-05,
+      "loss": 0.3124,
+      "step": 559
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.19669967418713558,
+      "learning_rate": 1.8584656084656086e-05,
+      "loss": 0.2893,
+      "step": 560
+    },
+    {
+      "epoch": 2.0035714285714286,
+      "grad_norm": 0.31757959736458097,
+      "learning_rate": 1.8518518518518518e-05,
+      "loss": 0.2429,
+      "step": 561
+    },
+    {
+      "epoch": 2.007142857142857,
+      "grad_norm": 0.24051287538208022,
+      "learning_rate": 1.8452380952380954e-05,
+      "loss": 0.2338,
+      "step": 562
+    },
+    {
+      "epoch": 2.0107142857142857,
+      "grad_norm": 0.35708701225608125,
+      "learning_rate": 1.8386243386243386e-05,
+      "loss": 0.2463,
+      "step": 563
+    },
+    {
+      "epoch": 2.0142857142857142,
+      "grad_norm": 0.27029145936725385,
+      "learning_rate": 1.8320105820105822e-05,
+      "loss": 0.2278,
+      "step": 564
+    },
+    {
+      "epoch": 2.017857142857143,
+      "grad_norm": 0.22108466496281431,
+      "learning_rate": 1.8253968253968254e-05,
+      "loss": 0.2297,
+      "step": 565
+    },
+    {
+      "epoch": 2.0214285714285714,
+      "grad_norm": 0.27054584770519335,
+      "learning_rate": 1.818783068783069e-05,
+      "loss": 0.243,
+      "step": 566
+    },
+    {
+      "epoch": 2.025,
+      "grad_norm": 0.2816697647623798,
+      "learning_rate": 1.8121693121693122e-05,
+      "loss": 0.2319,
+      "step": 567
+    },
+    {
+      "epoch": 2.0285714285714285,
+      "grad_norm": 0.21231465474509292,
+      "learning_rate": 1.8055555555555555e-05,
+      "loss": 0.2262,
+      "step": 568
+    },
+    {
+      "epoch": 2.032142857142857,
+      "grad_norm": 0.23068676723435452,
+      "learning_rate": 1.798941798941799e-05,
+      "loss": 0.2355,
+      "step": 569
+    },
+    {
+      "epoch": 2.0357142857142856,
+      "grad_norm": 0.2484338643311845,
+      "learning_rate": 1.7923280423280426e-05,
+      "loss": 0.2207,
+      "step": 570
+    },
+    {
+      "epoch": 2.039285714285714,
+      "grad_norm": 0.23315870455727414,
+      "learning_rate": 1.785714285714286e-05,
+      "loss": 0.228,
+      "step": 571
+    },
+    {
+      "epoch": 2.0428571428571427,
+      "grad_norm": 0.1787910887989252,
+      "learning_rate": 1.779100529100529e-05,
+      "loss": 0.2171,
+      "step": 572
+    },
+    {
+      "epoch": 2.0464285714285713,
+      "grad_norm": 0.210443149753908,
+      "learning_rate": 1.7724867724867723e-05,
+      "loss": 0.2449,
+      "step": 573
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.24488060995967575,
+      "learning_rate": 1.7658730158730162e-05,
+      "loss": 0.2348,
+      "step": 574
+    },
+    {
+      "epoch": 2.0535714285714284,
+      "grad_norm": 0.20124943813935747,
+      "learning_rate": 1.7592592592592595e-05,
+      "loss": 0.2261,
+      "step": 575
+    },
+    {
+      "epoch": 2.057142857142857,
+      "grad_norm": 0.192769699932563,
+      "learning_rate": 1.7526455026455027e-05,
+      "loss": 0.2329,
+      "step": 576
+    },
+    {
+      "epoch": 2.0607142857142855,
+      "grad_norm": 0.21306690878838788,
+      "learning_rate": 1.746031746031746e-05,
+      "loss": 0.2431,
+      "step": 577
+    },
+    {
+      "epoch": 2.064285714285714,
+      "grad_norm": 0.21315043216210092,
+      "learning_rate": 1.7394179894179895e-05,
+      "loss": 0.2288,
+      "step": 578
+    },
+    {
+      "epoch": 2.067857142857143,
+      "grad_norm": 0.21255867957603367,
+      "learning_rate": 1.732804232804233e-05,
+      "loss": 0.2177,
+      "step": 579
+    },
+    {
+      "epoch": 2.0714285714285716,
+      "grad_norm": 0.21778161733522775,
+      "learning_rate": 1.7261904761904763e-05,
+      "loss": 0.2408,
+      "step": 580
+    },
+    {
+      "epoch": 2.075,
+      "grad_norm": 0.22847292935036695,
+      "learning_rate": 1.7195767195767195e-05,
+      "loss": 0.2476,
+      "step": 581
+    },
+    {
+      "epoch": 2.0785714285714287,
+      "grad_norm": 0.20572630306089373,
+      "learning_rate": 1.712962962962963e-05,
+      "loss": 0.2344,
+      "step": 582
+    },
+    {
+      "epoch": 2.0821428571428573,
+      "grad_norm": 0.1942532897813614,
+      "learning_rate": 1.7063492063492063e-05,
+      "loss": 0.2257,
+      "step": 583
+    },
+    {
+      "epoch": 2.085714285714286,
+      "grad_norm": 0.20237427583072712,
+      "learning_rate": 1.69973544973545e-05,
+      "loss": 0.2386,
+      "step": 584
+    },
+    {
+      "epoch": 2.0892857142857144,
+      "grad_norm": 0.20801431825809052,
+      "learning_rate": 1.693121693121693e-05,
+      "loss": 0.2313,
+      "step": 585
+    },
+    {
+      "epoch": 2.092857142857143,
+      "grad_norm": 0.2074624765266775,
+      "learning_rate": 1.6865079365079367e-05,
+      "loss": 0.2348,
+      "step": 586
+    },
+    {
+      "epoch": 2.0964285714285715,
+      "grad_norm": 0.21248202404216116,
+      "learning_rate": 1.67989417989418e-05,
+      "loss": 0.232,
+      "step": 587
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.19008889402568285,
+      "learning_rate": 1.6732804232804232e-05,
+      "loss": 0.2353,
+      "step": 588
+    },
+    {
+      "epoch": 2.1035714285714286,
+      "grad_norm": 0.2095625504656774,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.2277,
+      "step": 589
+    },
+    {
+      "epoch": 2.107142857142857,
+      "grad_norm": 0.20811980077920417,
+      "learning_rate": 1.6600529100529103e-05,
+      "loss": 0.2307,
+      "step": 590
+    },
+    {
+      "epoch": 2.1107142857142858,
+      "grad_norm": 0.20187115179675214,
+      "learning_rate": 1.6534391534391536e-05,
+      "loss": 0.2304,
+      "step": 591
+    },
+    {
+      "epoch": 2.1142857142857143,
+      "grad_norm": 0.18568179415042824,
+      "learning_rate": 1.6468253968253968e-05,
+      "loss": 0.2231,
+      "step": 592
+    },
+    {
+      "epoch": 2.117857142857143,
+      "grad_norm": 0.1878024836970878,
+      "learning_rate": 1.6402116402116404e-05,
+      "loss": 0.2392,
+      "step": 593
+    },
+    {
+      "epoch": 2.1214285714285714,
+      "grad_norm": 0.21413929742492882,
+      "learning_rate": 1.633597883597884e-05,
+      "loss": 0.23,
+      "step": 594
+    },
+    {
+      "epoch": 2.125,
