diff --git "a/checkpoint-4540/trainer_state.json" "b/checkpoint-4540/trainer_state.json"
--- "a/checkpoint-4540/trainer_state.json"
+++ "b/checkpoint-4540/trainer_state.json"
@@ -10,17214 +10,17214 @@
   "log_history": [
     {
       "epoch": 0.0022002200220022,
-      "grad_norm": 3.53125,
+      "grad_norm": 2.671875,
       "learning_rate": 2e-05,
-      "loss": 1.4899,
+      "loss": 1.4517,
       "step": 1
     },
     {
       "epoch": 0.0022002200220022,
-      "eval_loss": 1.3368611335754395,
-      "eval_runtime": 10.223,
-      "eval_samples_per_second": 37.464,
-      "eval_steps_per_second": 4.695,
+      "eval_loss": 1.3369288444519043,
+      "eval_runtime": 11.0738,
+      "eval_samples_per_second": 34.586,
+      "eval_steps_per_second": 4.335,
       "step": 1
     },
     {
       "epoch": 0.0044004400440044,
-      "grad_norm": 2.921875,
+      "grad_norm": 2.515625,
       "learning_rate": 4e-05,
-      "loss": 1.5277,
+      "loss": 1.5206,
       "step": 2
     },
     {
       "epoch": 0.006600660066006601,
-      "grad_norm": 3.046875,
+      "grad_norm": 2.84375,
       "learning_rate": 6e-05,
-      "loss": 1.588,
+      "loss": 1.5721,
       "step": 3
     },
     {
       "epoch": 0.0088008800880088,
-      "grad_norm": 2.5,
+      "grad_norm": 2.265625,
       "learning_rate": 8e-05,
-      "loss": 1.4493,
+      "loss": 1.4437,
       "step": 4
     },
     {
       "epoch": 0.011001100110011002,
-      "grad_norm": 1.4375,
+      "grad_norm": 1.375,
       "learning_rate": 0.0001,
-      "loss": 1.1503,
+      "loss": 1.1402,
       "step": 5
     },
     {
       "epoch": 0.013201320132013201,
-      "grad_norm": 0.71484375,
+      "grad_norm": 0.67578125,
       "learning_rate": 0.00012,
-      "loss": 1.0286,
+      "loss": 1.0197,
       "step": 6
     },
     {
       "epoch": 0.015401540154015401,
-      "grad_norm": 0.455078125,
+      "grad_norm": 0.443359375,
       "learning_rate": 0.00014,
-      "loss": 1.0797,
+      "loss": 1.078,
       "step": 7
     },
     {
       "epoch": 0.0176017601760176,
-      "grad_norm": 0.52734375,
+      "grad_norm": 0.5234375,
       "learning_rate": 0.00016,
-      "loss": 1.1631,
+      "loss": 1.1628,
       "step": 8
     },
     {
       "epoch": 0.019801980198019802,
-      "grad_norm": 0.54296875,
+      "grad_norm": 0.55859375,
       "learning_rate": 0.00018,
-      "loss": 1.0174,
+      "loss": 1.0207,
       "step": 9
     },
     {
       "epoch": 0.022002200220022004,
-      "grad_norm": 0.51953125,
+      "grad_norm": 0.5234375,
       "learning_rate": 0.0002,
-      "loss": 1.0345,
+      "loss": 1.0358,
       "step": 10
     },
     {
       "epoch": 0.0242024202420242,
-      "grad_norm": 0.486328125,
+      "grad_norm": 0.490234375,
       "learning_rate": 0.00019999997595231204,
-      "loss": 0.9847,
+      "loss": 0.9862,
       "step": 11
     },
     {
       "epoch": 0.026402640264026403,
-      "grad_norm": 0.50390625,
+      "grad_norm": 0.4921875,
       "learning_rate": 0.00019999990380925966,
-      "loss": 0.9783,
+      "loss": 0.9772,
       "step": 12
     },
     {
       "epoch": 0.028602860286028604,
-      "grad_norm": 0.462890625,
+      "grad_norm": 0.45703125,
       "learning_rate": 0.00019999978357087758,
-      "loss": 0.8539,
+      "loss": 0.8544,
       "step": 13
     },
     {
       "epoch": 0.030803080308030802,
-      "grad_norm": 0.5078125,
+      "grad_norm": 0.49609375,
       "learning_rate": 0.00019999961523722363,
-      "loss": 0.9799,
+      "loss": 0.9794,
       "step": 14
     },
     {
       "epoch": 0.033003300330033,
       "grad_norm": 0.486328125,
       "learning_rate": 0.00019999939880837878,
-      "loss": 0.9257,
+      "loss": 0.9266,
       "step": 15
     },
     {
       "epoch": 0.0352035203520352,
-      "grad_norm": 0.455078125,
+      "grad_norm": 0.451171875,
       "learning_rate": 0.0001999991342844471,
-      "loss": 0.7979,
+      "loss": 0.7962,
       "step": 16
     },
     {
       "epoch": 0.0374037403740374,
-      "grad_norm": 0.4296875,
+      "grad_norm": 0.427734375,
       "learning_rate": 0.00019999882166555586,
-      "loss": 0.8652,
+      "loss": 0.8647,
       "step": 17
     },
     {
       "epoch": 0.039603960396039604,
-      "grad_norm": 0.41015625,
+      "grad_norm": 0.408203125,
       "learning_rate": 0.00019999846095185539,
-      "loss": 0.8488,
+      "loss": 0.8482,
       "step": 18
     },
     {
       "epoch": 0.041804180418041806,
-      "grad_norm": 0.453125,
+      "grad_norm": 0.4453125,
       "learning_rate": 0.00019999805214351914,
-      "loss": 0.8512,
+      "loss": 0.8504,
       "step": 19
     },
     {
       "epoch": 0.04400440044004401,
-      "grad_norm": 0.49609375,
+      "grad_norm": 0.494140625,
       "learning_rate": 0.00019999759524074376,
-      "loss": 0.8986,
+      "loss": 0.898,
       "step": 20
     },
     {
       "epoch": 0.0462046204620462,
-      "grad_norm": 0.546875,
+      "grad_norm": 0.54296875,
       "learning_rate": 0.000199997090243749,
-      "loss": 0.8535,
+      "loss": 0.855,
       "step": 21
     },
     {
       "epoch": 0.0484048404840484,
-      "grad_norm": 0.4609375,
+      "grad_norm": 0.470703125,
       "learning_rate": 0.00019999653715277774,
-      "loss": 0.7234,
+      "loss": 0.7236,
       "step": 22
     },
     {
       "epoch": 0.050605060506050605,
-      "grad_norm": 0.48046875,
+      "grad_norm": 0.482421875,
       "learning_rate": 0.000199995935968096,
-      "loss": 0.8115,
+      "loss": 0.813,
       "step": 23
     },
     {
       "epoch": 0.052805280528052806,
-      "grad_norm": 0.46484375,
+      "grad_norm": 0.447265625,
       "learning_rate": 0.0001999952866899929,
-      "loss": 0.8662,
+      "loss": 0.8651,
       "step": 24
     },
     {
       "epoch": 0.05500550055005501,
-      "grad_norm": 0.474609375,
+      "grad_norm": 0.470703125,
       "learning_rate": 0.00019999458931878073,
-      "loss": 0.9387,
+      "loss": 0.9397,
       "step": 25
     },
     {
       "epoch": 0.05720572057205721,
-      "grad_norm": 0.412109375,
+      "grad_norm": 0.4140625,
       "learning_rate": 0.0001999938438547949,
-      "loss": 0.8877,
+      "loss": 0.8888,
       "step": 26
     },
     {
       "epoch": 0.0594059405940594,
-      "grad_norm": 0.421875,
+      "grad_norm": 0.427734375,
       "learning_rate": 0.0001999930502983939,
-      "loss": 0.8817,
+      "loss": 0.884,
       "step": 27
     },
     {
       "epoch": 0.061606160616061605,
       "grad_norm": 0.412109375,
       "learning_rate": 0.00019999220864995942,
-      "loss": 0.9133,
+      "loss": 0.9118,
       "step": 28
     },
     {
       "epoch": 0.0638063806380638,
-      "grad_norm": 0.458984375,
+      "grad_norm": 0.46484375,
       "learning_rate": 0.00019999131890989627,
-      "loss": 0.8812,
+      "loss": 0.8826,
       "step": 29
     },
     {
       "epoch": 0.066006600660066,
-      "grad_norm": 0.4296875,
+      "grad_norm": 0.427734375,
       "learning_rate": 0.00019999038107863237,
-      "loss": 0.7786,
+      "loss": 0.778,
       "step": 30
     },
     {
       "epoch": 0.06820682068206821,
-      "grad_norm": 0.40234375,
+      "grad_norm": 0.400390625,
       "learning_rate": 0.00019998939515661877,
-      "loss": 0.7648,
+      "loss": 0.7674,
       "step": 31
     },
     {
       "epoch": 0.0704070407040704,
-      "grad_norm": 0.4375,
+      "grad_norm": 0.439453125,
       "learning_rate": 0.00019998836114432965,
-      "loss": 0.8475,
+      "loss": 0.8488,
       "step": 32
     },
     {
       "epoch": 0.07260726072607261,
-      "grad_norm": 0.4453125,
+      "grad_norm": 0.443359375,
       "learning_rate": 0.0001999872790422623,
-      "loss": 0.7568,
+      "loss": 0.7589,
       "step": 33
     },
     {
       "epoch": 0.0748074807480748,
-      "grad_norm": 0.45703125,
+      "grad_norm": 0.443359375,
       "learning_rate": 0.00019998614885093717,
-      "loss": 0.8363,
+      "loss": 0.834,
       "step": 34
     },
     {
       "epoch": 0.07700770077007701,
       "grad_norm": 0.44921875,
       "learning_rate": 0.00019998497057089788,
-      "loss": 0.8664,
+      "loss": 0.8657,
       "step": 35
     },
     {
       "epoch": 0.07920792079207921,
       "grad_norm": 0.447265625,
       "learning_rate": 0.00019998374420271108,
-      "loss": 0.8532,
+      "loss": 0.851,
       "step": 36
     },
     {
       "epoch": 0.0814081408140814,
       "grad_norm": 0.427734375,
       "learning_rate": 0.00019998246974696658,
-      "loss": 0.7901,
+      "loss": 0.7909,
       "step": 37
     },
     {
       "epoch": 0.08360836083608361,
       "grad_norm": 0.42578125,
       "learning_rate": 0.0001999811472042774,
-      "loss": 0.8607,
+      "loss": 0.8614,
       "step": 38
     },
     {
       "epoch": 0.0858085808580858,
-      "grad_norm": 0.453125,
+      "grad_norm": 0.455078125,
       "learning_rate": 0.00019997977657527956,
-      "loss": 0.8251,
+      "loss": 0.8275,
       "step": 39
     },
     {
       "epoch": 0.08800880088008801,
-      "grad_norm": 0.423828125,
+      "grad_norm": 0.41015625,
       "learning_rate": 0.0001999783578606323,
-      "loss": 0.8081,
+      "loss": 0.8065,
       "step": 40
     },
     {
       "epoch": 0.09020902090209021,
-      "grad_norm": 0.466796875,
+      "grad_norm": 0.46875,
       "learning_rate": 0.00019997689106101792,
-      "loss": 0.9046,
+      "loss": 0.9064,
       "step": 41
     },
     {
       "epoch": 0.0924092409240924,
-      "grad_norm": 0.4296875,
+      "grad_norm": 0.427734375,
       "learning_rate": 0.00019997537617714195,
-      "loss": 0.822,
+      "loss": 0.8216,
       "step": 42
     },
     {
       "epoch": 0.09460946094609461,
-      "grad_norm": 0.443359375,
+      "grad_norm": 0.435546875,
       "learning_rate": 0.00019997381320973294,
-      "loss": 0.8611,
+      "loss": 0.8629,
       "step": 43
     },
     {
       "epoch": 0.0968096809680968,
-      "grad_norm": 0.421875,
+      "grad_norm": 0.4140625,
       "learning_rate": 0.00019997220215954258,
-      "loss": 0.8363,
+      "loss": 0.8339,
       "step": 44
     },
     {
       "epoch": 0.09900990099009901,
-      "grad_norm": 0.453125,
+      "grad_norm": 0.443359375,
       "learning_rate": 0.00019997054302734574,
-      "loss": 0.8451,
+      "loss": 0.8467,
       "step": 45
     },
     {
       "epoch": 0.10121012101210121,
-      "grad_norm": 0.408203125,
+      "grad_norm": 0.40625,
       "learning_rate": 0.0001999688358139404,
-      "loss": 0.7696,
+      "loss": 0.7717,
       "step": 46
     },
     {
       "epoch": 0.1034103410341034,
-      "grad_norm": 0.439453125,
+      "grad_norm": 0.43359375,
       "learning_rate": 0.00019996708052014757,
-      "loss": 0.7629,
+      "loss": 0.7632,
       "step": 47
     },
     {
       "epoch": 0.10561056105610561,
-      "grad_norm": 0.443359375,
+      "grad_norm": 0.431640625,
       "learning_rate": 0.0001999652771468116,
-      "loss": 0.7053,
+      "loss": 0.7041,
       "step": 48
     },
     {
       "epoch": 0.1078107810781078,
-      "grad_norm": 0.5078125,
+      "grad_norm": 0.515625,
       "learning_rate": 0.00019996342569479972,
-      "loss": 0.8565,
+      "loss": 0.8605,
       "step": 49
     },
     {
       "epoch": 0.11001100110011001,
-      "grad_norm": 0.50390625,
+      "grad_norm": 0.5078125,
       "learning_rate": 0.00019996152616500243,
-      "loss": 0.8059,
+      "loss": 0.8061,
       "step": 50
     },
     {
       "epoch": 0.11221122112211221,
-      "grad_norm": 0.45703125,
+      "grad_norm": 0.451171875,
       "learning_rate": 0.00019995957855833334,
-      "loss": 0.8208,
+      "loss": 0.8207,
       "step": 51
     },
     {
       "epoch": 0.11441144114411442,
-      "grad_norm": 0.44140625,
+      "grad_norm": 0.43359375,
       "learning_rate": 0.0001999575828757291,
-      "loss": 0.7431,
+      "loss": 0.7457,
       "step": 52
     },
     {
       "epoch": 0.11661166116611661,
-      "grad_norm": 0.4453125,
+      "grad_norm": 0.447265625,
       "learning_rate": 0.00019995553911814962,
-      "loss": 0.6929,
+      "loss": 0.6914,
       "step": 53
     },
     {
       "epoch": 0.1188118811881188,
-      "grad_norm": 0.51171875,
+      "grad_norm": 0.49609375,
       "learning_rate": 0.00019995344728657773,
-      "loss": 0.73,
+      "loss": 0.7325,
       "step": 54
     },
     {
       "epoch": 0.12101210121012101,
-      "grad_norm": 0.431640625,
+      "grad_norm": 0.427734375,
       "learning_rate": 0.00019995130738201966,
-      "loss": 0.7784,
+      "loss": 0.783,
       "step": 55
     },
     {
       "epoch": 0.12321232123212321,
-      "grad_norm": 0.41796875,
+      "grad_norm": 0.4140625,
       "learning_rate": 0.0001999491194055045,
-      "loss": 0.7549,
+      "loss": 0.7532,
       "step": 56
     },
     {
       "epoch": 0.1254125412541254,
-      "grad_norm": 0.41796875,
+      "grad_norm": 0.4140625,
       "learning_rate": 0.00019994688335808457,
-      "loss": 0.7767,
+      "loss": 0.7778,
       "step": 57
     },
     {
       "epoch": 0.1276127612761276,
-      "grad_norm": 0.38671875,
+      "grad_norm": 0.384765625,
       "learning_rate": 0.00019994459924083536,
-      "loss": 0.7214,
+      "loss": 0.7187,
       "step": 58
     },
     {
       "epoch": 0.12981298129812982,
-      "grad_norm": 0.455078125,
+      "grad_norm": 0.462890625,
       "learning_rate": 0.00019994226705485538,
-      "loss": 0.8325,
+      "loss": 0.8316,
       "step": 59
     },
     {
       "epoch": 0.132013201320132,
-      "grad_norm": 0.50390625,
+      "grad_norm": 0.4921875,
       "learning_rate": 0.0001999398868012663,
-      "loss": 0.7242,
+      "loss": 0.7251,
       "step": 60
     },
     {
       "epoch": 0.1342134213421342,
-      "grad_norm": 0.419921875,
+      "grad_norm": 0.416015625,
       "learning_rate": 0.00019993745848121295,
-      "loss": 0.6839,
+      "loss": 0.6866,
       "step": 61
     },
     {
       "epoch": 0.13641364136413642,
       "grad_norm": 0.41796875,
       "learning_rate": 0.00019993498209586324,
-      "loss": 0.701,
+      "loss": 0.6995,
       "step": 62
     },
     {
       "epoch": 0.13861386138613863,
-      "grad_norm": 0.5,
+      "grad_norm": 0.48828125,
       "learning_rate": 0.00019993245764640812,
-      "loss": 0.838,
+      "loss": 0.8377,
       "step": 63
     },
     {
       "epoch": 0.1408140814081408,
-      "grad_norm": 0.419921875,
+      "grad_norm": 0.41015625,
       "learning_rate": 0.0001999298851340618,
-      "loss": 0.781,
+      "loss": 0.7809,
       "step": 64
     },
     {
       "epoch": 0.14301430143014301,
-      "grad_norm": 0.42578125,
+      "grad_norm": 0.416015625,
       "learning_rate": 0.00019992726456006157,
-      "loss": 0.8018,
+      "loss": 0.8005,
       "step": 65
     },
     {
       "epoch": 0.14521452145214522,
-      "grad_norm": 0.453125,
+      "grad_norm": 0.435546875,
       "learning_rate": 0.00019992459592566773,
-      "loss": 0.7916,
+      "loss": 0.7887,
       "step": 66
     },
     {
       "epoch": 0.1474147414741474,
-      "grad_norm": 0.4453125,
+      "grad_norm": 0.435546875,
       "learning_rate": 0.0001999218792321638,
-      "loss": 0.831,
+      "loss": 0.8311,
       "step": 67
     },
     {
       "epoch": 0.1496149614961496,
-      "grad_norm": 0.416015625,
+      "grad_norm": 0.40625,
       "learning_rate": 0.00019991911448085642,
-      "loss": 0.767,
+      "loss": 0.7683,
       "step": 68
     },
     {
       "epoch": 0.15181518151815182,
-      "grad_norm": 0.42578125,
+      "grad_norm": 0.41796875,
       "learning_rate": 0.00019991630167307523,
-      "loss": 0.7247,
+      "loss": 0.7251,
       "step": 69
     },
     {
       "epoch": 0.15401540154015403,
-      "grad_norm": 0.447265625,
+      "grad_norm": 0.43359375,
       "learning_rate": 0.0001999134408101731,
-      "loss": 0.7621,
+      "loss": 0.7618,
       "step": 70
     },
     {
       "epoch": 0.1562156215621562,
-      "grad_norm": 0.45703125,
+      "grad_norm": 0.451171875,
       "learning_rate": 0.00019991053189352602,
-      "loss": 0.7742,
+      "loss": 0.7717,
       "step": 71
     },
     {
       "epoch": 0.15841584158415842,
-      "grad_norm": 0.46484375,
+      "grad_norm": 0.455078125,
       "learning_rate": 0.00019990757492453298,
-      "loss": 0.7273,
+      "loss": 0.7253,
       "step": 72
     },
     {
       "epoch": 0.16061606160616063,
-      "grad_norm": 0.443359375,
+      "grad_norm": 0.4375,
       "learning_rate": 0.00019990456990461616,
-      "loss": 0.7675,
+      "loss": 0.7692,
       "step": 73
     },
     {
       "epoch": 0.1628162816281628,
-      "grad_norm": 0.44921875,
+      "grad_norm": 0.447265625,
       "learning_rate": 0.00019990151683522086,
-      "loss": 0.8182,
+      "loss": 0.8188,
       "step": 74
     },
     {
       "epoch": 0.16501650165016502,
-      "grad_norm": 0.427734375,
+      "grad_norm": 0.431640625,
       "learning_rate": 0.00019989841571781543,
-      "loss": 0.6707,
+      "loss": 0.6719,
       "step": 75
     },
     {
       "epoch": 0.16721672167216722,
-      "grad_norm": 0.458984375,
+      "grad_norm": 0.4609375,
       "learning_rate": 0.00019989526655389138,
-      "loss": 0.7961,
+      "loss": 0.795,
       "step": 76
     },
     {
       "epoch": 0.1694169416941694,
-      "grad_norm": 0.3984375,
+      "grad_norm": 0.40234375,
       "learning_rate": 0.00019989206934496333,
-      "loss": 0.7408,
+      "loss": 0.7406,
       "step": 77
     },
     {
       "epoch": 0.1716171617161716,
-      "grad_norm": 0.439453125,
+      "grad_norm": 0.44140625,
       "learning_rate": 0.00019988882409256896,
-      "loss": 0.7724,
+      "loss": 0.7712,
       "step": 78
     },
     {
       "epoch": 0.17381738173817382,
       "grad_norm": 0.439453125,
       "learning_rate": 0.00019988553079826912,
-      "loss": 0.7352,
+      "loss": 0.7375,
       "step": 79
     },
     {
       "epoch": 0.17601760176017603,
-      "grad_norm": 0.47265625,
+      "grad_norm": 0.482421875,
       "learning_rate": 0.0001998821894636477,
-      "loss": 0.6251,
+      "loss": 0.6229,
       "step": 80
     },
     {
       "epoch": 0.1782178217821782,
-      "grad_norm": 0.447265625,
+      "grad_norm": 0.45703125,
       "learning_rate": 0.00019987880009031172,
-      "loss": 0.7252,
+      "loss": 0.7251,
       "step": 81
     },
     {
       "epoch": 0.18041804180418042,
-      "grad_norm": 0.458984375,
+      "grad_norm": 0.453125,
       "learning_rate": 0.00019987536267989135,
-      "loss": 0.6399,
+      "loss": 0.6408,
       "step": 82
     },
     {
       "epoch": 0.18261826182618263,
-      "grad_norm": 0.56640625,
+      "grad_norm": 0.5625,
       "learning_rate": 0.00019987187723403981,
-      "loss": 0.8445,
+      "loss": 0.849,
       "step": 83
     },
     {
       "epoch": 0.1848184818481848,
-      "grad_norm": 0.494140625,
+      "grad_norm": 0.48828125,
       "learning_rate": 0.00019986834375443344,
-      "loss": 0.7356,
+      "loss": 0.7351,
       "step": 84
     },
     {
       "epoch": 0.18701870187018702,
-      "grad_norm": 0.482421875,
+      "grad_norm": 0.478515625,
       "learning_rate": 0.00019986476224277165,
-      "loss": 0.7571,
+      "loss": 0.7579,
       "step": 85
     },
     {
       "epoch": 0.18921892189218922,
-      "grad_norm": 0.4765625,
+      "grad_norm": 0.466796875,
       "learning_rate": 0.00019986113270077703,
-      "loss": 0.717,
+      "loss": 0.7193,
       "step": 86
     },
     {
       "epoch": 0.19141914191419143,
-      "grad_norm": 0.470703125,
+      "grad_norm": 0.4609375,
       "learning_rate": 0.0001998574551301952,
-      "loss": 0.7155,
+      "loss": 0.7132,
       "step": 87
     },
     {
       "epoch": 0.1936193619361936,
-      "grad_norm": 0.52734375,
+      "grad_norm": 0.53125,
       "learning_rate": 0.0001998537295327949,
-      "loss": 0.7495,
+      "loss": 0.7445,
       "step": 88
     },
     {
       "epoch": 0.19581958195819582,
-      "grad_norm": 0.55078125,
+      "grad_norm": 0.59375,
       "learning_rate": 0.00019984995591036797,
-      "loss": 0.7229,
+      "loss": 0.7212,
       "step": 89
     },
     {
       "epoch": 0.19801980198019803,
-      "grad_norm": 0.5390625,
+      "grad_norm": 0.53125,
       "learning_rate": 0.00019984613426472932,
-      "loss": 0.6987,
+      "loss": 0.7008,
       "step": 90
     },
     {
       "epoch": 0.2002200220022002,
-      "grad_norm": 0.46875,
+      "grad_norm": 0.4609375,
       "learning_rate": 0.00019984226459771704,
-      "loss": 0.6072,
+      "loss": 0.6075,
       "step": 91
     },
     {
       "epoch": 0.20242024202420242,
       "grad_norm": 0.435546875,
       "learning_rate": 0.00019983834691119224,
-      "loss": 0.5921,
+      "loss": 0.5908,
       "step": 92
     },
     {
       "epoch": 0.20462046204620463,
-      "grad_norm": 0.51171875,
+      "grad_norm": 0.50390625,
       "learning_rate": 0.00019983438120703913,
-      "loss": 0.6798,
+      "loss": 0.6755,
       "step": 93
     },
     {
       "epoch": 0.2068206820682068,
       "grad_norm": 0.5078125,
       "learning_rate": 0.00019983036748716504,
-      "loss": 0.719,
+      "loss": 0.7188,
       "step": 94
     },
     {
       "epoch": 0.20902090209020902,
-      "grad_norm": 0.54296875,
+      "grad_norm": 0.55078125,
       "learning_rate": 0.00019982630575350038,
-      "loss": 0.7543,
+      "loss": 0.7535,
       "step": 95
     },
     {
       "epoch": 0.21122112211221122,
-      "grad_norm": 0.458984375,
+      "grad_norm": 0.4609375,
       "learning_rate": 0.00019982219600799868,
-      "loss": 0.5874,
+      "loss": 0.5857,
       "step": 96
     },
     {
       "epoch": 0.21342134213421343,
-      "grad_norm": 0.54296875,
+      "grad_norm": 0.5390625,
       "learning_rate": 0.0001998180382526365,
-      "loss": 0.7416,
+      "loss": 0.7398,
       "step": 97
     },
     {
       "epoch": 0.2156215621562156,
-      "grad_norm": 0.58984375,
+      "grad_norm": 0.484375,
       "learning_rate": 0.00019981383248941355,
-      "loss": 0.6071,
+      "loss": 0.6067,
       "step": 98
     },
     {
       "epoch": 0.21782178217821782,
-      "grad_norm": 0.48046875,
+      "grad_norm": 0.486328125,
       "learning_rate": 0.00019980957872035258,
-      "loss": 0.5972,
+      "loss": 0.5988,
       "step": 99
     },
     {
       "epoch": 0.22002200220022003,
-      "grad_norm": 0.5078125,
+      "grad_norm": 0.5234375,
       "learning_rate": 0.00019980527694749952,
-      "loss": 0.6788,
+      "loss": 0.6792,
       "step": 100
     },
     {
       "epoch": 0.2222222222222222,
-      "grad_norm": 0.5234375,
+      "grad_norm": 0.53125,
       "learning_rate": 0.00019980092717292327,
-      "loss": 0.6443,
+      "loss": 0.6469,
       "step": 101
     },
     {
       "epoch": 0.22442244224422442,
-      "grad_norm": 0.5546875,
+      "grad_norm": 0.546875,
       "learning_rate": 0.00019979652939871586,
-      "loss": 0.6823,
+      "loss": 0.6841,
       "step": 102
     },
     {
       "epoch": 0.22662266226622663,
-      "grad_norm": 0.5859375,
+      "grad_norm": 0.55859375,
       "learning_rate": 0.00019979208362699247,
-      "loss": 0.6467,
+      "loss": 0.6498,
       "step": 103
     },
     {
       "epoch": 0.22882288228822883,
-      "grad_norm": 0.59765625,
+      "grad_norm": 0.56640625,
       "learning_rate": 0.00019978758985989128,
-      "loss": 0.6213,
+      "loss": 0.6171,
       "step": 104
     },
     {
       "epoch": 0.23102310231023102,
-      "grad_norm": 0.828125,
+      "grad_norm": 0.796875,
       "learning_rate": 0.00019978304809957357,
-      "loss": 0.6306,
+      "loss": 0.6278,
       "step": 105
     },
     {
       "epoch": 0.23322332233223322,
-      "grad_norm": 0.65234375,
+      "grad_norm": 0.64453125,
       "learning_rate": 0.0001997784583482237,
-      "loss": 0.6776,
+      "loss": 0.6742,
       "step": 106
     },
     {
       "epoch": 0.23542354235423543,
-      "grad_norm": 0.65234375,
+      "grad_norm": 0.66015625,
       "learning_rate": 0.00019977382060804922,
-      "loss": 0.6047,
+      "loss": 0.6057,
       "step": 107
     },
     {
       "epoch": 0.2376237623762376,
-      "grad_norm": 0.609375,
+      "grad_norm": 0.6328125,
       "learning_rate": 0.0001997691348812806,
-      "loss": 0.6722,
+      "loss": 0.6686,
       "step": 108
     },
     {
       "epoch": 0.23982398239823982,
-      "grad_norm": 0.4765625,
+      "grad_norm": 0.484375,
       "learning_rate": 0.00019976440117017144,
-      "loss": 0.6675,
+      "loss": 0.6644,
       "step": 109
     },
     {
       "epoch": 0.24202420242024203,
-      "grad_norm": 0.498046875,
+      "grad_norm": 0.494140625,
       "learning_rate": 0.00019975961947699846,
-      "loss": 0.6712,
+      "loss": 0.6714,
       "step": 110
     },
     {
       "epoch": 0.24422442244224424,
-      "grad_norm": 0.490234375,
+      "grad_norm": 0.49609375,
       "learning_rate": 0.00019975478980406144,
-      "loss": 0.6214,
+      "loss": 0.618,
       "step": 111
     },
     {
       "epoch": 0.24642464246424642,
-      "grad_norm": 0.5078125,
+      "grad_norm": 0.5,
       "learning_rate": 0.00019974991215368325,
-      "loss": 0.6583,
+      "loss": 0.6568,
       "step": 112
     },
     {
       "epoch": 0.24862486248624863,
-      "grad_norm": 0.462890625,
+      "grad_norm": 0.46875,
       "learning_rate": 0.00019974498652820975,
-      "loss": 0.6457,
+      "loss": 0.6504,
       "step": 113
     },
     {
       "epoch": 0.2508250825082508,
-      "grad_norm": 0.498046875,
+      "grad_norm": 0.4921875,
       "learning_rate": 0.00019974001293001002,
-      "loss": 0.6452,
+      "loss": 0.6431,
       "step": 114
     },
     {
       "epoch": 0.2508250825082508,
-      "eval_loss": 0.6273543238639832,
-      "eval_runtime": 10.1176,
-      "eval_samples_per_second": 37.855,
-      "eval_steps_per_second": 4.744,
+      "eval_loss": 0.6255941987037659,
+      "eval_runtime": 11.2106,
+      "eval_samples_per_second": 34.164,
+      "eval_steps_per_second": 4.282,
       "step": 114
     },
     {
       "epoch": 0.253025302530253,
-      "grad_norm": 0.498046875,
+      "grad_norm": 0.5078125,
       "learning_rate": 0.00019973499136147606,
-      "loss": 0.7141,
+      "loss": 0.7106,
       "step": 115
     },
     {
       "epoch": 0.2552255225522552,
-      "grad_norm": 0.484375,
+      "grad_norm": 0.5,
       "learning_rate": 0.00019972992182502302,
-      "loss": 0.5185,
+      "loss": 0.5194,
       "step": 116
     },
     {
       "epoch": 0.25742574257425743,
-      "grad_norm": 0.453125,
+      "grad_norm": 0.46484375,
       "learning_rate": 0.00019972480432308917,
-      "loss": 0.5353,
+      "loss": 0.5313,
       "step": 117
     },
     {
       "epoch": 0.25962596259625964,
-      "grad_norm": 0.52734375,
+      "grad_norm": 0.546875,
       "learning_rate": 0.00019971963885813575,
-      "loss": 0.6382,
+      "loss": 0.6405,
       "step": 118
     },
     {
       "epoch": 0.26182618261826185,
-      "grad_norm": 0.6015625,
+      "grad_norm": 0.625,
       "learning_rate": 0.00019971442543264712,
-      "loss": 0.7646,
+      "loss": 0.7635,
       "step": 119
     },
     {
       "epoch": 0.264026402640264,
-      "grad_norm": 0.609375,
+      "grad_norm": 0.62890625,
       "learning_rate": 0.00019970916404913068,
-      "loss": 0.6999,
+      "loss": 0.7007,
       "step": 120
     },
     {
       "epoch": 0.2662266226622662,
-      "grad_norm": 0.6171875,
+      "grad_norm": 0.58984375,
       "learning_rate": 0.0001997038547101169,
-      "loss": 0.659,
+      "loss": 0.6509,
       "step": 121
     },
     {
       "epoch": 0.2684268426842684,
-      "grad_norm": 0.55859375,
+      "grad_norm": 0.5546875,
       "learning_rate": 0.0001996984974181594,
-      "loss": 0.6741,
+      "loss": 0.6686,
       "step": 122
     },
     {
       "epoch": 0.2706270627062706,
       "grad_norm": 0.51953125,
       "learning_rate": 0.0001996930921758347,
-      "loss": 0.5667,
+      "loss": 0.566,
       "step": 123
     },
     {
       "epoch": 0.27282728272827284,
-      "grad_norm": 0.56640625,
+      "grad_norm": 0.55859375,
       "learning_rate": 0.00019968763898574253,
-      "loss": 0.6873,
+      "loss": 0.6888,
       "step": 124
     },
     {
       "epoch": 0.27502750275027504,
-      "grad_norm": 0.54296875,
+      "grad_norm": 0.53125,
       "learning_rate": 0.0001996821378505056,
-      "loss": 0.6297,
+      "loss": 0.6265,
       "step": 125
     },
     {
       "epoch": 0.27722772277227725,
-      "grad_norm": 0.53515625,
+      "grad_norm": 0.5390625,
       "learning_rate": 0.00019967658877276972,
-      "loss": 0.5374,
+      "loss": 0.5411,
       "step": 126
     },
     {
       "epoch": 0.2794279427942794,
-      "grad_norm": 0.5078125,
+      "grad_norm": 0.498046875,
       "learning_rate": 0.00019967099175520367,
-      "loss": 0.5409,
+      "loss": 0.5347,
       "step": 127
     },
     {
       "epoch": 0.2816281628162816,
-      "grad_norm": 0.671875,
+      "grad_norm": 0.66796875,
       "learning_rate": 0.00019966534680049947,
-      "loss": 0.5124,
+      "loss": 0.5073,
       "step": 128
     },
     {
       "epoch": 0.2838283828382838,
       "grad_norm": 0.58203125,
       "learning_rate": 0.000199659653911372,
-      "loss": 0.5845,
+      "loss": 0.5862,
       "step": 129
     },
     {
       "epoch": 0.28602860286028603,
-      "grad_norm": 0.61328125,
+      "grad_norm": 0.63671875,
       "learning_rate": 0.0001996539130905593,
-      "loss": 0.679,
+      "loss": 0.6786,
       "step": 130
     },
     {
       "epoch": 0.28822882288228824,
-      "grad_norm": 0.6328125,
+      "grad_norm": 0.65234375,
       "learning_rate": 0.00019964812434082244,
-      "loss": 0.6327,
+      "loss": 0.6341,
       "step": 131
     },
     {
       "epoch": 0.29042904290429045,
-      "grad_norm": 0.5859375,
+      "grad_norm": 0.59765625,
       "learning_rate": 0.00019964228766494556,
-      "loss": 0.5661,
+      "loss": 0.5611,
       "step": 132
     },
     {
       "epoch": 0.29262926292629265,
-      "grad_norm": 0.53125,
+      "grad_norm": 0.546875,
       "learning_rate": 0.0001996364030657358,
-      "loss": 0.5718,
+      "loss": 0.5736,
       "step": 133
     },
     {
       "epoch": 0.2948294829482948,
-      "grad_norm": 0.59375,
+      "grad_norm": 0.57421875,
       "learning_rate": 0.00019963047054602338,
-      "loss": 0.6293,
+      "loss": 0.6225,
       "step": 134
     },
     {
       "epoch": 0.297029702970297,
-      "grad_norm": 0.5703125,
+      "grad_norm": 0.578125,
       "learning_rate": 0.00019962449010866158,
-      "loss": 0.6433,
+      "loss": 0.6413,
       "step": 135
     },
     {
       "epoch": 0.2992299229922992,
-      "grad_norm": 0.56640625,
+      "grad_norm": 0.5859375,
       "learning_rate": 0.00019961846175652672,
-      "loss": 0.57,
+      "loss": 0.5744,
       "step": 136
     },
     {
       "epoch": 0.30143014301430143,
-      "grad_norm": 0.6328125,
+      "grad_norm": 0.55078125,
       "learning_rate": 0.00019961238549251814,
-      "loss": 0.4844,
+      "loss": 0.4756,
       "step": 137
     },
     {
       "epoch": 0.30363036303630364,
-      "grad_norm": 0.63671875,
+      "grad_norm": 0.625,
       "learning_rate": 0.00019960626131955826,
-      "loss": 0.5945,
+      "loss": 0.5914,
       "step": 138
     },
     {
       "epoch": 0.30583058305830585,
-      "grad_norm": 0.65234375,
+      "grad_norm": 0.63671875,
       "learning_rate": 0.00019960008924059254,
-      "loss": 0.6479,
+      "loss": 0.6449,
       "step": 139
     },
     {
       "epoch": 0.30803080308030806,
-      "grad_norm": 0.57421875,
+      "grad_norm": 0.5546875,
       "learning_rate": 0.00019959386925858942,
-      "loss": 0.6487,
+      "loss": 0.6484,
       "step": 140
     },
     {
       "epoch": 0.3102310231023102,
       "grad_norm": 0.53515625,
       "learning_rate": 0.00019958760137654046,
-      "loss": 0.5265,
+      "loss": 0.5275,
       "step": 141
     },
     {
       "epoch": 0.3124312431243124,
-      "grad_norm": 0.484375,
+      "grad_norm": 0.46484375,
       "learning_rate": 0.00019958128559746018,
-      "loss": 0.466,
+      "loss": 0.4687,
       "step": 142
     },
     {
       "epoch": 0.3146314631463146,
       "grad_norm": 0.57421875,
       "learning_rate": 0.00019957492192438622,
-      "loss": 0.5767,
+      "loss": 0.5726,
       "step": 143
     },
     {
       "epoch": 0.31683168316831684,
-      "grad_norm": 0.5390625,
+      "grad_norm": 0.5546875,
       "learning_rate": 0.0001995685103603792,
-      "loss": 0.5445,
+      "loss": 0.5361,
       "step": 144
     },
     {
       "epoch": 0.31903190319031904,
-      "grad_norm": 0.55078125,
+      "grad_norm": 0.546875,
       "learning_rate": 0.0001995620509085228,
-      "loss": 0.5724,
+      "loss": 0.5707,
       "step": 145
     },
     {
       "epoch": 0.32123212321232125,
       "grad_norm": 0.51953125,