+      "grad_norm": 0.22034940709172493,
+      "learning_rate": 1.626984126984127e-05,
+      "loss": 0.2351,
+      "step": 595
+    },
+    {
+      "epoch": 2.1285714285714286,
+      "grad_norm": 0.1869325238414086,
+      "learning_rate": 1.6203703703703704e-05,
+      "loss": 0.2292,
+      "step": 596
+    },
+    {
+      "epoch": 2.132142857142857,
+      "grad_norm": 0.2689584281465766,
+      "learning_rate": 1.6137566137566136e-05,
+      "loss": 0.2654,
+      "step": 597
+    },
+    {
+      "epoch": 2.1357142857142857,
+      "grad_norm": 0.2278321242448496,
+      "learning_rate": 1.6071428571428572e-05,
+      "loss": 0.2213,
+      "step": 598
+    },
+    {
+      "epoch": 2.1392857142857142,
+      "grad_norm": 0.18602657852015306,
+      "learning_rate": 1.6005291005291008e-05,
+      "loss": 0.22,
+      "step": 599
+    },
+    {
+      "epoch": 2.142857142857143,
+      "grad_norm": 0.2092726536964802,
+      "learning_rate": 1.593915343915344e-05,
+      "loss": 0.2583,
+      "step": 600
+    },
+    {
+      "epoch": 2.1464285714285714,
+      "grad_norm": 0.18689017521796633,
+      "learning_rate": 1.5873015873015872e-05,
+      "loss": 0.2437,
+      "step": 601
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.18430416782537248,
+      "learning_rate": 1.5806878306878308e-05,
+      "loss": 0.232,
+      "step": 602
+    },
+    {
+      "epoch": 2.1535714285714285,
+      "grad_norm": 0.2110638119891372,
+      "learning_rate": 1.574074074074074e-05,
+      "loss": 0.2447,
+      "step": 603
+    },
+    {
+      "epoch": 2.157142857142857,
+      "grad_norm": 0.1839045046799278,
+      "learning_rate": 1.5674603174603176e-05,
+      "loss": 0.2385,
+      "step": 604
+    },
+    {
+      "epoch": 2.1607142857142856,
+      "grad_norm": 0.19618663038770584,
+      "learning_rate": 1.560846560846561e-05,
+      "loss": 0.2158,
+      "step": 605
+    },
+    {
+      "epoch": 2.164285714285714,
+      "grad_norm": 0.23892399330993663,
+      "learning_rate": 1.5542328042328044e-05,
+      "loss": 0.252,
+      "step": 606
+    },
+    {
+      "epoch": 2.1678571428571427,
+      "grad_norm": 0.17355307018049443,
+      "learning_rate": 1.5476190476190476e-05,
+      "loss": 0.2209,
+      "step": 607
+    },
+    {
+      "epoch": 2.1714285714285713,
+      "grad_norm": 0.18191018299627368,
+      "learning_rate": 1.5410052910052912e-05,
+      "loss": 0.2253,
+      "step": 608
+    },
+    {
+      "epoch": 2.175,
+      "grad_norm": 0.17773372872210652,
+      "learning_rate": 1.5343915343915344e-05,
+      "loss": 0.2272,
+      "step": 609
+    },
+    {
+      "epoch": 2.1785714285714284,
+      "grad_norm": 0.1767151021069737,
+      "learning_rate": 1.527777777777778e-05,
+      "loss": 0.2173,
+      "step": 610
+    },
+    {
+      "epoch": 2.182142857142857,
+      "grad_norm": 0.1905216753089539,
+      "learning_rate": 1.5211640211640213e-05,
+      "loss": 0.228,
+      "step": 611
+    },
+    {
+      "epoch": 2.185714285714286,
+      "grad_norm": 0.18495310253582783,
+      "learning_rate": 1.5145502645502647e-05,
+      "loss": 0.2241,
+      "step": 612
+    },
+    {
+      "epoch": 2.189285714285714,
+      "grad_norm": 0.18491273540275943,
+      "learning_rate": 1.5079365079365079e-05,
+      "loss": 0.2288,
+      "step": 613
+    },
+    {
+      "epoch": 2.192857142857143,
+      "grad_norm": 0.17936903131585358,
+      "learning_rate": 1.5013227513227515e-05,
+      "loss": 0.2257,
+      "step": 614
+    },
+    {
+      "epoch": 2.1964285714285716,
+      "grad_norm": 0.21637527534432108,
+      "learning_rate": 1.4947089947089949e-05,
+      "loss": 0.2373,
+      "step": 615
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.18490618879345969,
+      "learning_rate": 1.4880952380952381e-05,
+      "loss": 0.225,
+      "step": 616
+    },
+    {
+      "epoch": 2.2035714285714287,
+      "grad_norm": 0.1813149978683926,
+      "learning_rate": 1.4814814814814815e-05,
+      "loss": 0.2215,
+      "step": 617
+    },
+    {
+      "epoch": 2.2071428571428573,
+      "grad_norm": 0.18875712618481552,
+      "learning_rate": 1.474867724867725e-05,
+      "loss": 0.2287,
+      "step": 618
+    },
+    {
+      "epoch": 2.210714285714286,
+      "grad_norm": 0.19542520096316007,
+      "learning_rate": 1.4682539682539683e-05,
+      "loss": 0.2187,
+      "step": 619
+    },
+    {
+      "epoch": 2.2142857142857144,
+      "grad_norm": 0.1770121220411708,
+      "learning_rate": 1.4616402116402117e-05,
+      "loss": 0.2211,
+      "step": 620
+    },
+    {
+      "epoch": 2.217857142857143,
+      "grad_norm": 0.18095568521540265,
+      "learning_rate": 1.455026455026455e-05,
+      "loss": 0.2278,
+      "step": 621
+    },
+    {
+      "epoch": 2.2214285714285715,
+      "grad_norm": 0.20368904281152592,
+      "learning_rate": 1.4484126984126987e-05,
+      "loss": 0.2375,
+      "step": 622
+    },
+    {
+      "epoch": 2.225,
+      "grad_norm": 0.18473093089055245,
+      "learning_rate": 1.4417989417989419e-05,
+      "loss": 0.2368,
+      "step": 623
+    },
+    {
+      "epoch": 2.2285714285714286,
+      "grad_norm": 0.18385819274311493,
+      "learning_rate": 1.4351851851851853e-05,
+      "loss": 0.2438,
+      "step": 624
+    },
+    {
+      "epoch": 2.232142857142857,
+      "grad_norm": 0.18116848921988157,
+      "learning_rate": 1.4285714285714285e-05,
+      "loss": 0.2123,
+      "step": 625
+    },
+    {
+      "epoch": 2.2357142857142858,
+      "grad_norm": 0.17582259461031796,
+      "learning_rate": 1.4219576719576721e-05,
+      "loss": 0.2178,
+      "step": 626
+    },
+    {
+      "epoch": 2.2392857142857143,
+      "grad_norm": 0.1893890859005074,
+      "learning_rate": 1.4153439153439155e-05,
+      "loss": 0.2418,
+      "step": 627
+    },
+    {
+      "epoch": 2.242857142857143,
+      "grad_norm": 0.19028412418223042,
+      "learning_rate": 1.4087301587301587e-05,
+      "loss": 0.2269,
+      "step": 628
+    },
+    {
+      "epoch": 2.2464285714285714,
+      "grad_norm": 0.17217433964632764,
+      "learning_rate": 1.4021164021164022e-05,
+      "loss": 0.2169,
+      "step": 629
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.17416078181980094,
+      "learning_rate": 1.3955026455026457e-05,