       "learning_rate": 0.0001995555435719237,
-      "loss": 0.5807,
+      "loss": 0.5763,
       "step": 146
     },
     {
       "epoch": 0.3234323432343234,
-      "grad_norm": 0.53515625,
+      "grad_norm": 0.54296875,
       "learning_rate": 0.0001995489883537116,
-      "loss": 0.4785,
+      "loss": 0.4774,
       "step": 147
     },
     {
       "epoch": 0.3256325632563256,
-      "grad_norm": 0.62109375,
+      "grad_norm": 0.66796875,
       "learning_rate": 0.00019954238525703933,
-      "loss": 0.6031,
+      "loss": 0.6063,
       "step": 148
     },
     {
       "epoch": 0.3278327832783278,
       "grad_norm": 0.6015625,
       "learning_rate": 0.0001995357342850826,
-      "loss": 0.5354,
+      "loss": 0.5387,
       "step": 149
     },
     {
       "epoch": 0.33003300330033003,
-      "grad_norm": 0.65234375,
+      "grad_norm": 0.69921875,
       "learning_rate": 0.00019952903544104026,
-      "loss": 0.5268,
+      "loss": 0.5244,
       "step": 150
     },
     {
       "epoch": 0.33223322332233224,
-      "grad_norm": 0.6484375,
+      "grad_norm": 0.60546875,
       "learning_rate": 0.0001995222887281341,
-      "loss": 0.6122,
+      "loss": 0.611,
       "step": 151
     },
     {
       "epoch": 0.33443344334433445,
-      "grad_norm": 0.640625,
+      "grad_norm": 0.64453125,
       "learning_rate": 0.00019951549414960905,
-      "loss": 0.5712,
+      "loss": 0.5676,
       "step": 152
     },
     {
       "epoch": 0.33663366336633666,
-      "grad_norm": 0.54296875,
+      "grad_norm": 0.53515625,
       "learning_rate": 0.00019950865170873295,
-      "loss": 0.6102,
+      "loss": 0.6142,
       "step": 153
     },
     {
       "epoch": 0.3388338833883388,
-      "grad_norm": 0.57421875,
+      "grad_norm": 0.5625,
       "learning_rate": 0.00019950176140879668,
-      "loss": 0.6033,
+      "loss": 0.6079,
       "step": 154
     },
     {
       "epoch": 0.341034103410341,
-      "grad_norm": 0.54296875,
+      "grad_norm": 0.52734375,
       "learning_rate": 0.00019949482325311418,
-      "loss": 0.5215,
+      "loss": 0.531,
       "step": 155
     },
     {
       "epoch": 0.3432343234323432,
-      "grad_norm": 0.578125,
+      "grad_norm": 0.5546875,
       "learning_rate": 0.00019948783724502237,
-      "loss": 0.5283,
+      "loss": 0.5292,
       "step": 156
     },
     {
       "epoch": 0.34543454345434543,
-      "grad_norm": 0.5703125,
+      "grad_norm": 0.5546875,
       "learning_rate": 0.0001994808033878812,
-      "loss": 0.5208,
+      "loss": 0.5207,
       "step": 157
     },
     {
       "epoch": 0.34763476347634764,
-      "grad_norm": 0.6328125,
+      "grad_norm": 0.62890625,
       "learning_rate": 0.00019947372168507366,
-      "loss": 0.5733,
+      "loss": 0.5677,
       "step": 158
     },
     {
       "epoch": 0.34983498349834985,
-      "grad_norm": 0.62109375,
+      "grad_norm": 0.625,
       "learning_rate": 0.00019946659214000568,
-      "loss": 0.6815,
+      "loss": 0.673,
       "step": 159
     },
     {
       "epoch": 0.35203520352035206,
       "grad_norm": 0.5234375,
       "learning_rate": 0.00019945941475610623,
-      "loss": 0.4203,
+      "loss": 0.4233,
       "step": 160
     },
     {
       "epoch": 0.3542354235423542,
-      "grad_norm": 0.58984375,
+      "grad_norm": 0.62109375,
       "learning_rate": 0.00019945218953682734,
-      "loss": 0.4789,
+      "loss": 0.4777,
       "step": 161
     },
     {
       "epoch": 0.3564356435643564,
-      "grad_norm": 0.828125,
+      "grad_norm": 0.83984375,
       "learning_rate": 0.000199444916485644,
-      "loss": 0.5825,
+      "loss": 0.5833,
       "step": 162
     },
     {
       "epoch": 0.3586358635863586,
-      "grad_norm": 0.703125,
+      "grad_norm": 0.73828125,
       "learning_rate": 0.0001994375956060542,
-      "loss": 0.5115,
+      "loss": 0.5146,
       "step": 163
     },
     {
       "epoch": 0.36083608360836084,
-      "grad_norm": 0.65625,
+      "grad_norm": 0.68359375,
       "learning_rate": 0.00019943022690157894,
-      "loss": 0.6393,
+      "loss": 0.6381,
       "step": 164
     },
     {
       "epoch": 0.36303630363036304,
-      "grad_norm": 0.58984375,
+      "grad_norm": 0.62890625,
       "learning_rate": 0.00019942281037576222,
-      "loss": 0.5961,
+      "loss": 0.6008,
       "step": 165
     },
     {
       "epoch": 0.36523652365236525,
-      "grad_norm": 0.58203125,
+      "grad_norm": 0.6015625,
       "learning_rate": 0.00019941534603217108,
-      "loss": 0.4918,
+      "loss": 0.487,
       "step": 166
     },
     {
       "epoch": 0.36743674367436746,
-      "grad_norm": 0.609375,
+      "grad_norm": 0.6171875,
       "learning_rate": 0.0001994078338743955,
-      "loss": 0.5496,
+      "loss": 0.5456,
       "step": 167
     },
     {
       "epoch": 0.3696369636963696,
-      "grad_norm": 0.6328125,
+      "grad_norm": 0.63671875,
       "learning_rate": 0.0001994002739060485,
-      "loss": 0.4807,
+      "loss": 0.4782,
       "step": 168
     },
     {
       "epoch": 0.3718371837183718,
-      "grad_norm": 0.62890625,
+      "grad_norm": 0.62109375,
       "learning_rate": 0.00019939266613076603,
-      "loss": 0.5536,
+      "loss": 0.5506,
       "step": 169
     },
     {
       "epoch": 0.37403740374037403,
-      "grad_norm": 0.5625,
+      "grad_norm": 0.5703125,
       "learning_rate": 0.00019938501055220711,
-      "loss": 0.505,
+      "loss": 0.5121,
       "step": 170
     },
     {
       "epoch": 0.37623762376237624,
-      "grad_norm": 0.546875,
+      "grad_norm": 0.5859375,
       "learning_rate": 0.00019937730717405373,
-      "loss": 0.4755,
+      "loss": 0.4761,
       "step": 171
     },
     {
       "epoch": 0.37843784378437845,
       "grad_norm": 0.609375,
       "learning_rate": 0.00019936955600001084,
-      "loss": 0.5409,
+      "loss": 0.5429,
       "step": 172
     },
     {
       "epoch": 0.38063806380638066,
-      "grad_norm": 0.56640625,
+      "grad_norm": 0.55859375,
       "learning_rate": 0.0001993617570338064,
-      "loss": 0.5537,
+      "loss": 0.5492,
       "step": 173
     },
     {
       "epoch": 0.38283828382838286,
-      "grad_norm": 0.68359375,
+      "grad_norm": 0.7421875,
       "learning_rate": 0.00019935391027919134,
-      "loss": 0.5191,
+      "loss": 0.5213,
       "step": 174
     },
     {
       "epoch": 0.385038503850385,
-      "grad_norm": 0.58984375,
+      "grad_norm": 0.578125,
       "learning_rate": 0.0001993460157399396,
-      "loss": 0.4316,
+      "loss": 0.4313,
       "step": 175
     },
     {
       "epoch": 0.3872387238723872,
-      "grad_norm": 0.7421875,
+      "grad_norm": 0.69921875,
       "learning_rate": 0.00019933807341984813,
-      "loss": 0.5529,
+      "loss": 0.5563,
       "step": 176
     },
     {
       "epoch": 0.38943894389438943,
-      "grad_norm": 0.671875,
+      "grad_norm": 0.6640625,
       "learning_rate": 0.00019933008332273674,
-      "loss": 0.498,
+      "loss": 0.5038,
       "step": 177
     },
     {
       "epoch": 0.39163916391639164,
-      "grad_norm": 0.734375,
+      "grad_norm": 0.73046875,
       "learning_rate": 0.00019932204545244835,
-      "loss": 0.5604,
+      "loss": 0.5607,
       "step": 178
     },
     {
       "epoch": 0.39383938393839385,
-      "grad_norm": 0.6875,
+      "grad_norm": 0.6640625,
       "learning_rate": 0.00019931395981284878,
-      "loss": 0.4456,
+      "loss": 0.4454,
       "step": 179
     },
     {
       "epoch": 0.39603960396039606,
-      "grad_norm": 0.7578125,
+      "grad_norm": 0.7265625,
       "learning_rate": 0.00019930582640782684,
-      "loss": 0.6052,
+      "loss": 0.6091,
       "step": 180
     },
     {
       "epoch": 0.39823982398239827,
-      "grad_norm": 0.6015625,
+      "grad_norm": 0.58984375,
       "learning_rate": 0.00019929764524129437,
-      "loss": 0.4087,
+      "loss": 0.4034,
       "step": 181
     },
     {
       "epoch": 0.4004400440044004,
-      "grad_norm": 0.64453125,
+      "grad_norm": 0.60546875,
       "learning_rate": 0.00019928941631718608,
-      "loss": 0.4851,
+      "loss": 0.4857,
       "step": 182
     },
     {
       "epoch": 0.40264026402640263,
-      "grad_norm": 0.546875,
+      "grad_norm": 0.53125,
       "learning_rate": 0.00019928113963945974,
-      "loss": 0.3949,
+      "loss": 0.4008,
       "step": 183
     },
     {
       "epoch": 0.40484048404840484,
-      "grad_norm": 0.54296875,
+      "grad_norm": 0.52734375,
       "learning_rate": 0.000199272815212096,
-      "loss": 0.4578,
+      "loss": 0.4603,
       "step": 184
     },
     {
       "epoch": 0.40704070407040704,
-      "grad_norm": 0.64453125,
+      "grad_norm": 0.671875,
       "learning_rate": 0.0001992644430390986,
-      "loss": 0.4893,
+      "loss": 0.5055,
       "step": 185
     },
     {
       "epoch": 0.40924092409240925,
-      "grad_norm": 0.7890625,
+      "grad_norm": 0.7734375,
       "learning_rate": 0.00019925602312449408,
-      "loss": 0.6331,
+      "loss": 0.631,
       "step": 186
     },
     {
       "epoch": 0.41144114411441146,
-      "grad_norm": 0.69921875,
+      "grad_norm": 0.71875,
       "learning_rate": 0.00019924755547233208,
-      "loss": 0.5551,
+      "loss": 0.5532,
       "step": 187
     },
     {
       "epoch": 0.4136413641364136,
-      "grad_norm": 0.58203125,
+      "grad_norm": 0.59375,
       "learning_rate": 0.00019923904008668517,
-      "loss": 0.432,
+      "loss": 0.4281,
       "step": 188
     },
     {
       "epoch": 0.4158415841584158,
-      "grad_norm": 0.6015625,
+      "grad_norm": 0.6484375,
       "learning_rate": 0.00019923047697164884,
-      "loss": 0.5318,
+      "loss": 0.5333,
       "step": 189
     },
     {
       "epoch": 0.41804180418041803,
-      "grad_norm": 0.70703125,
+      "grad_norm": 0.7421875,
       "learning_rate": 0.0001992218661313415,
-      "loss": 0.4624,
+      "loss": 0.4639,
       "step": 190
     },
     {
       "epoch": 0.42024202420242024,
-      "grad_norm": 0.65625,
+      "grad_norm": 0.67578125,
       "learning_rate": 0.00019921320756990466,
-      "loss": 0.5016,
+      "loss": 0.5084,
       "step": 191
     },
     {
       "epoch": 0.42244224422442245,
-      "grad_norm": 0.64453125,
+      "grad_norm": 0.6484375,
       "learning_rate": 0.0001992045012915026,
-      "loss": 0.5281,
+      "loss": 0.5213,
       "step": 192
     },
     {
       "epoch": 0.42464246424642466,
-      "grad_norm": 0.58203125,
+      "grad_norm": 0.58984375,
       "learning_rate": 0.0001991957473003227,
-      "loss": 0.502,
+      "loss": 0.4954,
       "step": 193
     },
     {
       "epoch": 0.42684268426842686,
-      "grad_norm": 0.6484375,
+      "grad_norm": 0.6640625,
       "learning_rate": 0.00019918694560057518,
-      "loss": 0.4656,
+      "loss": 0.4603,
       "step": 194
     },
     {
       "epoch": 0.429042904290429,
-      "grad_norm": 0.53125,
+      "grad_norm": 0.5390625,
       "learning_rate": 0.00019917809619649326,
-      "loss": 0.4035,
+      "loss": 0.4091,
       "step": 195
     },
     {
       "epoch": 0.4312431243124312,
-      "grad_norm": 0.59765625,
+      "grad_norm": 0.60546875,
       "learning_rate": 0.0001991691990923331,
-      "loss": 0.3667,
+      "loss": 0.3808,
       "step": 196
     },
     {
       "epoch": 0.43344334433443343,
-      "grad_norm": 0.60546875,
+      "grad_norm": 0.6015625,
       "learning_rate": 0.00019916025429237384,
-      "loss": 0.4985,
+      "loss": 0.4957,
       "step": 197
     },
     {
       "epoch": 0.43564356435643564,
-      "grad_norm": 0.59765625,
+      "grad_norm": 0.6796875,
       "learning_rate": 0.00019915126180091743,
-      "loss": 0.3472,
+      "loss": 0.3527,
       "step": 198
     },
     {
       "epoch": 0.43784378437843785,
-      "grad_norm": 0.7265625,
+      "grad_norm": 0.74609375,
       "learning_rate": 0.0001991422216222889,
-      "loss": 0.5408,
+      "loss": 0.5486,
       "step": 199
     },
     {
       "epoch": 0.44004400440044006,
-      "grad_norm": 0.64453125,
+      "grad_norm": 0.66015625,
       "learning_rate": 0.00019913313376083617,
-      "loss": 0.4638,
+      "loss": 0.4692,
       "step": 200
     },
     {
       "epoch": 0.44224422442244227,
-      "grad_norm": 0.81640625,
+      "grad_norm": 0.765625,
       "learning_rate": 0.00019912399822093004,
-      "loss": 0.4529,
+      "loss": 0.4496,
       "step": 201
     },
     {
       "epoch": 0.4444444444444444,
-      "grad_norm": 0.609375,
+      "grad_norm": 0.60546875,
       "learning_rate": 0.0001991148150069643,
-      "loss": 0.4917,
+      "loss": 0.4775,
       "step": 202
     },
     {
       "epoch": 0.44664466446644663,
-      "grad_norm": 0.828125,
+      "grad_norm": 0.83203125,
       "learning_rate": 0.00019910558412335565,
-      "loss": 0.4923,
+      "loss": 0.4813,
       "step": 203
     },
     {
       "epoch": 0.44884488448844884,
-      "grad_norm": 0.6796875,
+      "grad_norm": 0.68359375,
       "learning_rate": 0.0001990963055745437,
-      "loss": 0.4052,
+      "loss": 0.4041,
       "step": 204
     },
     {
       "epoch": 0.45104510451045104,
-      "grad_norm": 0.6796875,
+      "grad_norm": 0.671875,
       "learning_rate": 0.00019908697936499103,
-      "loss": 0.493,
+      "loss": 0.4845,
       "step": 205
     },
     {
       "epoch": 0.45324532453245325,
-      "grad_norm": 0.58984375,
+      "grad_norm": 0.6171875,
       "learning_rate": 0.0001990776054991831,
-      "loss": 0.4796,
+      "loss": 0.4737,
       "step": 206
     },
     {
       "epoch": 0.45544554455445546,
-      "grad_norm": 0.74609375,
+      "grad_norm": 0.77734375,
       "learning_rate": 0.00019906818398162835,
-      "loss": 0.6975,
+      "loss": 0.6988,
       "step": 207
     },
     {
       "epoch": 0.45764576457645767,
-      "grad_norm": 0.640625,
+      "grad_norm": 0.66796875,
       "learning_rate": 0.00019905871481685803,
-      "loss": 0.3441,
+      "loss": 0.3444,
       "step": 208
     },
     {
       "epoch": 0.4598459845984598,
-      "grad_norm": 0.62890625,
+      "grad_norm": 0.66796875,
       "learning_rate": 0.0001990491980094264,
-      "loss": 0.3954,
+      "loss": 0.4036,
       "step": 209
     },
     {
       "epoch": 0.46204620462046203,
-      "grad_norm": 0.703125,
+      "grad_norm": 0.71875,
       "learning_rate": 0.00019903963356391057,
-      "loss": 0.3972,
+      "loss": 0.398,
       "step": 210
     },
     {
       "epoch": 0.46424642464246424,
-      "grad_norm": 0.65625,
+      "grad_norm": 0.6953125,
       "learning_rate": 0.00019903002148491066,
-      "loss": 0.5089,
+      "loss": 0.5144,
       "step": 211
     },
     {
       "epoch": 0.46644664466446645,
-      "grad_norm": 0.62109375,
+      "grad_norm": 0.61328125,
       "learning_rate": 0.0001990203617770496,
-      "loss": 0.3819,
+      "loss": 0.3773,
       "step": 212
     },
     {
       "epoch": 0.46864686468646866,
-      "grad_norm": 0.7265625,
+      "grad_norm": 0.70703125,
       "learning_rate": 0.00019901065444497326,
-      "loss": 0.4381,
+      "loss": 0.4194,
       "step": 213
     },
     {
       "epoch": 0.47084708470847086,
-      "grad_norm": 0.9375,
+      "grad_norm": 0.76171875,
       "learning_rate": 0.00019900089949335042,
-      "loss": 0.4794,
+      "loss": 0.4776,
       "step": 214
     },
     {
       "epoch": 0.4730473047304731,
-      "grad_norm": 0.62109375,
+      "grad_norm": 0.59765625,
       "learning_rate": 0.00019899109692687278,
-      "loss": 0.4076,
+      "loss": 0.4016,
       "step": 215
     },
     {
       "epoch": 0.4752475247524752,
       "grad_norm": 0.5703125,
       "learning_rate": 0.0001989812467502549,
-      "loss": 0.4232,
+      "loss": 0.4201,
       "step": 216
     },
     {
       "epoch": 0.47744774477447743,
-      "grad_norm": 0.65234375,
+      "grad_norm": 0.6171875,
       "learning_rate": 0.00019897134896823426,
-      "loss": 0.3747,
+      "loss": 0.3726,
       "step": 217
     },
     {
       "epoch": 0.47964796479647964,
-      "grad_norm": 0.6953125,
+      "grad_norm": 0.66796875,
       "learning_rate": 0.00019896140358557126,
-      "loss": 0.3722,
+      "loss": 0.3733,
       "step": 218
     },
     {
       "epoch": 0.48184818481848185,
-      "grad_norm": 0.625,
+      "grad_norm": 0.58984375,
       "learning_rate": 0.00019895141060704912,
-      "loss": 0.3795,
+      "loss": 0.3723,
       "step": 219
     },
     {
       "epoch": 0.48404840484048406,
-      "grad_norm": 0.64453125,
+      "grad_norm": 0.71875,
       "learning_rate": 0.00019894137003747403,
-      "loss": 0.4071,
+      "loss": 0.4077,
       "step": 220
     },
     {
       "epoch": 0.48624862486248627,
-      "grad_norm": 0.73046875,
+      "grad_norm": 0.828125,
       "learning_rate": 0.00019893128188167505,
-      "loss": 0.4338,
+      "loss": 0.4448,
       "step": 221
     },
     {
       "epoch": 0.4884488448844885,
-      "grad_norm": 0.69921875,
+      "grad_norm": 0.67578125,
       "learning_rate": 0.00019892114614450412,
-      "loss": 0.4499,
+      "loss": 0.4435,
       "step": 222
     },
     {
       "epoch": 0.49064906490649063,
-      "grad_norm": 0.69140625,
+      "grad_norm": 0.6640625,
       "learning_rate": 0.00019891096283083603,
-      "loss": 0.3855,
+      "loss": 0.38,
       "step": 223
     },
     {
       "epoch": 0.49284928492849284,
-      "grad_norm": 0.60546875,
+      "grad_norm": 0.6484375,
       "learning_rate": 0.0001989007319455685,
-      "loss": 0.3656,
+      "loss": 0.364,
       "step": 224
     },
     {
       "epoch": 0.49504950495049505,
-      "grad_norm": 0.75390625,
+      "grad_norm": 0.8125,
       "learning_rate": 0.00019889045349362215,
-      "loss": 0.5003,
+      "loss": 0.498,
       "step": 225
     },
     {
       "epoch": 0.49724972497249725,
-      "grad_norm": 0.9296875,
+      "grad_norm": 0.94921875,
       "learning_rate": 0.00019888012747994033,
-      "loss": 0.5789,
+      "loss": 0.5783,
       "step": 226
     },
     {
       "epoch": 0.49944994499449946,
-      "grad_norm": 0.73828125,
+      "grad_norm": 0.75,
       "learning_rate": 0.0001988697539094895,
-      "loss": 0.3662,
+      "loss": 0.3584,
       "step": 227
     },
     {
       "epoch": 0.5016501650165016,
-      "grad_norm": 0.578125,
+      "grad_norm": 0.60546875,
       "learning_rate": 0.0001988593327872588,
-      "loss": 0.3926,
+      "loss": 0.3998,
       "step": 228
     },
     {
       "epoch": 0.5016501650165016,
-      "eval_loss": 0.4125826060771942,
-      "eval_runtime": 10.5935,
-      "eval_samples_per_second": 36.154,
-      "eval_steps_per_second": 4.531,
+      "eval_loss": 0.41308051347732544,
+      "eval_runtime": 10.1653,
+      "eval_samples_per_second": 37.677,
+      "eval_steps_per_second": 4.722,
       "step": 228
     },
     {
       "epoch": 0.5038503850385039,
-      "grad_norm": 0.62890625,
+      "grad_norm": 0.66796875,
       "learning_rate": 0.00019884886411826035,
-      "loss": 0.3295,
+      "loss": 0.3335,
       "step": 229
     },
     {
       "epoch": 0.506050605060506,
-      "grad_norm": 0.67578125,
+      "grad_norm": 0.6484375,
       "learning_rate": 0.00019883834790752905,
-      "loss": 0.4035,
+      "loss": 0.4043,
       "step": 230
     },
     {
       "epoch": 0.5082508250825083,
-      "grad_norm": 0.6953125,
+      "grad_norm": 0.734375,
       "learning_rate": 0.0001988277841601227,
-      "loss": 0.4319,
+      "loss": 0.4366,
       "step": 231
     },
     {
       "epoch": 0.5104510451045104,
-      "grad_norm": 0.67578125,
+      "grad_norm": 0.66796875,
       "learning_rate": 0.00019881717288112205,
-      "loss": 0.3752,
+      "loss": 0.367,
       "step": 232
     },
     {
       "epoch": 0.5126512651265126,
-      "grad_norm": 0.75,
+      "grad_norm": 0.76953125,
       "learning_rate": 0.00019880651407563056,
-      "loss": 0.4219,
+      "loss": 0.4173,
       "step": 233
     },
     {
       "epoch": 0.5148514851485149,
-      "grad_norm": 0.73828125,
+      "grad_norm": 0.7265625,
       "learning_rate": 0.0001987958077487747,
-      "loss": 0.3324,
+      "loss": 0.332,
       "step": 234
     },
     {
       "epoch": 0.517051705170517,
-      "grad_norm": 0.7890625,
+      "grad_norm": 0.84765625,
       "learning_rate": 0.00019878505390570362,
-      "loss": 0.3717,
+      "loss": 0.3915,
       "step": 235
     },
     {
       "epoch": 0.5192519251925193,
-      "grad_norm": 0.65625,
+      "grad_norm": 0.7265625,
       "learning_rate": 0.00019877425255158948,
-      "loss": 0.3897,
+      "loss": 0.3983,
       "step": 236
     },
     {
       "epoch": 0.5214521452145214,
-      "grad_norm": 0.70703125,
+      "grad_norm": 0.66015625,
       "learning_rate": 0.0001987634036916272,
-      "loss": 0.3024,
+      "loss": 0.3078,
       "step": 237
     },
     {
       "epoch": 0.5236523652365237,
-      "grad_norm": 0.65234375,
+      "grad_norm": 0.62109375,
       "learning_rate": 0.00019875250733103464,
-      "loss": 0.2787,
+      "loss": 0.2798,
       "step": 238
     },
     {
       "epoch": 0.5258525852585259,
-      "grad_norm": 0.63671875,
+      "grad_norm": 0.68359375,
       "learning_rate": 0.00019874156347505242,
-      "loss": 0.3774,
+      "loss": 0.3929,
       "step": 239
     },
     {
       "epoch": 0.528052805280528,
-      "grad_norm": 0.6796875,
+      "grad_norm": 0.61328125,
       "learning_rate": 0.00019873057212894398,
-      "loss": 0.3257,
+      "loss": 0.3255,
       "step": 240
     },
     {
       "epoch": 0.5302530253025303,
-      "grad_norm": 0.765625,
+      "grad_norm": 0.72265625,
       "learning_rate": 0.0001987195332979957,
-      "loss": 0.4009,
+      "loss": 0.3977,
       "step": 241
     },
     {
       "epoch": 0.5324532453245324,
-      "grad_norm": 0.7265625,
+      "grad_norm": 0.69921875,
       "learning_rate": 0.00019870844698751674,
-      "loss": 0.3143,
+      "loss": 0.3224,
       "step": 242
     },
     {
       "epoch": 0.5346534653465347,
-      "grad_norm": 0.76171875,
+      "grad_norm": 0.8359375,
       "learning_rate": 0.00019869731320283908,
-      "loss": 0.3511,
+      "loss": 0.3565,
       "step": 243
     },
     {
       "epoch": 0.5368536853685368,
-      "grad_norm": 0.671875,
+      "grad_norm": 0.64453125,
       "learning_rate": 0.0001986861319493176,
-      "loss": 0.3798,
+      "loss": 0.3873,
       "step": 244
     },
     {
       "epoch": 0.5390539053905391,
-      "grad_norm": 0.63671875,
+      "grad_norm": 0.609375,
       "learning_rate": 0.00019867490323232994,
-      "loss": 0.298,
+      "loss": 0.3043,
       "step": 245
     },
     {
       "epoch": 0.5412541254125413,
-      "grad_norm": 0.67578125,
+      "grad_norm": 0.609375,
       "learning_rate": 0.00019866362705727654,
-      "loss": 0.4185,
+      "loss": 0.4094,
       "step": 246
     },
     {
       "epoch": 0.5434543454345434,
-      "grad_norm": 0.66796875,
+      "grad_norm": 0.6328125,
       "learning_rate": 0.0001986523034295808,
-      "loss": 0.4291,
+      "loss": 0.4295,
       "step": 247
     },
     {
       "epoch": 0.5456545654565457,
-      "grad_norm": 0.62890625,
+      "grad_norm": 0.64453125,
       "learning_rate": 0.00019864093235468885,
-      "loss": 0.2955,
+      "loss": 0.301,
       "step": 248
     },
     {
       "epoch": 0.5478547854785478,
-      "grad_norm": 0.62109375,
+      "grad_norm": 0.64453125,
       "learning_rate": 0.0001986295138380696,
-      "loss": 0.3069,
+      "loss": 0.3091,
       "step": 249
     },
     {
       "epoch": 0.5500550055005501,
-      "grad_norm": 0.62109375,
+      "grad_norm": 0.6640625,
       "learning_rate": 0.00019861804788521493,
-      "loss": 0.2862,
+      "loss": 0.2778,
       "step": 250
     },
     {
       "epoch": 0.5522552255225522,
-      "grad_norm": 1.0234375,
+      "grad_norm": 0.79296875,
       "learning_rate": 0.00019860653450163928,
-      "loss": 0.407,
+      "loss": 0.4264,
       "step": 251
     },
     {
       "epoch": 0.5544554455445545,
-      "grad_norm": 0.75,
+      "grad_norm": 0.76953125,
       "learning_rate": 0.0001985949736928802,
-      "loss": 0.4085,
+      "loss": 0.3949,
       "step": 252
     },
     {
       "epoch": 0.5566556655665567,
-      "grad_norm": 0.6953125,
+      "grad_norm": 0.79296875,
       "learning_rate": 0.0001985833654644978,
-      "loss": 0.3571,
+      "loss": 0.3647,
       "step": 253
     },
     {
       "epoch": 0.5588558855885588,
-      "grad_norm": 0.78125,
+      "grad_norm": 0.7734375,
       "learning_rate": 0.00019857170982207518,
-      "loss": 0.4074,
+      "loss": 0.4127,
       "step": 254
     },
     {
       "epoch": 0.5610561056105611,
-      "grad_norm": 0.83203125,
+      "grad_norm": 0.953125,
       "learning_rate": 0.0001985600067712181,
-      "loss": 0.4818,
+      "loss": 0.4713,
       "step": 255
     },
     {
       "epoch": 0.5632563256325632,
-      "grad_norm": 0.64453125,
+      "grad_norm": 0.6640625,
       "learning_rate": 0.00019854825631755523,
-      "loss": 0.2706,
+      "loss": 0.2503,
       "step": 256
     },
     {
       "epoch": 0.5654565456545655,
-      "grad_norm": 0.578125,
+      "grad_norm": 0.59765625,
       "learning_rate": 0.000198536458466738,
-      "loss": 0.3287,
+      "loss": 0.3269,
       "step": 257
     },
     {
       "epoch": 0.5676567656765676,
-      "grad_norm": 0.61328125,
+      "grad_norm": 0.69140625,
       "learning_rate": 0.00019852461322444057,
-      "loss": 0.3655,
+      "loss": 0.3728,
       "step": 258
     },
     {
       "epoch": 0.5698569856985699,
       "grad_norm": 0.6875,
       "learning_rate": 0.00019851272059636003,
-      "loss": 0.3949,
+      "loss": 0.3783,
       "step": 259
     },
     {
       "epoch": 0.5720572057205721,
-      "grad_norm": 0.7109375,
+      "grad_norm": 0.6796875,
       "learning_rate": 0.00019850078058821614,
-      "loss": 0.4525,
+      "loss": 0.4469,
       "step": 260
     },
     {
       "epoch": 0.5742574257425742,
-      "grad_norm": 0.65234375,
+      "grad_norm": 0.65625,
       "learning_rate": 0.0001984887932057515,
-      "loss": 0.3975,
+      "loss": 0.3931,
       "step": 261
     },
     {
       "epoch": 0.5764576457645765,
-      "grad_norm": 0.73046875,
+      "grad_norm": 0.734375,
       "learning_rate": 0.00019847675845473148,
-      "loss": 0.3758,
+      "loss": 0.3675,
       "step": 262
     },
     {
       "epoch": 0.5786578657865786,
-      "grad_norm": 0.6796875,
+      "grad_norm": 0.734375,
       "learning_rate": 0.00019846467634094424,
-      "loss": 0.3622,
+      "loss": 0.3578,
       "step": 263
     },
     {
       "epoch": 0.5808580858085809,
-      "grad_norm": 0.640625,
+      "grad_norm": 0.66796875,
       "learning_rate": 0.00019845254687020077,
-      "loss": 0.4717,
+      "loss": 0.4702,
       "step": 264
     },
     {
       "epoch": 0.583058305830583,
-      "grad_norm": 0.671875,
+      "grad_norm": 0.76171875,
       "learning_rate": 0.00019844037004833473,
-      "loss": 0.3247,
+      "loss": 0.3208,
       "step": 265
     },
     {
       "epoch": 0.5852585258525853,
-      "grad_norm": 0.65625,
+      "grad_norm": 0.6875,
       "learning_rate": 0.00019842814588120258,
-      "loss": 0.3962,
+      "loss": 0.3941,
       "step": 266
     },
     {
       "epoch": 0.5874587458745875,
-      "grad_norm": 0.64453125,
+      "grad_norm": 0.6171875,
       "learning_rate": 0.00019841587437468362,
-      "loss": 0.3781,
+      "loss": 0.3789,
       "step": 267
     },
     {
       "epoch": 0.5896589658965896,
-      "grad_norm": 0.6796875,
+      "grad_norm": 0.68359375,
       "learning_rate": 0.0001984035555346799,
-      "loss": 0.3619,
+      "loss": 0.3474,
       "step": 268
     },
     {
       "epoch": 0.5918591859185919,
-      "grad_norm": 0.71484375,
+      "grad_norm": 0.625,
       "learning_rate": 0.0001983911893671162,
-      "loss": 0.3188,
+      "loss": 0.3114,
       "step": 269
     },
     {
       "epoch": 0.594059405940594,
-      "grad_norm": 0.70703125,
+      "grad_norm": 0.72265625,
       "learning_rate": 0.00019837877587794002,
-      "loss": 0.3344,
+      "loss": 0.3605,
       "step": 270
     },
     {
       "epoch": 0.5962596259625963,
-      "grad_norm": 0.734375,
+      "grad_norm": 0.6953125,
       "learning_rate": 0.00019836631507312173,
-      "loss": 0.315,
+      "loss": 0.3175,
       "step": 271
     },
     {
       "epoch": 0.5984598459845984,
-      "grad_norm": 0.76953125,
+      "grad_norm": 0.703125,
       "learning_rate": 0.0001983538069586544,
-      "loss": 0.3499,
+      "loss": 0.3372,
       "step": 272
     },
     {
       "epoch": 0.6006600660066007,
-      "grad_norm": 0.6875,
+      "grad_norm": 0.66015625,
       "learning_rate": 0.00019834125154055384,
-      "loss": 0.2922,
+      "loss": 0.2959,
       "step": 273
     },
     {
       "epoch": 0.6028602860286029,
-      "grad_norm": 0.9375,
+      "grad_norm": 0.87890625,
       "learning_rate": 0.00019832864882485862,
-      "loss": 0.3735,
+      "loss": 0.3852,
       "step": 274
     },
     {
       "epoch": 0.605060506050605,
-      "grad_norm": 0.83203125,
+      "grad_norm": 0.9140625,
       "learning_rate": 0.00019831599881763007,
-      "loss": 0.3075,
+      "loss": 0.3386,
       "step": 275
     },
     {
       "epoch": 0.6072607260726073,
-      "grad_norm": 0.71484375,
+      "grad_norm": 0.7578125,
       "learning_rate": 0.00019830330152495227,
-      "loss": 0.2863,
+      "loss": 0.3003,
       "step": 276
     },
     {
       "epoch": 0.6094609460946094,
-      "grad_norm": 0.6875,
+      "grad_norm": 0.7109375,
       "learning_rate": 0.00019829055695293202,
-      "loss": 0.2739,
+      "loss": 0.2803,
       "step": 277
     },
     {
       "epoch": 0.6116611661166117,
-      "grad_norm": 0.72265625,
+      "grad_norm": 0.734375,
       "learning_rate": 0.00019827776510769884,
-      "loss": 0.2876,
+      "loss": 0.3053,
       "step": 278
     },
     {
       "epoch": 0.6138613861386139,
-      "grad_norm": 0.75390625,
+      "grad_norm": 0.859375,
       "learning_rate": 0.00019826492599540508,
-      "loss": 0.3207,
+      "loss": 0.3301,
       "step": 279
     },
     {
       "epoch": 0.6160616061606161,
-      "grad_norm": 0.91796875,
+      "grad_norm": 0.87890625,
       "learning_rate": 0.00019825203962222572,
-      "loss": 0.3156,
+      "loss": 0.3155,
       "step": 280
     },
     {
       "epoch": 0.6182618261826183,
-      "grad_norm": 0.74609375,
+      "grad_norm": 0.6796875,
       "learning_rate": 0.0001982391059943585,
-      "loss": 0.2779,
+      "loss": 0.2761,
       "step": 281
     },
     {
       "epoch": 0.6204620462046204,
-      "grad_norm": 0.6640625,
+      "grad_norm": 0.65234375,
       "learning_rate": 0.00019822612511802393,
-      "loss": 0.2636,
+      "loss": 0.255,
       "step": 282
     },
     {
       "epoch": 0.6226622662266227,
-      "grad_norm": 0.6484375,
+      "grad_norm": 0.66796875,
       "learning_rate": 0.00019821309699946516,
-      "loss": 0.3003,
+      "loss": 0.2894,
       "step": 283
     },
     {
       "epoch": 0.6248624862486248,
-      "grad_norm": 0.65625,
+      "grad_norm": 0.671875,
       "learning_rate": 0.00019820002164494817,
-      "loss": 0.2927,
+      "loss": 0.2906,
       "step": 284
     },
     {
       "epoch": 0.6270627062706271,
-      "grad_norm": 0.74609375,
+      "grad_norm": 0.6953125,
       "learning_rate": 0.00019818689906076156,
-      "loss": 0.2854,
+      "loss": 0.3001,
       "step": 285
     },
     {
       "epoch": 0.6292629262926293,
-      "grad_norm": 0.8359375,
+      "grad_norm": 0.68359375,
       "learning_rate": 0.00019817372925321673,
-      "loss": 0.2833,
+      "loss": 0.2698,
       "step": 286
     },
     {
       "epoch": 0.6314631463146315,
-      "grad_norm": 0.70703125,
+      "grad_norm": 0.66796875,
       "learning_rate": 0.00019816051222864772,
-      "loss": 0.291,
+      "loss": 0.2903,
       "step": 287
     },
     {
       "epoch": 0.6336633663366337,
-      "grad_norm": 0.6953125,
+      "grad_norm": 0.74609375,
       "learning_rate": 0.0001981472479934113,
-      "loss": 0.3544,
+      "loss": 0.3616,
       "step": 288
     },
     {
       "epoch": 0.6358635863586358,
-      "grad_norm": 0.6171875,
+      "grad_norm": 0.6328125,
       "learning_rate": 0.00019813393655388696,
-      "loss": 0.3181,
+      "loss": 0.3193,
       "step": 289
     },
     {
       "epoch": 0.6380638063806381,
-      "grad_norm": 0.87890625,
+      "grad_norm": 0.8046875,
       "learning_rate": 0.00019812057791647686,
-      "loss": 0.4107,
+      "loss": 0.3991,
       "step": 290
     },
     {
       "epoch": 0.6402640264026402,
-      "grad_norm": 0.63671875,
+      "grad_norm": 0.70703125,
       "learning_rate": 0.00019810717208760593,
-      "loss": 0.3295,
+      "loss": 0.3441,
       "step": 291
     },
     {
       "epoch": 0.6424642464246425,
-      "grad_norm": 0.859375,
+      "grad_norm": 0.73828125,
       "learning_rate": 0.00019809371907372178,