+      "loss": 0.2236,
+      "step": 630
+    },
+    {
+      "epoch": 2.2535714285714286,
+      "grad_norm": 0.18819656596830817,
+      "learning_rate": 1.388888888888889e-05,
+      "loss": 0.2179,
+      "step": 631
+    },
+    {
+      "epoch": 2.257142857142857,
+      "grad_norm": 0.18045712351839374,
+      "learning_rate": 1.3822751322751324e-05,
+      "loss": 0.2352,
+      "step": 632
+    },
+    {
+      "epoch": 2.2607142857142857,
+      "grad_norm": 0.17913106320019745,
+      "learning_rate": 1.3756613756613756e-05,
+      "loss": 0.2219,
+      "step": 633
+    },
+    {
+      "epoch": 2.2642857142857142,
+      "grad_norm": 0.17764846957433142,
+      "learning_rate": 1.3690476190476192e-05,
+      "loss": 0.2233,
+      "step": 634
+    },
+    {
+      "epoch": 2.267857142857143,
+      "grad_norm": 0.18111175266100066,
+      "learning_rate": 1.3624338624338626e-05,
+      "loss": 0.2236,
+      "step": 635
+    },
+    {
+      "epoch": 2.2714285714285714,
+      "grad_norm": 0.18784388127514492,
+      "learning_rate": 1.3558201058201058e-05,
+      "loss": 0.2397,
+      "step": 636
+    },
+    {
+      "epoch": 2.275,
+      "grad_norm": 0.18636770724663787,
+      "learning_rate": 1.3492063492063492e-05,
+      "loss": 0.2344,
+      "step": 637
+    },
+    {
+      "epoch": 2.2785714285714285,
+      "grad_norm": 0.19027630118705968,
+      "learning_rate": 1.3425925925925928e-05,
+      "loss": 0.232,
+      "step": 638
+    },
+    {
+      "epoch": 2.282142857142857,
+      "grad_norm": 0.18847332911572906,
+      "learning_rate": 1.3359788359788362e-05,
+      "loss": 0.2334,
+      "step": 639
+    },
+    {
+      "epoch": 2.2857142857142856,
+      "grad_norm": 0.19926534275067434,
+      "learning_rate": 1.3293650793650794e-05,
+      "loss": 0.2423,
+      "step": 640
+    },
+    {
+      "epoch": 2.289285714285714,
+      "grad_norm": 0.1722155435575498,
+      "learning_rate": 1.3227513227513228e-05,
+      "loss": 0.2137,
+      "step": 641
+    },
+    {
+      "epoch": 2.2928571428571427,
+      "grad_norm": 0.19001647811035902,
+      "learning_rate": 1.3161375661375664e-05,
+      "loss": 0.2405,
+      "step": 642
+    },
+    {
+      "epoch": 2.2964285714285713,
+      "grad_norm": 0.17950155199753018,
+      "learning_rate": 1.3095238095238096e-05,
+      "loss": 0.2273,
+      "step": 643
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.18623589683984593,
+      "learning_rate": 1.302910052910053e-05,
+      "loss": 0.2233,
+      "step": 644
+    },
+    {
+      "epoch": 2.3035714285714284,
+      "grad_norm": 0.18041984508276784,
+      "learning_rate": 1.2962962962962962e-05,
+      "loss": 0.2322,
+      "step": 645
+    },
+    {
+      "epoch": 2.307142857142857,
+      "grad_norm": 0.18482676031328155,
+      "learning_rate": 1.2896825396825398e-05,
+      "loss": 0.2441,
+      "step": 646
+    },
+    {
+      "epoch": 2.310714285714286,
+      "grad_norm": 0.19015991372913651,
+      "learning_rate": 1.2830687830687832e-05,
+      "loss": 0.2303,
+      "step": 647
+    },
+    {
+      "epoch": 2.314285714285714,
+      "grad_norm": 0.18558270903762844,
+      "learning_rate": 1.2764550264550265e-05,
+      "loss": 0.2226,
+      "step": 648
+    },
+    {
+      "epoch": 2.317857142857143,
+      "grad_norm": 0.1749872002549765,
+      "learning_rate": 1.2698412698412699e-05,
+      "loss": 0.2231,
+      "step": 649
+    },
+    {
+      "epoch": 2.3214285714285716,
+      "grad_norm": 0.18598049636312228,
+      "learning_rate": 1.2632275132275134e-05,
+      "loss": 0.2456,
+      "step": 650
+    },
+    {
+      "epoch": 2.325,
+      "grad_norm": 0.17875368855456672,
+      "learning_rate": 1.2566137566137568e-05,
+      "loss": 0.2396,
+      "step": 651
+    },
+    {
+      "epoch": 2.3285714285714287,
+      "grad_norm": 0.18021223999715644,
+      "learning_rate": 1.25e-05,
+      "loss": 0.2398,
+      "step": 652
+    },
+    {
+      "epoch": 2.3321428571428573,
+      "grad_norm": 0.18333680314172154,
+      "learning_rate": 1.2433862433862433e-05,
+      "loss": 0.2433,
+      "step": 653
+    },
+    {
+      "epoch": 2.335714285714286,
+      "grad_norm": 0.1857422173200728,
+      "learning_rate": 1.2367724867724869e-05,
+      "loss": 0.2422,
+      "step": 654
+    },
+    {
+      "epoch": 2.3392857142857144,
+      "grad_norm": 0.1751418293891071,
+      "learning_rate": 1.2301587301587301e-05,
+      "loss": 0.2224,
+      "step": 655
+    },
+    {
+      "epoch": 2.342857142857143,
+      "grad_norm": 0.1898826300248994,
+      "learning_rate": 1.2235449735449737e-05,
+      "loss": 0.2363,
+      "step": 656
+    },
+    {
+      "epoch": 2.3464285714285715,
+      "grad_norm": 0.18347635752590621,
+      "learning_rate": 1.2169312169312169e-05,
+      "loss": 0.238,
+      "step": 657
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.18888548879332703,
+      "learning_rate": 1.2103174603174603e-05,
+      "loss": 0.2392,
+      "step": 658
+    },
+    {
+      "epoch": 2.3535714285714286,
+      "grad_norm": 0.18123258705601702,
+      "learning_rate": 1.2037037037037037e-05,
+      "loss": 0.2259,
+      "step": 659
+    },
+    {
+      "epoch": 2.357142857142857,
+      "grad_norm": 0.18551839713831378,
+      "learning_rate": 1.1970899470899471e-05,
+      "loss": 0.2507,
+      "step": 660
+    },
+    {
+      "epoch": 2.3607142857142858,
+      "grad_norm": 0.1848553914556543,
+      "learning_rate": 1.1904761904761905e-05,
+      "loss": 0.2289,
+      "step": 661
+    },
+    {
+      "epoch": 2.3642857142857143,
+      "grad_norm": 0.19821165678460764,
+      "learning_rate": 1.1838624338624339e-05,
+      "loss": 0.2507,
+      "step": 662
+    },
+    {
+      "epoch": 2.367857142857143,
+      "grad_norm": 0.1788376405395363,
+      "learning_rate": 1.1772486772486773e-05,
+      "loss": 0.2324,
+      "step": 663
+    },
+    {
+      "epoch": 2.3714285714285714,
+      "grad_norm": 0.17812414797025736,
+      "learning_rate": 1.1706349206349207e-05,
+      "loss": 0.2268,
+      "step": 664
+    },
+    {
+      "epoch": 2.375,
+      "grad_norm": 0.18106592672553284,
+      "learning_rate": 1.164021164021164e-05,
+      "loss": 0.2317,
+      "step": 665
+    },