-      "loss": 0.3001,
+      "loss": 0.3066,
       "step": 292
     },
     {
       "epoch": 0.6446644664466447,
-      "grad_norm": 0.7578125,
+      "grad_norm": 0.66015625,
       "learning_rate": 0.00019808021888129463,
-      "loss": 0.3079,
+      "loss": 0.2997,
       "step": 293
     },
     {
       "epoch": 0.6468646864686468,
-      "grad_norm": 0.58984375,
+      "grad_norm": 0.55078125,
       "learning_rate": 0.00019806667151681744,
-      "loss": 0.2687,
+      "loss": 0.2648,
       "step": 294
     },
     {
       "epoch": 0.6490649064906491,
-      "grad_norm": 0.75390625,
+      "grad_norm": 0.984375,
       "learning_rate": 0.0001980530769868059,
-      "loss": 0.3901,
+      "loss": 0.4033,
       "step": 295
     },
     {
       "epoch": 0.6512651265126512,
       "grad_norm": 0.70703125,
       "learning_rate": 0.00019803943529779838,
-      "loss": 0.2529,
+      "loss": 0.2437,
       "step": 296
     },
     {
       "epoch": 0.6534653465346535,
-      "grad_norm": 0.6953125,
+      "grad_norm": 0.60546875,
       "learning_rate": 0.00019802574645635582,
-      "loss": 0.2189,
+      "loss": 0.2096,
       "step": 297
     },
     {
       "epoch": 0.6556655665566556,
-      "grad_norm": 0.76171875,
+      "grad_norm": 0.73046875,
       "learning_rate": 0.00019801201046906197,
-      "loss": 0.3206,
+      "loss": 0.3144,
       "step": 298
     },
     {
       "epoch": 0.6578657865786579,
-      "grad_norm": 0.76953125,
+      "grad_norm": 0.7734375,
       "learning_rate": 0.0001979982273425232,
-      "loss": 0.3673,
+      "loss": 0.3588,
       "step": 299
     },
     {
       "epoch": 0.6600660066006601,
-      "grad_norm": 0.703125,
+      "grad_norm": 0.75,
       "learning_rate": 0.00019798439708336858,
-      "loss": 0.319,
+      "loss": 0.3077,
       "step": 300
     },
     {
       "epoch": 0.6622662266226622,
-      "grad_norm": 0.76171875,
+      "grad_norm": 0.7421875,
       "learning_rate": 0.00019797051969824977,
-      "loss": 0.2941,
+      "loss": 0.2865,
       "step": 301
     },
     {
       "epoch": 0.6644664466446645,
-      "grad_norm": 0.65625,
+      "grad_norm": 0.66796875,
       "learning_rate": 0.0001979565951938412,
-      "loss": 0.3089,
+      "loss": 0.3016,
       "step": 302
     },
     {
       "epoch": 0.6666666666666666,
-      "grad_norm": 0.734375,
+      "grad_norm": 0.68359375,
       "learning_rate": 0.00019794262357683987,
-      "loss": 0.275,
+      "loss": 0.2825,
       "step": 303
     },
     {
       "epoch": 0.6688668866886689,
-      "grad_norm": 0.59765625,
+      "grad_norm": 0.6796875,
       "learning_rate": 0.00019792860485396554,
-      "loss": 0.2254,
+      "loss": 0.2217,
       "step": 304
     },
     {
       "epoch": 0.671067106710671,
-      "grad_norm": 0.59765625,
+      "grad_norm": 0.6484375,
       "learning_rate": 0.00019791453903196052,
-      "loss": 0.2382,
+      "loss": 0.2444,
       "step": 305
     },
     {
       "epoch": 0.6732673267326733,
-      "grad_norm": 0.66015625,
+      "grad_norm": 0.72265625,
       "learning_rate": 0.00019790042611758982,
-      "loss": 0.2421,
+      "loss": 0.2436,
       "step": 306
     },
     {
       "epoch": 0.6754675467546755,
-      "grad_norm": 0.70703125,
+      "grad_norm": 0.7109375,
       "learning_rate": 0.0001978862661176411,
-      "loss": 0.2526,
+      "loss": 0.2426,
       "step": 307
     },
     {
       "epoch": 0.6776677667766776,
-      "grad_norm": 1.0390625,
+      "grad_norm": 1.125,
       "learning_rate": 0.0001978720590389247,
-      "loss": 0.3703,
+      "loss": 0.3646,
       "step": 308
     },
     {
       "epoch": 0.6798679867986799,
-      "grad_norm": 0.81640625,
+      "grad_norm": 0.8359375,
       "learning_rate": 0.00019785780488827356,
-      "loss": 0.2743,
+      "loss": 0.2623,
       "step": 309
     },
     {
       "epoch": 0.682068206820682,
       "grad_norm": 0.69140625,
       "learning_rate": 0.00019784350367254322,
-      "loss": 0.2524,
+      "loss": 0.2335,
       "step": 310
     },
     {
       "epoch": 0.6842684268426843,
-      "grad_norm": 1.0234375,
+      "grad_norm": 0.91015625,
       "learning_rate": 0.00019782915539861193,
-      "loss": 0.313,
+      "loss": 0.2921,
       "step": 311
     },
     {
       "epoch": 0.6864686468646864,
-      "grad_norm": 0.7109375,
+      "grad_norm": 0.75,
       "learning_rate": 0.00019781476007338058,
-      "loss": 0.2326,
+      "loss": 0.2334,
       "step": 312
     },
     {
       "epoch": 0.6886688668866887,
-      "grad_norm": 0.53515625,
+      "grad_norm": 0.60546875,
       "learning_rate": 0.00019780031770377262,
-      "loss": 0.1716,
+      "loss": 0.1795,
       "step": 313
     },
     {
       "epoch": 0.6908690869086909,
-      "grad_norm": 0.64453125,
+      "grad_norm": 0.62890625,
       "learning_rate": 0.00019778582829673414,
-      "loss": 0.2357,
+      "loss": 0.2299,
       "step": 314
     },
     {
       "epoch": 0.693069306930693,
-      "grad_norm": 0.74609375,
+      "grad_norm": 0.8828125,
       "learning_rate": 0.00019777129185923395,
-      "loss": 0.294,
+      "loss": 0.301,
       "step": 315
     },
     {
       "epoch": 0.6952695269526953,
-      "grad_norm": 0.59765625,
+      "grad_norm": 0.62890625,
       "learning_rate": 0.00019775670839826336,
-      "loss": 0.2071,
+      "loss": 0.207,
       "step": 316
     },
     {
       "epoch": 0.6974697469746974,
-      "grad_norm": 0.6796875,
+      "grad_norm": 0.6640625,
       "learning_rate": 0.00019774207792083628,
-      "loss": 0.2649,
+      "loss": 0.2565,
       "step": 317
     },
     {
       "epoch": 0.6996699669966997,
       "grad_norm": 0.828125,
       "learning_rate": 0.00019772740043398938,
-      "loss": 0.3169,
+      "loss": 0.3088,
       "step": 318
     },
     {
       "epoch": 0.7018701870187019,
-      "grad_norm": 0.91796875,
+      "grad_norm": 0.92578125,
       "learning_rate": 0.00019771267594478184,
-      "loss": 0.3754,
+      "loss": 0.3798,
       "step": 319
     },
     {
       "epoch": 0.7040704070407041,
-      "grad_norm": 0.7734375,
+      "grad_norm": 0.8046875,
       "learning_rate": 0.00019769790446029544,
-      "loss": 0.3233,
+      "loss": 0.3155,
       "step": 320
     },
     {
       "epoch": 0.7062706270627063,
-      "grad_norm": 0.6171875,
+      "grad_norm": 0.65234375,
       "learning_rate": 0.00019768308598763459,
-      "loss": 0.2271,
+      "loss": 0.224,
       "step": 321
     },
     {
       "epoch": 0.7084708470847084,
-      "grad_norm": 0.78125,
+      "grad_norm": 0.71875,
       "learning_rate": 0.0001976682205339263,
-      "loss": 0.3091,
+      "loss": 0.3031,
       "step": 322
     },
     {
       "epoch": 0.7106710671067107,
-      "grad_norm": 0.64453125,
+      "grad_norm": 0.703125,
       "learning_rate": 0.0001976533081063201,
-      "loss": 0.2162,
+      "loss": 0.2161,
       "step": 323
     },
     {
       "epoch": 0.7128712871287128,
-      "grad_norm": 0.6875,
+      "grad_norm": 0.59375,
       "learning_rate": 0.00019763834871198825,
-      "loss": 0.2329,
+      "loss": 0.2191,
       "step": 324
     },
     {
       "epoch": 0.7150715071507151,
-      "grad_norm": 0.625,
+      "grad_norm": 0.64453125,
       "learning_rate": 0.0001976233423581255,
-      "loss": 0.1917,
+      "loss": 0.1995,
       "step": 325
     },
     {
       "epoch": 0.7172717271727173,
       "grad_norm": 0.53125,
       "learning_rate": 0.00019760828905194923,
-      "loss": 0.1592,
+      "loss": 0.1577,
       "step": 326
     },
     {
       "epoch": 0.7194719471947195,
-      "grad_norm": 0.78515625,
+      "grad_norm": 0.75,
       "learning_rate": 0.00019759318880069936,
-      "loss": 0.269,
+      "loss": 0.2553,
       "step": 327
     },
     {
       "epoch": 0.7216721672167217,
-      "grad_norm": 0.7578125,
+      "grad_norm": 0.734375,
       "learning_rate": 0.00019757804161163844,
-      "loss": 0.2371,
+      "loss": 0.2334,
       "step": 328
     },
     {
       "epoch": 0.7238723872387238,
-      "grad_norm": 0.71484375,
+      "grad_norm": 0.74609375,
       "learning_rate": 0.00019756284749205153,
-      "loss": 0.2163,
+      "loss": 0.212,
       "step": 329
     },
     {
       "epoch": 0.7260726072607261,
-      "grad_norm": 0.796875,
+      "grad_norm": 0.78515625,
       "learning_rate": 0.00019754760644924636,
-      "loss": 0.2614,
+      "loss": 0.2523,
       "step": 330
     },
     {
       "epoch": 0.7282728272827282,
-      "grad_norm": 0.76171875,
+      "grad_norm": 0.82421875,
       "learning_rate": 0.0001975323184905531,
-      "loss": 0.2837,
+      "loss": 0.2987,
       "step": 331
     },
     {
       "epoch": 0.7304730473047305,
-      "grad_norm": 0.78125,
+      "grad_norm": 0.765625,
       "learning_rate": 0.00019751698362332457,
-      "loss": 0.2581,
+      "loss": 0.2524,
       "step": 332
     },
     {
       "epoch": 0.7326732673267327,
-      "grad_norm": 0.5234375,
+      "grad_norm": 0.59375,
       "learning_rate": 0.00019750160185493616,
-      "loss": 0.1676,
+      "loss": 0.1719,
       "step": 333
     },
     {
       "epoch": 0.7348734873487349,
-      "grad_norm": 0.6328125,
+      "grad_norm": 0.64453125,
       "learning_rate": 0.00019748617319278577,
-      "loss": 0.1725,
+      "loss": 0.1779,
       "step": 334
     },
     {
       "epoch": 0.7370737073707371,
-      "grad_norm": 0.80859375,
+      "grad_norm": 0.70703125,
       "learning_rate": 0.00019747069764429383,
-      "loss": 0.2374,
+      "loss": 0.2308,
       "step": 335
     },
     {
       "epoch": 0.7392739273927392,
-      "grad_norm": 0.69921875,
+      "grad_norm": 0.6953125,
       "learning_rate": 0.00019745517521690346,
-      "loss": 0.1675,
+      "loss": 0.1573,
       "step": 336
     },
     {
       "epoch": 0.7414741474147415,
-      "grad_norm": 0.546875,
+      "grad_norm": 0.59765625,
       "learning_rate": 0.00019743960591808016,
-      "loss": 0.1753,
+      "loss": 0.1732,
       "step": 337
     },
     {
       "epoch": 0.7436743674367436,
-      "grad_norm": 0.72265625,
+      "grad_norm": 0.6796875,
       "learning_rate": 0.00019742398975531206,
-      "loss": 0.2295,
+      "loss": 0.226,
       "step": 338
     },
     {
       "epoch": 0.7458745874587459,
-      "grad_norm": 0.5234375,
+      "grad_norm": 0.63671875,
       "learning_rate": 0.0001974083267361098,
-      "loss": 0.162,
+      "loss": 0.1635,
       "step": 339
     },
     {
       "epoch": 0.7480748074807481,
-      "grad_norm": 0.734375,
+      "grad_norm": 0.75390625,
       "learning_rate": 0.0001973926168680066,
-      "loss": 0.1918,
+      "loss": 0.1811,
       "step": 340
     },
     {
       "epoch": 0.7502750275027503,
-      "grad_norm": 0.8046875,
+      "grad_norm": 0.67578125,
       "learning_rate": 0.00019737686015855815,
-      "loss": 0.2746,
+      "loss": 0.2449,
       "step": 341
     },
     {
       "epoch": 0.7524752475247525,
-      "grad_norm": 0.65234375,
+      "grad_norm": 0.6796875,
       "learning_rate": 0.0001973610566153427,
-      "loss": 0.1778,
+      "loss": 0.1741,
       "step": 342
     },
     {
       "epoch": 0.7524752475247525,
-      "eval_loss": 0.2298581600189209,
-      "eval_runtime": 10.3962,
-      "eval_samples_per_second": 36.841,
-      "eval_steps_per_second": 4.617,
+      "eval_loss": 0.2321537882089615,
+      "eval_runtime": 10.267,
+      "eval_samples_per_second": 37.304,
+      "eval_steps_per_second": 4.675,
       "step": 342
     },
     {
       "epoch": 0.7546754675467546,
-      "grad_norm": 0.8125,
+      "grad_norm": 0.734375,
       "learning_rate": 0.00019734520624596104,
-      "loss": 0.2494,
+      "loss": 0.2528,
       "step": 343
     },
     {
       "epoch": 0.7568756875687569,
-      "grad_norm": 0.55859375,
+      "grad_norm": 0.64453125,
       "learning_rate": 0.00019732930905803643,
-      "loss": 0.1559,
+      "loss": 0.1679,
       "step": 344
     },
     {
       "epoch": 0.759075907590759,
-      "grad_norm": 0.66015625,
+      "grad_norm": 0.70703125,
       "learning_rate": 0.00019731336505921475,
-      "loss": 0.207,
+      "loss": 0.1995,
       "step": 345
     },
     {
       "epoch": 0.7612761276127613,
-      "grad_norm": 0.68359375,
+      "grad_norm": 0.7109375,
       "learning_rate": 0.00019729737425716425,
-      "loss": 0.201,
+      "loss": 0.2122,
       "step": 346
     },
     {
       "epoch": 0.7634763476347635,
-      "grad_norm": 0.70703125,
+      "grad_norm": 0.8671875,
       "learning_rate": 0.0001972813366595758,
-      "loss": 0.2001,
+      "loss": 0.1892,
       "step": 347
     },
     {
       "epoch": 0.7656765676567657,
-      "grad_norm": 0.85546875,
+      "grad_norm": 0.71484375,
       "learning_rate": 0.00019726525227416278,
-      "loss": 0.259,
+      "loss": 0.2434,
       "step": 348
     },
     {
       "epoch": 0.7678767876787679,
-      "grad_norm": 0.7890625,
+      "grad_norm": 0.68359375,
       "learning_rate": 0.00019724912110866098,
-      "loss": 0.213,
+      "loss": 0.2119,
       "step": 349
     },
     {
       "epoch": 0.77007700770077,
       "grad_norm": 0.75,
       "learning_rate": 0.00019723294317082878,
-      "loss": 0.2581,
+      "loss": 0.2562,
       "step": 350
     },
     {
       "epoch": 0.7722772277227723,
-      "grad_norm": 0.66015625,
+      "grad_norm": 0.81640625,
       "learning_rate": 0.00019721671846844699,
-      "loss": 0.191,
+      "loss": 0.2086,
       "step": 351
     },
     {
       "epoch": 0.7744774477447744,
-      "grad_norm": 0.76953125,
+      "grad_norm": 0.73046875,
       "learning_rate": 0.00019720044700931893,
-      "loss": 0.2418,
+      "loss": 0.2387,
       "step": 352
     },
     {
       "epoch": 0.7766776677667767,
-      "grad_norm": 0.66015625,
+      "grad_norm": 0.6640625,
       "learning_rate": 0.00019718412880127047,
-      "loss": 0.1799,
+      "loss": 0.1721,
       "step": 353
     },
     {
       "epoch": 0.7788778877887789,
-      "grad_norm": 0.6015625,
+      "grad_norm": 0.59765625,
       "learning_rate": 0.0001971677638521499,
-      "loss": 0.1689,
+      "loss": 0.174,
       "step": 354
     },
     {
       "epoch": 0.7810781078107811,
-      "grad_norm": 0.54296875,
+      "grad_norm": 0.6015625,
       "learning_rate": 0.00019715135216982798,
-      "loss": 0.161,
+      "loss": 0.1687,
       "step": 355
     },
     {
       "epoch": 0.7832783278327833,
-      "grad_norm": 0.671875,
+      "grad_norm": 0.61328125,
       "learning_rate": 0.000197134893762198,
-      "loss": 0.2368,
+      "loss": 0.2047,
       "step": 356
     },
     {
       "epoch": 0.7854785478547854,
-      "grad_norm": 0.77734375,
+      "grad_norm": 0.75,
       "learning_rate": 0.00019711838863717564,
-      "loss": 0.2406,
+      "loss": 0.2452,
       "step": 357
     },
     {
       "epoch": 0.7876787678767877,
-      "grad_norm": 0.5703125,
+      "grad_norm": 0.62109375,
       "learning_rate": 0.00019710183680269916,
-      "loss": 0.1759,
+      "loss": 0.177,
       "step": 358
     },
     {
       "epoch": 0.7898789878987899,
-      "grad_norm": 0.6796875,
+      "grad_norm": 0.71875,
       "learning_rate": 0.0001970852382667292,
-      "loss": 0.168,
+      "loss": 0.1786,
       "step": 359
     },
     {
       "epoch": 0.7920792079207921,
-      "grad_norm": 0.60546875,
+      "grad_norm": 0.65625,
       "learning_rate": 0.0001970685930372489,
-      "loss": 0.1585,
+      "loss": 0.1539,
       "step": 360
     },
     {
       "epoch": 0.7942794279427943,
-      "grad_norm": 0.703125,
+      "grad_norm": 0.76953125,
       "learning_rate": 0.00019705190112226384,
-      "loss": 0.2344,
+      "loss": 0.2409,
       "step": 361
     },
     {
       "epoch": 0.7964796479647965,
-      "grad_norm": 0.765625,
+      "grad_norm": 0.79296875,
       "learning_rate": 0.00019703516252980206,
-      "loss": 0.2145,
+      "loss": 0.2061,
       "step": 362
     },
     {
       "epoch": 0.7986798679867987,
-      "grad_norm": 0.65234375,
+      "grad_norm": 0.69140625,
       "learning_rate": 0.00019701837726791404,
-      "loss": 0.1159,
+      "loss": 0.1197,
       "step": 363
     },
     {
       "epoch": 0.8008800880088008,
-      "grad_norm": 0.7734375,
+      "grad_norm": 0.6953125,
       "learning_rate": 0.00019700154534467272,
-      "loss": 0.1728,
+      "loss": 0.1613,
       "step": 364
     },
     {
       "epoch": 0.8030803080308031,
       "grad_norm": 0.765625,
       "learning_rate": 0.00019698466676817348,
-      "loss": 0.1855,
+      "loss": 0.1813,
       "step": 365
     },
     {
       "epoch": 0.8052805280528053,
-      "grad_norm": 0.9375,
+      "grad_norm": 0.94140625,
       "learning_rate": 0.00019696774154653415,
-      "loss": 0.1967,
+      "loss": 0.2163,
       "step": 366
     },
     {
       "epoch": 0.8074807480748075,
-      "grad_norm": 0.7265625,
+      "grad_norm": 0.79296875,
       "learning_rate": 0.00019695076968789495,
-      "loss": 0.2298,
+      "loss": 0.2358,
       "step": 367
     },
     {
       "epoch": 0.8096809680968097,
-      "grad_norm": 0.71875,
+      "grad_norm": 0.71484375,
       "learning_rate": 0.00019693375120041857,
-      "loss": 0.1722,
+      "loss": 0.1642,
       "step": 368
     },
     {
       "epoch": 0.8118811881188119,
-      "grad_norm": 0.734375,
+      "grad_norm": 0.71875,
       "learning_rate": 0.0001969166860922901,
-      "loss": 0.1879,
+      "loss": 0.1824,
       "step": 369
     },
     {
       "epoch": 0.8140814081408141,
-      "grad_norm": 0.78515625,
+      "grad_norm": 0.78125,
       "learning_rate": 0.0001968995743717171,
-      "loss": 0.2209,
+      "loss": 0.2229,
       "step": 370
     },
     {
       "epoch": 0.8162816281628162,
-      "grad_norm": 0.6875,
+      "grad_norm": 0.63671875,
       "learning_rate": 0.00019688241604692952,
-      "loss": 0.1622,
+      "loss": 0.1636,
       "step": 371
     },
     {
       "epoch": 0.8184818481848185,
-      "grad_norm": 0.703125,
+      "grad_norm": 0.69921875,
       "learning_rate": 0.00019686521112617967,
-      "loss": 0.2404,
+      "loss": 0.2271,
       "step": 372
     },
     {
       "epoch": 0.8206820682068207,
-      "grad_norm": 0.7890625,
+      "grad_norm": 0.859375,
       "learning_rate": 0.00019684795961774239,
-      "loss": 0.2463,
+      "loss": 0.2685,
       "step": 373
     },
     {
       "epoch": 0.8228822882288229,
-      "grad_norm": 0.6796875,
+      "grad_norm": 0.5859375,
       "learning_rate": 0.00019683066152991477,
-      "loss": 0.1968,
+      "loss": 0.1837,
       "step": 374
     },
     {
       "epoch": 0.8250825082508251,
-      "grad_norm": 0.59375,
+      "grad_norm": 0.6640625,
       "learning_rate": 0.0001968133168710165,
-      "loss": 0.1617,
+      "loss": 0.1626,
       "step": 375
     },
     {
       "epoch": 0.8272827282728272,
-      "grad_norm": 0.78515625,
+      "grad_norm": 0.74609375,
       "learning_rate": 0.00019679592564938946,
-      "loss": 0.1902,
+      "loss": 0.1866,
       "step": 376
     },
     {
       "epoch": 0.8294829482948295,
-      "grad_norm": 0.53125,
+      "grad_norm": 0.55859375,
       "learning_rate": 0.00019677848787339809,
-      "loss": 0.1679,
+      "loss": 0.1721,
       "step": 377
     },
     {
       "epoch": 0.8316831683168316,
-      "grad_norm": 0.6953125,
+      "grad_norm": 0.62109375,
       "learning_rate": 0.0001967610035514291,
-      "loss": 0.1717,
+      "loss": 0.1745,
       "step": 378
     },
     {
       "epoch": 0.8338833883388339,
-      "grad_norm": 0.6875,
+      "grad_norm": 0.625,
       "learning_rate": 0.00019674347269189172,
-      "loss": 0.1929,
+      "loss": 0.1851,
       "step": 379
     },
     {
       "epoch": 0.8360836083608361,
-      "grad_norm": 0.609375,
+      "grad_norm": 0.640625,
       "learning_rate": 0.00019672589530321738,
-      "loss": 0.1588,
+      "loss": 0.1648,
       "step": 380
     },
     {
       "epoch": 0.8382838283828383,
-      "grad_norm": 0.65625,
+      "grad_norm": 0.69140625,
       "learning_rate": 0.00019670827139386008,
-      "loss": 0.2087,
+      "loss": 0.2198,
       "step": 381
     },
     {
       "epoch": 0.8404840484048405,
-      "grad_norm": 0.70703125,
+      "grad_norm": 0.76953125,
       "learning_rate": 0.00019669060097229603,
-      "loss": 0.2054,
+      "loss": 0.2043,
       "step": 382
     },
     {
       "epoch": 0.8426842684268426,
-      "grad_norm": 0.859375,
+      "grad_norm": 0.8046875,
       "learning_rate": 0.00019667288404702396,
-      "loss": 0.2153,
+      "loss": 0.221,
       "step": 383
     },
     {
       "epoch": 0.8448844884488449,
-      "grad_norm": 0.640625,
+      "grad_norm": 0.6640625,
       "learning_rate": 0.00019665512062656481,
-      "loss": 0.2234,
+      "loss": 0.2377,
       "step": 384
     },
     {
       "epoch": 0.847084708470847,
-      "grad_norm": 0.6171875,
+      "grad_norm": 0.55078125,
       "learning_rate": 0.00019663731071946206,
-      "loss": 0.1467,
+      "loss": 0.1461,
       "step": 385
     },
     {
       "epoch": 0.8492849284928493,
-      "grad_norm": 0.734375,
+      "grad_norm": 0.6328125,
       "learning_rate": 0.00019661945433428136,
-      "loss": 0.1701,
+      "loss": 0.1662,
       "step": 386
     },
     {
       "epoch": 0.8514851485148515,
-      "grad_norm": 0.609375,
+      "grad_norm": 0.53515625,
       "learning_rate": 0.0001966015514796109,
-      "loss": 0.1538,
+      "loss": 0.1385,
       "step": 387
     },
     {
       "epoch": 0.8536853685368537,
-      "grad_norm": 0.6796875,
+      "grad_norm": 0.59765625,
       "learning_rate": 0.00019658360216406103,
-      "loss": 0.1574,
+      "loss": 0.1528,
       "step": 388
     },
     {
       "epoch": 0.8558855885588559,
-      "grad_norm": 0.80859375,
+      "grad_norm": 0.9375,
       "learning_rate": 0.00019656560639626455,
-      "loss": 0.2132,
+      "loss": 0.2194,
       "step": 389
     },
     {
       "epoch": 0.858085808580858,
-      "grad_norm": 0.6328125,
+      "grad_norm": 0.73828125,
       "learning_rate": 0.00019654756418487667,
-      "loss": 0.1869,
+      "loss": 0.1898,
       "step": 390
     },
     {
       "epoch": 0.8602860286028603,
-      "grad_norm": 0.86328125,
+      "grad_norm": 0.77734375,
       "learning_rate": 0.00019652947553857483,
-      "loss": 0.2361,
+      "loss": 0.2046,
       "step": 391
     },
     {
       "epoch": 0.8624862486248625,
-      "grad_norm": 0.53515625,
+      "grad_norm": 0.71875,
       "learning_rate": 0.0001965113404660588,
-      "loss": 0.1633,
+      "loss": 0.1577,
       "step": 392
     },
     {
       "epoch": 0.8646864686468647,
-      "grad_norm": 0.73828125,
+      "grad_norm": 0.734375,
       "learning_rate": 0.00019649315897605073,
-      "loss": 0.1962,
+      "loss": 0.1822,
       "step": 393
     },
     {
       "epoch": 0.8668866886688669,
-      "grad_norm": 0.8203125,
+      "grad_norm": 0.8125,
       "learning_rate": 0.00019647493107729505,
-      "loss": 0.1893,
+      "loss": 0.1871,
       "step": 394
     },
     {
       "epoch": 0.8690869086908691,
-      "grad_norm": 0.73828125,
+      "grad_norm": 0.640625,
       "learning_rate": 0.00019645665677855859,
-      "loss": 0.1412,
+      "loss": 0.1235,
       "step": 395
     },
     {
       "epoch": 0.8712871287128713,
-      "grad_norm": 0.71875,
+      "grad_norm": 0.71484375,
       "learning_rate": 0.00019643833608863037,
-      "loss": 0.1999,
+      "loss": 0.1978,
       "step": 396
     },
     {
       "epoch": 0.8734873487348734,
-      "grad_norm": 0.578125,
+      "grad_norm": 0.66015625,
       "learning_rate": 0.00019641996901632189,
-      "loss": 0.1562,
+      "loss": 0.173,
       "step": 397
     },
     {
       "epoch": 0.8756875687568757,
-      "grad_norm": 0.7265625,
+      "grad_norm": 0.80078125,
       "learning_rate": 0.0001964015555704668,
-      "loss": 0.2117,
+      "loss": 0.2042,
       "step": 398
     },
     {
       "epoch": 0.8778877887788779,
-      "grad_norm": 0.63671875,
+      "grad_norm": 0.76953125,
       "learning_rate": 0.0001963830957599211,
-      "loss": 0.195,
+      "loss": 0.2085,
       "step": 399
     },
     {
       "epoch": 0.8800880088008801,
-      "grad_norm": 0.703125,
+      "grad_norm": 0.86328125,
       "learning_rate": 0.00019636458959356316,
-      "loss": 0.1638,
+      "loss": 0.1895,
       "step": 400
     },
     {
       "epoch": 0.8822882288228823,
-      "grad_norm": 0.6640625,
+      "grad_norm": 0.6875,
       "learning_rate": 0.00019634603708029355,
-      "loss": 0.1513,
+      "loss": 0.1539,
       "step": 401
     },
     {
       "epoch": 0.8844884488448845,
-      "grad_norm": 0.7421875,
+      "grad_norm": 0.81640625,
       "learning_rate": 0.00019632743822903517,
-      "loss": 0.1911,
+      "loss": 0.1938,
       "step": 402
     },
     {
       "epoch": 0.8866886688668867,
       "grad_norm": 0.74609375,
       "learning_rate": 0.00019630879304873323,
-      "loss": 0.1732,
+      "loss": 0.1681,
       "step": 403
     },
     {
       "epoch": 0.8888888888888888,
-      "grad_norm": 0.671875,
+      "grad_norm": 0.5625,
       "learning_rate": 0.0001962901015483552,
-      "loss": 0.1339,
+      "loss": 0.1305,
       "step": 404
     },
     {
       "epoch": 0.8910891089108911,
-      "grad_norm": 0.7578125,
+      "grad_norm": 0.875,
       "learning_rate": 0.00019627136373689082,
-      "loss": 0.1926,
+      "loss": 0.1963,
       "step": 405
     },
     {
       "epoch": 0.8932893289328933,
-      "grad_norm": 0.72265625,
+      "grad_norm": 0.6796875,
       "learning_rate": 0.00019625257962335209,
-      "loss": 0.1452,
+      "loss": 0.1416,
       "step": 406
     },
     {
       "epoch": 0.8954895489548955,
-      "grad_norm": 0.7890625,
+      "grad_norm": 0.7421875,
       "learning_rate": 0.00019623374921677334,
-      "loss": 0.1681,
+      "loss": 0.1544,
       "step": 407
     },
     {
       "epoch": 0.8976897689768977,
-      "grad_norm": 0.828125,
+      "grad_norm": 0.8046875,
       "learning_rate": 0.0001962148725262111,
-      "loss": 0.1994,
+      "loss": 0.2095,
       "step": 408
     },
     {
       "epoch": 0.8998899889988999,
-      "grad_norm": 0.62890625,
+      "grad_norm": 0.59375,
       "learning_rate": 0.00019619594956074416,
-      "loss": 0.1631,
+      "loss": 0.1624,
       "step": 409
     },
     {
       "epoch": 0.9020902090209021,
-      "grad_norm": 0.7265625,
+      "grad_norm": 0.625,
       "learning_rate": 0.00019617698032947366,
-      "loss": 0.1773,
+      "loss": 0.17,
       "step": 410
     },
     {
       "epoch": 0.9042904290429042,
-      "grad_norm": 0.74609375,
+      "grad_norm": 0.640625,
       "learning_rate": 0.00019615796484152284,
-      "loss": 0.1733,
+      "loss": 0.1476,
       "step": 411
     },
     {
       "epoch": 0.9064906490649065,
-      "grad_norm": 0.625,
+      "grad_norm": 0.62109375,
       "learning_rate": 0.00019613890310603735,
-      "loss": 0.1625,
+      "loss": 0.1607,
       "step": 412
     },
     {
       "epoch": 0.9086908690869087,
-      "grad_norm": 0.58984375,
+      "grad_norm": 0.5546875,
       "learning_rate": 0.00019611979513218492,
-      "loss": 0.1209,
+      "loss": 0.1137,
       "step": 413
     },
     {
       "epoch": 0.9108910891089109,
-      "grad_norm": 0.6484375,
+      "grad_norm": 0.703125,
       "learning_rate": 0.0001961006409291557,
-      "loss": 0.1826,
+      "loss": 0.1838,
       "step": 414
     },
     {
       "epoch": 0.9130913091309131,
-      "grad_norm": 0.65234375,
+      "grad_norm": 0.7890625,
       "learning_rate": 0.0001960814405061619,
-      "loss": 0.1654,
+      "loss": 0.1695,
       "step": 415
     },
     {
       "epoch": 0.9152915291529153,
       "grad_norm": 0.7109375,
       "learning_rate": 0.00019606219387243806,
-      "loss": 0.1685,
+      "loss": 0.1551,
       "step": 416
     },
     {
       "epoch": 0.9174917491749175,
-      "grad_norm": 1.0078125,
+      "grad_norm": 0.470703125,
       "learning_rate": 0.00019604290103724092,
-      "loss": 0.1173,
+      "loss": 0.0806,
       "step": 417
     },
     {
       "epoch": 0.9196919691969196,
-      "grad_norm": 0.703125,
+      "grad_norm": 1.46875,
       "learning_rate": 0.00019602356200984947,
-      "loss": 0.1322,
+      "loss": 0.1182,
       "step": 418
     },
     {
       "epoch": 0.9218921892189219,
-      "grad_norm": 0.53515625,
+      "grad_norm": 0.62890625,
       "learning_rate": 0.00019600417679956485,
-      "loss": 0.1149,
+      "loss": 0.1147,
       "step": 419
     },
     {
       "epoch": 0.9240924092409241,
-      "grad_norm": 0.8125,
+      "grad_norm": 0.8359375,
       "learning_rate": 0.00019598474541571043,
-      "loss": 0.1976,
+      "loss": 0.2119,
       "step": 420
     },
     {
       "epoch": 0.9262926292629263,
-      "grad_norm": 0.65625,
+      "grad_norm": 0.609375,
       "learning_rate": 0.00019596526786763184,
-      "loss": 0.1127,
+      "loss": 0.1034,
       "step": 421
     },
     {
       "epoch": 0.9284928492849285,
-      "grad_norm": 0.5859375,
+      "grad_norm": 0.6171875,
       "learning_rate": 0.0001959457441646969,
-      "loss": 0.1295,
+      "loss": 0.1243,
       "step": 422
     },
     {
       "epoch": 0.9306930693069307,
-      "grad_norm": 0.734375,
+      "grad_norm": 0.74609375,
       "learning_rate": 0.00019592617431629558,
-      "loss": 0.1626,
+      "loss": 0.1688,
       "step": 423
     },
     {
       "epoch": 0.9328932893289329,
-      "grad_norm": 0.72265625,
+      "grad_norm": 0.69140625,
       "learning_rate": 0.00019590655833184008,
-      "loss": 0.1271,
+      "loss": 0.109,
       "step": 424
     },
     {
       "epoch": 0.935093509350935,
-      "grad_norm": 0.60546875,
+      "grad_norm": 0.52734375,
       "learning_rate": 0.00019588689622076477,
-      "loss": 0.101,
+      "loss": 0.0831,
       "step": 425
     },
     {
       "epoch": 0.9372937293729373,
-      "grad_norm": 0.80078125,
+      "grad_norm": 0.828125,
       "learning_rate": 0.00019586718799252624,
-      "loss": 0.144,
+      "loss": 0.1397,
       "step": 426
     },
     {
       "epoch": 0.9394939493949395,
-      "grad_norm": 0.50390625,
+      "grad_norm": 0.54296875,
       "learning_rate": 0.00019584743365660322,
-      "loss": 0.0884,
+      "loss": 0.0853,
       "step": 427
     },
     {
       "epoch": 0.9416941694169417,
-      "grad_norm": 0.65625,
+      "grad_norm": 0.61328125,
       "learning_rate": 0.0001958276332224966,
-      "loss": 0.1142,
+      "loss": 0.1183,
       "step": 428
     },
     {
       "epoch": 0.9438943894389439,
-      "grad_norm": 0.546875,
+      "grad_norm": 0.60546875,
       "learning_rate": 0.00019580778669972958,
-      "loss": 0.0947,
+      "loss": 0.091,
       "step": 429
     },
     {
       "epoch": 0.9460946094609461,
-      "grad_norm": 0.71875,
+      "grad_norm": 0.921875,
       "learning_rate": 0.00019578789409784727,
-      "loss": 0.1275,
+      "loss": 0.1507,
       "step": 430
     },
     {
       "epoch": 0.9482948294829483,
-      "grad_norm": 0.63671875,
+      "grad_norm": 0.78125,
       "learning_rate": 0.00019576795542641722,
-      "loss": 0.1263,
+      "loss": 0.1381,
       "step": 431
     },
     {
       "epoch": 0.9504950495049505,
-      "grad_norm": 0.6484375,
+      "grad_norm": 0.62109375,
       "learning_rate": 0.00019574797069502894,
-      "loss": 0.1451,
+      "loss": 0.1302,
       "step": 432
     },
     {
       "epoch": 0.9526952695269527,
-      "grad_norm": 0.83203125,
+      "grad_norm": 0.60546875,
       "learning_rate": 0.00019572793991329417,
-      "loss": 0.1201,
+      "loss": 0.1052,
       "step": 433
     },
     {
       "epoch": 0.9548954895489549,
-      "grad_norm": 0.70703125,
+      "grad_norm": 0.71875,
       "learning_rate": 0.0001957078630908468,
-      "loss": 0.108,
+      "loss": 0.1154,
       "step": 434
     },
     {
       "epoch": 0.9570957095709571,
-      "grad_norm": 0.8203125,
+      "grad_norm": 0.765625,
       "learning_rate": 0.00019568774023734288,
-      "loss": 0.1679,
+      "loss": 0.1552,
       "step": 435
     },
     {
       "epoch": 0.9592959295929593,
       "grad_norm": 0.478515625,
       "learning_rate": 0.00019566757136246053,
-      "loss": 0.0975,
+      "loss": 0.0918,
       "step": 436
     },
     {
       "epoch": 0.9614961496149615,
-      "grad_norm": 0.83984375,
+      "grad_norm": 0.75390625,
       "learning_rate": 0.00019564735647590003,
-      "loss": 0.1949,
+      "loss": 0.1841,
       "step": 437
     },
     {
       "epoch": 0.9636963696369637,
-      "grad_norm": 0.61328125,
+      "grad_norm": 0.66796875,
       "learning_rate": 0.00019562709558738388,
-      "loss": 0.1232,
+      "loss": 0.1473,
       "step": 438
     },
     {
       "epoch": 0.9658965896589659,
-      "grad_norm": 0.57421875,
+      "grad_norm": 0.640625,
       "learning_rate": 0.00019560678870665657,
-      "loss": 0.1018,
+      "loss": 0.1126,
       "step": 439
     },
     {
       "epoch": 0.9680968096809681,
-      "grad_norm": 0.609375,
+      "grad_norm": 0.62109375,
       "learning_rate": 0.00019558643584348476,
-      "loss": 0.1458,
+      "loss": 0.1483,
       "step": 440
     },
     {
       "epoch": 0.9702970297029703,
-      "grad_norm": 0.69921875,
+      "grad_norm": 0.6328125,
       "learning_rate": 0.0001955660370076573,
-      "loss": 0.1537,
+      "loss": 0.138,
       "step": 441
     },
     {
       "epoch": 0.9724972497249725,
-      "grad_norm": 0.8125,
+      "grad_norm": 0.72265625,
       "learning_rate": 0.00019554559220898504,
-      "loss": 0.1464,
+      "loss": 0.1281,
       "step": 442
     },
     {
       "epoch": 0.9746974697469747,
-      "grad_norm": 0.4609375,
+      "grad_norm": 0.455078125,
       "learning_rate": 0.000195525101457301,
-      "loss": 0.0828,
+      "loss": 0.0786,
       "step": 443
     },
     {
       "epoch": 0.976897689768977,
-      "grad_norm": 0.69921875,
+      "grad_norm": 0.76171875,
       "learning_rate": 0.00019550456476246026,
-      "loss": 0.1187,
+      "loss": 0.1138,
       "step": 444
     },
     {
       "epoch": 0.9790979097909791,
-      "grad_norm": 0.69140625,
+      "grad_norm": 0.7578125,
       "learning_rate": 0.00019548398213434007,
-      "loss": 0.1352,
+      "loss": 0.1348,
       "step": 445
     },
     {
       "epoch": 0.9812981298129813,
-      "grad_norm": 0.5703125,
+      "grad_norm": 0.54296875,
       "learning_rate": 0.00019546335358283968,
-      "loss": 0.1081,
+      "loss": 0.0986,
       "step": 446
     },
     {
       "epoch": 0.9834983498349835,
-      "grad_norm": 0.71875,
+      "grad_norm": 0.7265625,
       "learning_rate": 0.00019544267911788046,
-      "loss": 0.1267,
+      "loss": 0.1218,
       "step": 447
     },
     {
       "epoch": 0.9856985698569857,
-      "grad_norm": 0.640625,
+      "grad_norm": 0.62109375,
       "learning_rate": 0.0001954219587494059,
-      "loss": 0.0838,
+      "loss": 0.0796,
       "step": 448
     },
     {
       "epoch": 0.9878987898789879,
-      "grad_norm": 0.55078125,
+      "grad_norm": 0.703125,
       "learning_rate": 0.00019540119248738152,
-      "loss": 0.1027,
+      "loss": 0.116,
       "step": 449
     },
     {
       "epoch": 0.9900990099009901,
-      "grad_norm": 0.84765625,
+      "grad_norm": 0.5625,
       "learning_rate": 0.00019538038034179497,
-      "loss": 0.1097,
+      "loss": 0.1012,
       "step": 450
     },
     {
       "epoch": 0.9922992299229924,
-      "grad_norm": 0.81640625,
+      "grad_norm": 0.7421875,
       "learning_rate": 0.00019535952232265587,
-      "loss": 0.1199,
+      "loss": 0.1129,
       "step": 451
     },
     {
       "epoch": 0.9944994499449945,
-      "grad_norm": 0.44140625,
+      "grad_norm": 0.55859375,
       "learning_rate": 0.00019533861843999602,
-      "loss": 0.056,
+      "loss": 0.0669,
       "step": 452
     },
     {
       "epoch": 0.9966996699669967,
-      "grad_norm": 0.64453125,
+      "grad_norm": 0.73828125,
       "learning_rate": 0.0001953176687038692,
-      "loss": 0.1256,
+      "loss": 0.1259,
       "step": 453
     },
     {
       "epoch": 0.9988998899889989,
-      "grad_norm": 0.62109375,
+      "grad_norm": 0.69140625,
       "learning_rate": 0.00019529667312435123,
-      "loss": 0.1116,
+      "loss": 0.114,
       "step": 454
     },
     {
       "epoch": 1.001100110011001,
-      "grad_norm": 0.6328125,
+      "grad_norm": 0.640625,
       "learning_rate": 0.00019527563171154006,
-      "loss": 0.0821,
+      "loss": 0.092,
       "step": 455
     },
     {
       "epoch": 1.0033003300330032,
-      "grad_norm": 0.6328125,
+      "grad_norm": 0.9453125,
       "learning_rate": 0.00019525454447555564,
-      "loss": 0.0819,
+      "loss": 0.0913,
       "step": 456
     },
     {
       "epoch": 1.0033003300330032,
-      "eval_loss": 0.12458635121583939,
-      "eval_runtime": 11.1909,
-      "eval_samples_per_second": 34.224,
-      "eval_steps_per_second": 4.289,
+      "eval_loss": 0.1268334686756134,
+      "eval_runtime": 10.0265,
+      "eval_samples_per_second": 38.199,
+      "eval_steps_per_second": 4.787,
       "step": 456
     },
     {
       "epoch": 1.0055005500550056,
-      "grad_norm": 0.578125,
+      "grad_norm": 0.58984375,
       "learning_rate": 0.00019523341142653987,
-      "loss": 0.0722,
+      "loss": 0.0735,
       "step": 457
     },
     {
       "epoch": 1.0077007700770078,
-      "grad_norm": 0.6328125,
+      "grad_norm": 0.5078125,
       "learning_rate": 0.00019521223257465688,
-      "loss": 0.0877,
+      "loss": 0.0779,
       "step": 458
     },
     {
       "epoch": 1.00990099009901,
-      "grad_norm": 0.59765625,
+      "grad_norm": 0.5234375,
       "learning_rate": 0.00019519100793009267,
-      "loss": 0.0778,
+      "loss": 0.0672,
       "step": 459
     },
     {
       "epoch": 1.012101210121012,
-      "grad_norm": 0.53515625,
+      "grad_norm": 0.671875,
       "learning_rate": 0.00019516973750305532,
-      "loss": 0.0712,
+      "loss": 0.0784,
       "step": 460
     },
     {
       "epoch": 1.0143014301430142,
-      "grad_norm": 0.77734375,
+      "grad_norm": 0.6015625,
       "learning_rate": 0.0001951484213037749,
-      "loss": 0.0868,
+      "loss": 0.0772,
       "step": 461
     },
     {
       "epoch": 1.0165016501650166,
-      "grad_norm": 0.671875,
+      "grad_norm": 0.6796875,
       "learning_rate": 0.00019512705934250356,
-      "loss": 0.0914,
+      "loss": 0.1012,
       "step": 462
     },
     {
       "epoch": 1.0187018701870187,
-      "grad_norm": 0.6875,
+      "grad_norm": 0.62109375,
       "learning_rate": 0.00019510565162951537,
-      "loss": 0.1461,
+      "loss": 0.1335,
       "step": 463
     },
     {
       "epoch": 1.020902090209021,
-      "grad_norm": 0.6015625,
+      "grad_norm": 0.58984375,
       "learning_rate": 0.00019508419817510647,
-      "loss": 0.1027,
+      "loss": 0.1006,
       "step": 464
     },
     {
       "epoch": 1.023102310231023,
-      "grad_norm": 0.48828125,
+      "grad_norm": 0.63671875,
       "learning_rate": 0.000195062698989595,
-      "loss": 0.0687,
+      "loss": 0.0653,
       "step": 465
     },
     {
       "epoch": 1.0253025302530252,
-      "grad_norm": 0.51953125,
+      "grad_norm": 0.55859375,
       "learning_rate": 0.00019504115408332102,
-      "loss": 0.0691,
+      "loss": 0.0725,
       "step": 466
     },
     {
       "epoch": 1.0275027502750276,
-      "grad_norm": 0.81640625,
+      "grad_norm": 0.6328125,
       "learning_rate": 0.00019501956346664668,
-      "loss": 0.0953,
+      "loss": 0.0827,
       "step": 467
     },
     {
       "epoch": 1.0297029702970297,
-      "grad_norm": 0.5546875,
+      "grad_norm": 0.55078125,
       "learning_rate": 0.00019499792714995607,
-      "loss": 0.1053,
+      "loss": 0.1033,
       "step": 468
     },
     {
       "epoch": 1.0319031903190319,
-      "grad_norm": 0.64453125,
+      "grad_norm": 0.55078125,
       "learning_rate": 0.0001949762451436552,
-      "loss": 0.095,
+      "loss": 0.0729,
       "step": 469
     },
     {
       "epoch": 1.034103410341034,
-      "grad_norm": 0.48828125,
+      "grad_norm": 0.74609375,
       "learning_rate": 0.00019495451745817218,
-      "loss": 0.0748,
+      "loss": 0.0911,
       "step": 470
     },
     {
       "epoch": 1.0363036303630364,
-      "grad_norm": 0.412109375,
+      "grad_norm": 0.4921875,
       "learning_rate": 0.000194932744103957,
-      "loss": 0.0578,
+      "loss": 0.066,
       "step": 471
     },
     {
       "epoch": 1.0385038503850386,
-      "grad_norm": 0.65625,
+      "grad_norm": 0.59765625,
       "learning_rate": 0.00019491092509148162,
-      "loss": 0.0979,
+      "loss": 0.1045,
       "step": 472
     },
     {
       "epoch": 1.0407040704070407,
-      "grad_norm": 0.62109375,
+      "grad_norm": 0.8203125,
       "learning_rate": 0.00019488906043124002,
-      "loss": 0.0962,
+      "loss": 0.1031,
       "step": 473
     },
     {
       "epoch": 1.0429042904290429,
-      "grad_norm": 0.60546875,
+      "grad_norm": 0.55859375,
       "learning_rate": 0.00019486715013374803,
-      "loss": 0.0825,
+      "loss": 0.0817,
       "step": 474
     },
     {
       "epoch": 1.045104510451045,
-      "grad_norm": 0.6015625,
+      "grad_norm": 0.578125,
       "learning_rate": 0.00019484519420954354,
-      "loss": 0.092,
+      "loss": 0.0926,
       "step": 475
     },
     {
       "epoch": 1.0473047304730474,
-      "grad_norm": 0.5078125,
+      "grad_norm": 0.447265625,
       "learning_rate": 0.0001948231926691863,
-      "loss": 0.0596,
+      "loss": 0.0522,
       "step": 476
     },
     {
       "epoch": 1.0495049504950495,
-      "grad_norm": 0.5234375,
+      "grad_norm": 0.54296875,
       "learning_rate": 0.00019480114552325806,
-      "loss": 0.0761,
+      "loss": 0.0803,
       "step": 477
     },
     {
       "epoch": 1.0517051705170517,
-      "grad_norm": 0.5390625,
+      "grad_norm": 0.5546875,
       "learning_rate": 0.00019477905278236242,
-      "loss": 0.0696,
+      "loss": 0.0735,
       "step": 478
     },
     {
       "epoch": 1.0539053905390539,
-      "grad_norm": 0.52734375,
+      "grad_norm": 0.5859375,
       "learning_rate": 0.00019475691445712507,
-      "loss": 0.0607,
+      "loss": 0.0758,
       "step": 479
     },
     {
       "epoch": 1.056105610561056,
-      "grad_norm": 0.57421875,
+      "grad_norm": 0.67578125,
       "learning_rate": 0.00019473473055819346,
-      "loss": 0.0664,
+      "loss": 0.0854,
       "step": 480
     },
     {
       "epoch": 1.0583058305830584,
-      "grad_norm": 0.609375,
+      "grad_norm": 0.51171875,
       "learning_rate": 0.000194712501096237,
-      "loss": 0.0616,
+      "loss": 0.0545,
       "step": 481
     },
     {
       "epoch": 1.0605060506050605,
-      "grad_norm": 0.6484375,
+      "grad_norm": 0.6640625,
       "learning_rate": 0.00019469022608194708,
-      "loss": 0.0982,
+      "loss": 0.0941,
       "step": 482
     },
     {
       "epoch": 1.0627062706270627,
-      "grad_norm": 0.734375,
+      "grad_norm": 0.51953125,
       "learning_rate": 0.0001946679055260369,
-      "loss": 0.0857,
+      "loss": 0.0754,
       "step": 483
     },
     {
       "epoch": 1.0649064906490648,
-      "grad_norm": 0.51171875,
+      "grad_norm": 0.5703125,
       "learning_rate": 0.00019464553943924164,
-      "loss": 0.0682,
+      "loss": 0.0743,
       "step": 484
     },
     {
       "epoch": 1.0671067106710672,
-      "grad_norm": 0.5625,
+      "grad_norm": 0.59765625,
       "learning_rate": 0.00019462312783231838,
-      "loss": 0.0725,
+      "loss": 0.0759,
       "step": 485
     },
     {
       "epoch": 1.0693069306930694,
-      "grad_norm": 0.52734375,
+      "grad_norm": 0.51171875,
       "learning_rate": 0.00019460067071604603,
-      "loss": 0.0589,
+      "loss": 0.0648,
       "step": 486
     },
     {
       "epoch": 1.0715071507150715,
-      "grad_norm": 0.66015625,
+      "grad_norm": 0.5625,
       "learning_rate": 0.00019457816810122545,
-      "loss": 0.0795,
+      "loss": 0.0705,
       "step": 487
     },
     {
       "epoch": 1.0737073707370737,
-      "grad_norm": 0.57421875,
+      "grad_norm": 0.56640625,
       "learning_rate": 0.00019455561999867933,
-      "loss": 0.0828,
+      "loss": 0.0752,
       "step": 488
     },
     {
       "epoch": 1.0759075907590758,
-      "grad_norm": 0.69921875,
+      "grad_norm": 0.51171875,
       "learning_rate": 0.00019453302641925227,
-      "loss": 0.1052,
+      "loss": 0.0797,
       "step": 489
     },
     {
       "epoch": 1.0781078107810782,
-      "grad_norm": 0.5625,
+      "grad_norm": 0.55859375,
       "learning_rate": 0.00019451038737381077,
-      "loss": 0.0676,
+      "loss": 0.0603,
       "step": 490
     },
     {
       "epoch": 1.0803080308030804,
-      "grad_norm": 0.56640625,
+      "grad_norm": 0.5703125,
       "learning_rate": 0.00019448770287324314,
-      "loss": 0.0906,
+      "loss": 0.0845,
       "step": 491
     },
     {
       "epoch": 1.0825082508250825,
-      "grad_norm": 0.5390625,
+      "grad_norm": 0.5703125,
       "learning_rate": 0.00019446497292845954,
-      "loss": 0.0719,
+      "loss": 0.0818,
       "step": 492
     },
     {
       "epoch": 1.0847084708470847,
-      "grad_norm": 0.62109375,
+      "grad_norm": 0.6640625,
       "learning_rate": 0.0001944421975503921,
-      "loss": 0.0999,
+      "loss": 0.1026,
       "step": 493
     },
     {
       "epoch": 1.0869086908690868,
-      "grad_norm": 0.65625,
+      "grad_norm": 0.6171875,
       "learning_rate": 0.00019441937674999468,
-      "loss": 0.0807,
+      "loss": 0.0848,
       "step": 494
     },
     {
       "epoch": 1.0891089108910892,
-      "grad_norm": 0.43359375,
+      "grad_norm": 0.5078125,
       "learning_rate": 0.00019439651053824303,
-      "loss": 0.0492,
+      "loss": 0.052,
       "step": 495
     },
     {
       "epoch": 1.0913091309130913,
-      "grad_norm": 0.65234375,
+      "grad_norm": 0.69921875,
       "learning_rate": 0.00019437359892613472,
-      "loss": 0.0871,
+      "loss": 0.0754,
       "step": 496
     },
     {
       "epoch": 1.0935093509350935,
-      "grad_norm": 0.486328125,
+      "grad_norm": 0.6484375,
       "learning_rate": 0.00019435064192468922,
-      "loss": 0.0669,
+      "loss": 0.0814,
       "step": 497
     },
     {
       "epoch": 1.0957095709570956,
-      "grad_norm": 0.62109375,
+      "grad_norm": 0.53515625,
       "learning_rate": 0.00019432763954494776,
-      "loss": 0.115,
+      "loss": 0.1052,
       "step": 498
     },
     {
       "epoch": 1.0979097909790978,
-      "grad_norm": 0.78515625,
+      "grad_norm": 0.78125,
       "learning_rate": 0.00019430459179797343,
-      "loss": 0.1089,
+      "loss": 0.1152,
       "step": 499
     },
     {
       "epoch": 1.1001100110011002,
-      "grad_norm": 0.55078125,
+      "grad_norm": 0.578125,
       "learning_rate": 0.00019428149869485113,
-      "loss": 0.0751,
+      "loss": 0.0772,
       "step": 500
     },
     {
       "epoch": 1.1023102310231023,
-      "grad_norm": 0.5625,
+      "grad_norm": 0.50390625,
       "learning_rate": 0.00019425836024668757,
-      "loss": 0.0693,
+      "loss": 0.0697,
       "step": 501
     },
     {
       "epoch": 1.1045104510451045,
-      "grad_norm": 0.578125,
+      "grad_norm": 0.52734375,
       "learning_rate": 0.00019423517646461127,
-      "loss": 0.104,
+      "loss": 0.0892,
       "step": 502
     },
     {
       "epoch": 1.1067106710671066,
-      "grad_norm": 0.478515625,
+      "grad_norm": 0.53125,
       "learning_rate": 0.00019421194735977254,
-      "loss": 0.0486,
+      "loss": 0.0538,
       "step": 503
     },
     {
       "epoch": 1.108910891089109,
-      "grad_norm": 0.5546875,
+      "grad_norm": 0.640625,
       "learning_rate": 0.00019418867294334355,
-      "loss": 0.082,
+      "loss": 0.0892,
       "step": 504
     },
     {
       "epoch": 1.1111111111111112,
-      "grad_norm": 0.5,
+      "grad_norm": 0.4375,
       "learning_rate": 0.00019416535322651818,
-      "loss": 0.0671,
+      "loss": 0.061,
       "step": 505
     },
     {
       "epoch": 1.1133113311331133,
-      "grad_norm": 0.53125,
+      "grad_norm": 0.66015625,
       "learning_rate": 0.00019414198822051217,
-      "loss": 0.0753,
+      "loss": 0.0846,
       "step": 506
     },
     {
       "epoch": 1.1155115511551155,
-      "grad_norm": 0.5,
+      "grad_norm": 0.5546875,
       "learning_rate": 0.000194118577936563,
-      "loss": 0.0628,
+      "loss": 0.0668,
       "step": 507
     },
     {
       "epoch": 1.1177117711771176,
-      "grad_norm": 0.55078125,
+      "grad_norm": 0.62890625,
       "learning_rate": 0.0001940951223859299,
-      "loss": 0.0917,
+      "loss": 0.0828,
       "step": 508
     },
     {
       "epoch": 1.11991199119912,
-      "grad_norm": 0.53515625,
+      "grad_norm": 0.6953125,
       "learning_rate": 0.00019407162157989393,
-      "loss": 0.0822,
+      "loss": 0.0871,
       "step": 509
     },
     {
       "epoch": 1.1221122112211221,
-      "grad_norm": 0.55078125,
+      "grad_norm": 0.53125,
       "learning_rate": 0.0001940480755297579,
-      "loss": 0.07,
+      "loss": 0.0687,
       "step": 510
     },
     {
       "epoch": 1.1243124312431243,
-      "grad_norm": 0.5234375,
+      "grad_norm": 0.5546875,
       "learning_rate": 0.00019402448424684634,
-      "loss": 0.0873,
+      "loss": 0.0911,
       "step": 511
     },
     {
       "epoch": 1.1265126512651265,
-      "grad_norm": 0.546875,
+      "grad_norm": 0.5234375,
       "learning_rate": 0.00019400084774250562,
-      "loss": 0.0955,
+      "loss": 0.0953,
       "step": 512
     },
     {
       "epoch": 1.1287128712871288,
-      "grad_norm": 0.484375,
+      "grad_norm": 0.412109375,
       "learning_rate": 0.00019397716602810374,
-      "loss": 0.0586,
+      "loss": 0.0508,
       "step": 513
     },
     {
       "epoch": 1.130913091309131,
-      "grad_norm": 0.4609375,
+      "grad_norm": 0.50390625,
       "learning_rate": 0.00019395343911503057,
-      "loss": 0.0562,
+      "loss": 0.0581,
       "step": 514
     },
     {
       "epoch": 1.1331133113311331,
-      "grad_norm": 0.87109375,
+      "grad_norm": 0.77734375,
       "learning_rate": 0.0001939296670146976,
-      "loss": 0.1075,
+      "loss": 0.1191,
       "step": 515
     },
     {
       "epoch": 1.1353135313531353,
-      "grad_norm": 0.400390625,
+      "grad_norm": 0.54296875,
       "learning_rate": 0.00019390584973853815,
-      "loss": 0.0493,
+      "loss": 0.07,
       "step": 516
     },
     {
       "epoch": 1.1375137513751374,
-      "grad_norm": 0.7109375,
+      "grad_norm": 0.5546875,
       "learning_rate": 0.0001938819872980072,
-      "loss": 0.0789,
+      "loss": 0.0642,
       "step": 517
     },
     {
       "epoch": 1.1397139713971396,
-      "grad_norm": 0.68359375,
+      "grad_norm": 0.5234375,
       "learning_rate": 0.00019385807970458154,
-      "loss": 0.0767,
+      "loss": 0.0693,
       "step": 518
     },
     {
       "epoch": 1.141914191419142,
-      "grad_norm": 0.5625,
+      "grad_norm": 0.62109375,
       "learning_rate": 0.00019383412696975956,
-      "loss": 0.0678,
+      "loss": 0.0566,
       "step": 519
     },
     {
       "epoch": 1.1441144114411441,
-      "grad_norm": 0.61328125,
+      "grad_norm": 0.53125,
       "learning_rate": 0.00019381012910506146,
-      "loss": 0.0684,
+      "loss": 0.0655,
       "step": 520
     },
     {
       "epoch": 1.1463146314631463,
-      "grad_norm": 0.4375,
+      "grad_norm": 0.36328125,
       "learning_rate": 0.00019378608612202901,
-      "loss": 0.0438,
+      "loss": 0.0434,
       "step": 521
     },
     {
       "epoch": 1.1485148514851484,
-      "grad_norm": 0.5234375,
+      "grad_norm": 0.4453125,
       "learning_rate": 0.00019376199803222588,
-      "loss": 0.0707,
+      "loss": 0.0581,
       "step": 522
     },
     {
       "epoch": 1.1507150715071508,
-      "grad_norm": 0.93359375,
+      "grad_norm": 0.6953125,
       "learning_rate": 0.00019373786484723731,
-      "loss": 0.0964,
+      "loss": 0.0977,
       "step": 523
     },
     {
       "epoch": 1.152915291529153,
-      "grad_norm": 0.5390625,
+      "grad_norm": 0.5859375,
       "learning_rate": 0.0001937136865786702,
-      "loss": 0.0816,
+      "loss": 0.0754,
       "step": 524
     },
     {
       "epoch": 1.155115511551155,
-      "grad_norm": 0.59765625,
+      "grad_norm": 0.5078125,
       "learning_rate": 0.0001936894632381532,
-      "loss": 0.0724,
+      "loss": 0.0557,
       "step": 525
     },
     {
       "epoch": 1.1573157315731573,
-      "grad_norm": 0.6484375,
+      "grad_norm": 0.59765625,
       "learning_rate": 0.00019366519483733662,
-      "loss": 0.0818,
+      "loss": 0.0657,
       "step": 526
     },
     {
       "epoch": 1.1595159515951594,
-      "grad_norm": 0.703125,
+      "grad_norm": 0.671875,
       "learning_rate": 0.00019364088138789244,
-      "loss": 0.0692,
+      "loss": 0.0663,
       "step": 527
     },
     {
       "epoch": 1.1617161716171618,
-      "grad_norm": 0.5390625,
+      "grad_norm": 0.65234375,
       "learning_rate": 0.00019361652290151432,
-      "loss": 0.0642,
+      "loss": 0.0799,
       "step": 528
     },
     {
       "epoch": 1.163916391639164,
-      "grad_norm": 0.6484375,
+      "grad_norm": 0.58203125,
       "learning_rate": 0.00019359211938991755,
-      "loss": 0.0854,
+      "loss": 0.0828,
       "step": 529
     },
     {
       "epoch": 1.166116611661166,
-      "grad_norm": 0.55859375,
+      "grad_norm": 0.58984375,
       "learning_rate": 0.00019356767086483906,
-      "loss": 0.074,
+      "loss": 0.0804,
       "step": 530
     },
     {
       "epoch": 1.1683168316831682,
-      "grad_norm": 0.486328125,
+      "grad_norm": 0.482421875,
       "learning_rate": 0.00019354317733803752,
-      "loss": 0.0833,
+      "loss": 0.0619,
       "step": 531
     },
     {
       "epoch": 1.1705170517051706,
-      "grad_norm": 0.79296875,
+      "grad_norm": 0.5625,
       "learning_rate": 0.00019351863882129312,
-      "loss": 0.0897,
+      "loss": 0.064,
       "step": 532
     },
     {
       "epoch": 1.1727172717271728,
-      "grad_norm": 0.62109375,
+      "grad_norm": 0.671875,
       "learning_rate": 0.00019349405532640782,
-      "loss": 0.093,
+      "loss": 0.0901,
       "step": 533
     },
     {
       "epoch": 1.174917491749175,
-      "grad_norm": 0.5078125,
+      "grad_norm": 0.451171875,
       "learning_rate": 0.0001934694268652051,
-      "loss": 0.0643,
+      "loss": 0.0539,
       "step": 534
     },
     {
       "epoch": 1.177117711771177,
-      "grad_norm": 0.462890625,
+      "grad_norm": 0.5,
       "learning_rate": 0.00019344475344953012,
-      "loss": 0.0595,
+      "loss": 0.0543,
       "step": 535
     },
     {
       "epoch": 1.1793179317931792,
-      "grad_norm": 0.546875,
+      "grad_norm": 0.5703125,
       "learning_rate": 0.00019342003509124965,
-      "loss": 0.0966,
+      "loss": 0.0997,
       "step": 536
     },
     {
       "epoch": 1.1815181518151816,
-      "grad_norm": 0.361328125,
+      "grad_norm": 0.458984375,
       "learning_rate": 0.00019339527180225208,
-      "loss": 0.0426,
+      "loss": 0.0565,
       "step": 537
     },
     {
       "epoch": 1.1837183718371838,
-      "grad_norm": 0.65625,
+      "grad_norm": 0.796875,
       "learning_rate": 0.00019337046359444742,
-      "loss": 0.0925,
+      "loss": 0.1032,
       "step": 538
     },
     {
       "epoch": 1.185918591859186,
-      "grad_norm": 0.546875,
+      "grad_norm": 0.5625,
       "learning_rate": 0.00019334561047976723,
-      "loss": 0.0527,
+      "loss": 0.059,
       "step": 539
     },
     {
       "epoch": 1.188118811881188,
-      "grad_norm": 0.64453125,
+      "grad_norm": 0.4375,
       "learning_rate": 0.00019332071247016476,
-      "loss": 0.0458,
+      "loss": 0.0492,
       "step": 540
     },
     {
       "epoch": 1.1903190319031904,
-      "grad_norm": 0.51953125,
+      "grad_norm": 0.498046875,
       "learning_rate": 0.00019329576957761475,
-      "loss": 0.0618,
+      "loss": 0.0597,
       "step": 541
     },
     {
       "epoch": 1.1925192519251926,
-      "grad_norm": 0.5625,
+      "grad_norm": 0.59375,
       "learning_rate": 0.00019327078181411364,
-      "loss": 0.089,
+      "loss": 0.095,
       "step": 542
     },
     {
       "epoch": 1.1947194719471947,
-      "grad_norm": 0.59375,
+      "grad_norm": 0.421875,
       "learning_rate": 0.00019324574919167935,
-      "loss": 0.0751,
+      "loss": 0.0642,
       "step": 543
     },
     {
       "epoch": 1.196919691969197,
-      "grad_norm": 0.5234375,
+      "grad_norm": 0.625,
       "learning_rate": 0.00019322067172235138,
-      "loss": 0.0747,
+      "loss": 0.0887,
       "step": 544
     },
     {
       "epoch": 1.199119911991199,
-      "grad_norm": 0.62890625,
+      "grad_norm": 0.59765625,
       "learning_rate": 0.0001931955494181909,
-      "loss": 0.0534,
+      "loss": 0.0694,
       "step": 545
     },
     {
       "epoch": 1.2013201320132012,
-      "grad_norm": 0.50390625,
+      "grad_norm": 0.54296875,
       "learning_rate": 0.00019317038229128058,
-      "loss": 0.0655,
+      "loss": 0.0632,
       "step": 546
     },
     {
       "epoch": 1.2035203520352036,
-      "grad_norm": 0.431640625,
+      "grad_norm": 0.64453125,
       "learning_rate": 0.0001931451703537246,
-      "loss": 0.0567,
+      "loss": 0.073,
       "step": 547
     },
     {
       "epoch": 1.2057205720572057,
-      "grad_norm": 0.515625,
+      "grad_norm": 0.443359375,
       "learning_rate": 0.00019311991361764872,
-      "loss": 0.052,
+      "loss": 0.0532,
       "step": 548
     },
     {
       "epoch": 1.2079207920792079,
-      "grad_norm": 0.59375,
+      "grad_norm": 0.5859375,
       "learning_rate": 0.0001930946120952003,
-      "loss": 0.0844,
+      "loss": 0.0815,
       "step": 549
     },
     {
       "epoch": 1.21012101210121,
-      "grad_norm": 0.345703125,
+      "grad_norm": 0.3125,
       "learning_rate": 0.00019306926579854821,
-      "loss": 0.0289,
+      "loss": 0.026,
       "step": 550
     },
     {
       "epoch": 1.2123212321232124,
-      "grad_norm": 0.671875,
+      "grad_norm": 0.5078125,
       "learning_rate": 0.00019304387473988282,
-      "loss": 0.06,
+      "loss": 0.0646,
       "step": 551
     },
     {
       "epoch": 1.2145214521452146,
-      "grad_norm": 0.765625,
+      "grad_norm": 0.62109375,
       "learning_rate": 0.00019301843893141608,
-      "loss": 0.0682,
+      "loss": 0.0588,
       "step": 552
     },
     {
       "epoch": 1.2167216721672167,
-      "grad_norm": 0.5234375,
+      "grad_norm": 0.625,
       "learning_rate": 0.00019299295838538138,
-      "loss": 0.0647,
+      "loss": 0.0665,
       "step": 553
     },
     {
       "epoch": 1.2189218921892189,
-      "grad_norm": 0.52734375,
+      "grad_norm": 0.53125,
       "learning_rate": 0.00019296743311403376,
-      "loss": 0.0587,
+      "loss": 0.0554,
       "step": 554
     },
     {
       "epoch": 1.221122112211221,
-      "grad_norm": 0.4765625,
+      "grad_norm": 0.49609375,
       "learning_rate": 0.00019294186312964967,
-      "loss": 0.0642,
+      "loss": 0.0499,
       "step": 555
     },
     {
       "epoch": 1.2233223322332234,
-      "grad_norm": 0.64453125,
+      "grad_norm": 0.66796875,
       "learning_rate": 0.00019291624844452706,
-      "loss": 0.0791,
+      "loss": 0.0906,
       "step": 556
     },
     {
       "epoch": 1.2255225522552256,
-      "grad_norm": 0.70703125,
+      "grad_norm": 0.50390625,
       "learning_rate": 0.00019289058907098544,
-      "loss": 0.057,
+      "loss": 0.0528,
       "step": 557
     },
     {
       "epoch": 1.2277227722772277,
-      "grad_norm": 0.5,
+      "grad_norm": 0.482421875,
       "learning_rate": 0.00019286488502136577,
-      "loss": 0.0643,
+      "loss": 0.07,
       "step": 558
     },
     {
       "epoch": 1.2299229922992299,
-      "grad_norm": 0.55859375,
+      "grad_norm": 0.64453125,
       "learning_rate": 0.00019283913630803055,
-      "loss": 0.075,
+      "loss": 0.0819,
       "step": 559
     },
     {
       "epoch": 1.2321232123212322,
-      "grad_norm": 0.5625,
+      "grad_norm": 0.46875,
       "learning_rate": 0.00019281334294336363,
-      "loss": 0.0728,
+      "loss": 0.0631,
       "step": 560
     },
     {
       "epoch": 1.2343234323432344,
-      "grad_norm": 0.42578125,
+      "grad_norm": 0.63671875,
       "learning_rate": 0.0001927875049397705,
-      "loss": 0.0616,
+      "loss": 0.072,
       "step": 561
     },
     {
       "epoch": 1.2365236523652365,
-      "grad_norm": 0.63671875,
+      "grad_norm": 0.484375,
       "learning_rate": 0.00019276162230967803,
-      "loss": 0.0668,
+      "loss": 0.0714,
       "step": 562
     },
     {
       "epoch": 1.2387238723872387,
-      "grad_norm": 0.390625,
+      "grad_norm": 0.41796875,
       "learning_rate": 0.00019273569506553456,
-      "loss": 0.0349,
+      "loss": 0.0364,
       "step": 563
     },
     {
       "epoch": 1.2409240924092408,
-      "grad_norm": 0.4609375,
+      "grad_norm": 0.400390625,
       "learning_rate": 0.00019270972321980991,
-      "loss": 0.0441,
+      "loss": 0.0394,
       "step": 564
     },
     {
       "epoch": 1.2431243124312432,
-      "grad_norm": 0.470703125,
+      "grad_norm": 0.4765625,
       "learning_rate": 0.00019268370678499533,
-      "loss": 0.0556,
+      "loss": 0.0547,
       "step": 565
     },
     {
       "epoch": 1.2453245324532454,
-      "grad_norm": 0.62890625,
+      "grad_norm": 0.5,
       "learning_rate": 0.00019265764577360348,
-      "loss": 0.0726,
+      "loss": 0.0628,
       "step": 566
     },
     {
       "epoch": 1.2475247524752475,
-      "grad_norm": 0.365234375,
+      "grad_norm": 0.439453125,
       "learning_rate": 0.00019263154019816855,
-      "loss": 0.046,
+      "loss": 0.0474,
       "step": 567
     },
     {
       "epoch": 1.2497249724972497,
-      "grad_norm": 0.7265625,
+      "grad_norm": 0.79296875,
       "learning_rate": 0.00019260539007124613,
-      "loss": 0.1122,
+      "loss": 0.12,
       "step": 568
     },
     {
       "epoch": 1.251925192519252,
-      "grad_norm": 0.51171875,
+      "grad_norm": 0.51953125,
       "learning_rate": 0.0001925791954054132,
-      "loss": 0.0624,
+      "loss": 0.0635,
       "step": 569
     },
     {
       "epoch": 1.2541254125412542,
-      "grad_norm": 0.482421875,
+      "grad_norm": 0.51953125,
       "learning_rate": 0.00019255295621326815,
-      "loss": 0.0559,
+      "loss": 0.0679,
       "step": 570
     },
     {
       "epoch": 1.2541254125412542,
-      "eval_loss": 0.08770393580198288,
-      "eval_runtime": 10.3627,
-      "eval_samples_per_second": 36.959,
-      "eval_steps_per_second": 4.632,
+      "eval_loss": 0.08089210838079453,
+      "eval_runtime": 10.1638,
+      "eval_samples_per_second": 37.683,
+      "eval_steps_per_second": 4.723,
       "step": 570
     },
     {
       "epoch": 1.2563256325632564,
-      "grad_norm": 0.5703125,
+      "grad_norm": 0.474609375,
       "learning_rate": 0.00019252667250743084,
-      "loss": 0.0801,
+      "loss": 0.0617,
       "step": 571
     },
     {
       "epoch": 1.2585258525852585,
-      "grad_norm": 0.65625,
+      "grad_norm": 0.64453125,
       "learning_rate": 0.00019250034430054255,
-      "loss": 0.0831,
+      "loss": 0.0734,
       "step": 572
     },
     {
       "epoch": 1.2607260726072607,
-      "grad_norm": 0.53125,
+      "grad_norm": 0.4375,
       "learning_rate": 0.0001924739716052659,
-      "loss": 0.0598,
+      "loss": 0.0551,
       "step": 573
     },
     {
       "epoch": 1.2629262926292628,
-      "grad_norm": 0.427734375,
+      "grad_norm": 0.404296875,
       "learning_rate": 0.00019244755443428494,
-      "loss": 0.0451,
+      "loss": 0.0431,
       "step": 574
     },
     {
       "epoch": 1.2651265126512652,
-      "grad_norm": 0.68359375,
+      "grad_norm": 0.46875,
       "learning_rate": 0.00019242109280030509,
-      "loss": 0.0733,
+      "loss": 0.0612,
       "step": 575
     },
     {
       "epoch": 1.2673267326732673,
-      "grad_norm": 0.40625,
+      "grad_norm": 0.3984375,
       "learning_rate": 0.0001923945867160532,
-      "loss": 0.0398,
+      "loss": 0.0463,
       "step": 576
     },
     {
       "epoch": 1.2695269526952695,
-      "grad_norm": 0.80078125,
+      "grad_norm": 0.74609375,
       "learning_rate": 0.00019236803619427745,
-      "loss": 0.0722,
+      "loss": 0.0796,
       "step": 577
     },
     {
       "epoch": 1.2717271727172716,
-      "grad_norm": 0.65234375,
+      "grad_norm": 0.58984375,
       "learning_rate": 0.00019234144124774747,
-      "loss": 0.0819,
+      "loss": 0.0659,
       "step": 578
     },
     {
       "epoch": 1.273927392739274,
-      "grad_norm": 0.5859375,
+      "grad_norm": 0.54296875,
       "learning_rate": 0.00019231480188925412,
-      "loss": 0.0706,
+      "loss": 0.0631,
       "step": 579
     },
     {
       "epoch": 1.2761276127612762,
-      "grad_norm": 0.46875,
+      "grad_norm": 0.439453125,
       "learning_rate": 0.0001922881181316097,
-      "loss": 0.0436,
+      "loss": 0.0458,
       "step": 580
     },
     {
       "epoch": 1.2783278327832783,
-      "grad_norm": 0.462890625,
+      "grad_norm": 0.57421875,
       "learning_rate": 0.00019226138998764796,
-      "loss": 0.0469,
+      "loss": 0.0635,
       "step": 581
     },
     {
       "epoch": 1.2805280528052805,
-      "grad_norm": 0.59765625,
+      "grad_norm": 0.5625,
       "learning_rate": 0.00019223461747022383,
-      "loss": 0.0625,
+      "loss": 0.0526,
       "step": 582
     },
     {
       "epoch": 1.2827282728272826,
-      "grad_norm": 0.494140625,
+      "grad_norm": 0.671875,
       "learning_rate": 0.00019220780059221363,
-      "loss": 0.0421,
+      "loss": 0.0578,
       "step": 583
     },
     {
       "epoch": 1.284928492849285,
-      "grad_norm": 0.498046875,