+    {
+      "epoch": 2.3785714285714286,
+      "grad_norm": 0.18288704755186977,
+      "learning_rate": 1.1574074074074075e-05,
+      "loss": 0.2416,
+      "step": 666
+    },
+    {
+      "epoch": 2.382142857142857,
+      "grad_norm": 0.17602698506739747,
+      "learning_rate": 1.1507936507936508e-05,
+      "loss": 0.2211,
+      "step": 667
+    },
+    {
+      "epoch": 2.3857142857142857,
+      "grad_norm": 0.1816085869921426,
+      "learning_rate": 1.1441798941798943e-05,
+      "loss": 0.2313,
+      "step": 668
+    },
+    {
+      "epoch": 2.3892857142857142,
+      "grad_norm": 0.194775343340367,
+      "learning_rate": 1.1375661375661376e-05,
+      "loss": 0.2356,
+      "step": 669
+    },
+    {
+      "epoch": 2.392857142857143,
+      "grad_norm": 0.17077540878707073,
+      "learning_rate": 1.130952380952381e-05,
+      "loss": 0.2298,
+      "step": 670
+    },
+    {
+      "epoch": 2.3964285714285714,
+      "grad_norm": 0.175279045522553,
+      "learning_rate": 1.1243386243386244e-05,
+      "loss": 0.2396,
+      "step": 671
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.18286590484502516,
+      "learning_rate": 1.1177248677248678e-05,
+      "loss": 0.2275,
+      "step": 672
+    },
+    {
+      "epoch": 2.4035714285714285,
+      "grad_norm": 0.18753106406671288,
+      "learning_rate": 1.1111111111111112e-05,
+      "loss": 0.2236,
+      "step": 673
+    },
+    {
+      "epoch": 2.407142857142857,
+      "grad_norm": 0.18464216388906515,
+      "learning_rate": 1.1044973544973546e-05,
+      "loss": 0.2355,
+      "step": 674
+    },
+    {
+      "epoch": 2.4107142857142856,
+      "grad_norm": 0.17102011507564926,
+      "learning_rate": 1.0978835978835978e-05,
+      "loss": 0.229,
+      "step": 675
+    },
+    {
+      "epoch": 2.414285714285714,
+      "grad_norm": 0.18264098873146276,
+      "learning_rate": 1.0912698412698414e-05,
+      "loss": 0.2312,
+      "step": 676
+    },
+    {
+      "epoch": 2.4178571428571427,
+      "grad_norm": 0.1785039596777418,
+      "learning_rate": 1.0846560846560846e-05,
+      "loss": 0.2249,
+      "step": 677
+    },
+    {
+      "epoch": 2.4214285714285713,
+      "grad_norm": 0.1784008774316628,
+      "learning_rate": 1.0780423280423282e-05,
+      "loss": 0.226,
+      "step": 678
+    },
+    {
+      "epoch": 2.425,
+      "grad_norm": 0.1832540020874472,
+      "learning_rate": 1.0714285714285714e-05,
+      "loss": 0.2501,
+      "step": 679
+    },
+    {
+      "epoch": 2.4285714285714284,
+      "grad_norm": 0.20377146992108572,
+      "learning_rate": 1.0648148148148148e-05,
+      "loss": 0.2543,
+      "step": 680
+    },
+    {
+      "epoch": 2.432142857142857,
+      "grad_norm": 0.1731480289620937,
+      "learning_rate": 1.0582010582010582e-05,
+      "loss": 0.2251,
+      "step": 681
+    },
+    {
+      "epoch": 2.435714285714286,
+      "grad_norm": 0.18630618209870797,
+      "learning_rate": 1.0515873015873016e-05,
+      "loss": 0.2305,
+      "step": 682
+    },
+    {
+      "epoch": 2.439285714285714,
+      "grad_norm": 0.18313051849271003,
+      "learning_rate": 1.044973544973545e-05,
+      "loss": 0.2473,
+      "step": 683
+    },
+    {
+      "epoch": 2.442857142857143,
+      "grad_norm": 0.1798100157959098,
+      "learning_rate": 1.0383597883597884e-05,
+      "loss": 0.2267,
+      "step": 684
+    },
+    {
+      "epoch": 2.4464285714285716,
+      "grad_norm": 0.17955916787907394,
+      "learning_rate": 1.0317460317460318e-05,
+      "loss": 0.233,
+      "step": 685
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.17135533263303424,
+      "learning_rate": 1.0251322751322752e-05,
+      "loss": 0.226,
+      "step": 686
+    },
+    {
+      "epoch": 2.4535714285714287,
+      "grad_norm": 0.19248968000765487,
+      "learning_rate": 1.0185185185185185e-05,
+      "loss": 0.2366,
+      "step": 687
+    },
+    {
+      "epoch": 2.4571428571428573,
+      "grad_norm": 0.17861904795307337,
+      "learning_rate": 1.011904761904762e-05,
+      "loss": 0.2478,
+      "step": 688
+    },
+    {
+      "epoch": 2.460714285714286,
+      "grad_norm": 0.18505579363830563,
+      "learning_rate": 1.0052910052910053e-05,
+      "loss": 0.2372,
+      "step": 689
+    },
+    {
+      "epoch": 2.4642857142857144,
+      "grad_norm": 0.19616447828403222,
+      "learning_rate": 9.986772486772487e-06,
+      "loss": 0.2506,
+      "step": 690
+    },
+    {
+      "epoch": 2.467857142857143,
+      "grad_norm": 0.17707491322859925,
+      "learning_rate": 9.92063492063492e-06,
+      "loss": 0.2397,
+      "step": 691
+    },
+    {
+      "epoch": 2.4714285714285715,
+      "grad_norm": 0.1828939717458825,
+      "learning_rate": 9.854497354497355e-06,
+      "loss": 0.2253,
+      "step": 692
+    },
+    {
+      "epoch": 2.475,
+      "grad_norm": 0.18676097595764674,
+      "learning_rate": 9.788359788359789e-06,
+      "loss": 0.2346,
+      "step": 693
+    },
+    {
+      "epoch": 2.4785714285714286,
+      "grad_norm": 0.17789007644392743,
+      "learning_rate": 9.722222222222223e-06,
+      "loss": 0.2251,
+      "step": 694
+    },
+    {
+      "epoch": 2.482142857142857,
+      "grad_norm": 0.17227042169157322,
+      "learning_rate": 9.656084656084657e-06,
+      "loss": 0.2392,
+      "step": 695
+    },
+    {
+      "epoch": 2.4857142857142858,
+      "grad_norm": 0.18299432310216956,
+      "learning_rate": 9.58994708994709e-06,
+      "loss": 0.2297,
+      "step": 696
+    },
+    {
+      "epoch": 2.4892857142857143,
+      "grad_norm": 0.17987825853874717,
+      "learning_rate": 9.523809523809523e-06,
+      "loss": 0.2329,
+      "step": 697
+    },
+    {
+      "epoch": 2.492857142857143,
+      "grad_norm": 0.1847284392995962,
+      "learning_rate": 9.457671957671959e-06,
+      "loss": 0.2373,
+      "step": 698
+    },
+    {
+      "epoch": 2.4964285714285714,
+      "grad_norm": 0.19108475054530752,
+      "learning_rate": 9.391534391534391e-06,
+      "loss": 0.2443,
+      "step": 699
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.17762233929660595,
+      "learning_rate": 9.325396825396827e-06,
+      "loss": 0.249,
+      "step": 700
+    },
+    {
+      "epoch": 2.5035714285714286,
+      "grad_norm": 0.18577159200065121,
+      "learning_rate": 9.259259259259259e-06,
+      "loss": 0.2347,
+      "step": 701
+    },
+    {
+      "epoch": 2.507142857142857,
+      "grad_norm": 0.18583389854713103,
+      "learning_rate": 9.193121693121693e-06,
+      "loss": 0.2345,
+      "step": 702
+    },
+    {
+      "epoch": 2.5107142857142857,
+      "grad_norm": 0.1821008677728882,
+      "learning_rate": 9.126984126984127e-06,
+      "loss": 0.2327,
+      "step": 703
+    },
+    {
+      "epoch": 2.5142857142857142,
+      "grad_norm": 0.16583075461713662,
+      "learning_rate": 9.060846560846561e-06,
+      "loss": 0.2231,
+      "step": 704
+    },
+    {
+      "epoch": 2.517857142857143,
+      "grad_norm": 0.2001880594583591,
+      "learning_rate": 8.994708994708995e-06,
+      "loss": 0.2419,
+      "step": 705
+    },
+    {
+      "epoch": 2.5214285714285714,
+      "grad_norm": 0.17888017007754797,
+      "learning_rate": 8.92857142857143e-06,
+      "loss": 0.2314,
+      "step": 706
+    },
+    {
+      "epoch": 2.525,
+      "grad_norm": 0.17485962007226807,
+      "learning_rate": 8.862433862433862e-06,
+      "loss": 0.2302,
+      "step": 707
+    },
+    {
+      "epoch": 2.5285714285714285,
+      "grad_norm": 0.16959073643160472,
+      "learning_rate": 8.796296296296297e-06,
+      "loss": 0.2226,
+      "step": 708
+    },
+    {
+      "epoch": 2.532142857142857,
+      "grad_norm": 0.20007752443201304,
+      "learning_rate": 8.73015873015873e-06,
+      "loss": 0.2233,
+      "step": 709
+    },
+    {
+      "epoch": 2.5357142857142856,
+      "grad_norm": 0.19356372651426598,
+      "learning_rate": 8.664021164021165e-06,
+      "loss": 0.2372,
+      "step": 710
+    },
+    {
+      "epoch": 2.539285714285714,
+      "grad_norm": 0.18451290392787992,
+      "learning_rate": 8.597883597883598e-06,
+      "loss": 0.2322,
+      "step": 711
+    },
+    {
+      "epoch": 2.5428571428571427,
+      "grad_norm": 0.18163860169783583,
+      "learning_rate": 8.531746031746032e-06,
+      "loss": 0.223,
+      "step": 712
+    },
+    {
+      "epoch": 2.5464285714285713,
+      "grad_norm": 0.1911909031133387,
+      "learning_rate": 8.465608465608466e-06,
+      "loss": 0.2301,
+      "step": 713
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.18714038250476822,
+      "learning_rate": 8.3994708994709e-06,
+      "loss": 0.228,
+      "step": 714
+    },
+    {
+      "epoch": 2.553571428571429,
+      "grad_norm": 0.18434708405402636,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.2222,
+      "step": 715
+    },
+    {
+      "epoch": 2.557142857142857,
+      "grad_norm": 0.17987444791378004,
+      "learning_rate": 8.267195767195768e-06,
+      "loss": 0.2322,
+      "step": 716
+    },
+    {
+      "epoch": 2.560714285714286,
+      "grad_norm": 0.17618364721072877,
+      "learning_rate": 8.201058201058202e-06,
+      "loss": 0.2174,
+      "step": 717
+    },
+    {
+      "epoch": 2.564285714285714,
+      "grad_norm": 0.1785776238199109,
+      "learning_rate": 8.134920634920636e-06,
+      "loss": 0.2241,
+      "step": 718
+    },
+    {
+      "epoch": 2.567857142857143,
+      "grad_norm": 0.17867927941204922,
+      "learning_rate": 8.068783068783068e-06,
+      "loss": 0.2379,
+      "step": 719
+    },
+    {
+      "epoch": 2.571428571428571,
+      "grad_norm": 0.1840689506687409,
+      "learning_rate": 8.002645502645504e-06,
+      "loss": 0.2369,
+      "step": 720
+    },
+    {
+      "epoch": 2.575,
+      "grad_norm": 0.17808544655427985,
+      "learning_rate": 7.936507936507936e-06,
+      "loss": 0.2311,
+      "step": 721
+    },
+    {
+      "epoch": 2.5785714285714287,
+      "grad_norm": 0.17047547537082877,
+      "learning_rate": 7.87037037037037e-06,
+      "loss": 0.2201,
+      "step": 722
+    },
+    {
+      "epoch": 2.5821428571428573,
+      "grad_norm": 0.18592554309922818,
+      "learning_rate": 7.804232804232804e-06,
+      "loss": 0.2309,
+      "step": 723
+    },
+    {
+      "epoch": 2.585714285714286,
+      "grad_norm": 0.17378197486805952,
+      "learning_rate": 7.738095238095238e-06,
+      "loss": 0.2188,
+      "step": 724
+    },
+    {
+      "epoch": 2.5892857142857144,
+      "grad_norm": 0.18604038917406307,
+      "learning_rate": 7.671957671957672e-06,
+      "loss": 0.2314,
+      "step": 725
+    },
+    {
+      "epoch": 2.592857142857143,
+      "grad_norm": 0.18113006559215097,
+      "learning_rate": 7.605820105820106e-06,
+      "loss": 0.2331,
+      "step": 726
+    },
+    {
+      "epoch": 2.5964285714285715,
+      "grad_norm": 0.17238546627215254,
+      "learning_rate": 7.5396825396825394e-06,
+      "loss": 0.2335,
+      "step": 727
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.17095178191244506,
+      "learning_rate": 7.473544973544974e-06,
+      "loss": 0.2273,
+      "step": 728
+    },
+    {
+      "epoch": 2.6035714285714286,
+      "grad_norm": 0.17207400390689426,
+      "learning_rate": 7.4074074074074075e-06,
+      "loss": 0.2175,
+      "step": 729
+    },
+    {
+      "epoch": 2.607142857142857,
+      "grad_norm": 0.18956864252915798,
+      "learning_rate": 7.3412698412698415e-06,
+      "loss": 0.2399,
+      "step": 730
+    },
+    {
+      "epoch": 2.6107142857142858,
+      "grad_norm": 0.17204353445976817,
+      "learning_rate": 7.275132275132275e-06,
+      "loss": 0.2431,
+      "step": 731
+    },
+    {
+      "epoch": 2.6142857142857143,
+      "grad_norm": 0.17793246134971735,
+      "learning_rate": 7.2089947089947095e-06,
+      "loss": 0.2247,
+      "step": 732
+    },
+    {
+      "epoch": 2.617857142857143,
+      "grad_norm": 0.18196174543524102,
+      "learning_rate": 7.142857142857143e-06,
+      "loss": 0.2361,
+      "step": 733
+    },
+    {
+      "epoch": 2.6214285714285714,
+      "grad_norm": 0.16145155829354535,
+      "learning_rate": 7.076719576719578e-06,
+      "loss": 0.2148,
+      "step": 734
+    },
+    {
+      "epoch": 2.625,
+      "grad_norm": 0.1759931623536279,
+      "learning_rate": 7.010582010582011e-06,
+      "loss": 0.2287,
+      "step": 735
+    },
+    {
+      "epoch": 2.6285714285714286,
+      "grad_norm": 0.1701646786770766,
+      "learning_rate": 6.944444444444445e-06,
+      "loss": 0.2275,
+      "step": 736