+      "grad_norm": 0.53125,
       "learning_rate": 0.0001921809393665151,
-      "loss": 0.0576,
+      "loss": 0.0563,
       "step": 584
     },
     {
       "epoch": 1.2871287128712872,
-      "grad_norm": 0.671875,
+      "grad_norm": 0.84765625,
       "learning_rate": 0.0001921540338060472,
-      "loss": 0.0759,
+      "loss": 0.0659,
       "step": 585
     },
     {
       "epoch": 1.2893289328932893,
-      "grad_norm": 0.3828125,
+      "grad_norm": 0.36328125,
       "learning_rate": 0.00019212708392375027,
-      "loss": 0.0363,
+      "loss": 0.0331,
       "step": 586
     },
     {
       "epoch": 1.2915291529152915,
-      "grad_norm": 0.52734375,
+      "grad_norm": 0.80859375,
       "learning_rate": 0.00019210008973258598,
-      "loss": 0.0629,
+      "loss": 0.0668,
       "step": 587
     },
     {
       "epoch": 1.2937293729372938,
-      "grad_norm": 0.498046875,
+      "grad_norm": 0.59375,
       "learning_rate": 0.00019207305124553726,
-      "loss": 0.0615,
+      "loss": 0.0633,
       "step": 588
     },
     {
       "epoch": 1.295929592959296,
-      "grad_norm": 0.5625,
+      "grad_norm": 0.51953125,
       "learning_rate": 0.0001920459684756084,
-      "loss": 0.069,
+      "loss": 0.0556,
       "step": 589
     },
     {
       "epoch": 1.2981298129812981,
-      "grad_norm": 0.52734375,
+      "grad_norm": 0.33984375,
       "learning_rate": 0.00019201884143582495,
-      "loss": 0.0664,
+      "loss": 0.0372,
       "step": 590
     },
     {
       "epoch": 1.3003300330033003,
-      "grad_norm": 0.578125,
+      "grad_norm": 0.5390625,
       "learning_rate": 0.00019199167013923374,
-      "loss": 0.0618,
+      "loss": 0.0557,
       "step": 591
     },
     {
       "epoch": 1.3025302530253025,
-      "grad_norm": 0.56640625,
+      "grad_norm": 0.63671875,
       "learning_rate": 0.00019196445459890294,
-      "loss": 0.0645,
+      "loss": 0.0786,
       "step": 592
     },
     {
       "epoch": 1.3047304730473046,
-      "grad_norm": 0.609375,
+      "grad_norm": 0.66015625,
       "learning_rate": 0.0001919371948279219,
-      "loss": 0.0733,
+      "loss": 0.0621,
       "step": 593
     },
     {
       "epoch": 1.306930693069307,
-      "grad_norm": 0.7265625,
+      "grad_norm": 0.57421875,
       "learning_rate": 0.00019190989083940144,
-      "loss": 0.0897,
+      "loss": 0.0835,
       "step": 594
     },
     {
       "epoch": 1.3091309130913091,
-      "grad_norm": 0.455078125,
+      "grad_norm": 0.5,
       "learning_rate": 0.00019188254264647337,
-      "loss": 0.0405,
+      "loss": 0.038,
       "step": 595
     },
     {
       "epoch": 1.3113311331133113,
-      "grad_norm": 0.427734375,
+      "grad_norm": 0.70703125,
       "learning_rate": 0.00019185515026229103,
-      "loss": 0.0718,
+      "loss": 0.0915,
       "step": 596
     },
     {
       "epoch": 1.3135313531353137,
-      "grad_norm": 0.41796875,
+      "grad_norm": 0.44921875,
       "learning_rate": 0.00019182771370002878,
-      "loss": 0.0462,
+      "loss": 0.0398,
       "step": 597
     },
     {
       "epoch": 1.3157315731573158,
       "grad_norm": 0.62109375,
       "learning_rate": 0.00019180023297288244,
-      "loss": 0.0639,
+      "loss": 0.0512,
       "step": 598
     },
     {
       "epoch": 1.317931793179318,
-      "grad_norm": 0.43359375,
+      "grad_norm": 0.51171875,
       "learning_rate": 0.00019177270809406886,
-      "loss": 0.0565,
+      "loss": 0.0537,
       "step": 599
     },
     {
       "epoch": 1.3201320132013201,
-      "grad_norm": 0.5234375,
+      "grad_norm": 0.494140625,
       "learning_rate": 0.00019174513907682634,
-      "loss": 0.0543,
+      "loss": 0.0463,
       "step": 600
     },
     {
       "epoch": 1.3223322332233223,
-      "grad_norm": 0.34765625,
+      "grad_norm": 0.44921875,
       "learning_rate": 0.00019171752593441423,
-      "loss": 0.0411,
+      "loss": 0.0488,
       "step": 601
     },
     {
       "epoch": 1.3245324532453244,
-      "grad_norm": 0.38671875,
+      "grad_norm": 0.4375,
       "learning_rate": 0.0001916898686801132,
-      "loss": 0.0494,
+      "loss": 0.0495,
       "step": 602
     },
     {
       "epoch": 1.3267326732673268,
-      "grad_norm": 0.3359375,
+      "grad_norm": 0.373046875,
       "learning_rate": 0.0001916621673272251,
-      "loss": 0.0412,
+      "loss": 0.0391,
       "step": 603
     },
     {
       "epoch": 1.328932893289329,
-      "grad_norm": 0.419921875,
+      "grad_norm": 0.53515625,
       "learning_rate": 0.00019163442188907306,
-      "loss": 0.0368,
+      "loss": 0.0407,
       "step": 604
     },
     {
       "epoch": 1.331133113311331,
-      "grad_norm": 0.349609375,
+      "grad_norm": 0.458984375,
       "learning_rate": 0.00019160663237900125,
-      "loss": 0.0488,
+      "loss": 0.0477,
       "step": 605
     },
     {
       "epoch": 1.3333333333333333,
-      "grad_norm": 0.5546875,
+      "grad_norm": 0.578125,
       "learning_rate": 0.00019157879881037522,
       "loss": 0.0356,
       "step": 606
     },
     {
       "epoch": 1.3355335533553356,
-      "grad_norm": 0.55859375,
+      "grad_norm": 0.45703125,
       "learning_rate": 0.0001915509211965816,
-      "loss": 0.0576,
+      "loss": 0.053,
       "step": 607
     },
     {
       "epoch": 1.3377337733773378,
-      "grad_norm": 0.478515625,
+      "grad_norm": 0.458984375,
       "learning_rate": 0.00019152299955102822,
-      "loss": 0.0542,
+      "loss": 0.0449,
       "step": 608
     },
     {
       "epoch": 1.33993399339934,
       "grad_norm": 0.609375,
       "learning_rate": 0.00019149503388714414,
-      "loss": 0.0525,
+      "loss": 0.0504,
       "step": 609
     },
     {
       "epoch": 1.342134213421342,
-      "grad_norm": 0.427734375,
+      "grad_norm": 0.5078125,
       "learning_rate": 0.0001914670242183795,
-      "loss": 0.0443,
+      "loss": 0.0417,
       "step": 610
     },
     {
       "epoch": 1.3443344334433442,
-      "grad_norm": 0.47265625,
+      "grad_norm": 0.50390625,
       "learning_rate": 0.0001914389705582057,
-      "loss": 0.0502,
+      "loss": 0.056,
       "step": 611
     },
     {
       "epoch": 1.3465346534653464,
-      "grad_norm": 0.67578125,
+      "grad_norm": 0.64453125,
       "learning_rate": 0.00019141087292011525,
-      "loss": 0.0903,
+      "loss": 0.0759,
       "step": 612
     },
     {
       "epoch": 1.3487348734873488,
-      "grad_norm": 0.671875,
+      "grad_norm": 0.546875,
       "learning_rate": 0.00019138273131762175,
-      "loss": 0.0549,
+      "loss": 0.0483,
       "step": 613
     },
     {
       "epoch": 1.350935093509351,
-      "grad_norm": 0.58984375,
+      "grad_norm": 0.6875,
       "learning_rate": 0.0001913545457642601,
-      "loss": 0.0536,
+      "loss": 0.0605,
       "step": 614
     },
     {
       "epoch": 1.353135313531353,
-      "grad_norm": 0.3671875,
+      "grad_norm": 0.41015625,
       "learning_rate": 0.0001913263162735862,
-      "loss": 0.0411,
+      "loss": 0.0487,
       "step": 615
     },
     {
       "epoch": 1.3553355335533555,
-      "grad_norm": 0.7578125,
+      "grad_norm": 0.49609375,
       "learning_rate": 0.00019129804285917713,
-      "loss": 0.0497,
+      "loss": 0.0588,
       "step": 616
     },
     {
       "epoch": 1.3575357535753576,
-      "grad_norm": 0.5234375,
+      "grad_norm": 0.5546875,
       "learning_rate": 0.0001912697255346311,
-      "loss": 0.04,
+      "loss": 0.0411,
       "step": 617
     },
     {
       "epoch": 1.3597359735973598,
-      "grad_norm": 0.53515625,
+      "grad_norm": 0.4609375,
       "learning_rate": 0.00019124136431356744,
-      "loss": 0.057,
+      "loss": 0.0554,
       "step": 618
     },
     {
       "epoch": 1.361936193619362,
-      "grad_norm": 0.5,
+      "grad_norm": 0.447265625,
       "learning_rate": 0.00019121295920962662,
-      "loss": 0.049,
+      "loss": 0.0505,
       "step": 619
     },
     {
       "epoch": 1.364136413641364,
-      "grad_norm": 0.451171875,
+      "grad_norm": 0.7109375,
       "learning_rate": 0.00019118451023647008,
-      "loss": 0.0525,
+      "loss": 0.0588,
       "step": 620
     },
     {
       "epoch": 1.3663366336633662,
-      "grad_norm": 0.474609375,
+      "grad_norm": 0.349609375,
       "learning_rate": 0.00019115601740778056,
-      "loss": 0.0477,
+      "loss": 0.0494,
       "step": 621
     },
     {
       "epoch": 1.3685368536853686,
-      "grad_norm": 0.765625,
+      "grad_norm": 0.63671875,
       "learning_rate": 0.00019112748073726174,
-      "loss": 0.0743,
+      "loss": 0.0742,
       "step": 622
     },
     {
       "epoch": 1.3707370737073707,
-      "grad_norm": 0.51171875,
+      "grad_norm": 0.3828125,
       "learning_rate": 0.00019109890023863847,
-      "loss": 0.0505,
+      "loss": 0.0375,
       "step": 623
     },
     {
       "epoch": 1.372937293729373,
-      "grad_norm": 0.57421875,
+      "grad_norm": 0.58984375,
       "learning_rate": 0.00019107027592565662,
-      "loss": 0.1039,
+      "loss": 0.0863,
       "step": 624
     },
     {
       "epoch": 1.3751375137513753,
-      "grad_norm": 0.53515625,
+      "grad_norm": 0.6328125,
       "learning_rate": 0.0001910416078120832,
-      "loss": 0.1001,
+      "loss": 0.1015,
       "step": 625
     },
     {
       "epoch": 1.3773377337733774,
-      "grad_norm": 0.443359375,
+      "grad_norm": 0.4765625,
       "learning_rate": 0.00019101289591170618,
-      "loss": 0.0422,
+      "loss": 0.0466,
       "step": 626
     },
     {
       "epoch": 1.3795379537953796,
-      "grad_norm": 0.419921875,
+      "grad_norm": 0.6328125,
       "learning_rate": 0.00019098414023833472,
-      "loss": 0.0568,
+      "loss": 0.0511,
       "step": 627
     },
     {
       "epoch": 1.3817381738173817,
-      "grad_norm": 0.421875,
+      "grad_norm": 0.443359375,
       "learning_rate": 0.00019095534080579895,
-      "loss": 0.0383,
+      "loss": 0.0477,
       "step": 628
     },
     {
       "epoch": 1.3839383938393839,
-      "grad_norm": 0.451171875,
+      "grad_norm": 0.5234375,
       "learning_rate": 0.00019092649762795009,
-      "loss": 0.0481,
+      "loss": 0.0494,
       "step": 629
     },
     {
       "epoch": 1.386138613861386,
-      "grad_norm": 0.65625,
+      "grad_norm": 0.79296875,
       "learning_rate": 0.0001908976107186603,
-      "loss": 0.0661,
+      "loss": 0.0843,
       "step": 630
     },
     {
       "epoch": 1.3883388338833884,
-      "grad_norm": 0.498046875,
+      "grad_norm": 0.34765625,
       "learning_rate": 0.00019086868009182293,
-      "loss": 0.042,
+      "loss": 0.0327,
       "step": 631
     },
     {
       "epoch": 1.3905390539053906,
-      "grad_norm": 0.61328125,
+      "grad_norm": 0.54296875,
       "learning_rate": 0.0001908397057613522,
-      "loss": 0.0707,
+      "loss": 0.0695,
       "step": 632
     },
     {
       "epoch": 1.3927392739273927,
-      "grad_norm": 0.65625,
+      "grad_norm": 0.412109375,
       "learning_rate": 0.00019081068774118353,
-      "loss": 0.0713,
+      "loss": 0.0419,
       "step": 633
     },
     {
       "epoch": 1.3949394939493949,
-      "grad_norm": 0.3671875,
+      "grad_norm": 0.37890625,
       "learning_rate": 0.00019078162604527313,
-      "loss": 0.0425,
+      "loss": 0.0423,
       "step": 634
     },
     {
       "epoch": 1.3971397139713972,
-      "grad_norm": 0.78515625,
+      "grad_norm": 0.58984375,
       "learning_rate": 0.00019075252068759838,
-      "loss": 0.0959,
+      "loss": 0.0742,
       "step": 635
     },
     {
       "epoch": 1.3993399339933994,
-      "grad_norm": 0.41796875,
+      "grad_norm": 0.51171875,
       "learning_rate": 0.0001907233716821576,
-      "loss": 0.0513,
+      "loss": 0.0527,
       "step": 636
     },
     {
       "epoch": 1.4015401540154016,
-      "grad_norm": 0.5234375,
+      "grad_norm": 0.458984375,
       "learning_rate": 0.00019069417904297016,
-      "loss": 0.0489,
+      "loss": 0.0385,
       "step": 637
     },
     {
       "epoch": 1.4037403740374037,
-      "grad_norm": 0.57421875,
+      "grad_norm": 0.5546875,
       "learning_rate": 0.00019066494278407631,
-      "loss": 0.0536,
+      "loss": 0.0506,
       "step": 638
     },
     {
       "epoch": 1.4059405940594059,
-      "grad_norm": 0.404296875,
+      "grad_norm": 0.439453125,
       "learning_rate": 0.00019063566291953739,
-      "loss": 0.0401,
+      "loss": 0.0471,
       "step": 639
     },
     {
       "epoch": 1.408140814081408,
-      "grad_norm": 0.330078125,
+      "grad_norm": 0.421875,
       "learning_rate": 0.0001906063394634356,
-      "loss": 0.0545,
+      "loss": 0.0484,
       "step": 640
     },
     {
       "epoch": 1.4103410341034104,
-      "grad_norm": 0.52734375,
+      "grad_norm": 0.6015625,
       "learning_rate": 0.00019057697242987423,
-      "loss": 0.062,
+      "loss": 0.0552,
       "step": 641
     },
     {
       "epoch": 1.4125412541254125,
-      "grad_norm": 0.5859375,
+      "grad_norm": 0.7734375,
       "learning_rate": 0.00019054756183297742,
-      "loss": 0.0429,
+      "loss": 0.0467,
       "step": 642
     },
     {
       "epoch": 1.4147414741474147,
-      "grad_norm": 0.462890625,
+      "grad_norm": 0.419921875,
       "learning_rate": 0.00019051810768689034,
-      "loss": 0.0505,
+      "loss": 0.0369,
       "step": 643
     },
     {
       "epoch": 1.416941694169417,
-      "grad_norm": 0.45703125,
+      "grad_norm": 0.4453125,
       "learning_rate": 0.00019048861000577904,
-      "loss": 0.0403,
+      "loss": 0.0371,
       "step": 644
     },
     {
       "epoch": 1.4191419141914192,
-      "grad_norm": 0.427734375,
+      "grad_norm": 0.451171875,
       "learning_rate": 0.00019045906880383055,
-      "loss": 0.041,
+      "loss": 0.0437,
       "step": 645
     },
     {
       "epoch": 1.4213421342134214,
-      "grad_norm": 0.73046875,
+      "grad_norm": 0.48046875,
       "learning_rate": 0.00019042948409525283,
-      "loss": 0.0776,
+      "loss": 0.0609,
       "step": 646
     },
     {
       "epoch": 1.4235423542354235,
-      "grad_norm": 0.345703125,
+      "grad_norm": 0.361328125,
       "learning_rate": 0.00019039985589427478,
-      "loss": 0.0498,
+      "loss": 0.0467,
       "step": 647
     },
     {
       "epoch": 1.4257425742574257,
-      "grad_norm": 0.515625,
+      "grad_norm": 0.4609375,
       "learning_rate": 0.00019037018421514613,
-      "loss": 0.0432,
+      "loss": 0.038,
       "step": 648
     },
     {
       "epoch": 1.4279427942794278,
-      "grad_norm": 0.53125,
+      "grad_norm": 0.4765625,
       "learning_rate": 0.00019034046907213768,
-      "loss": 0.0658,
+      "loss": 0.06,
       "step": 649
     },
     {
       "epoch": 1.4301430143014302,
-      "grad_norm": 0.5625,
+      "grad_norm": 0.56640625,
       "learning_rate": 0.00019031071047954093,
-      "loss": 0.0456,
+      "loss": 0.0465,
       "step": 650
     },
     {
       "epoch": 1.4323432343234324,
-      "grad_norm": 0.494140625,
+      "grad_norm": 0.54296875,
       "learning_rate": 0.0001902809084516685,
-      "loss": 0.0338,
+      "loss": 0.0365,
       "step": 651
     },
     {
       "epoch": 1.4345434543454345,
-      "grad_norm": 0.466796875,
+      "grad_norm": 0.5390625,
       "learning_rate": 0.00019025106300285367,
-      "loss": 0.0489,
+      "loss": 0.0511,
       "step": 652
     },
     {
       "epoch": 1.4367436743674367,
-      "grad_norm": 0.5078125,
+      "grad_norm": 0.46484375,
       "learning_rate": 0.0001902211741474508,
-      "loss": 0.0511,
+      "loss": 0.0391,
       "step": 653
     },
     {
       "epoch": 1.438943894389439,
-      "grad_norm": 0.349609375,
+      "grad_norm": 0.341796875,
       "learning_rate": 0.00019019124189983502,
-      "loss": 0.037,
+      "loss": 0.035,
       "step": 654
     },
     {
       "epoch": 1.4411441144114412,
-      "grad_norm": 0.703125,
+      "grad_norm": 0.96875,
       "learning_rate": 0.00019016126627440237,
-      "loss": 0.0705,
+      "loss": 0.0858,
       "step": 655
     },
     {
       "epoch": 1.4433443344334433,
-      "grad_norm": 0.337890625,
+      "grad_norm": 0.419921875,
       "learning_rate": 0.00019013124728556977,
-      "loss": 0.0391,
+      "loss": 0.044,
       "step": 656
     },
     {
       "epoch": 1.4455445544554455,
-      "grad_norm": 0.396484375,
+      "grad_norm": 0.46484375,
       "learning_rate": 0.0001901011849477749,
-      "loss": 0.0334,
+      "loss": 0.0322,
       "step": 657
     },
     {
       "epoch": 1.4477447744774476,
-      "grad_norm": 0.7109375,
+      "grad_norm": 0.53125,
       "learning_rate": 0.0001900710792754764,
-      "loss": 0.066,
+      "loss": 0.0601,
       "step": 658
     },
     {
       "epoch": 1.44994499449945,
-      "grad_norm": 0.263671875,
+      "grad_norm": 0.43359375,
       "learning_rate": 0.00019004093028315367,
-      "loss": 0.0279,
+      "loss": 0.0408,
       "step": 659
     },
     {
       "epoch": 1.4521452145214522,
-      "grad_norm": 0.56640625,
+      "grad_norm": 0.59375,
       "learning_rate": 0.00019001073798530698,
-      "loss": 0.0694,
+      "loss": 0.0849,
       "step": 660
     },
     {
       "epoch": 1.4543454345434543,
-      "grad_norm": 0.546875,
+      "grad_norm": 0.62109375,
       "learning_rate": 0.0001899805023964575,
-      "loss": 0.0428,
+      "loss": 0.0507,
       "step": 661
     },
     {
       "epoch": 1.4565456545654565,
-      "grad_norm": 0.5625,
+      "grad_norm": 0.51171875,
       "learning_rate": 0.00018995022353114708,
-      "loss": 0.066,
+      "loss": 0.0603,
       "step": 662
     },
     {
       "epoch": 1.4587458745874589,
-      "grad_norm": 0.46875,
+      "grad_norm": 0.423828125,
       "learning_rate": 0.00018991990140393848,
-      "loss": 0.0592,
+      "loss": 0.0551,
       "step": 663
     },
     {
       "epoch": 1.460946094609461,
-      "grad_norm": 0.53515625,
+      "grad_norm": 0.43359375,
       "learning_rate": 0.00018988953602941522,
-      "loss": 0.0458,
+      "loss": 0.0425,
       "step": 664
     },
     {
       "epoch": 1.4631463146314632,
-      "grad_norm": 0.384765625,
+      "grad_norm": 0.38671875,
       "learning_rate": 0.00018985912742218165,
-      "loss": 0.042,
+      "loss": 0.0384,
       "step": 665
     },
     {
       "epoch": 1.4653465346534653,
-      "grad_norm": 0.4453125,
+      "grad_norm": 0.466796875,
       "learning_rate": 0.00018982867559686294,
-      "loss": 0.0399,
+      "loss": 0.0484,
       "step": 666
     },
     {
       "epoch": 1.4675467546754675,
-      "grad_norm": 0.435546875,
+      "grad_norm": 0.6484375,
       "learning_rate": 0.00018979818056810496,
-      "loss": 0.0482,
+      "loss": 0.0565,
       "step": 667
     },
     {
       "epoch": 1.4697469746974696,
-      "grad_norm": 0.5078125,
+      "grad_norm": 0.431640625,
       "learning_rate": 0.00018976764235057445,
-      "loss": 0.0371,
+      "loss": 0.0344,
       "step": 668
     },
     {
       "epoch": 1.471947194719472,
-      "grad_norm": 0.49609375,
+      "grad_norm": 0.4921875,
       "learning_rate": 0.00018973706095895887,
-      "loss": 0.0461,
+      "loss": 0.0468,
       "step": 669
     },
     {
       "epoch": 1.4741474147414741,
-      "grad_norm": 0.4765625,
+      "grad_norm": 0.34765625,
       "learning_rate": 0.00018970643640796642,
-      "loss": 0.0455,
+      "loss": 0.0378,
       "step": 670
     },
     {
       "epoch": 1.4763476347634763,
-      "grad_norm": 0.498046875,
+      "grad_norm": 0.5546875,
       "learning_rate": 0.0001896757687123261,
-      "loss": 0.0499,
+      "loss": 0.0522,
       "step": 671
     },
     {
       "epoch": 1.4785478547854787,
-      "grad_norm": 0.5625,
+      "grad_norm": 0.462890625,
       "learning_rate": 0.0001896450578867877,
-      "loss": 0.0579,
+      "loss": 0.0461,
       "step": 672
     },
     {
       "epoch": 1.4807480748074808,
-      "grad_norm": 0.36328125,
+      "grad_norm": 0.4765625,
       "learning_rate": 0.0001896143039461217,
-      "loss": 0.0471,
+      "loss": 0.0537,
       "step": 673
     },
     {
       "epoch": 1.482948294829483,
-      "grad_norm": 0.5234375,
+      "grad_norm": 0.53515625,
       "learning_rate": 0.00018958350690511928,
-      "loss": 0.0689,
+      "loss": 0.0676,
       "step": 674
     },
     {
       "epoch": 1.4851485148514851,
-      "grad_norm": 0.3515625,
+      "grad_norm": 0.33984375,
       "learning_rate": 0.0001895526667785924,
-      "loss": 0.0274,
+      "loss": 0.0253,
       "step": 675
     },
     {
       "epoch": 1.4873487348734873,
-      "grad_norm": 0.60546875,
+      "grad_norm": 0.49609375,
       "learning_rate": 0.00018952178358137378,
-      "loss": 0.047,
+      "loss": 0.0404,
       "step": 676
     },
     {
       "epoch": 1.4895489548954894,
-      "grad_norm": 0.6328125,
+      "grad_norm": 0.69140625,
       "learning_rate": 0.00018949085732831674,
-      "loss": 0.0534,
+      "loss": 0.0633,
       "step": 677
     },
     {
       "epoch": 1.4917491749174918,
-      "grad_norm": 0.51953125,
+      "grad_norm": 0.490234375,
       "learning_rate": 0.00018945988803429546,
-      "loss": 0.0417,
+      "loss": 0.0373,
       "step": 678
     },
     {
       "epoch": 1.493949394939494,
-      "grad_norm": 0.447265625,
+      "grad_norm": 0.478515625,
       "learning_rate": 0.00018942887571420469,
-      "loss": 0.0374,
+      "loss": 0.0334,
       "step": 679
     },
     {
       "epoch": 1.4961496149614961,
-      "grad_norm": 0.3828125,
+      "grad_norm": 0.359375,
       "learning_rate": 0.0001893978203829599,
-      "loss": 0.0415,
+      "loss": 0.0312,
       "step": 680
     },
     {
       "epoch": 1.4983498349834983,
-      "grad_norm": 0.52734375,
+      "grad_norm": 0.44921875,
       "learning_rate": 0.00018936672205549732,
-      "loss": 0.0395,
+      "loss": 0.0377,
       "step": 681
     },
     {
       "epoch": 1.5005500550055006,
-      "grad_norm": 0.474609375,
+      "grad_norm": 0.578125,
       "learning_rate": 0.00018933558074677378,
-      "loss": 0.0525,
+      "loss": 0.0594,
       "step": 682
     },
     {
       "epoch": 1.5027502750275028,
-      "grad_norm": 0.6015625,
+      "grad_norm": 0.6328125,
       "learning_rate": 0.0001893043964717668,
-      "loss": 0.0526,
+      "loss": 0.0635,
       "step": 683
     },
     {
       "epoch": 1.504950495049505,
-      "grad_norm": 0.55078125,
+      "grad_norm": 0.50390625,
       "learning_rate": 0.0001892731692454746,
-      "loss": 0.0531,
+      "loss": 0.0503,
       "step": 684
     },
     {
       "epoch": 1.504950495049505,
-      "eval_loss": 0.06403707712888718,
-      "eval_runtime": 10.2594,
-      "eval_samples_per_second": 37.332,
-      "eval_steps_per_second": 4.679,
+      "eval_loss": 0.060496263206005096,
+      "eval_runtime": 10.1466,
+      "eval_samples_per_second": 37.747,
+      "eval_steps_per_second": 4.731,
       "step": 684
     },
     {
       "epoch": 1.507150715071507,
-      "grad_norm": 0.484375,
+      "grad_norm": 0.408203125,
       "learning_rate": 0.000189241899082916,
-      "loss": 0.041,
+      "loss": 0.0318,
       "step": 685
     },
     {
       "epoch": 1.5093509350935093,
-      "grad_norm": 0.4140625,
+      "grad_norm": 0.482421875,
       "learning_rate": 0.00018921058599913055,
-      "loss": 0.0512,
+      "loss": 0.0491,
       "step": 686
     },
     {
       "epoch": 1.5115511551155114,
-      "grad_norm": 0.419921875,
+      "grad_norm": 0.5078125,
       "learning_rate": 0.00018917923000917836,
-      "loss": 0.032,
+      "loss": 0.0389,
       "step": 687
     },
     {
       "epoch": 1.5137513751375138,
-      "grad_norm": 0.345703125,
+      "grad_norm": 0.310546875,
       "learning_rate": 0.0001891478311281402,
-      "loss": 0.0321,
+      "loss": 0.0306,
       "step": 688
     },
     {
       "epoch": 1.515951595159516,
-      "grad_norm": 0.5859375,
+      "grad_norm": 0.59765625,
       "learning_rate": 0.0001891163893711175,
-      "loss": 0.0538,
+      "loss": 0.0533,
       "step": 689
     },
     {
       "epoch": 1.5181518151815183,
-      "grad_norm": 0.478515625,
+      "grad_norm": 0.44140625,
       "learning_rate": 0.00018908490475323233,
-      "loss": 0.0482,
+      "loss": 0.038,
       "step": 690
     },
     {
       "epoch": 1.5203520352035205,
       "grad_norm": 0.52734375,
       "learning_rate": 0.00018905337728962726,
-      "loss": 0.0503,
+      "loss": 0.0481,
       "step": 691
     },
     {
       "epoch": 1.5225522552255226,
-      "grad_norm": 0.40625,
+      "grad_norm": 0.416015625,
       "learning_rate": 0.00018902180699546556,
-      "loss": 0.048,
+      "loss": 0.0398,
       "step": 692
     },
     {
       "epoch": 1.5247524752475248,
-      "grad_norm": 0.390625,
+      "grad_norm": 0.56640625,
       "learning_rate": 0.00018899019388593115,
-      "loss": 0.041,
+      "loss": 0.0549,
       "step": 693
     },
     {
       "epoch": 1.526952695269527,
-      "grad_norm": 0.6796875,
+      "grad_norm": 0.55078125,
       "learning_rate": 0.00018895853797622837,
-      "loss": 0.0357,
+      "loss": 0.0413,
       "step": 694
     },
     {
       "epoch": 1.529152915291529,
-      "grad_norm": 0.51953125,
+      "grad_norm": 0.49609375,
       "learning_rate": 0.00018892683928158232,
-      "loss": 0.0585,
+      "loss": 0.0518,
       "step": 695
     },
     {
       "epoch": 1.5313531353135312,
-      "grad_norm": 0.4609375,
+      "grad_norm": 0.46484375,
       "learning_rate": 0.0001888950978172386,
-      "loss": 0.0351,
+      "loss": 0.0272,
       "step": 696
     },
     {
       "epoch": 1.5335533553355336,
-      "grad_norm": 0.60546875,
+      "grad_norm": 0.490234375,
       "learning_rate": 0.00018886331359846336,
-      "loss": 0.0578,
+      "loss": 0.0546,
       "step": 697
     },
     {
       "epoch": 1.5357535753575358,
-      "grad_norm": 0.5625,
+      "grad_norm": 0.5859375,
       "learning_rate": 0.00018883148664054335,
-      "loss": 0.0469,
+      "loss": 0.0381,
       "step": 698
     },
     {
       "epoch": 1.537953795379538,
-      "grad_norm": 0.41015625,
+      "grad_norm": 0.4453125,
       "learning_rate": 0.00018879961695878586,
-      "loss": 0.0386,
+      "loss": 0.0377,
       "step": 699
     },
     {
       "epoch": 1.5401540154015403,
-      "grad_norm": 0.40234375,
+      "grad_norm": 0.447265625,
       "learning_rate": 0.00018876770456851877,
       "loss": 0.0311,
       "step": 700
     },
     {
       "epoch": 1.5423542354235424,
-      "grad_norm": 0.578125,
+      "grad_norm": 0.498046875,
       "learning_rate": 0.00018873574948509046,
-      "loss": 0.0582,
+      "loss": 0.0515,
       "step": 701
     },
     {
       "epoch": 1.5445544554455446,
-      "grad_norm": 0.37890625,
+      "grad_norm": 0.3828125,
       "learning_rate": 0.00018870375172386976,
-      "loss": 0.0357,
+      "loss": 0.0295,
       "step": 702
     },
     {
       "epoch": 1.5467546754675467,
-      "grad_norm": 0.486328125,
+      "grad_norm": 0.400390625,
       "learning_rate": 0.00018867171130024623,
-      "loss": 0.0462,
+      "loss": 0.0383,
       "step": 703
     },
     {
       "epoch": 1.548954895489549,
-      "grad_norm": 0.380859375,
+      "grad_norm": 0.3359375,
       "learning_rate": 0.00018863962822962974,
-      "loss": 0.0337,
+      "loss": 0.0303,
       "step": 704
     },
     {
       "epoch": 1.551155115511551,
-      "grad_norm": 0.40234375,
+      "grad_norm": 0.3203125,
       "learning_rate": 0.00018860750252745084,
-      "loss": 0.0332,
+      "loss": 0.0247,
       "step": 705
     },
     {
       "epoch": 1.5533553355335532,
-      "grad_norm": 0.43359375,
+      "grad_norm": 0.546875,
       "learning_rate": 0.00018857533420916042,
-      "loss": 0.0448,
+      "loss": 0.0427,
       "step": 706
     },
     {
       "epoch": 1.5555555555555556,
-      "grad_norm": 0.41796875,
+      "grad_norm": 0.416015625,
       "learning_rate": 0.00018854312329023002,
-      "loss": 0.0411,
+      "loss": 0.0445,
       "step": 707
     },
     {
       "epoch": 1.5577557755775577,
-      "grad_norm": 0.400390625,
+      "grad_norm": 0.404296875,
       "learning_rate": 0.00018851086978615163,
-      "loss": 0.0388,
+      "loss": 0.0321,
       "step": 708
     },
     {
       "epoch": 1.55995599559956,
-      "grad_norm": 0.578125,
+      "grad_norm": 0.427734375,
       "learning_rate": 0.00018847857371243762,
-      "loss": 0.0457,
+      "loss": 0.0349,
       "step": 709
     },
     {
       "epoch": 1.5621562156215623,
-      "grad_norm": 0.34765625,
+      "grad_norm": 0.455078125,
       "learning_rate": 0.00018844623508462092,
-      "loss": 0.032,
+      "loss": 0.0405,
       "step": 710
     },
     {
       "epoch": 1.5643564356435644,
-      "grad_norm": 0.4765625,
+      "grad_norm": 0.53515625,
       "learning_rate": 0.00018841385391825494,
-      "loss": 0.0419,
+      "loss": 0.0481,
       "step": 711
     },
     {
       "epoch": 1.5665566556655666,
-      "grad_norm": 0.41015625,
+      "grad_norm": 0.51953125,
       "learning_rate": 0.00018838143022891355,
-      "loss": 0.0253,
+      "loss": 0.0269,
       "step": 712
     },
     {
       "epoch": 1.5687568756875687,
-      "grad_norm": 0.314453125,
+      "grad_norm": 0.349609375,
       "learning_rate": 0.00018834896403219098,
-      "loss": 0.0363,
+      "loss": 0.0383,
       "step": 713
     },
     {
       "epoch": 1.5709570957095709,
-      "grad_norm": 0.447265625,
+      "grad_norm": 0.427734375,
       "learning_rate": 0.000188316455343702,
-      "loss": 0.0407,
+      "loss": 0.0381,
       "step": 714
     },
     {
       "epoch": 1.573157315731573,
-      "grad_norm": 0.6328125,
+      "grad_norm": 0.41796875,
       "learning_rate": 0.0001882839041790818,
-      "loss": 0.0437,
+      "loss": 0.0325,
       "step": 715
     },
     {
       "epoch": 1.5753575357535754,
-      "grad_norm": 0.5,
+      "grad_norm": 0.427734375,
       "learning_rate": 0.00018825131055398597,
-      "loss": 0.0468,
+      "loss": 0.0494,
       "step": 716
     },
     {
       "epoch": 1.5775577557755776,
-      "grad_norm": 0.4140625,
+      "grad_norm": 0.51953125,
       "learning_rate": 0.00018821867448409055,
-      "loss": 0.0357,
+      "loss": 0.0417,
       "step": 717
     },
     {
       "epoch": 1.5797579757975797,
-      "grad_norm": 0.388671875,
+      "grad_norm": 0.396484375,
       "learning_rate": 0.00018818599598509196,
-      "loss": 0.0491,
+      "loss": 0.0391,
       "step": 718
     },
     {
       "epoch": 1.581958195819582,
-      "grad_norm": 0.55078125,
+      "grad_norm": 0.65625,
       "learning_rate": 0.00018815327507270703,
-      "loss": 0.0486,
+      "loss": 0.0444,
       "step": 719
     },
     {
       "epoch": 1.5841584158415842,
-      "grad_norm": 0.416015625,
+      "grad_norm": 0.322265625,
       "learning_rate": 0.00018812051176267307,
-      "loss": 0.0261,
+      "loss": 0.0243,
       "step": 720
     },
     {
       "epoch": 1.5863586358635864,
-      "grad_norm": 0.40625,
+      "grad_norm": 0.380859375,
       "learning_rate": 0.0001880877060707477,
-      "loss": 0.0316,
+      "loss": 0.0289,
       "step": 721
     },
     {
       "epoch": 1.5885588558855885,
-      "grad_norm": 0.431640625,
+      "grad_norm": 0.578125,
       "learning_rate": 0.00018805485801270887,
-      "loss": 0.0427,
+      "loss": 0.0531,
       "step": 722
     },
     {
       "epoch": 1.5907590759075907,
-      "grad_norm": 0.296875,
+      "grad_norm": 0.318359375,
       "learning_rate": 0.00018802196760435503,
-      "loss": 0.0265,
+      "loss": 0.027,
       "step": 723
     },
     {
       "epoch": 1.5929592959295928,
-      "grad_norm": 0.4296875,
+      "grad_norm": 0.466796875,
       "learning_rate": 0.00018798903486150494,
-      "loss": 0.0453,
+      "loss": 0.0502,
       "step": 724
     },
     {
       "epoch": 1.595159515951595,
-      "grad_norm": 0.267578125,
+      "grad_norm": 0.48046875,
       "learning_rate": 0.00018795605979999775,
-      "loss": 0.0221,
+      "loss": 0.0338,
       "step": 725
     },
     {
       "epoch": 1.5973597359735974,
-      "grad_norm": 0.404296875,
+      "grad_norm": 0.439453125,
       "learning_rate": 0.00018792304243569292,
-      "loss": 0.0425,
+      "loss": 0.0431,
       "step": 726
     },
     {
       "epoch": 1.5995599559955995,
-      "grad_norm": 0.69140625,
+      "grad_norm": 0.53515625,
       "learning_rate": 0.00018788998278447026,
-      "loss": 0.0561,
+      "loss": 0.0509,
       "step": 727
     },
     {
       "epoch": 1.601760176017602,
-      "grad_norm": 0.6640625,
+      "grad_norm": 0.341796875,
       "learning_rate": 0.00018785688086222997,
-      "loss": 0.0372,
+      "loss": 0.028,
       "step": 728
     },
     {
       "epoch": 1.603960396039604,
-      "grad_norm": 0.482421875,
+      "grad_norm": 0.365234375,
       "learning_rate": 0.0001878237366848925,
-      "loss": 0.0347,
+      "loss": 0.0289,
       "step": 729
     },
     {
       "epoch": 1.6061606160616062,