+    },
+    {
+      "epoch": 2.632142857142857,
+      "grad_norm": 0.17779081050600287,
+      "learning_rate": 6.878306878306878e-06,
+      "loss": 0.2316,
+      "step": 737
+    },
+    {
+      "epoch": 2.6357142857142857,
+      "grad_norm": 0.18362696292213518,
+      "learning_rate": 6.812169312169313e-06,
+      "loss": 0.238,
+      "step": 738
+    },
+    {
+      "epoch": 2.6392857142857142,
+      "grad_norm": 0.17406500211549078,
+      "learning_rate": 6.746031746031746e-06,
+      "loss": 0.2281,
+      "step": 739
+    },
+    {
+      "epoch": 2.642857142857143,
+      "grad_norm": 0.17589179855071144,
+      "learning_rate": 6.679894179894181e-06,
+      "loss": 0.233,
+      "step": 740
+    },
+    {
+      "epoch": 2.6464285714285714,
+      "grad_norm": 0.1822352689709617,
+      "learning_rate": 6.613756613756614e-06,
+      "loss": 0.2237,
+      "step": 741
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.1733948397887049,
+      "learning_rate": 6.547619047619048e-06,
+      "loss": 0.2314,
+      "step": 742
+    },
+    {
+      "epoch": 2.6535714285714285,
+      "grad_norm": 0.1786629505603967,
+      "learning_rate": 6.481481481481481e-06,
+      "loss": 0.2277,
+      "step": 743
+    },
+    {
+      "epoch": 2.657142857142857,
+      "grad_norm": 0.17628172776632087,
+      "learning_rate": 6.415343915343916e-06,
+      "loss": 0.2348,
+      "step": 744
+    },
+    {
+      "epoch": 2.6607142857142856,
+      "grad_norm": 0.1733767829020336,
+      "learning_rate": 6.349206349206349e-06,
+      "loss": 0.2218,
+      "step": 745
+    },
+    {
+      "epoch": 2.664285714285714,
+      "grad_norm": 0.18544515134482925,
+      "learning_rate": 6.283068783068784e-06,
+      "loss": 0.2457,
+      "step": 746
+    },
+    {
+      "epoch": 2.6678571428571427,
+      "grad_norm": 0.17389055902822748,
+      "learning_rate": 6.2169312169312165e-06,
+      "loss": 0.2207,
+      "step": 747
+    },
+    {
+      "epoch": 2.6714285714285713,
+      "grad_norm": 0.17967519183170957,
+      "learning_rate": 6.1507936507936505e-06,
+      "loss": 0.234,
+      "step": 748
+    },
+    {
+      "epoch": 2.675,
+      "grad_norm": 0.17829447583538188,
+      "learning_rate": 6.0846560846560845e-06,
+      "loss": 0.2364,
+      "step": 749
+    },
+    {
+      "epoch": 2.678571428571429,
+      "grad_norm": 0.18456609239506558,
+      "learning_rate": 6.0185185185185185e-06,
+      "loss": 0.2496,
+      "step": 750
+    },
+    {
+      "epoch": 2.682142857142857,
+      "grad_norm": 0.17865798673149355,
+      "learning_rate": 5.9523809523809525e-06,
+      "loss": 0.2364,
+      "step": 751
+    },
+    {
+      "epoch": 2.685714285714286,
+      "grad_norm": 0.17635857473397207,
+      "learning_rate": 5.8862433862433866e-06,
+      "loss": 0.2213,
+      "step": 752
+    },
+    {
+      "epoch": 2.689285714285714,
+      "grad_norm": 0.1822372859229297,
+      "learning_rate": 5.82010582010582e-06,
+      "loss": 0.2403,
+      "step": 753
+    },
+    {
+      "epoch": 2.692857142857143,
+      "grad_norm": 0.17030059066015693,
+      "learning_rate": 5.753968253968254e-06,
+      "loss": 0.2413,
+      "step": 754
+    },
+    {
+      "epoch": 2.696428571428571,
+      "grad_norm": 0.17248590430393515,
+      "learning_rate": 5.687830687830688e-06,
+      "loss": 0.2153,
+      "step": 755
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.17460725813586495,
+      "learning_rate": 5.621693121693122e-06,
+      "loss": 0.2173,
+      "step": 756
+    },
+    {
+      "epoch": 2.7035714285714287,
+      "grad_norm": 0.17059438171912317,
+      "learning_rate": 5.555555555555556e-06,
+      "loss": 0.2233,
+      "step": 757
+    },
+    {
+      "epoch": 2.7071428571428573,
+      "grad_norm": 0.21659560008854054,
+      "learning_rate": 5.489417989417989e-06,
+      "loss": 0.2441,
+      "step": 758
+    },
+    {
+      "epoch": 2.710714285714286,
+      "grad_norm": 0.1845238565391778,
+      "learning_rate": 5.423280423280423e-06,
+      "loss": 0.2375,
+      "step": 759
+    },
+    {
+      "epoch": 2.7142857142857144,
+      "grad_norm": 0.17889563032728956,
+      "learning_rate": 5.357142857142857e-06,
+      "loss": 0.2421,
+      "step": 760
+    },
+    {
+      "epoch": 2.717857142857143,
+      "grad_norm": 0.170080292427315,
+      "learning_rate": 5.291005291005291e-06,
+      "loss": 0.2186,
+      "step": 761
+    },
+    {
+      "epoch": 2.7214285714285715,
+      "grad_norm": 0.1717847765729961,
+      "learning_rate": 5.224867724867725e-06,
+      "loss": 0.2337,
+      "step": 762
+    },
+    {
+      "epoch": 2.725,
+      "grad_norm": 0.1755900225838068,
+      "learning_rate": 5.158730158730159e-06,
+      "loss": 0.2447,
+      "step": 763
+    },
+    {
+      "epoch": 2.7285714285714286,
+      "grad_norm": 0.17817670849952222,
+      "learning_rate": 5.092592592592592e-06,
+      "loss": 0.2447,
+      "step": 764
+    },
+    {
+      "epoch": 2.732142857142857,
+      "grad_norm": 0.17639708065354745,
+      "learning_rate": 5.026455026455026e-06,
+      "loss": 0.2265,
+      "step": 765
+    },
+    {
+      "epoch": 2.7357142857142858,
+      "grad_norm": 0.16296605804635478,
+      "learning_rate": 4.96031746031746e-06,
+      "loss": 0.2188,
+      "step": 766
+    },
+    {
+      "epoch": 2.7392857142857143,
+      "grad_norm": 0.18383269192673646,
+      "learning_rate": 4.894179894179894e-06,
+      "loss": 0.2441,
+      "step": 767
+    },
+    {
+      "epoch": 2.742857142857143,
+      "grad_norm": 0.1643414102128402,
+      "learning_rate": 4.828042328042328e-06,
+      "loss": 0.2257,
+      "step": 768
+    },
+    {
+      "epoch": 2.7464285714285714,
+      "grad_norm": 0.1723088685929767,
+      "learning_rate": 4.7619047619047615e-06,
+      "loss": 0.2242,
+      "step": 769
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.1753495983884268,
+      "learning_rate": 4.6957671957671955e-06,
+      "loss": 0.2278,
+      "step": 770
+    },
+    {
+      "epoch": 2.7535714285714286,
+      "grad_norm": 0.18095021857472,
+      "learning_rate": 4.6296296296296296e-06,
+      "loss": 0.2292,
+      "step": 771
+    },
+    {
+      "epoch": 2.757142857142857,
+      "grad_norm": 0.17392056173666068,