-      "grad_norm": 0.45703125,
+      "grad_norm": 0.50390625,
       "learning_rate": 0.00018779055026839868,
-      "loss": 0.0343,
+      "loss": 0.0483,
       "step": 730
     },
     {
       "epoch": 1.6083608360836084,
-      "grad_norm": 0.45703125,
+      "grad_norm": 0.19921875,
       "learning_rate": 0.00018775732162870968,
-      "loss": 0.0329,
+      "loss": 0.0243,
       "step": 731
     },
     {
       "epoch": 1.6105610561056105,
-      "grad_norm": 0.40234375,
+      "grad_norm": 0.4609375,
       "learning_rate": 0.0001877240507818069,
-      "loss": 0.0312,
+      "loss": 0.033,
       "step": 732
     },
     {
       "epoch": 1.6127612761276127,
-      "grad_norm": 0.51953125,
+      "grad_norm": 0.326171875,
       "learning_rate": 0.00018769073774369208,
-      "loss": 0.0308,
+      "loss": 0.0256,
       "step": 733
     },
     {
       "epoch": 1.6149614961496148,
-      "grad_norm": 0.474609375,
+      "grad_norm": 0.4453125,
       "learning_rate": 0.00018765738253038726,
-      "loss": 0.0315,
+      "loss": 0.0417,
       "step": 734
     },
     {
       "epoch": 1.6171617161716172,
-      "grad_norm": 0.38671875,
+      "grad_norm": 0.51171875,
       "learning_rate": 0.00018762398515793477,
-      "loss": 0.0436,
+      "loss": 0.0476,
       "step": 735
     },
     {
       "epoch": 1.6193619361936193,
-      "grad_norm": 0.291015625,
+      "grad_norm": 0.375,
       "learning_rate": 0.00018759054564239716,
-      "loss": 0.0262,
+      "loss": 0.0325,
       "step": 736
     },
     {
       "epoch": 1.6215621562156217,
-      "grad_norm": 0.369140625,
+      "grad_norm": 0.193359375,
       "learning_rate": 0.00018755706399985734,
-      "loss": 0.0284,
+      "loss": 0.0262,
       "step": 737
     },
     {
       "epoch": 1.6237623762376239,
-      "grad_norm": 0.27734375,
+      "grad_norm": 0.345703125,
       "learning_rate": 0.00018752354024641842,
-      "loss": 0.0321,
+      "loss": 0.0386,
       "step": 738
     },
     {
       "epoch": 1.625962596259626,
-      "grad_norm": 0.62109375,
+      "grad_norm": 0.458984375,
       "learning_rate": 0.00018748997439820372,
-      "loss": 0.0452,
+      "loss": 0.0431,
       "step": 739
     },
     {
       "epoch": 1.6281628162816282,
-      "grad_norm": 0.388671875,
+      "grad_norm": 0.466796875,
       "learning_rate": 0.00018745636647135694,
-      "loss": 0.041,
+      "loss": 0.0456,
       "step": 740
     },
     {
       "epoch": 1.6303630363036303,
-      "grad_norm": 0.41796875,
+      "grad_norm": 0.408203125,
       "learning_rate": 0.0001874227164820419,
-      "loss": 0.0292,
+      "loss": 0.037,
       "step": 741
     },
     {
       "epoch": 1.6325632563256325,
-      "grad_norm": 0.671875,
+      "grad_norm": 0.56640625,
       "learning_rate": 0.0001873890244464427,
-      "loss": 0.0626,
+      "loss": 0.0533,
       "step": 742
     },
     {
       "epoch": 1.6347634763476346,
-      "grad_norm": 0.384765625,
+      "grad_norm": 0.337890625,
       "learning_rate": 0.00018735529038076362,
-      "loss": 0.0434,
+      "loss": 0.0324,
       "step": 743
     },
     {
       "epoch": 1.636963696369637,
-      "grad_norm": 0.21484375,
+      "grad_norm": 0.578125,
       "learning_rate": 0.0001873215143012292,
-      "loss": 0.0169,
+      "loss": 0.0251,
       "step": 744
     },
     {
       "epoch": 1.6391639163916392,
-      "grad_norm": 0.74609375,
+      "grad_norm": 0.5390625,
       "learning_rate": 0.00018728769622408423,
-      "loss": 0.0331,
+      "loss": 0.0379,
       "step": 745
     },
     {
       "epoch": 1.6413641364136413,
-      "grad_norm": 0.48828125,
+      "grad_norm": 0.375,
       "learning_rate": 0.00018725383616559354,
-      "loss": 0.0377,
+      "loss": 0.0325,
       "step": 746
     },
     {
       "epoch": 1.6435643564356437,
-      "grad_norm": 0.35546875,
+      "grad_norm": 0.322265625,
       "learning_rate": 0.00018721993414204236,
-      "loss": 0.0483,
+      "loss": 0.0352,
       "step": 747
     },
     {
       "epoch": 1.6457645764576458,
-      "grad_norm": 0.431640625,
+      "grad_norm": 0.40234375,
       "learning_rate": 0.00018718599016973592,
-      "loss": 0.0297,
+      "loss": 0.0283,
       "step": 748
     },
     {
       "epoch": 1.647964796479648,
-      "grad_norm": 0.41015625,
+      "grad_norm": 0.462890625,
       "learning_rate": 0.00018715200426499973,
-      "loss": 0.0406,
+      "loss": 0.0392,
       "step": 749
     },
     {
       "epoch": 1.6501650165016502,
-      "grad_norm": 0.474609375,
+      "grad_norm": 0.44140625,
       "learning_rate": 0.0001871179764441794,
-      "loss": 0.031,
+      "loss": 0.0319,
       "step": 750
     },
     {
       "epoch": 1.6523652365236523,
-      "grad_norm": 0.38671875,
+      "grad_norm": 0.30078125,
       "learning_rate": 0.0001870839067236408,
-      "loss": 0.0331,
+      "loss": 0.0298,
       "step": 751
     },
     {
       "epoch": 1.6545654565456545,
-      "grad_norm": 0.43359375,
+      "grad_norm": 0.353515625,
       "learning_rate": 0.00018704979511976983,
-      "loss": 0.0435,
+      "loss": 0.0391,
       "step": 752
     },
     {
       "epoch": 1.6567656765676566,
-      "grad_norm": 0.498046875,
+      "grad_norm": 0.3984375,
       "learning_rate": 0.00018701564164897265,
-      "loss": 0.0352,
+      "loss": 0.0307,
       "step": 753
     },
     {
       "epoch": 1.658965896589659,
-      "grad_norm": 0.490234375,
+      "grad_norm": 0.546875,
       "learning_rate": 0.00018698144632767547,
-      "loss": 0.0333,
+      "loss": 0.034,
       "step": 754
     },
     {
       "epoch": 1.6611661166116611,
-      "grad_norm": 0.38671875,
+      "grad_norm": 0.298828125,
       "learning_rate": 0.00018694720917232467,
-      "loss": 0.0291,
+      "loss": 0.0211,
       "step": 755
     },
     {
       "epoch": 1.6633663366336635,
-      "grad_norm": 0.404296875,
+      "grad_norm": 0.318359375,
       "learning_rate": 0.00018691293019938674,
-      "loss": 0.0358,
+      "loss": 0.0272,
       "step": 756
     },
     {
       "epoch": 1.6655665566556657,
-      "grad_norm": 0.25390625,
+      "grad_norm": 0.380859375,
       "learning_rate": 0.00018687860942534826,
-      "loss": 0.0294,
+      "loss": 0.0353,
       "step": 757
     },
     {
       "epoch": 1.6677667766776678,
-      "grad_norm": 0.57421875,
+      "grad_norm": 0.5859375,
       "learning_rate": 0.0001868442468667159,
-      "loss": 0.0614,
+      "loss": 0.0583,
       "step": 758
     },
     {
       "epoch": 1.66996699669967,
-      "grad_norm": 0.4609375,
+      "grad_norm": 0.45703125,
       "learning_rate": 0.00018680984254001656,
-      "loss": 0.0427,
+      "loss": 0.0482,
       "step": 759
     },
     {
       "epoch": 1.6721672167216721,
-      "grad_norm": 0.39453125,
+      "grad_norm": 0.294921875,
       "learning_rate": 0.00018677539646179707,
-      "loss": 0.0373,
+      "loss": 0.0285,
       "step": 760
     },
     {
       "epoch": 1.6743674367436743,
-      "grad_norm": 0.45703125,
+      "grad_norm": 0.462890625,
       "learning_rate": 0.00018674090864862443,
-      "loss": 0.0281,
+      "loss": 0.0282,
       "step": 761
     },
     {
       "epoch": 1.6765676567656764,
-      "grad_norm": 0.376953125,
+      "grad_norm": 0.392578125,
       "learning_rate": 0.00018670637911708565,
-      "loss": 0.0307,
+      "loss": 0.0301,
       "step": 762
     },
     {
       "epoch": 1.6787678767876788,
-      "grad_norm": 0.640625,
+      "grad_norm": 0.384765625,
       "learning_rate": 0.00018667180788378782,
-      "loss": 0.0475,
+      "loss": 0.0285,
       "step": 763
     },
     {
       "epoch": 1.680968096809681,
-      "grad_norm": 0.62109375,
+      "grad_norm": 0.58984375,
       "learning_rate": 0.00018663719496535817,
-      "loss": 0.0535,
+      "loss": 0.058,
       "step": 764
     },
     {
       "epoch": 1.6831683168316833,
-      "grad_norm": 0.318359375,
+      "grad_norm": 0.2138671875,
       "learning_rate": 0.00018660254037844388,
-      "loss": 0.0261,
+      "loss": 0.0224,
       "step": 765
     },
     {
       "epoch": 1.6853685368536855,
-      "grad_norm": 0.296875,
+      "grad_norm": 0.3359375,
       "learning_rate": 0.0001865678441397122,
-      "loss": 0.0259,
+      "loss": 0.024,
       "step": 766
     },
     {
       "epoch": 1.6875687568756876,
-      "grad_norm": 0.40234375,
+      "grad_norm": 0.333984375,
       "learning_rate": 0.0001865331062658504,
-      "loss": 0.0263,
+      "loss": 0.0245,
       "step": 767
     },
     {
       "epoch": 1.6897689768976898,
-      "grad_norm": 0.44140625,
+      "grad_norm": 0.40625,
       "learning_rate": 0.00018649832677356584,
-      "loss": 0.0313,
+      "loss": 0.0303,
       "step": 768
     },
     {
       "epoch": 1.691969196919692,
-      "grad_norm": 0.375,
+      "grad_norm": 0.4453125,
       "learning_rate": 0.00018646350567958582,
-      "loss": 0.0257,
+      "loss": 0.0254,
       "step": 769
     },
     {
       "epoch": 1.694169416941694,
-      "grad_norm": 0.337890625,
+      "grad_norm": 0.466796875,
       "learning_rate": 0.00018642864300065767,
-      "loss": 0.0344,
+      "loss": 0.0377,
       "step": 770
     },
     {
       "epoch": 1.6963696369636962,
-      "grad_norm": 0.5078125,
+      "grad_norm": 0.443359375,
       "learning_rate": 0.00018639373875354868,
-      "loss": 0.0445,
+      "loss": 0.0331,
       "step": 771
     },
     {
       "epoch": 1.6985698569856986,
-      "grad_norm": 0.361328125,
+      "grad_norm": 0.400390625,
       "learning_rate": 0.0001863587929550463,
-      "loss": 0.0312,
+      "loss": 0.0406,
       "step": 772
     },
     {
       "epoch": 1.7007700770077008,
-      "grad_norm": 0.4921875,
+      "grad_norm": 0.4140625,
       "learning_rate": 0.00018632380562195772,
-      "loss": 0.0312,
+      "loss": 0.036,
       "step": 773
     },
     {
       "epoch": 1.702970297029703,
-      "grad_norm": 0.455078125,
+      "grad_norm": 0.58984375,
       "learning_rate": 0.0001862887767711103,
-      "loss": 0.0359,
+      "loss": 0.0364,
       "step": 774
     },
     {
       "epoch": 1.7051705170517053,
-      "grad_norm": 0.34765625,
+      "grad_norm": 0.380859375,
       "learning_rate": 0.00018625370641935129,
-      "loss": 0.0378,
+      "loss": 0.0416,
       "step": 775
     },
     {
       "epoch": 1.7073707370737075,
-      "grad_norm": 0.3515625,
+      "grad_norm": 0.341796875,
       "learning_rate": 0.0001862185945835479,
-      "loss": 0.038,
+      "loss": 0.0335,
       "step": 776
     },
     {
       "epoch": 1.7095709570957096,
-      "grad_norm": 0.2890625,
+      "grad_norm": 0.29296875,
       "learning_rate": 0.0001861834412805873,
-      "loss": 0.0293,
+      "loss": 0.0303,
       "step": 777
     },
     {
       "epoch": 1.7117711771177118,
-      "grad_norm": 0.361328125,
+      "grad_norm": 0.37109375,
       "learning_rate": 0.00018614824652737658,
-      "loss": 0.0279,
+      "loss": 0.0269,
       "step": 778
     },
     {
       "epoch": 1.713971397139714,
-      "grad_norm": 0.345703125,
+      "grad_norm": 0.41796875,
       "learning_rate": 0.00018611301034084283,
-      "loss": 0.0352,
+      "loss": 0.043,
       "step": 779
     },
     {
       "epoch": 1.716171617161716,
-      "grad_norm": 0.58984375,
+      "grad_norm": 0.2890625,
       "learning_rate": 0.00018607773273793298,
-      "loss": 0.0428,
+      "loss": 0.0252,
       "step": 780
     },
     {
       "epoch": 1.7183718371837182,
-      "grad_norm": 0.39453125,
+      "grad_norm": 0.427734375,
       "learning_rate": 0.00018604241373561397,
-      "loss": 0.0268,
+      "loss": 0.0287,
       "step": 781
     },
     {
       "epoch": 1.7205720572057206,
-      "grad_norm": 0.341796875,
+      "grad_norm": 0.279296875,
       "learning_rate": 0.0001860070533508726,
-      "loss": 0.0372,
+      "loss": 0.035,
       "step": 782
     },
     {
       "epoch": 1.7227722772277227,
-      "grad_norm": 0.341796875,
+      "grad_norm": 0.3671875,
       "learning_rate": 0.00018597165160071555,
-      "loss": 0.0251,
+      "loss": 0.0222,
       "step": 783
     },
     {
       "epoch": 1.7249724972497251,
-      "grad_norm": 0.375,
+      "grad_norm": 0.32421875,
       "learning_rate": 0.00018593620850216943,
-      "loss": 0.0255,
+      "loss": 0.0262,
       "step": 784
     },
     {
       "epoch": 1.7271727172717273,
-      "grad_norm": 0.451171875,
+      "grad_norm": 0.40234375,
       "learning_rate": 0.00018590072407228075,
-      "loss": 0.0326,
+      "loss": 0.0267,
       "step": 785
     },
     {
       "epoch": 1.7293729372937294,
-      "grad_norm": 0.51171875,
+      "grad_norm": 0.388671875,
       "learning_rate": 0.0001858651983281159,
-      "loss": 0.0319,
+      "loss": 0.0211,
       "step": 786
     },
     {
       "epoch": 1.7315731573157316,
-      "grad_norm": 0.349609375,
+      "grad_norm": 0.4375,
       "learning_rate": 0.00018582963128676108,
-      "loss": 0.0221,
+      "loss": 0.0313,
       "step": 787
     },
     {
       "epoch": 1.7337733773377337,
-      "grad_norm": 0.384765625,
+      "grad_norm": 0.455078125,
       "learning_rate": 0.0001857940229653224,
       "loss": 0.0415,
       "step": 788
     },
     {
       "epoch": 1.7359735973597359,
-      "grad_norm": 0.458984375,
+      "grad_norm": 0.400390625,
       "learning_rate": 0.00018575837338092582,
-      "loss": 0.0365,
+      "loss": 0.036,
       "step": 789
     },
     {
       "epoch": 1.738173817381738,
-      "grad_norm": 0.57421875,
+      "grad_norm": 0.515625,
       "learning_rate": 0.00018572268255071718,
-      "loss": 0.0415,
+      "loss": 0.0364,
       "step": 790
     },
     {
       "epoch": 1.7403740374037404,
-      "grad_norm": 0.326171875,
+      "grad_norm": 0.25390625,
       "learning_rate": 0.00018568695049186204,
-      "loss": 0.0203,
+      "loss": 0.0142,
       "step": 791
     },
     {
       "epoch": 1.7425742574257426,
-      "grad_norm": 0.466796875,
+      "grad_norm": 0.3515625,
       "learning_rate": 0.00018565117722154595,
-      "loss": 0.0271,
+      "loss": 0.0277,
       "step": 792
     },
     {
       "epoch": 1.7447744774477447,
-      "grad_norm": 0.1923828125,
+      "grad_norm": 0.3359375,
       "learning_rate": 0.00018561536275697415,
-      "loss": 0.021,
+      "loss": 0.0234,
       "step": 793
     },
     {
       "epoch": 1.746974697469747,
-      "grad_norm": 0.3359375,
+      "grad_norm": 0.3984375,
       "learning_rate": 0.00018557950711537173,
-      "loss": 0.0285,
+      "loss": 0.027,
       "step": 794
     },
     {
       "epoch": 1.7491749174917492,
-      "grad_norm": 0.25,
+      "grad_norm": 0.404296875,
       "learning_rate": 0.00018554361031398363,
-      "loss": 0.0263,
+      "loss": 0.0327,
       "step": 795
     },
     {
       "epoch": 1.7513751375137514,
-      "grad_norm": 0.408203125,
+      "grad_norm": 0.451171875,
       "learning_rate": 0.00018550767237007453,
-      "loss": 0.0287,
+      "loss": 0.0338,
       "step": 796
     },
     {
       "epoch": 1.7535753575357536,
-      "grad_norm": 0.302734375,
+      "grad_norm": 0.6875,
       "learning_rate": 0.00018547169330092894,
-      "loss": 0.0319,
+      "loss": 0.0427,
       "step": 797
     },
     {
       "epoch": 1.7557755775577557,
-      "grad_norm": 0.32421875,
+      "grad_norm": 0.515625,
       "learning_rate": 0.00018543567312385113,
-      "loss": 0.0313,
+      "loss": 0.0476,
       "step": 798
     },
     {
       "epoch": 1.7557755775577557,
-      "eval_loss": 0.04786590486764908,
-      "eval_runtime": 10.1329,
-      "eval_samples_per_second": 37.798,
-      "eval_steps_per_second": 4.737,
+      "eval_loss": 0.04843483120203018,
+      "eval_runtime": 10.0716,
+      "eval_samples_per_second": 38.028,
+      "eval_steps_per_second": 4.766,
       "step": 798
     },
     {
       "epoch": 1.7579757975797579,
-      "grad_norm": 0.52734375,
+      "grad_norm": 0.59375,
       "learning_rate": 0.0001853996118561651,
-      "loss": 0.0319,
+      "loss": 0.0363,
       "step": 799
     },
     {
       "epoch": 1.76017601760176,
-      "grad_norm": 0.33984375,
+      "grad_norm": 0.28125,
       "learning_rate": 0.00018536350951521468,
-      "loss": 0.034,
+      "loss": 0.0359,
       "step": 800
     },
     {
       "epoch": 1.7623762376237624,
-      "grad_norm": 0.640625,
+      "grad_norm": 0.5,
       "learning_rate": 0.00018532736611836343,
-      "loss": 0.0259,
+      "loss": 0.0388,
       "step": 801
     },
     {
       "epoch": 1.7645764576457645,
-      "grad_norm": 0.42578125,
+      "grad_norm": 0.37109375,
       "learning_rate": 0.00018529118168299466,
-      "loss": 0.0271,
+      "loss": 0.0256,
       "step": 802
     },
     {
       "epoch": 1.766776677667767,
-      "grad_norm": 0.44140625,
+      "grad_norm": 0.359375,
       "learning_rate": 0.0001852549562265114,
-      "loss": 0.0226,
+      "loss": 0.0255,
       "step": 803
     },
     {
       "epoch": 1.768976897689769,
-      "grad_norm": 0.36328125,
+      "grad_norm": 0.421875,
       "learning_rate": 0.0001852186897663364,
-      "loss": 0.0226,
+      "loss": 0.0377,
       "step": 804
     },
     {
       "epoch": 1.7711771177117712,
-      "grad_norm": 0.443359375,
+      "grad_norm": 0.439453125,
       "learning_rate": 0.00018518238231991218,
-      "loss": 0.0252,
+      "loss": 0.0284,
       "step": 805
     },
     {
       "epoch": 1.7733773377337734,
-      "grad_norm": 0.52734375,
+      "grad_norm": 0.416015625,
       "learning_rate": 0.00018514603390470092,
-      "loss": 0.0324,
+      "loss": 0.0292,
       "step": 806
     },
     {
       "epoch": 1.7755775577557755,
-      "grad_norm": 0.353515625,
+      "grad_norm": 0.470703125,
       "learning_rate": 0.00018510964453818458,
-      "loss": 0.0328,
+      "loss": 0.0366,
       "step": 807
     },
     {
       "epoch": 1.7777777777777777,
-      "grad_norm": 0.5234375,
+      "grad_norm": 0.396484375,
       "learning_rate": 0.00018507321423786465,
-      "loss": 0.0362,
+      "loss": 0.0334,
       "step": 808
     },
     {
       "epoch": 1.7799779977997798,
-      "grad_norm": 0.310546875,
+      "grad_norm": 0.3984375,
       "learning_rate": 0.00018503674302126254,
-      "loss": 0.0274,
+      "loss": 0.0233,
       "step": 809
     },
     {
       "epoch": 1.7821782178217822,
-      "grad_norm": 0.36328125,
+      "grad_norm": 0.498046875,
       "learning_rate": 0.00018500023090591917,
-      "loss": 0.0233,
+      "loss": 0.038,
       "step": 810
     },
     {
       "epoch": 1.7843784378437844,
-      "grad_norm": 0.25390625,
+      "grad_norm": 0.279296875,
       "learning_rate": 0.00018496367790939516,
-      "loss": 0.0235,
+      "loss": 0.0305,
       "step": 811
     },
     {
       "epoch": 1.7865786578657867,
-      "grad_norm": 0.2578125,
+      "grad_norm": 0.341796875,
       "learning_rate": 0.0001849270840492708,
-      "loss": 0.0214,
+      "loss": 0.0297,
       "step": 812
     },
     {
       "epoch": 1.7887788778877889,
-      "grad_norm": 0.490234375,
+      "grad_norm": 0.423828125,
       "learning_rate": 0.00018489044934314613,
-      "loss": 0.0363,
+      "loss": 0.0385,
       "step": 813
     },
     {
       "epoch": 1.790979097909791,
-      "grad_norm": 0.57421875,
+      "grad_norm": 0.2060546875,
       "learning_rate": 0.00018485377380864069,
-      "loss": 0.0384,
+      "loss": 0.0255,
       "step": 814
     },
     {
       "epoch": 1.7931793179317932,
-      "grad_norm": 0.57421875,
+      "grad_norm": 0.3984375,
       "learning_rate": 0.0001848170574633937,
-      "loss": 0.0262,
+      "loss": 0.0288,
       "step": 815
     },
     {
       "epoch": 1.7953795379537953,
-      "grad_norm": 0.515625,
+      "grad_norm": 0.259765625,
       "learning_rate": 0.00018478030032506403,
-      "loss": 0.0333,
+      "loss": 0.0315,
       "step": 816
     },
     {
       "epoch": 1.7975797579757975,
-      "grad_norm": 0.33984375,
+      "grad_norm": 0.41015625,
       "learning_rate": 0.0001847435024113302,
-      "loss": 0.0227,
+      "loss": 0.0308,
       "step": 817
     },
     {
       "epoch": 1.7997799779977997,
-      "grad_norm": 0.396484375,
+      "grad_norm": 0.39453125,
       "learning_rate": 0.00018470666373989028,
-      "loss": 0.0274,
+      "loss": 0.032,
       "step": 818
     },
     {
       "epoch": 1.801980198019802,
-      "grad_norm": 0.447265625,
+      "grad_norm": 0.404296875,
       "learning_rate": 0.00018466978432846198,
-      "loss": 0.029,
+      "loss": 0.0319,
       "step": 819
     },
     {
       "epoch": 1.8041804180418042,
-      "grad_norm": 0.37890625,
+      "grad_norm": 0.333984375,
       "learning_rate": 0.00018463286419478255,
-      "loss": 0.0169,
+      "loss": 0.0166,
       "step": 820
     },
     {
       "epoch": 1.8063806380638063,
-      "grad_norm": 0.3984375,
+      "grad_norm": 0.392578125,
       "learning_rate": 0.00018459590335660894,
-      "loss": 0.0259,
+      "loss": 0.0225,
       "step": 821
     },
     {
       "epoch": 1.8085808580858087,
-      "grad_norm": 0.53125,
+      "grad_norm": 0.314453125,
       "learning_rate": 0.00018455890183171754,
-      "loss": 0.0335,
+      "loss": 0.0245,
       "step": 822
     },
     {
       "epoch": 1.8107810781078109,
-      "grad_norm": 0.53125,
+      "grad_norm": 0.494140625,
       "learning_rate": 0.0001845218596379044,
-      "loss": 0.0261,
+      "loss": 0.03,
       "step": 823
     },
     {
       "epoch": 1.812981298129813,
-      "grad_norm": 0.4140625,
+      "grad_norm": 0.3515625,
       "learning_rate": 0.00018448477679298508,
-      "loss": 0.0269,
+      "loss": 0.0229,
       "step": 824
     },
     {
       "epoch": 1.8151815181518152,
-      "grad_norm": 0.423828125,
+      "grad_norm": 0.255859375,
       "learning_rate": 0.00018444765331479475,
-      "loss": 0.0375,
+      "loss": 0.0324,
       "step": 825
     },
     {
       "epoch": 1.8173817381738173,
-      "grad_norm": 0.375,
+      "grad_norm": 0.484375,
       "learning_rate": 0.00018441048922118806,
-      "loss": 0.0393,
+      "loss": 0.0362,
       "step": 826
     },
     {
       "epoch": 1.8195819581958195,
-      "grad_norm": 0.2392578125,
+      "grad_norm": 0.6015625,
       "learning_rate": 0.00018437328453003922,
-      "loss": 0.0198,
+      "loss": 0.0325,
       "step": 827
     },
     {
       "epoch": 1.8217821782178216,
-      "grad_norm": 0.36328125,
+      "grad_norm": 0.333984375,
       "learning_rate": 0.00018433603925924202,
-      "loss": 0.0395,
+      "loss": 0.0416,
       "step": 828
     },
     {
       "epoch": 1.823982398239824,
-      "grad_norm": 0.46484375,
+      "grad_norm": 0.72265625,
       "learning_rate": 0.00018429875342670964,
-      "loss": 0.03,
+      "loss": 0.0287,
       "step": 829
     },
     {
       "epoch": 1.8261826182618262,
-      "grad_norm": 0.408203125,
+      "grad_norm": 0.400390625,
       "learning_rate": 0.00018426142705037487,
-      "loss": 0.0303,
+      "loss": 0.0232,
       "step": 830
     },
     {
       "epoch": 1.8283828382838285,
-      "grad_norm": 0.453125,
+      "grad_norm": 0.3515625,
       "learning_rate": 0.00018422406014818993,
-      "loss": 0.026,
+      "loss": 0.0198,
       "step": 831
     },
     {
       "epoch": 1.8305830583058307,
-      "grad_norm": 0.55078125,
+      "grad_norm": 0.447265625,
       "learning_rate": 0.00018418665273812668,
-      "loss": 0.0308,
+      "loss": 0.0364,
       "step": 832
     },
     {
       "epoch": 1.8327832783278328,
-      "grad_norm": 0.296875,
+      "grad_norm": 0.35546875,
       "learning_rate": 0.00018414920483817625,
-      "loss": 0.0221,
+      "loss": 0.0188,
       "step": 833
     },
     {
       "epoch": 1.834983498349835,
-      "grad_norm": 0.97265625,
+      "grad_norm": 0.396484375,
       "learning_rate": 0.00018411171646634937,
-      "loss": 0.0555,
+      "loss": 0.0356,
       "step": 834
     },
     {
       "epoch": 1.8371837183718371,
-      "grad_norm": 0.3828125,
+      "grad_norm": 0.359375,
       "learning_rate": 0.00018407418764067627,
-      "loss": 0.0312,
+      "loss": 0.036,
       "step": 835
     },
     {
       "epoch": 1.8393839383938393,
-      "grad_norm": 0.72265625,
+      "grad_norm": 0.4609375,
       "learning_rate": 0.0001840366183792065,
-      "loss": 0.0378,
+      "loss": 0.0296,
       "step": 836
     },
     {
       "epoch": 1.8415841584158414,
-      "grad_norm": 0.435546875,
+      "grad_norm": 0.4375,
       "learning_rate": 0.0001839990087000092,
-      "loss": 0.0315,
+      "loss": 0.0291,
       "step": 837
     },
     {
       "epoch": 1.8437843784378438,
-      "grad_norm": 0.359375,
+      "grad_norm": 0.26171875,
       "learning_rate": 0.00018396135862117286,
-      "loss": 0.0281,
+      "loss": 0.0227,
       "step": 838
     },
     {
       "epoch": 1.845984598459846,
-      "grad_norm": 0.44921875,
+      "grad_norm": 0.56640625,
       "learning_rate": 0.00018392366816080542,
-      "loss": 0.042,
+      "loss": 0.0438,
       "step": 839
     },
     {
       "epoch": 1.8481848184818483,
-      "grad_norm": 0.384765625,
+      "grad_norm": 0.447265625,
       "learning_rate": 0.00018388593733703428,
-      "loss": 0.0249,
+      "loss": 0.0353,
       "step": 840
     },
     {
       "epoch": 1.8503850385038505,
-      "grad_norm": 0.462890625,
+      "grad_norm": 0.609375,
       "learning_rate": 0.00018384816616800619,
-      "loss": 0.0406,
+      "loss": 0.0342,
       "step": 841
     },
     {
       "epoch": 1.8525852585258527,
-      "grad_norm": 0.267578125,
+      "grad_norm": 0.326171875,
       "learning_rate": 0.0001838103546718873,
-      "loss": 0.033,
+      "loss": 0.03,
       "step": 842
     },
     {
       "epoch": 1.8547854785478548,
-      "grad_norm": 0.40234375,
+      "grad_norm": 0.451171875,
       "learning_rate": 0.00018377250286686332,
-      "loss": 0.0315,
+      "loss": 0.0301,
       "step": 843
     },
     {
       "epoch": 1.856985698569857,
-      "grad_norm": 0.3359375,
+      "grad_norm": 0.203125,
       "learning_rate": 0.00018373461077113908,
-      "loss": 0.0237,
+      "loss": 0.015,
       "step": 844
     },
     {
       "epoch": 1.859185918591859,
-      "grad_norm": 0.48046875,
+      "grad_norm": 0.6015625,
       "learning_rate": 0.00018369667840293898,
-      "loss": 0.0505,
+      "loss": 0.045,
       "step": 845
     },
     {
       "epoch": 1.8613861386138613,
-      "grad_norm": 0.26171875,
+      "grad_norm": 0.263671875,
       "learning_rate": 0.00018365870578050677,
-      "loss": 0.0264,
+      "loss": 0.0251,
       "step": 846
     },
     {
       "epoch": 1.8635863586358636,
-      "grad_norm": 0.58203125,
+      "grad_norm": 0.353515625,
       "learning_rate": 0.00018362069292210544,
-      "loss": 0.0497,
+      "loss": 0.0295,
       "step": 847
     },
     {
       "epoch": 1.8657865786578658,
-      "grad_norm": 0.361328125,
+      "grad_norm": 0.482421875,
       "learning_rate": 0.0001835826398460175,
-      "loss": 0.0299,
+      "loss": 0.0476,
       "step": 848
     },
     {
       "epoch": 1.867986798679868,
-      "grad_norm": 0.392578125,
+      "grad_norm": 0.279296875,
       "learning_rate": 0.00018354454657054469,
-      "loss": 0.0468,
+      "loss": 0.0294,
       "step": 849
     },
     {
       "epoch": 1.8701870187018703,
-      "grad_norm": 0.37109375,
+      "grad_norm": 0.3515625,
       "learning_rate": 0.00018350641311400812,
       "loss": 0.0302,
       "step": 850
     },
     {
       "epoch": 1.8723872387238725,
-      "grad_norm": 0.2734375,
+      "grad_norm": 0.267578125,
       "learning_rate": 0.0001834682394947482,
-      "loss": 0.0258,
+      "loss": 0.0235,
       "step": 851
     },
     {
       "epoch": 1.8745874587458746,
-      "grad_norm": 0.359375,
+      "grad_norm": 0.353515625,
       "learning_rate": 0.00018343002573112464,
-      "loss": 0.0226,
+      "loss": 0.0286,
       "step": 852
     },
     {
       "epoch": 1.8767876787678768,
-      "grad_norm": 0.419921875,
+      "grad_norm": 0.33203125,
       "learning_rate": 0.00018339177184151662,
-      "loss": 0.0375,
+      "loss": 0.0361,
       "step": 853
     },
     {
       "epoch": 1.878987898789879,
-      "grad_norm": 0.30078125,
+      "grad_norm": 0.431640625,
       "learning_rate": 0.00018335347784432236,
-      "loss": 0.0327,
+      "loss": 0.0409,
       "step": 854
     },
     {
       "epoch": 1.881188118811881,
-      "grad_norm": 0.380859375,
+      "grad_norm": 0.345703125,
       "learning_rate": 0.00018331514375795959,
-      "loss": 0.0243,
+      "loss": 0.0297,
       "step": 855
     },
     {
       "epoch": 1.8833883388338832,
-      "grad_norm": 0.435546875,
+      "grad_norm": 0.3984375,
       "learning_rate": 0.00018327676960086517,
-      "loss": 0.0259,
+      "loss": 0.0249,
       "step": 856
     },
     {
       "epoch": 1.8855885588558856,
-      "grad_norm": 0.341796875,
+      "grad_norm": 0.40234375,
       "learning_rate": 0.00018323835539149532,
-      "loss": 0.0319,
+      "loss": 0.0359,
       "step": 857
     },
     {
       "epoch": 1.8877887788778878,
-      "grad_norm": 0.267578125,
+      "grad_norm": 0.203125,
       "learning_rate": 0.0001831999011483255,
-      "loss": 0.0253,
+      "loss": 0.024,
       "step": 858
     },
     {
       "epoch": 1.8899889988998901,
-      "grad_norm": 0.41015625,
+      "grad_norm": 0.46484375,
       "learning_rate": 0.00018316140688985047,
-      "loss": 0.0281,
+      "loss": 0.0321,
       "step": 859
     },
     {
       "epoch": 1.8921892189218923,
-      "grad_norm": 0.337890625,
+      "grad_norm": 0.4921875,
       "learning_rate": 0.00018312287263458409,
-      "loss": 0.0278,
+      "loss": 0.0372,
       "step": 860
     },
     {
       "epoch": 1.8943894389438944,
-      "grad_norm": 0.39453125,
+      "grad_norm": 0.3671875,
       "learning_rate": 0.00018308429840105964,
-      "loss": 0.0269,
+      "loss": 0.0313,
       "step": 861
     },
     {
       "epoch": 1.8965896589658966,
-      "grad_norm": 0.41015625,
+      "grad_norm": 0.55859375,
       "learning_rate": 0.00018304568420782948,
-      "loss": 0.0266,
+      "loss": 0.036,
       "step": 862
     },
     {
       "epoch": 1.8987898789878987,
-      "grad_norm": 0.3125,
+      "grad_norm": 0.3359375,
       "learning_rate": 0.0001830070300734653,
-      "loss": 0.0241,
+      "loss": 0.026,
       "step": 863
     },
     {
       "epoch": 1.900990099009901,
-      "grad_norm": 0.4921875,
+      "grad_norm": 0.314453125,
       "learning_rate": 0.00018296833601655794,
-      "loss": 0.028,
+      "loss": 0.0289,
       "step": 864
     },
     {
       "epoch": 1.903190319031903,
-      "grad_norm": 0.5234375,
+      "grad_norm": 0.43359375,
       "learning_rate": 0.0001829296020557174,
-      "loss": 0.037,
+      "loss": 0.0321,
       "step": 865
     },
     {
       "epoch": 1.9053905390539054,
-      "grad_norm": 0.388671875,
+      "grad_norm": 0.408203125,
       "learning_rate": 0.000182890828209573,
-      "loss": 0.0264,
+      "loss": 0.0276,
       "step": 866
     },
     {
       "epoch": 1.9075907590759076,
-      "grad_norm": 0.47265625,
+      "grad_norm": 0.341796875,
       "learning_rate": 0.00018285201449677317,
-      "loss": 0.0448,
+      "loss": 0.0305,
       "step": 867
     },
     {
       "epoch": 1.9097909790979097,
-      "grad_norm": 0.56640625,
+      "grad_norm": 0.458984375,
       "learning_rate": 0.00018281316093598544,
-      "loss": 0.0348,
+      "loss": 0.0333,
       "step": 868
     },
     {
       "epoch": 1.911991199119912,
-      "grad_norm": 0.41796875,
+      "grad_norm": 0.484375,
       "learning_rate": 0.0001827742675458966,
-      "loss": 0.0284,
+      "loss": 0.0325,
       "step": 869
     },
     {
       "epoch": 1.9141914191419143,
-      "grad_norm": 0.5078125,
+      "grad_norm": 0.3828125,
       "learning_rate": 0.00018273533434521263,
-      "loss": 0.0383,
+      "loss": 0.0324,
       "step": 870
     },
     {
       "epoch": 1.9163916391639164,
-      "grad_norm": 0.296875,
+      "grad_norm": 0.41015625,
       "learning_rate": 0.00018269636135265853,
-      "loss": 0.0213,
+      "loss": 0.0311,
       "step": 871
     },
     {
       "epoch": 1.9185918591859186,
-      "grad_norm": 0.494140625,
+      "grad_norm": 0.68359375,
       "learning_rate": 0.0001826573485869785,
-      "loss": 0.0286,
+      "loss": 0.0442,
       "step": 872
     },
     {
       "epoch": 1.9207920792079207,
-      "grad_norm": 0.486328125,
+      "grad_norm": 0.3203125,
       "learning_rate": 0.00018261829606693595,
-      "loss": 0.0251,
+      "loss": 0.022,
       "step": 873
     },
     {
       "epoch": 1.9229922992299229,
-      "grad_norm": 0.54296875,
+      "grad_norm": 0.4375,
       "learning_rate": 0.00018257920381131327,
-      "loss": 0.0516,
+      "loss": 0.0386,
       "step": 874
     },
     {
       "epoch": 1.925192519251925,
-      "grad_norm": 0.353515625,
+      "grad_norm": 0.2890625,
       "learning_rate": 0.0001825400718389121,
-      "loss": 0.0186,
+      "loss": 0.0193,
       "step": 875
     },
     {
       "epoch": 1.9273927392739274,
-      "grad_norm": 0.478515625,
+      "grad_norm": 0.376953125,
       "learning_rate": 0.000182500900168553,
-      "loss": 0.0335,
+      "loss": 0.03,
       "step": 876
     },
     {
       "epoch": 1.9295929592959296,
-      "grad_norm": 0.396484375,
+      "grad_norm": 0.265625,
       "learning_rate": 0.0001824616888190758,
-      "loss": 0.0315,
+      "loss": 0.019,
       "step": 877
     },
     {
       "epoch": 1.931793179317932,
-      "grad_norm": 0.4921875,
+      "grad_norm": 0.296875,
       "learning_rate": 0.00018242243780933936,
-      "loss": 0.0306,
+      "loss": 0.0239,
       "step": 878
     },
     {
       "epoch": 1.933993399339934,
-      "grad_norm": 0.6171875,
+      "grad_norm": 0.3828125,
       "learning_rate": 0.00018238314715822158,
-      "loss": 0.0462,
+      "loss": 0.0297,
       "step": 879
     },
     {
       "epoch": 1.9361936193619362,
-      "grad_norm": 0.25,
+      "grad_norm": 0.40234375,
       "learning_rate": 0.00018234381688461942,
-      "loss": 0.022,
+      "loss": 0.0294,
       "step": 880
     },
     {
       "epoch": 1.9383938393839384,
-      "grad_norm": 0.57421875,
+      "grad_norm": 0.36328125,
       "learning_rate": 0.000182304447007449,
-      "loss": 0.0346,
+      "loss": 0.038,
       "step": 881
     },
     {
       "epoch": 1.9405940594059405,
-      "grad_norm": 0.361328125,
+      "grad_norm": 0.2412109375,
       "learning_rate": 0.0001822650375456453,
-      "loss": 0.0341,
+      "loss": 0.0272,
       "step": 882
     },
     {
       "epoch": 1.9427942794279427,
-      "grad_norm": 0.49609375,
+      "grad_norm": 0.478515625,
       "learning_rate": 0.00018222558851816256,
-      "loss": 0.0328,
+      "loss": 0.0261,
       "step": 883
     },
     {
       "epoch": 1.9449944994499448,
-      "grad_norm": 0.33984375,
+      "grad_norm": 0.296875,
       "learning_rate": 0.00018218609994397387,
-      "loss": 0.0277,
+      "loss": 0.0243,
       "step": 884
     },
     {
       "epoch": 1.9471947194719472,
-      "grad_norm": 0.48828125,
+      "grad_norm": 0.53515625,
       "learning_rate": 0.0001821465718420714,
-      "loss": 0.0259,
+      "loss": 0.0269,
       "step": 885
     },
     {
       "epoch": 1.9493949394939494,
-      "grad_norm": 0.44921875,
+      "grad_norm": 0.248046875,
       "learning_rate": 0.00018210700423146637,
-      "loss": 0.0304,
+      "loss": 0.0189,
       "step": 886
     },
     {
       "epoch": 1.9515951595159517,
-      "grad_norm": 0.27734375,
+      "grad_norm": 0.294921875,
       "learning_rate": 0.000182067397131189,
-      "loss": 0.0268,
+      "loss": 0.0272,
       "step": 887
     },
     {
       "epoch": 1.953795379537954,
-      "grad_norm": 0.447265625,
+      "grad_norm": 0.37109375,
       "learning_rate": 0.0001820277505602884,
-      "loss": 0.0355,
+      "loss": 0.0286,
       "step": 888
     },
     {
       "epoch": 1.955995599559956,
-      "grad_norm": 0.42578125,
+      "grad_norm": 0.48828125,
       "learning_rate": 0.0001819880645378328,
-      "loss": 0.0435,
+      "loss": 0.0377,
       "step": 889
     },
     {
       "epoch": 1.9581958195819582,
-      "grad_norm": 0.447265625,
+      "grad_norm": 0.32421875,
       "learning_rate": 0.00018194833908290932,
-      "loss": 0.036,
+      "loss": 0.0261,
       "step": 890
     },
     {
       "epoch": 1.9603960396039604,
-      "grad_norm": 0.32421875,
+      "grad_norm": 0.453125,
       "learning_rate": 0.00018190857421462403,
-      "loss": 0.022,
+      "loss": 0.0205,
       "step": 891
     },
     {
       "epoch": 1.9625962596259625,
-      "grad_norm": 0.283203125,
+      "grad_norm": 0.310546875,
       "learning_rate": 0.00018186876995210204,
-      "loss": 0.0261,
+      "loss": 0.0276,
       "step": 892
     },
     {
       "epoch": 1.9647964796479647,
-      "grad_norm": 0.341796875,
+      "grad_norm": 0.2294921875,
       "learning_rate": 0.00018182892631448736,
-      "loss": 0.0272,
+      "loss": 0.0273,
       "step": 893
     },
     {
       "epoch": 1.966996699669967,
-      "grad_norm": 0.466796875,
+      "grad_norm": 0.57421875,
       "learning_rate": 0.00018178904332094293,
-      "loss": 0.0382,
+      "loss": 0.0358,
       "step": 894
     },
     {
       "epoch": 1.9691969196919692,
-      "grad_norm": 0.3125,
+      "grad_norm": 0.349609375,
       "learning_rate": 0.0001817491209906506,
-      "loss": 0.0369,
+      "loss": 0.0361,
       "step": 895
     },
     {
       "epoch": 1.9713971397139713,
-      "grad_norm": 0.439453125,
+      "grad_norm": 0.337890625,
       "learning_rate": 0.00018170915934281122,
-      "loss": 0.0272,
+      "loss": 0.0205,
       "step": 896
     },
     {
       "epoch": 1.9735973597359737,
-      "grad_norm": 0.322265625,
+      "grad_norm": 0.2890625,
       "learning_rate": 0.00018166915839664443,
-      "loss": 0.0196,
+      "loss": 0.0179,
       "step": 897
     },
     {
       "epoch": 1.9757975797579759,
-      "grad_norm": 0.357421875,
+      "grad_norm": 0.34375,
       "learning_rate": 0.00018162911817138884,
-      "loss": 0.0289,
+      "loss": 0.0294,
       "step": 898
     },
     {
       "epoch": 1.977997799779978,
-      "grad_norm": 0.232421875,
+      "grad_norm": 0.267578125,
       "learning_rate": 0.00018158903868630203,
-      "loss": 0.0161,
+      "loss": 0.0266,
       "step": 899
     },
     {
       "epoch": 1.9801980198019802,
-      "grad_norm": 0.443359375,
+      "grad_norm": 0.294921875,
       "learning_rate": 0.0001815489199606603,
-      "loss": 0.0273,
+      "loss": 0.0327,
       "step": 900
     },
     {
       "epoch": 1.9823982398239823,
-      "grad_norm": 0.37890625,
+      "grad_norm": 0.306640625,
       "learning_rate": 0.0001815087620137589,
-      "loss": 0.0332,
+      "loss": 0.0274,
       "step": 901
     },
     {
       "epoch": 1.9845984598459845,
-      "grad_norm": 0.37109375,
+      "grad_norm": 0.408203125,
       "learning_rate": 0.000181468564864912,
-      "loss": 0.0253,
+      "loss": 0.0394,
       "step": 902
     },
     {
       "epoch": 1.9867986798679866,
-      "grad_norm": 0.314453125,
+      "grad_norm": 0.2314453125,
       "learning_rate": 0.00018142832853345253,
-      "loss": 0.0252,
+      "loss": 0.0258,
       "step": 903
     },
     {
       "epoch": 1.988998899889989,
-      "grad_norm": 0.4375,
+      "grad_norm": 0.408203125,
       "learning_rate": 0.0001813880530387323,
-      "loss": 0.0358,
+      "loss": 0.0332,
       "step": 904
     },
     {
       "epoch": 1.9911991199119912,
-      "grad_norm": 0.318359375,
+      "grad_norm": 0.447265625,
       "learning_rate": 0.000181347738400122,
-      "loss": 0.0215,
+      "loss": 0.036,
       "step": 905
     },
     {
       "epoch": 1.9933993399339935,
-      "grad_norm": 0.35546875,
+      "grad_norm": 0.416015625,
       "learning_rate": 0.00018130738463701103,
-      "loss": 0.0233,
+      "loss": 0.0246,
       "step": 906
     },
     {
       "epoch": 1.9955995599559957,
-      "grad_norm": 0.494140625,
+      "grad_norm": 0.3125,
       "learning_rate": 0.00018126699176880778,
-      "loss": 0.0422,
+      "loss": 0.0292,
       "step": 907
     },
     {
       "epoch": 1.9977997799779978,
-      "grad_norm": 0.3046875,
+      "grad_norm": 0.2890625,
       "learning_rate": 0.0001812265598149393,
-      "loss": 0.0219,
+      "loss": 0.0223,
       "step": 908
     },
     {
       "epoch": 2.0,
-      "grad_norm": 0.333984375,
+      "grad_norm": 0.2734375,
       "learning_rate": 0.0001811860887948515,
-      "loss": 0.0221,
+      "loss": 0.0211,
       "step": 909
     },
     {
       "epoch": 2.002200220022002,
-      "grad_norm": 0.400390625,
+      "grad_norm": 0.296875,
       "learning_rate": 0.00018114557872800905,
-      "loss": 0.0212,
+      "loss": 0.0165,
       "step": 910
     },
     {
       "epoch": 2.0044004400440043,
-      "grad_norm": 0.51171875,
+      "grad_norm": 0.2734375,
       "learning_rate": 0.00018110502963389547,
-      "loss": 0.038,
+      "loss": 0.0253,
       "step": 911
     },
     {
       "epoch": 2.0066006600660065,
-      "grad_norm": 0.3515625,
+      "grad_norm": 0.17578125,
       "learning_rate": 0.00018106444153201292,
-      "loss": 0.0134,
+      "loss": 0.0084,
       "step": 912
     },
     {
       "epoch": 2.0066006600660065,
-      "eval_loss": 0.04637401923537254,
-      "eval_runtime": 10.8204,
-      "eval_samples_per_second": 35.396,
-      "eval_steps_per_second": 4.436,
+      "eval_loss": 0.041667137295007706,
+      "eval_runtime": 10.1026,
+      "eval_samples_per_second": 37.911,
+      "eval_steps_per_second": 4.751,
       "step": 912
     },
     {
       "epoch": 2.0088008800880086,
-      "grad_norm": 0.2060546875,
+      "grad_norm": 0.30078125,
       "learning_rate": 0.00018102381444188244,
-      "loss": 0.0203,
+      "loss": 0.0243,
       "step": 913
     },
     {
       "epoch": 2.011001100110011,
-      "grad_norm": 0.46484375,
+      "grad_norm": 0.314453125,
       "learning_rate": 0.00018098314838304382,
-      "loss": 0.034,
+      "loss": 0.0202,
       "step": 914
     },
     {
       "epoch": 2.0132013201320134,
-      "grad_norm": 0.435546875,
+      "grad_norm": 0.2177734375,
       "learning_rate": 0.0001809424433750555,
-      "loss": 0.0272,
+      "loss": 0.016,
       "step": 915
     },
     {
       "epoch": 2.0154015401540155,
-      "grad_norm": 0.427734375,
+      "grad_norm": 0.41015625,
       "learning_rate": 0.00018090169943749476,
-      "loss": 0.028,
+      "loss": 0.0281,
       "step": 916
     },
     {
       "epoch": 2.0176017601760177,
-      "grad_norm": 0.27734375,
+      "grad_norm": 0.279296875,
       "learning_rate": 0.00018086091658995747,
-      "loss": 0.0168,
+      "loss": 0.0166,
       "step": 917
     },
     {
       "epoch": 2.01980198019802,
-      "grad_norm": 0.390625,
+      "grad_norm": 0.373046875,
       "learning_rate": 0.00018082009485205838,
-      "loss": 0.0233,
+      "loss": 0.0245,
       "step": 918
     },
     {
       "epoch": 2.022002200220022,
-      "grad_norm": 0.384765625,
+      "grad_norm": 0.2734375,
       "learning_rate": 0.00018077923424343083,
-      "loss": 0.0308,
+      "loss": 0.0252,
       "step": 919
     },
     {
       "epoch": 2.024202420242024,
-      "grad_norm": 0.466796875,
+      "grad_norm": 0.259765625,
       "learning_rate": 0.0001807383347837268,
-      "loss": 0.0304,
+      "loss": 0.0193,
       "step": 920
     },
     {
       "epoch": 2.0264026402640263,
-      "grad_norm": 0.40625,
+      "grad_norm": 0.146484375,
       "learning_rate": 0.00018069739649261718,
-      "loss": 0.019,
+      "loss": 0.0142,
       "step": 921
     },
     {
       "epoch": 2.0286028602860284,
-      "grad_norm": 0.259765625,
+      "grad_norm": 0.30859375,
       "learning_rate": 0.0001806564193897913,
-      "loss": 0.0173,
+      "loss": 0.0184,
       "step": 922
     },
     {
       "epoch": 2.0308030803080306,
-      "grad_norm": 0.28515625,
+      "grad_norm": 0.240234375,
       "learning_rate": 0.0001806154034949573,
-      "loss": 0.0197,
+      "loss": 0.0213,
       "step": 923
     },
     {
       "epoch": 2.033003300330033,
-      "grad_norm": 0.419921875,
+      "grad_norm": 0.328125,
       "learning_rate": 0.00018057434882784188,
-      "loss": 0.0268,
+      "loss": 0.0281,
       "step": 924
     },
     {
       "epoch": 2.0352035203520353,
-      "grad_norm": 0.3203125,
+      "grad_norm": 0.181640625,
       "learning_rate": 0.00018053325540819045,
-      "loss": 0.0329,
+      "loss": 0.0204,
       "step": 925
     },
     {
       "epoch": 2.0374037403740375,
-      "grad_norm": 0.337890625,
+      "grad_norm": 0.59765625,
       "learning_rate": 0.0001804921232557671,
-      "loss": 0.027,
+      "loss": 0.0373,
       "step": 926
     },
     {
       "epoch": 2.0396039603960396,
-      "grad_norm": 0.451171875,
+      "grad_norm": 0.2265625,
       "learning_rate": 0.0001804509523903544,
-      "loss": 0.0387,
+      "loss": 0.0227,
       "step": 927
     },
     {
       "epoch": 2.041804180418042,
-      "grad_norm": 0.28125,
+      "grad_norm": 0.294921875,
       "learning_rate": 0.00018040974283175374,
-      "loss": 0.0176,
+      "loss": 0.0185,
       "step": 928
     },
     {
       "epoch": 2.044004400440044,
-      "grad_norm": 0.3046875,
+      "grad_norm": 0.291015625,
       "learning_rate": 0.00018036849459978493,
-      "loss": 0.0288,
+      "loss": 0.0221,
       "step": 929
     },
     {
       "epoch": 2.046204620462046,
-      "grad_norm": 0.396484375,
+      "grad_norm": 0.205078125,
       "learning_rate": 0.00018032720771428648,
-      "loss": 0.0251,
+      "loss": 0.0195,
       "step": 930
     },
     {
       "epoch": 2.0484048404840483,
-      "grad_norm": 0.33203125,
+      "grad_norm": 0.474609375,
       "learning_rate": 0.0001802858821951155,
-      "loss": 0.026,
+      "loss": 0.0267,
       "step": 931
     },
     {
       "epoch": 2.0506050605060504,
-      "grad_norm": 0.40625,
+      "grad_norm": 0.240234375,
       "learning_rate": 0.00018024451806214763,
-      "loss": 0.0331,
+      "loss": 0.0221,
       "step": 932
     },
     {
       "epoch": 2.052805280528053,
-      "grad_norm": 0.259765625,
+      "grad_norm": 0.171875,
       "learning_rate": 0.0001802031153352771,
-      "loss": 0.0178,
+      "loss": 0.0176,
       "step": 933
     },
     {
       "epoch": 2.055005500550055,
-      "grad_norm": 0.294921875,
+      "grad_norm": 0.421875,
       "learning_rate": 0.00018016167403441674,
-      "loss": 0.0174,
+      "loss": 0.0191,
       "step": 934
     },
     {
       "epoch": 2.0572057205720573,
-      "grad_norm": 0.447265625,
+      "grad_norm": 0.2890625,
       "learning_rate": 0.00018012019417949784,
-      "loss": 0.0334,
+      "loss": 0.0273,
       "step": 935
     },
     {
       "epoch": 2.0594059405940595,
-      "grad_norm": 0.2421875,
+      "grad_norm": 0.32421875,
       "learning_rate": 0.00018007867579047034,
-      "loss": 0.0192,
+      "loss": 0.0212,
       "step": 936
     },
     {
       "epoch": 2.0616061606160616,
-      "grad_norm": 0.37109375,
+      "grad_norm": 0.427734375,
       "learning_rate": 0.00018003711888730267,
-      "loss": 0.034,
+      "loss": 0.0371,
       "step": 937
     },
     {
       "epoch": 2.0638063806380638,
-      "grad_norm": 0.283203125,
+      "grad_norm": 0.298828125,
       "learning_rate": 0.00017999552348998176,
-      "loss": 0.0166,
+      "loss": 0.0147,
       "step": 938
     },
     {
       "epoch": 2.066006600660066,
-      "grad_norm": 0.2314453125,
+      "grad_norm": 0.2890625,
       "learning_rate": 0.00017995388961851308,
-      "loss": 0.0118,
+      "loss": 0.0147,
       "step": 939
     },
     {
       "epoch": 2.068206820682068,
-      "grad_norm": 0.26953125,
+      "grad_norm": 0.2578125,
       "learning_rate": 0.0001799122172929206,
-      "loss": 0.018,
+      "loss": 0.0188,
       "step": 940
     },
     {
       "epoch": 2.0704070407040702,
       "grad_norm": 0.486328125,
       "learning_rate": 0.00017987050653324674,
-      "loss": 0.0202,
+      "loss": 0.0278,
       "step": 941
     },
     {
       "epoch": 2.072607260726073,
-      "grad_norm": 0.443359375,
+      "grad_norm": 0.25390625,
       "learning_rate": 0.0001798287573595525,
-      "loss": 0.0255,
+      "loss": 0.0184,
       "step": 942
     },
     {
       "epoch": 2.074807480748075,
-      "grad_norm": 0.25,
+      "grad_norm": 0.3046875,
       "learning_rate": 0.00017978696979191727,
-      "loss": 0.0223,
+      "loss": 0.0197,
       "step": 943
     },
     {
       "epoch": 2.077007700770077,
-      "grad_norm": 0.2470703125,
+      "grad_norm": 0.2734375,
       "learning_rate": 0.00017974514385043897,
-      "loss": 0.0242,
+      "loss": 0.0272,
       "step": 944
     },
     {
       "epoch": 2.0792079207920793,
-      "grad_norm": 0.44140625,
+      "grad_norm": 0.35546875,
       "learning_rate": 0.0001797032795552339,
-      "loss": 0.0207,
+      "loss": 0.0177,
       "step": 945
     },
     {
       "epoch": 2.0814081408140814,
-      "grad_norm": 0.40234375,
+      "grad_norm": 0.33984375,
       "learning_rate": 0.0001796613769264369,
-      "loss": 0.0246,
+      "loss": 0.0214,
       "step": 946
     },
     {
       "epoch": 2.0836083608360836,
-      "grad_norm": 0.291015625,
+      "grad_norm": 0.294921875,
       "learning_rate": 0.00017961943598420116,
-      "loss": 0.0226,
+      "loss": 0.0209,
       "step": 947
     },
     {
       "epoch": 2.0858085808580857,
-      "grad_norm": 0.443359375,
+      "grad_norm": 0.294921875,
       "learning_rate": 0.00017957745674869837,
-      "loss": 0.0242,
+      "loss": 0.0197,
       "step": 948
     },
     {
       "epoch": 2.088008800880088,
-      "grad_norm": 0.4140625,
+      "grad_norm": 0.275390625,
       "learning_rate": 0.00017953543924011854,
-      "loss": 0.0211,
+      "loss": 0.0152,
       "step": 949
     },
     {
       "epoch": 2.09020902090209,
-      "grad_norm": 0.447265625,
+      "grad_norm": 0.59375,
       "learning_rate": 0.00017949338347867023,
-      "loss": 0.0223,
+      "loss": 0.0242,
       "step": 950
     },
     {
       "epoch": 2.092409240924092,
-      "grad_norm": 0.51171875,
+      "grad_norm": 0.263671875,
       "learning_rate": 0.00017945128948458023,
-      "loss": 0.0431,
+      "loss": 0.0319,
       "step": 951
     },
     {
       "epoch": 2.094609460946095,
-      "grad_norm": 0.2099609375,
+      "grad_norm": 0.310546875,
       "learning_rate": 0.00017940915727809387,
-      "loss": 0.0224,
+      "loss": 0.022,
       "step": 952
     },
     {
       "epoch": 2.096809680968097,
-      "grad_norm": 0.302734375,
+      "grad_norm": 0.275390625,
       "learning_rate": 0.00017936698687947476,
-      "loss": 0.0209,
+      "loss": 0.0183,
       "step": 953
     },
     {
       "epoch": 2.099009900990099,
-      "grad_norm": 0.37109375,
+      "grad_norm": 0.193359375,
       "learning_rate": 0.00017932477830900494,
-      "loss": 0.0236,
+      "loss": 0.0194,
       "step": 954
     },
     {
       "epoch": 2.1012101210121013,
-      "grad_norm": 0.251953125,
+      "grad_norm": 0.306640625,
       "learning_rate": 0.00017928253158698473,
-      "loss": 0.0243,
+      "loss": 0.0267,
       "step": 955
     },
     {
       "epoch": 2.1034103410341034,
-      "grad_norm": 0.19921875,
+      "grad_norm": 0.4765625,
       "learning_rate": 0.0001792402467337329,
-      "loss": 0.0184,
+      "loss": 0.0212,
       "step": 956
     },
     {
       "epoch": 2.1056105610561056,
-      "grad_norm": 0.36328125,
+      "grad_norm": 0.32421875,
       "learning_rate": 0.00017919792376958649,
-      "loss": 0.0192,
+      "loss": 0.0148,
       "step": 957
     },
     {
       "epoch": 2.1078107810781077,
-      "grad_norm": 0.255859375,
+      "grad_norm": 0.357421875,
       "learning_rate": 0.0001791555627149009,
-      "loss": 0.0231,
+      "loss": 0.0277,
       "step": 958
     },
     {
       "epoch": 2.11001100110011,
-      "grad_norm": 0.34765625,
+      "grad_norm": 0.28515625,
       "learning_rate": 0.00017911316359004982,
-      "loss": 0.0231,
+      "loss": 0.0241,
       "step": 959
     },
     {
       "epoch": 2.112211221122112,
-      "grad_norm": 0.3359375,
+      "grad_norm": 0.2265625,
       "learning_rate": 0.00017907072641542527,
-      "loss": 0.0226,
+      "loss": 0.017,
       "step": 960
     },
     {
       "epoch": 2.1144114411441146,
-      "grad_norm": 0.2890625,
+      "grad_norm": 0.30859375,
       "learning_rate": 0.00017902825121143758,
-      "loss": 0.0181,
+      "loss": 0.0175,
       "step": 961
     },
     {
       "epoch": 2.1166116611661168,
-      "grad_norm": 0.279296875,
+      "grad_norm": 0.451171875,
       "learning_rate": 0.00017898573799851535,
-      "loss": 0.0225,
+      "loss": 0.0266,
       "step": 962
     },
     {
       "epoch": 2.118811881188119,
-      "grad_norm": 0.4296875,
+      "grad_norm": 0.330078125,
       "learning_rate": 0.00017894318679710546,
-      "loss": 0.0229,
+      "loss": 0.0173,
       "step": 963
     },
     {
       "epoch": 2.121012101210121,
-      "grad_norm": 0.41796875,
+      "grad_norm": 0.298828125,
       "learning_rate": 0.0001789005976276731,
-      "loss": 0.0212,
+      "loss": 0.0176,
       "step": 964
     },
     {
       "epoch": 2.1232123212321232,
-      "grad_norm": 0.396484375,
+      "grad_norm": 0.28515625,
       "learning_rate": 0.0001788579705107017,
-      "loss": 0.0165,
+      "loss": 0.0106,
       "step": 965
     },
     {
       "epoch": 2.1254125412541254,
-      "grad_norm": 0.2392578125,
+      "grad_norm": 0.291015625,
       "learning_rate": 0.00017881530546669288,
-      "loss": 0.009,
+      "loss": 0.0092,
       "step": 966
     },
     {
       "epoch": 2.1276127612761275,
-      "grad_norm": 0.30859375,
+      "grad_norm": 0.2431640625,
       "learning_rate": 0.0001787726025161666,
-      "loss": 0.0194,
+      "loss": 0.0206,
       "step": 967
     },
     {
       "epoch": 2.1298129812981297,
-      "grad_norm": 0.419921875,
+      "grad_norm": 0.24609375,
       "learning_rate": 0.00017872986167966098,
-      "loss": 0.0206,
+      "loss": 0.0168,
       "step": 968
     },
     {
       "epoch": 2.132013201320132,
-      "grad_norm": 0.306640625,
+      "grad_norm": 0.3125,
       "learning_rate": 0.00017868708297773237,
-      "loss": 0.0291,
+      "loss": 0.0222,
       "step": 969
     },
     {
       "epoch": 2.1342134213421344,
-      "grad_norm": 0.388671875,
+      "grad_norm": 0.302734375,
       "learning_rate": 0.0001786442664309554,
-      "loss": 0.0272,
+      "loss": 0.0289,
       "step": 970
     },
     {
       "epoch": 2.1364136413641366,
-      "grad_norm": 0.453125,
+      "grad_norm": 0.31640625,
       "learning_rate": 0.00017860141205992277,
-      "loss": 0.0291,
+      "loss": 0.0203,
       "step": 971
     },
     {
       "epoch": 2.1386138613861387,
-      "grad_norm": 0.236328125,
+      "grad_norm": 0.337890625,
       "learning_rate": 0.00017855851988524555,
-      "loss": 0.0107,
+      "loss": 0.015,
       "step": 972
     },
     {
       "epoch": 2.140814081408141,
-      "grad_norm": 0.384765625,
+      "grad_norm": 0.10888671875,
       "learning_rate": 0.00017851558992755285,
-      "loss": 0.0266,
+      "loss": 0.021,
       "step": 973
     },
     {
       "epoch": 2.143014301430143,
-      "grad_norm": 0.189453125,
+      "grad_norm": 0.1923828125,
       "learning_rate": 0.00017847262220749196,
-      "loss": 0.0229,
+      "loss": 0.0203,
       "step": 974
     },
     {
       "epoch": 2.145214521452145,
-      "grad_norm": 0.35546875,
+      "grad_norm": 0.1982421875,
       "learning_rate": 0.00017842961674572835,
-      "loss": 0.0281,
+      "loss": 0.0234,
       "step": 975
     },
     {
       "epoch": 2.1474147414741473,
-      "grad_norm": 0.275390625,
+      "grad_norm": 0.330078125,
       "learning_rate": 0.00017838657356294578,
-      "loss": 0.023,
+      "loss": 0.0228,
       "step": 976
     },
     {
       "epoch": 2.1496149614961495,
-      "grad_norm": 0.302734375,
+      "grad_norm": 0.287109375,
       "learning_rate": 0.00017834349267984592,
-      "loss": 0.0152,
+      "loss": 0.0132,
       "step": 977
     },
     {
       "epoch": 2.1518151815181517,
-      "grad_norm": 0.55859375,
+      "grad_norm": 0.1533203125,
       "learning_rate": 0.00017830037411714872,
-      "loss": 0.0316,
+      "loss": 0.0229,
       "step": 978
     },
     {
       "epoch": 2.1540154015401543,
-      "grad_norm": 0.1796875,
+      "grad_norm": 0.30078125,
       "learning_rate": 0.00017825721789559217,
-      "loss": 0.0128,
+      "loss": 0.0126,
       "step": 979
     },
     {
       "epoch": 2.1562156215621564,
-      "grad_norm": 0.2158203125,
+      "grad_norm": 0.255859375,
       "learning_rate": 0.0001782140240359325,
-      "loss": 0.0165,
+      "loss": 0.0293,
       "step": 980
     },
     {
       "epoch": 2.1584158415841586,
-      "grad_norm": 0.283203125,
+      "grad_norm": 0.2216796875,
       "learning_rate": 0.0001781707925589439,
-      "loss": 0.0256,
+      "loss": 0.0217,
       "step": 981
     },
     {
       "epoch": 2.1606160616061607,
-      "grad_norm": 0.2470703125,
+      "grad_norm": 0.32421875,
       "learning_rate": 0.00017812752348541873,
-      "loss": 0.0215,
+      "loss": 0.0235,
       "step": 982
     },
     {
       "epoch": 2.162816281628163,
-      "grad_norm": 0.298828125,
+      "grad_norm": 0.30859375,
       "learning_rate": 0.00017808421683616744,
-      "loss": 0.0176,
+      "loss": 0.014,
       "step": 983
     },
     {
       "epoch": 2.165016501650165,
-      "grad_norm": 0.369140625,
+      "grad_norm": 0.384765625,
       "learning_rate": 0.00017804087263201845,
-      "loss": 0.0132,
+      "loss": 0.0144,
       "step": 984
     },
     {
       "epoch": 2.167216721672167,
-      "grad_norm": 0.2734375,
+      "grad_norm": 0.287109375,
       "learning_rate": 0.0001779974908938184,
-      "loss": 0.0148,
+      "loss": 0.0132,
       "step": 985
     },
     {
       "epoch": 2.1694169416941693,
-      "grad_norm": 0.4609375,
+      "grad_norm": 0.2890625,
       "learning_rate": 0.00017795407164243184,
-      "loss": 0.0285,
+      "loss": 0.0189,
       "step": 986
     },
     {
       "epoch": 2.1716171617161715,
-      "grad_norm": 0.310546875,
+      "grad_norm": 0.376953125,
       "learning_rate": 0.00017791061489874144,
-      "loss": 0.0167,
+      "loss": 0.0244,
       "step": 987
     },
     {
       "epoch": 2.1738173817381736,
-      "grad_norm": 0.3359375,
+      "grad_norm": 0.33203125,
       "learning_rate": 0.00017786712068364793,
-      "loss": 0.0194,
+      "loss": 0.0153,
       "step": 988
     },
     {
       "epoch": 2.1760176017601762,
-      "grad_norm": 0.408203125,
+      "grad_norm": 0.203125,
       "learning_rate": 0.00017782358901806994,
-      "loss": 0.0119,
+      "loss": 0.0075,
       "step": 989
     },
     {
       "epoch": 2.1782178217821784,
-      "grad_norm": 0.35546875,
+      "grad_norm": 0.4375,
       "learning_rate": 0.00017778001992294426,
-      "loss": 0.0201,
+      "loss": 0.0251,
       "step": 990
     },
     {
       "epoch": 2.1804180418041805,
-      "grad_norm": 0.458984375,
+      "grad_norm": 0.267578125,
       "learning_rate": 0.00017773641341922556,
-      "loss": 0.0175,
+      "loss": 0.0143,
       "step": 991
     },
     {
       "epoch": 2.1826182618261827,
-      "grad_norm": 0.59765625,
+      "grad_norm": 0.30078125,
       "learning_rate": 0.00017769276952788655,
-      "loss": 0.0336,
+      "loss": 0.0202,
       "step": 992
     },
     {
       "epoch": 2.184818481848185,
-      "grad_norm": 0.25390625,
+      "grad_norm": 0.19140625,
       "learning_rate": 0.00017764908826991798,
-      "loss": 0.0152,
+      "loss": 0.0132,
       "step": 993
     },
     {
       "epoch": 2.187018701870187,
-      "grad_norm": 0.47265625,
+      "grad_norm": 0.2412109375,
       "learning_rate": 0.00017760536966632842,
-      "loss": 0.0266,
+      "loss": 0.0172,
       "step": 994
     },
     {
       "epoch": 2.189218921892189,
-      "grad_norm": 0.4375,
+      "grad_norm": 0.234375,
       "learning_rate": 0.0001775616137381446,
-      "loss": 0.0287,
+      "loss": 0.0231,
       "step": 995
     },
     {
       "epoch": 2.1914191419141913,
-      "grad_norm": 0.44140625,
+      "grad_norm": 0.271484375,
       "learning_rate": 0.00017751782050641105,
-      "loss": 0.0323,
+      "loss": 0.0166,
       "step": 996
     },
     {
       "epoch": 2.1936193619361934,
-      "grad_norm": 0.5234375,
+      "grad_norm": 0.341796875,
       "learning_rate": 0.00017747398999219026,
-      "loss": 0.0279,
+      "loss": 0.0137,
       "step": 997
     },
     {
       "epoch": 2.1958195819581956,
-      "grad_norm": 0.7421875,
+      "grad_norm": 0.6484375,
       "learning_rate": 0.00017743012221656276,
-      "loss": 0.0327,
+      "loss": 0.0368,
       "step": 998
     },
     {
       "epoch": 2.198019801980198,
-      "grad_norm": 0.169921875,
+      "grad_norm": 0.30078125,
       "learning_rate": 0.0001773862172006268,
-      "loss": 0.0166,
+      "loss": 0.0203,
       "step": 999
     },
     {
       "epoch": 2.2002200220022003,
-      "grad_norm": 0.2109375,
+      "grad_norm": 0.2578125,
       "learning_rate": 0.0001773422749654988,
-      "loss": 0.0149,
+      "loss": 0.0157,
       "step": 1000
     },
     {
       "epoch": 2.2024202420242025,
-      "grad_norm": 0.361328125,
+      "grad_norm": 0.62890625,
       "learning_rate": 0.00017729829553231286,
-      "loss": 0.0288,
+      "loss": 0.0328,
       "step": 1001
     },
     {
       "epoch": 2.2046204620462047,
-      "grad_norm": 0.3359375,
+      "grad_norm": 0.37109375,
       "learning_rate": 0.00017725427892222107,
-      "loss": 0.0236,
+      "loss": 0.0242,
       "step": 1002
     },
     {
       "epoch": 2.206820682068207,
-      "grad_norm": 0.341796875,
+      "grad_norm": 0.408203125,
       "learning_rate": 0.00017721022515639339,
-      "loss": 0.0273,
+      "loss": 0.0256,
       "step": 1003
     },
     {
       "epoch": 2.209020902090209,
-      "grad_norm": 0.26953125,
+      "grad_norm": 0.263671875,
       "learning_rate": 0.00017716613425601763,
-      "loss": 0.0299,
+      "loss": 0.0281,
       "step": 1004
     },
     {
       "epoch": 2.211221122112211,
-      "grad_norm": 0.259765625,
+      "grad_norm": 0.28515625,
       "learning_rate": 0.0001771220062422995,
-      "loss": 0.0219,
+      "loss": 0.0163,
       "step": 1005
     },
     {
       "epoch": 2.2134213421342133,
-      "grad_norm": 0.2314453125,
+      "grad_norm": 0.1552734375,
       "learning_rate": 0.0001770778411364625,
-      "loss": 0.0242,
+      "loss": 0.0188,
       "step": 1006
     },
     {
       "epoch": 2.2156215621562154,
-      "grad_norm": 0.19921875,
+      "grad_norm": 0.234375,
       "learning_rate": 0.00017703363895974804,
-      "loss": 0.0158,
+      "loss": 0.0185,
       "step": 1007
     },
     {
       "epoch": 2.217821782178218,
-      "grad_norm": 0.25390625,
+      "grad_norm": 0.19921875,
       "learning_rate": 0.0001769893997334153,
-      "loss": 0.0247,
+      "loss": 0.0193,
       "step": 1008
     },
     {
       "epoch": 2.22002200220022,
-      "grad_norm": 0.3125,
+      "grad_norm": 0.365234375,
       "learning_rate": 0.00017694512347874133,
-      "loss": 0.0249,
+      "loss": 0.0298,
       "step": 1009
     },
     {
       "epoch": 2.2222222222222223,
-      "grad_norm": 0.3515625,
+      "grad_norm": 0.1884765625,
       "learning_rate": 0.0001769008102170209,
-      "loss": 0.0227,
+      "loss": 0.022,
       "step": 1010
     },
     {
       "epoch": 2.2244224422442245,
-      "grad_norm": 0.275390625,
+      "grad_norm": 0.28125,
       "learning_rate": 0.0001768564599695667,
-      "loss": 0.0223,
+      "loss": 0.0219,
       "step": 1011
     },
     {
       "epoch": 2.2266226622662266,
-      "grad_norm": 0.365234375,
+      "grad_norm": 0.30078125,
       "learning_rate": 0.00017681207275770913,
-      "loss": 0.0252,
+      "loss": 0.0224,
       "step": 1012
     },
     {
       "epoch": 2.228822882288229,
-      "grad_norm": 0.326171875,
+      "grad_norm": 0.21484375,
       "learning_rate": 0.00017676764860279637,
-      "loss": 0.0095,
+      "loss": 0.0105,
       "step": 1013
     },
     {
       "epoch": 2.231023102310231,
-      "grad_norm": 0.279296875,
+      "grad_norm": 0.412109375,
       "learning_rate": 0.0001767231875261944,
-      "loss": 0.0247,
+      "loss": 0.0238,
       "step": 1014
     },
     {
       "epoch": 2.233223322332233,
-      "grad_norm": 0.25,
+      "grad_norm": 0.40625,
       "learning_rate": 0.00017667868954928694,
-      "loss": 0.0119,
+      "loss": 0.0233,
       "step": 1015
     },
     {
       "epoch": 2.2354235423542352,
-      "grad_norm": 0.2041015625,
+      "grad_norm": 0.32421875,
       "learning_rate": 0.00017663415469347548,
-      "loss": 0.0146,
+      "loss": 0.0147,
       "step": 1016
     },
     {
       "epoch": 2.237623762376238,
-      "grad_norm": 0.5,
+      "grad_norm": 0.26171875,
       "learning_rate": 0.0001765895829801792,
-      "loss": 0.0177,
+      "loss": 0.0224,
       "step": 1017
     },
     {
       "epoch": 2.23982398239824,
-      "grad_norm": 0.296875,
+      "grad_norm": 0.328125,
       "learning_rate": 0.00017654497443083503,
-      "loss": 0.0169,
+      "loss": 0.0269,
       "step": 1018
     },
     {
       "epoch": 2.242024202420242,
-      "grad_norm": 0.431640625,
+      "grad_norm": 0.4453125,
       "learning_rate": 0.00017650032906689763,
-      "loss": 0.0166,
+      "loss": 0.0136,
       "step": 1019
     },
     {
       "epoch": 2.2442244224422443,
-      "grad_norm": 0.482421875,