+      "learning_rate": 4.563492063492064e-06,
+      "loss": 0.2319,
+      "step": 772
+    },
+    {
+      "epoch": 2.7607142857142857,
+      "grad_norm": 0.18469344249167743,
+      "learning_rate": 4.497354497354498e-06,
+      "loss": 0.2325,
+      "step": 773
+    },
+    {
+      "epoch": 2.7642857142857142,
+      "grad_norm": 0.17728939991953163,
+      "learning_rate": 4.431216931216931e-06,
+      "loss": 0.2297,
+      "step": 774
+    },
+    {
+      "epoch": 2.767857142857143,
+      "grad_norm": 0.16544092309779834,
+      "learning_rate": 4.365079365079365e-06,
+      "loss": 0.226,
+      "step": 775
+    },
+    {
+      "epoch": 2.7714285714285714,
+      "grad_norm": 0.16584515529789243,
+      "learning_rate": 4.298941798941799e-06,
+      "loss": 0.2141,
+      "step": 776
+    },
+    {
+      "epoch": 2.775,
+      "grad_norm": 0.16782058126563995,
+      "learning_rate": 4.232804232804233e-06,
+      "loss": 0.2288,
+      "step": 777
+    },
+    {
+      "epoch": 2.7785714285714285,
+      "grad_norm": 0.16665185876634223,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 0.2228,
+      "step": 778
+    },
+    {
+      "epoch": 2.782142857142857,
+      "grad_norm": 0.1703696442982876,
+      "learning_rate": 4.100529100529101e-06,
+      "loss": 0.2306,
+      "step": 779
+    },
+    {
+      "epoch": 2.7857142857142856,
+      "grad_norm": 0.18191610892251986,
+      "learning_rate": 4.034391534391534e-06,
+      "loss": 0.2394,
+      "step": 780
+    },
+    {
+      "epoch": 2.789285714285714,
+      "grad_norm": 0.16810039134914587,
+      "learning_rate": 3.968253968253968e-06,
+      "loss": 0.2278,
+      "step": 781
+    },
+    {
+      "epoch": 2.7928571428571427,
+      "grad_norm": 0.16806662085997198,
+      "learning_rate": 3.902116402116402e-06,
+      "loss": 0.225,
+      "step": 782
+    },
+    {
+      "epoch": 2.7964285714285713,
+      "grad_norm": 0.18068067088931247,
+      "learning_rate": 3.835978835978836e-06,
+      "loss": 0.2414,
+      "step": 783
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.16921363983680413,
+      "learning_rate": 3.7698412698412697e-06,
+      "loss": 0.2272,
+      "step": 784
+    },
+    {
+      "epoch": 2.803571428571429,
+      "grad_norm": 0.16846202822257286,
+      "learning_rate": 3.7037037037037037e-06,
+      "loss": 0.2362,
+      "step": 785
+    },
+    {
+      "epoch": 2.807142857142857,
+      "grad_norm": 0.16612631374892517,
+      "learning_rate": 3.6375661375661373e-06,
+      "loss": 0.2157,
+      "step": 786
+    },
+    {
+      "epoch": 2.810714285714286,
+      "grad_norm": 0.17676456261916748,
+      "learning_rate": 3.5714285714285714e-06,
+      "loss": 0.2266,
+      "step": 787
+    },
+    {
+      "epoch": 2.814285714285714,
+      "grad_norm": 0.18340833300307588,
+      "learning_rate": 3.5052910052910054e-06,
+      "loss": 0.2302,
+      "step": 788
+    },
+    {
+      "epoch": 2.817857142857143,
+      "grad_norm": 0.16589799732508684,
+      "learning_rate": 3.439153439153439e-06,
+      "loss": 0.226,
+      "step": 789
+    },
+    {
+      "epoch": 2.821428571428571,
+      "grad_norm": 0.1747575485237352,
+      "learning_rate": 3.373015873015873e-06,
+      "loss": 0.2473,
+      "step": 790
+    },
+    {
+      "epoch": 2.825,
+      "grad_norm": 0.163506420611649,
+      "learning_rate": 3.306878306878307e-06,
+      "loss": 0.2261,
+      "step": 791
+    },
+    {
+      "epoch": 2.8285714285714287,
+      "grad_norm": 0.17456333444950714,
+      "learning_rate": 3.2407407407407406e-06,
+      "loss": 0.2327,
+      "step": 792
+    },
+    {
+      "epoch": 2.8321428571428573,
+      "grad_norm": 0.17665107151037676,
+      "learning_rate": 3.1746031746031746e-06,
+      "loss": 0.2451,
+      "step": 793
+    },
+    {
+      "epoch": 2.835714285714286,
+      "grad_norm": 0.1664317067287685,
+      "learning_rate": 3.1084656084656082e-06,
+      "loss": 0.2263,
+      "step": 794
+    },
+    {
+      "epoch": 2.8392857142857144,
+      "grad_norm": 0.17068712106095152,
+      "learning_rate": 3.0423280423280423e-06,
+      "loss": 0.2319,
+      "step": 795
+    },
+    {
+      "epoch": 2.842857142857143,
+      "grad_norm": 0.1751420980495918,
+      "learning_rate": 2.9761904761904763e-06,
+      "loss": 0.2518,
+      "step": 796
+    },
+    {
+      "epoch": 2.8464285714285715,
+      "grad_norm": 0.1769172205565068,
+      "learning_rate": 2.91005291005291e-06,
+      "loss": 0.2269,
+      "step": 797
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.16502481589490442,
+      "learning_rate": 2.843915343915344e-06,
+      "loss": 0.2232,
+      "step": 798
+    },
+    {
+      "epoch": 2.8535714285714286,
+      "grad_norm": 0.16678543733046433,
+      "learning_rate": 2.777777777777778e-06,
+      "loss": 0.227,
+      "step": 799
+    },
+    {
+      "epoch": 2.857142857142857,
+      "grad_norm": 0.19475718949453422,
+      "learning_rate": 2.7116402116402115e-06,
+      "loss": 0.2333,
+      "step": 800
+    },
+    {
+      "epoch": 2.8607142857142858,
+      "grad_norm": 0.2114953877136492,
+      "learning_rate": 2.6455026455026455e-06,
+      "loss": 0.2417,
+      "step": 801
+    },
+    {
+      "epoch": 2.8642857142857143,
+      "grad_norm": 0.17928112632430857,
+      "learning_rate": 2.5793650793650795e-06,
+      "loss": 0.2252,
+      "step": 802
+    },
+    {
+      "epoch": 2.867857142857143,
+      "grad_norm": 0.16897295278875885,
+      "learning_rate": 2.513227513227513e-06,
+      "loss": 0.2283,
+      "step": 803
+    },
+    {
+      "epoch": 2.8714285714285714,
+      "grad_norm": 0.16421852840448375,
+      "learning_rate": 2.447089947089947e-06,
+      "loss": 0.2321,
+      "step": 804
+    },
+    {
+      "epoch": 2.875,
+      "grad_norm": 0.17429929348761858,
+      "learning_rate": 2.3809523809523808e-06,
+      "loss": 0.234,
+      "step": 805
+    },
+    {
+      "epoch": 2.8785714285714286,
+      "grad_norm": 0.1670240072940453,
+      "learning_rate": 2.3148148148148148e-06,
+      "loss": 0.2316,
+      "step": 806
+    },
+    {
+      "epoch": 2.882142857142857,
+      "grad_norm": 0.16815116176705477,
+      "learning_rate": 2.248677248677249e-06,
+      "loss": 0.234,
+      "step": 807
+    },
+    {
+      "epoch": 2.8857142857142857,
+      "grad_norm": 0.17097749002195808,
+      "learning_rate": 2.1825396825396824e-06,
+      "loss": 0.2331,
+      "step": 808
+    },
+    {
+      "epoch": 2.8892857142857142,
+      "grad_norm": 0.16902704656821627,
+      "learning_rate": 2.1164021164021164e-06,
+      "loss": 0.2341,
+      "step": 809
+    },
+    {
+      "epoch": 2.892857142857143,
+      "grad_norm": 0.16676465645249983,
+      "learning_rate": 2.0502645502645504e-06,
+      "loss": 0.2118,
+      "step": 810
+    },
+    {
+      "epoch": 2.8964285714285714,
+      "grad_norm": 0.17299227382270477,
+      "learning_rate": 1.984126984126984e-06,
+      "loss": 0.2309,
+      "step": 811
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.15968977326118708,
+      "learning_rate": 1.917989417989418e-06,
+      "loss": 0.214,
+      "step": 812
+    },
+    {
+      "epoch": 2.9035714285714285,
+      "grad_norm": 0.2613061702934918,
+      "learning_rate": 1.8518518518518519e-06,
+      "loss": 0.2334,
+      "step": 813
+    },
+    {
+      "epoch": 2.907142857142857,
+      "grad_norm": 0.2284100545821048,
+      "learning_rate": 1.7857142857142857e-06,
+      "loss": 0.2228,
+      "step": 814
+    },
+    {
+      "epoch": 2.9107142857142856,
+      "grad_norm": 0.16497119392682538,
+      "learning_rate": 1.7195767195767195e-06,
+      "loss": 0.2229,
+      "step": 815
+    },
+    {
+      "epoch": 2.914285714285714,
+      "grad_norm": 0.1896153073364442,
+      "learning_rate": 1.6534391534391535e-06,
+      "loss": 0.2298,
+      "step": 816
+    },
+    {
+      "epoch": 2.9178571428571427,
+      "grad_norm": 0.1696950074061391,
+      "learning_rate": 1.5873015873015873e-06,
+      "loss": 0.2427,
+      "step": 817
+    },
+    {
+      "epoch": 2.9214285714285713,
+      "grad_norm": 0.16828194080218858,
+      "learning_rate": 1.5211640211640211e-06,
+      "loss": 0.2284,
+      "step": 818
+    },
+    {
+      "epoch": 2.925,
+      "grad_norm": 0.16462336212584547,
+      "learning_rate": 1.455026455026455e-06,
+      "loss": 0.2272,
+      "step": 819
+    },
+    {
+      "epoch": 2.928571428571429,
+      "grad_norm": 0.16880979188889203,
+      "learning_rate": 1.388888888888889e-06,
+      "loss": 0.2308,
+      "step": 820
+    },
+    {
+      "epoch": 2.932142857142857,
+      "grad_norm": 0.1732300474561914,
+      "learning_rate": 1.3227513227513228e-06,
+      "loss": 0.2381,
+      "step": 821
+    },
+    {
+      "epoch": 2.935714285714286,
+      "grad_norm": 0.17587716422009395,
+      "learning_rate": 1.2566137566137566e-06,
+      "loss": 0.2384,
+      "step": 822
+    },
+    {
+      "epoch": 2.939285714285714,
+      "grad_norm": 0.16732454461603405,
+      "learning_rate": 1.1904761904761904e-06,
+      "loss": 0.2413,
+      "step": 823
+    },
+    {
+      "epoch": 2.942857142857143,
+      "grad_norm": 0.1662788419168319,
+      "learning_rate": 1.1243386243386244e-06,
+      "loss": 0.2285,
+      "step": 824
+    },
+    {
+      "epoch": 2.946428571428571,
+      "grad_norm": 0.16883388484015832,
+      "learning_rate": 1.0582010582010582e-06,
+      "loss": 0.2365,
+      "step": 825
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.16230089132019374,
+      "learning_rate": 9.92063492063492e-07,
+      "loss": 0.2236,
+      "step": 826
+    },
+    {
+      "epoch": 2.9535714285714287,
+      "grad_norm": 0.16482273737685368,
+      "learning_rate": 9.259259259259259e-07,
+      "loss": 0.2202,
+      "step": 827
+    },
+    {
+      "epoch": 2.9571428571428573,
+      "grad_norm": 0.16466066450329664,
+      "learning_rate": 8.597883597883597e-07,
+      "loss": 0.242,
+      "step": 828
+    },
+    {
+      "epoch": 2.960714285714286,
+      "grad_norm": 0.16812467687183735,
+      "learning_rate": 7.936507936507937e-07,
+      "loss": 0.2348,
+      "step": 829
+    },
+    {
+      "epoch": 2.9642857142857144,
+      "grad_norm": 0.1675787940302276,
+      "learning_rate": 7.275132275132275e-07,
+      "loss": 0.2344,
+      "step": 830
+    },
+    {
+      "epoch": 2.967857142857143,
+      "grad_norm": 0.16562515924074495,
+      "learning_rate": 6.613756613756614e-07,
+      "loss": 0.219,
+      "step": 831
+    },
+    {
+      "epoch": 2.9714285714285715,
+      "grad_norm": 0.164004183577236,
+      "learning_rate": 5.952380952380952e-07,
+      "loss": 0.2263,
+      "step": 832
+    },
+    {
+      "epoch": 2.975,
+      "grad_norm": 0.16828300869294466,
+      "learning_rate": 5.291005291005291e-07,
+      "loss": 0.2347,
+      "step": 833
+    },
+    {
+      "epoch": 2.9785714285714286,
+      "grad_norm": 0.19223571198056752,
+      "learning_rate": 4.6296296296296297e-07,
+      "loss": 0.2412,
+      "step": 834
+    },
+    {
+      "epoch": 2.982142857142857,
+      "grad_norm": 0.15992135865774978,
+      "learning_rate": 3.9682539682539683e-07,
+      "loss": 0.2186,
+      "step": 835
+    },
+    {
+      "epoch": 2.9857142857142858,
+      "grad_norm": 0.1650982711635989,
+      "learning_rate": 3.306878306878307e-07,
+      "loss": 0.2341,
+      "step": 836
+    },
+    {
+      "epoch": 2.9892857142857143,
+      "grad_norm": 0.15638979563559063,
+      "learning_rate": 2.6455026455026455e-07,
+      "loss": 0.2167,
+      "step": 837
+    },
+    {
+      "epoch": 2.992857142857143,
+      "grad_norm": 0.1624211828693843,
+      "learning_rate": 1.9841269841269841e-07,
+      "loss": 0.2304,
+      "step": 838
+    },
+    {
+      "epoch": 2.9964285714285714,
+      "grad_norm": 0.16635082301815485,
+      "learning_rate": 1.3227513227513228e-07,
+      "loss": 0.2287,
+      "step": 839
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.1590571177775105,
+      "learning_rate": 6.613756613756614e-08,
+      "loss": 0.2115,
+      "step": 840
+    },
+    {
+      "epoch": 3.0,
+      "step": 840,
+      "total_flos": 9.342043600298443e+18,
+      "train_loss": 0.4384442511236384,
+      "train_runtime": 28369.0979,
+      "train_samples_per_second": 0.473,
+      "train_steps_per_second": 0.03
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 840,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 9.342043600298443e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}