+      "grad_norm": 0.35546875,
       "learning_rate": 0.00017645564690983937,
-      "loss": 0.0282,
+      "loss": 0.0225,
       "step": 1020
     },
     {
       "epoch": 2.2464246424642464,
-      "grad_norm": 0.3515625,
+      "grad_norm": 0.33984375,
       "learning_rate": 0.00017641092798115023,
-      "loss": 0.0378,
+      "loss": 0.0328,
       "step": 1021
     },
     {
       "epoch": 2.2486248624862486,
-      "grad_norm": 0.251953125,
+      "grad_norm": 0.259765625,
       "learning_rate": 0.00017636617230233807,
-      "loss": 0.0203,
+      "loss": 0.0187,
       "step": 1022
     },
     {
       "epoch": 2.2508250825082508,
-      "grad_norm": 0.408203125,
+      "grad_norm": 0.388671875,
       "learning_rate": 0.0001763213798949282,
-      "loss": 0.0257,
+      "loss": 0.0233,
       "step": 1023
     },
     {
       "epoch": 2.253025302530253,
-      "grad_norm": 0.62890625,
+      "grad_norm": 0.18359375,
       "learning_rate": 0.00017627655078046375,
-      "loss": 0.0213,
+      "loss": 0.0126,
       "step": 1024
     },
     {
       "epoch": 2.255225522552255,
-      "grad_norm": 0.318359375,
+      "grad_norm": 0.373046875,
       "learning_rate": 0.00017623168498050542,
-      "loss": 0.0213,
+      "loss": 0.0409,
       "step": 1025
     },
     {
       "epoch": 2.2574257425742577,
-      "grad_norm": 0.318359375,
+      "grad_norm": 0.466796875,
       "learning_rate": 0.0001761867825166316,
-      "loss": 0.0161,
+      "loss": 0.0273,
       "step": 1026
     },
     {
       "epoch": 2.2574257425742577,
-      "eval_loss": 0.041678767651319504,
-      "eval_runtime": 10.5618,
-      "eval_samples_per_second": 36.263,
-      "eval_steps_per_second": 4.545,
+      "eval_loss": 0.04095015674829483,
+      "eval_runtime": 10.0798,
+      "eval_samples_per_second": 37.997,
+      "eval_steps_per_second": 4.762,
       "step": 1026
     },
     {
       "epoch": 2.25962596259626,
-      "grad_norm": 0.36328125,
+      "grad_norm": 0.3671875,
       "learning_rate": 0.0001761418434104383,
-      "loss": 0.0268,
+      "loss": 0.0248,
       "step": 1027
     },
     {
       "epoch": 2.261826182618262,
-      "grad_norm": 0.29296875,
+      "grad_norm": 0.349609375,
       "learning_rate": 0.00017609686768353913,
-      "loss": 0.0263,
+      "loss": 0.0225,
       "step": 1028
     },
     {
       "epoch": 2.264026402640264,
-      "grad_norm": 0.244140625,
+      "grad_norm": 0.302734375,
       "learning_rate": 0.00017605185535756536,
-      "loss": 0.0222,
+      "loss": 0.0278,
       "step": 1029
     },
     {
       "epoch": 2.2662266226622663,
-      "grad_norm": 0.2197265625,
+      "grad_norm": 0.2001953125,
       "learning_rate": 0.00017600680645416583,
-      "loss": 0.0165,
+      "loss": 0.0172,
       "step": 1030
     },
     {
       "epoch": 2.2684268426842684,
-      "grad_norm": 0.314453125,
+      "grad_norm": 0.31640625,
       "learning_rate": 0.00017596172099500697,
-      "loss": 0.0265,
+      "loss": 0.0197,
       "step": 1031
     },
     {
       "epoch": 2.2706270627062706,
-      "grad_norm": 0.4296875,
+      "grad_norm": 0.310546875,
       "learning_rate": 0.00017591659900177285,
-      "loss": 0.0254,
+      "loss": 0.018,
       "step": 1032
     },
     {
       "epoch": 2.2728272827282727,
-      "grad_norm": 0.3984375,
+      "grad_norm": 0.5,
       "learning_rate": 0.00017587144049616499,
-      "loss": 0.0229,
+      "loss": 0.0384,
       "step": 1033
     },
     {
       "epoch": 2.275027502750275,
-      "grad_norm": 0.291015625,
+      "grad_norm": 0.1826171875,
       "learning_rate": 0.0001758262454999026,
-      "loss": 0.028,
+      "loss": 0.0176,
       "step": 1034
     },
     {
       "epoch": 2.2772277227722775,
-      "grad_norm": 0.263671875,
+      "grad_norm": 0.373046875,
       "learning_rate": 0.00017578101403472235,
-      "loss": 0.014,
+      "loss": 0.0261,
       "step": 1035
     },
     {
       "epoch": 2.279427942794279,
-      "grad_norm": 0.421875,
+      "grad_norm": 0.298828125,
       "learning_rate": 0.00017573574612237848,
-      "loss": 0.022,
+      "loss": 0.021,
       "step": 1036
     },
     {
       "epoch": 2.281628162816282,
-      "grad_norm": 0.2255859375,
+      "grad_norm": 0.2353515625,
       "learning_rate": 0.0001756904417846428,
-      "loss": 0.0098,
+      "loss": 0.0141,
       "step": 1037
     },
     {
       "epoch": 2.283828382838284,
-      "grad_norm": 0.4453125,
+      "grad_norm": 0.421875,
       "learning_rate": 0.00017564510104330457,
-      "loss": 0.0299,
+      "loss": 0.0237,
       "step": 1038
     },
     {
       "epoch": 2.286028602860286,
-      "grad_norm": 0.357421875,
+      "grad_norm": 0.373046875,
       "learning_rate": 0.00017559972392017058,
-      "loss": 0.0314,
+      "loss": 0.0271,
       "step": 1039
     },
     {
       "epoch": 2.2882288228822882,
-      "grad_norm": 0.39453125,
+      "grad_norm": 0.29296875,
       "learning_rate": 0.00017555431043706515,
-      "loss": 0.0166,
+      "loss": 0.0172,
       "step": 1040
     },
     {
       "epoch": 2.2904290429042904,
-      "grad_norm": 0.1923828125,
+      "grad_norm": 0.2333984375,
       "learning_rate": 0.00017550886061583007,
-      "loss": 0.0122,
+      "loss": 0.0147,
       "step": 1041
     },
     {
       "epoch": 2.2926292629262925,
-      "grad_norm": 0.32421875,
+      "grad_norm": 0.369140625,
       "learning_rate": 0.0001754633744783246,
-      "loss": 0.0236,
+      "loss": 0.0307,
       "step": 1042
     },
     {
       "epoch": 2.2948294829482947,
-      "grad_norm": 0.1748046875,
+      "grad_norm": 0.458984375,
       "learning_rate": 0.00017541785204642544,
-      "loss": 0.0175,
+      "loss": 0.0234,
       "step": 1043
     },
     {
       "epoch": 2.297029702970297,
-      "grad_norm": 0.39453125,
+      "grad_norm": 0.2353515625,
       "learning_rate": 0.00017537229334202683,
-      "loss": 0.0246,
+      "loss": 0.0153,
       "step": 1044
     },
     {
       "epoch": 2.299229922992299,
-      "grad_norm": 0.56640625,
+      "grad_norm": 0.287109375,
       "learning_rate": 0.00017532669838704035,
-      "loss": 0.0362,
+      "loss": 0.0301,
       "step": 1045
     },
     {
       "epoch": 2.3014301430143016,
-      "grad_norm": 0.46875,
+      "grad_norm": 0.1767578125,
       "learning_rate": 0.00017528106720339506,
-      "loss": 0.0219,
+      "loss": 0.0133,
       "step": 1046
     },
     {
       "epoch": 2.3036303630363038,
-      "grad_norm": 0.2734375,
+      "grad_norm": 0.27734375,
       "learning_rate": 0.0001752353998130375,
-      "loss": 0.0179,
+      "loss": 0.0207,
       "step": 1047
     },
     {
       "epoch": 2.305830583058306,
-      "grad_norm": 0.3203125,
+      "grad_norm": 0.453125,
       "learning_rate": 0.00017518969623793153,
-      "loss": 0.026,
+      "loss": 0.0408,
       "step": 1048
     },
     {
       "epoch": 2.308030803080308,
-      "grad_norm": 0.376953125,
+      "grad_norm": 0.41015625,
       "learning_rate": 0.0001751439565000585,
-      "loss": 0.0211,
+      "loss": 0.0176,
       "step": 1049
     },
     {
       "epoch": 2.31023102310231,
-      "grad_norm": 0.32421875,
+      "grad_norm": 0.30859375,
       "learning_rate": 0.00017509818062141702,
-      "loss": 0.0199,
+      "loss": 0.029,
       "step": 1050
     },
     {
       "epoch": 2.3124312431243124,
-      "grad_norm": 0.1806640625,
+      "grad_norm": 0.2177734375,
       "learning_rate": 0.00017505236862402326,
-      "loss": 0.0155,
+      "loss": 0.0124,
       "step": 1051
     },
     {
       "epoch": 2.3146314631463145,
-      "grad_norm": 0.48046875,
+      "grad_norm": 0.32421875,
       "learning_rate": 0.0001750065205299107,
-      "loss": 0.0321,
+      "loss": 0.0297,
       "step": 1052
     },
     {
       "epoch": 2.3168316831683167,
-      "grad_norm": 0.2109375,
+      "grad_norm": 0.25390625,
       "learning_rate": 0.00017496063636113002,
-      "loss": 0.0161,
+      "loss": 0.0193,
       "step": 1053
     },
     {
       "epoch": 2.319031903190319,
-      "grad_norm": 0.1435546875,
+      "grad_norm": 0.2333984375,
       "learning_rate": 0.00017491471613974947,
-      "loss": 0.0165,
+      "loss": 0.0161,
       "step": 1054
     },
     {
       "epoch": 2.3212321232123214,
-      "grad_norm": 0.27734375,
+      "grad_norm": 0.259765625,
       "learning_rate": 0.00017486875988785455,
-      "loss": 0.0197,
+      "loss": 0.0134,
       "step": 1055
     },
     {
       "epoch": 2.3234323432343236,
-      "grad_norm": 0.2119140625,
+      "grad_norm": 0.36328125,
       "learning_rate": 0.0001748227676275481,
-      "loss": 0.0083,
+      "loss": 0.0087,
       "step": 1056
     },
     {
       "epoch": 2.3256325632563257,
-      "grad_norm": 0.306640625,
+      "grad_norm": 0.34375,
       "learning_rate": 0.00017477673938095026,
-      "loss": 0.0196,
+      "loss": 0.0214,
       "step": 1057
     },
     {
       "epoch": 2.327832783278328,
-      "grad_norm": 0.189453125,
+      "grad_norm": 0.462890625,
       "learning_rate": 0.00017473067517019845,
-      "loss": 0.0088,
+      "loss": 0.0185,
       "step": 1058
     },
     {
       "epoch": 2.33003300330033,
-      "grad_norm": 0.376953125,
+      "grad_norm": 0.416015625,
       "learning_rate": 0.00017468457501744749,
-      "loss": 0.0178,
+      "loss": 0.0309,
       "step": 1059
     },
     {
       "epoch": 2.332233223322332,
-      "grad_norm": 0.2294921875,
+      "grad_norm": 0.37109375,
       "learning_rate": 0.00017463843894486937,
-      "loss": 0.0131,
+      "loss": 0.0181,
       "step": 1060
     },
     {
       "epoch": 2.3344334433443343,
-      "grad_norm": 0.474609375,
+      "grad_norm": 0.2158203125,
       "learning_rate": 0.00017459226697465346,
-      "loss": 0.0171,
+      "loss": 0.0118,
       "step": 1061
     },
     {
       "epoch": 2.3366336633663365,
-      "grad_norm": 0.2470703125,
+      "grad_norm": 0.298828125,
       "learning_rate": 0.00017454605912900628,
-      "loss": 0.0139,
+      "loss": 0.0164,
       "step": 1062
     },
     {
       "epoch": 2.3388338833883386,
-      "grad_norm": 0.271484375,
+      "grad_norm": 0.453125,
       "learning_rate": 0.0001744998154301517,
-      "loss": 0.0205,
+      "loss": 0.028,
       "step": 1063
     },
     {
       "epoch": 2.3410341034103412,
-      "grad_norm": 0.43359375,
+      "grad_norm": 0.279296875,
       "learning_rate": 0.0001744535359003308,
-      "loss": 0.0191,
+      "loss": 0.0165,
       "step": 1064
     },
     {
       "epoch": 2.3432343234323434,
-      "grad_norm": 0.3828125,
+      "grad_norm": 0.30859375,
       "learning_rate": 0.0001744072205618019,
-      "loss": 0.019,
+      "loss": 0.0163,
       "step": 1065
     },
     {
       "epoch": 2.3454345434543455,
-      "grad_norm": 0.2890625,
+      "grad_norm": 0.51953125,
       "learning_rate": 0.0001743608694368405,
-      "loss": 0.0191,
+      "loss": 0.0267,
       "step": 1066
     },
     {
       "epoch": 2.3476347634763477,
-      "grad_norm": 0.248046875,
+      "grad_norm": 0.373046875,
       "learning_rate": 0.00017431448254773944,
-      "loss": 0.0121,
+      "loss": 0.0184,
       "step": 1067
     },
     {
       "epoch": 2.34983498349835,
-      "grad_norm": 0.2890625,
+      "grad_norm": 0.337890625,
       "learning_rate": 0.00017426805991680857,
-      "loss": 0.021,
+      "loss": 0.0225,
       "step": 1068
     },
     {
       "epoch": 2.352035203520352,
-      "grad_norm": 0.306640625,
+      "grad_norm": 0.4453125,
       "learning_rate": 0.00017422160156637507,
-      "loss": 0.0208,
+      "loss": 0.0379,
       "step": 1069
     },
     {
       "epoch": 2.354235423542354,
       "grad_norm": 0.353515625,
       "learning_rate": 0.00017417510751878323,
-      "loss": 0.0136,
+      "loss": 0.0134,
       "step": 1070
     },
     {
       "epoch": 2.3564356435643563,
-      "grad_norm": 0.365234375,
+      "grad_norm": 0.373046875,
       "learning_rate": 0.0001741285777963946,
-      "loss": 0.0241,
+      "loss": 0.0169,
       "step": 1071
     },
     {
       "epoch": 2.3586358635863585,
-      "grad_norm": 0.490234375,
+      "grad_norm": 0.3203125,
       "learning_rate": 0.00017408201242158776,
-      "loss": 0.0133,
+      "loss": 0.0153,
       "step": 1072
     },
     {
       "epoch": 2.360836083608361,
-      "grad_norm": 0.275390625,
+      "grad_norm": 0.201171875,
       "learning_rate": 0.0001740354114167585,
-      "loss": 0.0089,
+      "loss": 0.0076,
       "step": 1073
     },
     {
       "epoch": 2.363036303630363,
-      "grad_norm": 0.310546875,
+      "grad_norm": 0.3125,
       "learning_rate": 0.0001739887748043198,
-      "loss": 0.0241,
+      "loss": 0.0204,
       "step": 1074
     },
     {
       "epoch": 2.3652365236523654,
-      "grad_norm": 0.2138671875,
+      "grad_norm": 0.26171875,
       "learning_rate": 0.0001739421026067017,
-      "loss": 0.012,
+      "loss": 0.0169,
       "step": 1075
     },
     {
       "epoch": 2.3674367436743675,
-      "grad_norm": 0.1875,
+      "grad_norm": 0.1884765625,
       "learning_rate": 0.00017389539484635136,
-      "loss": 0.0151,
+      "loss": 0.0118,
       "step": 1076
     },
     {
       "epoch": 2.3696369636963697,
-      "grad_norm": 0.26953125,
+      "grad_norm": 0.2451171875,
       "learning_rate": 0.00017384865154573302,
-      "loss": 0.0118,
+      "loss": 0.0161,
       "step": 1077
     },
     {
       "epoch": 2.371837183718372,
-      "grad_norm": 0.251953125,
+      "grad_norm": 0.224609375,
       "learning_rate": 0.0001738018727273281,
-      "loss": 0.0164,
+      "loss": 0.0156,
       "step": 1078
     },
     {
       "epoch": 2.374037403740374,
-      "grad_norm": 0.32421875,
+      "grad_norm": 0.375,
       "learning_rate": 0.00017375505841363503,
-      "loss": 0.0255,
+      "loss": 0.0326,
       "step": 1079
     },
     {
       "epoch": 2.376237623762376,
-      "grad_norm": 0.1962890625,
+      "grad_norm": 0.44140625,
       "learning_rate": 0.0001737082086271693,
-      "loss": 0.0195,
+      "loss": 0.0321,
       "step": 1080
     },
     {
       "epoch": 2.3784378437843783,
-      "grad_norm": 0.3984375,
+      "grad_norm": 0.5078125,
       "learning_rate": 0.00017366132339046355,
-      "loss": 0.033,
+      "loss": 0.0403,
       "step": 1081
     },
     {
       "epoch": 2.380638063806381,
-      "grad_norm": 0.162109375,
+      "grad_norm": 0.26953125,
       "learning_rate": 0.00017361440272606736,
-      "loss": 0.0107,
+      "loss": 0.0093,
       "step": 1082
     },
     {
       "epoch": 2.382838283828383,
-      "grad_norm": 0.310546875,
+      "grad_norm": 0.478515625,
       "learning_rate": 0.00017356744665654743,
-      "loss": 0.0171,
+      "loss": 0.0261,
       "step": 1083
     },
     {
       "epoch": 2.385038503850385,
-      "grad_norm": 0.423828125,
+      "grad_norm": 0.3203125,
       "learning_rate": 0.00017352045520448742,
-      "loss": 0.0234,
+      "loss": 0.0279,
       "step": 1084
     },
     {
       "epoch": 2.3872387238723873,
-      "grad_norm": 0.345703125,
+      "grad_norm": 0.396484375,
       "learning_rate": 0.0001734734283924881,
-      "loss": 0.0236,
+      "loss": 0.0243,
       "step": 1085
     },
     {
       "epoch": 2.3894389438943895,
-      "grad_norm": 0.314453125,
+      "grad_norm": 0.30859375,
       "learning_rate": 0.00017342636624316716,
-      "loss": 0.013,
+      "loss": 0.0163,
       "step": 1086
     },
     {
       "epoch": 2.3916391639163916,
-      "grad_norm": 0.333984375,
+      "grad_norm": 0.296875,
       "learning_rate": 0.0001733792687791593,
-      "loss": 0.0309,
+      "loss": 0.0327,
       "step": 1087
     },
     {
       "epoch": 2.393839383938394,
-      "grad_norm": 0.294921875,
+      "grad_norm": 0.26171875,
       "learning_rate": 0.00017333213602311627,
-      "loss": 0.0255,
+      "loss": 0.0212,
       "step": 1088
     },
     {
       "epoch": 2.396039603960396,
-      "grad_norm": 0.408203125,
+      "grad_norm": 0.353515625,
       "learning_rate": 0.0001732849679977067,
-      "loss": 0.0262,
+      "loss": 0.0188,
       "step": 1089
     },
     {
       "epoch": 2.398239823982398,
-      "grad_norm": 0.36328125,
+      "grad_norm": 0.20703125,
       "learning_rate": 0.00017323776472561627,
-      "loss": 0.0167,
+      "loss": 0.0163,
       "step": 1090
     },
     {
       "epoch": 2.4004400440044003,
-      "grad_norm": 0.345703125,
+      "grad_norm": 0.37109375,
       "learning_rate": 0.00017319052622954753,
-      "loss": 0.0169,
+      "loss": 0.0203,
       "step": 1091
     },
     {
       "epoch": 2.4026402640264024,
-      "grad_norm": 0.25,
+      "grad_norm": 0.3515625,
       "learning_rate": 0.00017314325253222002,
-      "loss": 0.0301,
+      "loss": 0.0331,
       "step": 1092
     },
     {
       "epoch": 2.404840484048405,
-      "grad_norm": 0.36328125,
+      "grad_norm": 0.1796875,
       "learning_rate": 0.00017309594365637026,
-      "loss": 0.0288,
+      "loss": 0.0256,
       "step": 1093
     },
     {
       "epoch": 2.407040704070407,
-      "grad_norm": 0.40625,
+      "grad_norm": 0.21875,
       "learning_rate": 0.00017304859962475152,
-      "loss": 0.0214,
+      "loss": 0.0168,
       "step": 1094
     },
     {
       "epoch": 2.4092409240924093,
-      "grad_norm": 0.2373046875,
+      "grad_norm": 0.28125,
       "learning_rate": 0.00017300122046013422,
-      "loss": 0.0205,
+      "loss": 0.0186,
       "step": 1095
     },
     {
       "epoch": 2.4114411441144115,
-      "grad_norm": 0.267578125,
+      "grad_norm": 0.470703125,
       "learning_rate": 0.00017295380618530547,
-      "loss": 0.0228,
+      "loss": 0.0331,
       "step": 1096
     },
     {
       "epoch": 2.4136413641364136,
-      "grad_norm": 0.154296875,
+      "grad_norm": 0.279296875,
       "learning_rate": 0.00017290635682306932,
-      "loss": 0.0105,
+      "loss": 0.0115,
       "step": 1097
     },
     {
       "epoch": 2.4158415841584158,
-      "grad_norm": 0.255859375,
+      "grad_norm": 0.458984375,
       "learning_rate": 0.00017285887239624679,
-      "loss": 0.0175,
+      "loss": 0.0298,
       "step": 1098
     },
     {
       "epoch": 2.418041804180418,
-      "grad_norm": 0.48046875,
+      "grad_norm": 0.259765625,
       "learning_rate": 0.00017281135292767565,
-      "loss": 0.0292,
+      "loss": 0.0151,
       "step": 1099
     },
     {
       "epoch": 2.42024202420242,
-      "grad_norm": 0.31640625,
+      "grad_norm": 0.25,
       "learning_rate": 0.00017276379844021058,
-      "loss": 0.0251,
+      "loss": 0.0186,
       "step": 1100
     },
     {
       "epoch": 2.4224422442244222,
-      "grad_norm": 0.357421875,
+      "grad_norm": 0.3828125,
       "learning_rate": 0.0001727162089567231,
-      "loss": 0.0158,
+      "loss": 0.0133,
       "step": 1101
     },
     {
       "epoch": 2.424642464246425,
-      "grad_norm": 0.2109375,
+      "grad_norm": 0.28125,
       "learning_rate": 0.00017266858450010154,
-      "loss": 0.0119,
+      "loss": 0.0162,
       "step": 1102
     },
     {
       "epoch": 2.426842684268427,
-      "grad_norm": 0.412109375,
+      "grad_norm": 0.380859375,
       "learning_rate": 0.000172620925093251,
-      "loss": 0.0242,
+      "loss": 0.0219,
       "step": 1103
     },
     {
       "epoch": 2.429042904290429,
-      "grad_norm": 0.5546875,
+      "grad_norm": 0.42578125,
       "learning_rate": 0.00017257323075909359,
-      "loss": 0.0236,
+      "loss": 0.0109,
       "step": 1104
     },
     {
       "epoch": 2.4312431243124313,
-      "grad_norm": 0.373046875,
+      "grad_norm": 0.330078125,
       "learning_rate": 0.00017252550152056795,
-      "loss": 0.0162,
+      "loss": 0.0139,
       "step": 1105
     },
     {
       "epoch": 2.4334433443344334,
-      "grad_norm": 0.189453125,
+      "grad_norm": 0.2431640625,
       "learning_rate": 0.00017247773740062972,
-      "loss": 0.0156,
+      "loss": 0.0147,
       "step": 1106
     },
     {
       "epoch": 2.4356435643564356,
-      "grad_norm": 0.310546875,
+      "grad_norm": 0.54296875,
       "learning_rate": 0.00017242993842225116,
-      "loss": 0.0194,
+      "loss": 0.0182,
       "step": 1107
     },
     {
       "epoch": 2.4378437843784377,
-      "grad_norm": 0.28125,
+      "grad_norm": 0.515625,
       "learning_rate": 0.00017238210460842143,
-      "loss": 0.0137,
+      "loss": 0.015,
       "step": 1108
     },
     {
       "epoch": 2.44004400440044,
-      "grad_norm": 0.373046875,
+      "grad_norm": 0.48046875,
       "learning_rate": 0.00017233423598214635,
-      "loss": 0.0351,
+      "loss": 0.0267,
       "step": 1109
     },
     {
       "epoch": 2.442244224422442,
-      "grad_norm": 0.33984375,
+      "grad_norm": 0.427734375,
       "learning_rate": 0.00017228633256644854,
-      "loss": 0.0165,
+      "loss": 0.0211,
       "step": 1110
     },
     {
       "epoch": 2.4444444444444446,
-      "grad_norm": 0.375,
+      "grad_norm": 0.294921875,
       "learning_rate": 0.00017223839438436727,
-      "loss": 0.0165,
+      "loss": 0.0208,
       "step": 1111
     },
     {
       "epoch": 2.446644664466447,
-      "grad_norm": 0.255859375,
+      "grad_norm": 0.24609375,
       "learning_rate": 0.0001721904214589587,
-      "loss": 0.0167,
+      "loss": 0.0161,
       "step": 1112
     },
     {
       "epoch": 2.448844884488449,
-      "grad_norm": 0.1884765625,
+      "grad_norm": 0.375,
       "learning_rate": 0.0001721424138132955,
-      "loss": 0.0206,
+      "loss": 0.0288,
       "step": 1113
     },
     {
       "epoch": 2.451045104510451,
-      "grad_norm": 0.2578125,
+      "grad_norm": 0.578125,
       "learning_rate": 0.00017209437147046715,
-      "loss": 0.0268,
+      "loss": 0.0267,
       "step": 1114
     },
     {
       "epoch": 2.4532453245324533,
-      "grad_norm": 0.2314453125,
+      "grad_norm": 0.3046875,
       "learning_rate": 0.0001720462944535798,
-      "loss": 0.0269,
+      "loss": 0.0274,
       "step": 1115
     },
     {
       "epoch": 2.4554455445544554,
-      "grad_norm": 0.294921875,
+      "grad_norm": 0.39453125,
       "learning_rate": 0.00017199818278575623,
-      "loss": 0.0256,
+      "loss": 0.0229,
       "step": 1116
     },
     {
       "epoch": 2.4576457645764576,
-      "grad_norm": 0.390625,
+      "grad_norm": 0.25390625,
       "learning_rate": 0.00017195003649013598,
-      "loss": 0.0153,
+      "loss": 0.0148,
       "step": 1117
     },
     {
       "epoch": 2.4598459845984597,
-      "grad_norm": 0.23046875,
+      "grad_norm": 0.427734375,
       "learning_rate": 0.0001719018555898752,
-      "loss": 0.015,
+      "loss": 0.0178,
       "step": 1118
     },
     {
       "epoch": 2.462046204620462,
-      "grad_norm": 0.3046875,
+      "grad_norm": 0.62109375,
       "learning_rate": 0.0001718536401081466,
-      "loss": 0.0237,
+      "loss": 0.0346,
       "step": 1119
     },
     {
       "epoch": 2.4642464246424645,
-      "grad_norm": 0.443359375,
+      "grad_norm": 0.53515625,
       "learning_rate": 0.0001718053900681397,
-      "loss": 0.0298,
+      "loss": 0.0256,
       "step": 1120
     },
     {
       "epoch": 2.4664466446644666,
-      "grad_norm": 0.375,
+      "grad_norm": 0.28125,
       "learning_rate": 0.00017175710549306047,
-      "loss": 0.0273,
+      "loss": 0.0245,
       "step": 1121
     },
     {
       "epoch": 2.4686468646864688,
-      "grad_norm": 0.173828125,
+      "grad_norm": 0.287109375,
       "learning_rate": 0.00017170878640613158,
-      "loss": 0.0154,
+      "loss": 0.0149,
       "step": 1122
     },
     {
       "epoch": 2.470847084708471,
-      "grad_norm": 0.52734375,
+      "grad_norm": 0.412109375,
       "learning_rate": 0.00017166043283059227,
-      "loss": 0.0266,
+      "loss": 0.0309,
       "step": 1123
     },
     {
       "epoch": 2.473047304730473,
       "grad_norm": 0.2578125,
       "learning_rate": 0.00017161204478969837,
-      "loss": 0.0137,
+      "loss": 0.0192,
       "step": 1124
     },
     {
       "epoch": 2.4752475247524752,
-      "grad_norm": 0.33984375,
+      "grad_norm": 0.2578125,
       "learning_rate": 0.0001715636223067223,
-      "loss": 0.021,
+      "loss": 0.0185,
       "step": 1125
     },
     {
       "epoch": 2.4774477447744774,
-      "grad_norm": 0.30078125,
+      "grad_norm": 0.357421875,
       "learning_rate": 0.00017151516540495304,
-      "loss": 0.0172,
+      "loss": 0.0197,
       "step": 1126
     },
     {
       "epoch": 2.4796479647964795,
-      "grad_norm": 0.4140625,
+      "grad_norm": 0.359375,
       "learning_rate": 0.0001714666741076961,
-      "loss": 0.0267,
+      "loss": 0.0256,
       "step": 1127
     },
     {
       "epoch": 2.4818481848184817,
-      "grad_norm": 0.330078125,
+      "grad_norm": 0.396484375,
       "learning_rate": 0.00017141814843827353,
-      "loss": 0.0166,
+      "loss": 0.0182,
       "step": 1128
     },
     {
       "epoch": 2.4840484048404843,
-      "grad_norm": 0.306640625,
+      "grad_norm": 0.341796875,
       "learning_rate": 0.00017136958842002401,
-      "loss": 0.0161,
+      "loss": 0.0177,
       "step": 1129
     },
     {
       "epoch": 2.4862486248624864,
-      "grad_norm": 0.18359375,
+      "grad_norm": 0.314453125,
       "learning_rate": 0.0001713209940763026,
-      "loss": 0.0141,
+      "loss": 0.0199,
       "step": 1130
     },
     {
       "epoch": 2.4884488448844886,
-      "grad_norm": 0.53515625,
+      "grad_norm": 0.52734375,
       "learning_rate": 0.00017127236543048094,
-      "loss": 0.0242,
+      "loss": 0.0253,
       "step": 1131
     },
     {
       "epoch": 2.4906490649064907,
-      "grad_norm": 0.373046875,
+      "grad_norm": 0.357421875,
       "learning_rate": 0.00017122370250594718,
-      "loss": 0.0177,
+      "loss": 0.021,
       "step": 1132
     },
     {
       "epoch": 2.492849284928493,
-      "grad_norm": 0.3046875,
+      "grad_norm": 0.40625,
       "learning_rate": 0.00017117500532610593,
-      "loss": 0.0138,
+      "loss": 0.0171,
       "step": 1133
     },
     {
       "epoch": 2.495049504950495,
-      "grad_norm": 0.314453125,
+      "grad_norm": 0.380859375,
       "learning_rate": 0.00017112627391437828,
-      "loss": 0.0252,
+      "loss": 0.0276,
       "step": 1134
     },
     {
       "epoch": 2.497249724972497,
-      "grad_norm": 0.220703125,
+      "grad_norm": 0.314453125,
       "learning_rate": 0.00017107750829420176,
-      "loss": 0.0212,
+      "loss": 0.0211,
       "step": 1135
     },
     {
       "epoch": 2.4994499449944994,
-      "grad_norm": 0.44140625,
+      "grad_norm": 0.39453125,
       "learning_rate": 0.00017102870848903043,
-      "loss": 0.0236,
+      "loss": 0.034,
       "step": 1136
     },
     {
       "epoch": 2.5016501650165015,
-      "grad_norm": 0.255859375,
+      "grad_norm": 0.23828125,
       "learning_rate": 0.00017097987452233473,
-      "loss": 0.0231,
+      "loss": 0.0223,
       "step": 1137
     },
     {
       "epoch": 2.503850385038504,
-      "grad_norm": 0.1962890625,
+      "grad_norm": 0.43359375,
       "learning_rate": 0.00017093100641760146,
-      "loss": 0.011,
+      "loss": 0.0254,
       "step": 1138
     },
     {
       "epoch": 2.506050605060506,
-      "grad_norm": 0.3359375,
+      "grad_norm": 0.294921875,
       "learning_rate": 0.00017088210419833404,
-      "loss": 0.0202,
+      "loss": 0.0166,
       "step": 1139
     },
     {
       "epoch": 2.5082508250825084,
-      "grad_norm": 0.34765625,
+      "grad_norm": 0.4296875,
       "learning_rate": 0.0001708331678880521,
-      "loss": 0.0203,
+      "loss": 0.0296,
       "step": 1140
     },
     {
       "epoch": 2.5082508250825084,
-      "eval_loss": 0.03608192875981331,
-      "eval_runtime": 10.593,
-      "eval_samples_per_second": 36.156,
-      "eval_steps_per_second": 4.531,
+      "eval_loss": 0.038366567343473434,
+      "eval_runtime": 10.151,
+      "eval_samples_per_second": 37.73,
+      "eval_steps_per_second": 4.729,
       "step": 1140
     },
     {
       "epoch": 2.5104510451045106,
-      "grad_norm": 0.26171875,
+      "grad_norm": 0.3125,
       "learning_rate": 0.00017078419751029173,
-      "loss": 0.0146,
+      "loss": 0.0155,
       "step": 1141
     },
     {
       "epoch": 2.5126512651265127,
-      "grad_norm": 0.4140625,
+      "grad_norm": 0.396484375,
       "learning_rate": 0.00017073519308860547,
       "loss": 0.0252,
       "step": 1142
     },
     {
       "epoch": 2.514851485148515,
-      "grad_norm": 0.328125,
+      "grad_norm": 0.484375,
       "learning_rate": 0.00017068615464656216,
-      "loss": 0.0101,
+      "loss": 0.0169,
       "step": 1143
     },
     {
       "epoch": 2.517051705170517,
-      "grad_norm": 0.2578125,
+      "grad_norm": 0.296875,
       "learning_rate": 0.00017063708220774702,
-      "loss": 0.0147,
+      "loss": 0.0146,
       "step": 1144
     },
     {
       "epoch": 2.519251925192519,
-      "grad_norm": 0.34375,
+      "grad_norm": 0.267578125,
       "learning_rate": 0.00017058797579576161,
-      "loss": 0.0214,
+      "loss": 0.0157,
       "step": 1145
     },
     {
       "epoch": 2.5214521452145213,
-      "grad_norm": 0.36328125,
+      "grad_norm": 0.3203125,
       "learning_rate": 0.0001705388354342239,
-      "loss": 0.0201,
+      "loss": 0.0167,
       "step": 1146
     },
     {
       "epoch": 2.523652365236524,
-      "grad_norm": 0.1865234375,
+      "grad_norm": 0.28125,
       "learning_rate": 0.00017048966114676804,
-      "loss": 0.0135,
+      "loss": 0.0175,
       "step": 1147
     },
     {
       "epoch": 2.5258525852585256,
-      "grad_norm": 0.294921875,
+      "grad_norm": 0.345703125,
       "learning_rate": 0.00017044045295704468,
-      "loss": 0.0161,
+      "loss": 0.0252,
       "step": 1148
     },
     {
       "epoch": 2.5280528052805282,
-      "grad_norm": 0.439453125,
+      "grad_norm": 0.296875,
       "learning_rate": 0.00017039121088872062,
-      "loss": 0.0209,
+      "loss": 0.0179,
       "step": 1149
     },
     {
       "epoch": 2.5302530253025304,
-      "grad_norm": 0.310546875,
+      "grad_norm": 0.3203125,
       "learning_rate": 0.00017034193496547902,
-      "loss": 0.0088,
+      "loss": 0.0142,
       "step": 1150
     },
     {
       "epoch": 2.5324532453245325,
-      "grad_norm": 0.30859375,
+      "grad_norm": 0.26953125,
       "learning_rate": 0.00017029262521101935,
-      "loss": 0.0202,
+      "loss": 0.0174,
       "step": 1151
     },
     {
       "epoch": 2.5346534653465347,
-      "grad_norm": 0.1357421875,
+      "grad_norm": 0.251953125,
       "learning_rate": 0.00017024328164905734,
-      "loss": 0.0108,
+      "loss": 0.0208,
       "step": 1152
     },
     {
       "epoch": 2.536853685368537,
-      "grad_norm": 0.337890625,
+      "grad_norm": 0.318359375,
       "learning_rate": 0.00017019390430332495,
-      "loss": 0.0277,
+      "loss": 0.0248,
       "step": 1153
     },
     {
       "epoch": 2.539053905390539,
-      "grad_norm": 0.185546875,
+      "grad_norm": 0.2890625,
       "learning_rate": 0.0001701444931975703,
-      "loss": 0.0154,
+      "loss": 0.0241,
       "step": 1154
     },
     {
       "epoch": 2.541254125412541,
-      "grad_norm": 0.2392578125,
+      "grad_norm": 0.1865234375,
       "learning_rate": 0.000170095048355558,
-      "loss": 0.0115,
+      "loss": 0.0133,
       "step": 1155
     },
     {
       "epoch": 2.5434543454345433,
-      "grad_norm": 0.400390625,
+      "grad_norm": 0.36328125,
       "learning_rate": 0.00017004556980106865,
-      "loss": 0.0209,
+      "loss": 0.0231,
       "step": 1156
     },
     {
       "epoch": 2.5456545654565454,
-      "grad_norm": 0.24609375,
+      "grad_norm": 0.2275390625,
       "learning_rate": 0.00016999605755789912,
-      "loss": 0.0179,
+      "loss": 0.0185,
       "step": 1157
     },
     {
       "epoch": 2.547854785478548,
-      "grad_norm": 0.341796875,
+      "grad_norm": 0.2421875,
       "learning_rate": 0.00016994651164986262,
-      "loss": 0.0237,
+      "loss": 0.0192,
       "step": 1158
     },
     {
       "epoch": 2.55005500550055,
-      "grad_norm": 0.205078125,
+      "grad_norm": 0.259765625,
       "learning_rate": 0.00016989693210078835,
-      "loss": 0.019,
+      "loss": 0.0175,
       "step": 1159
     },
     {
       "epoch": 2.5522552255225524,
-      "grad_norm": 0.2021484375,
+      "grad_norm": 0.4296875,
       "learning_rate": 0.00016984731893452174,
-      "loss": 0.013,
+      "loss": 0.0203,
       "step": 1160
     },
     {
       "epoch": 2.5544554455445545,
-      "grad_norm": 0.298828125,
+      "grad_norm": 0.46484375,
       "learning_rate": 0.00016979767217492454,
-      "loss": 0.0244,
+      "loss": 0.0317,
       "step": 1161
     },
     {
       "epoch": 2.5566556655665567,
-      "grad_norm": 0.359375,
+      "grad_norm": 0.419921875,
       "learning_rate": 0.00016974799184587451,
-      "loss": 0.0173,
+      "loss": 0.0277,
       "step": 1162
     },
     {
       "epoch": 2.558855885588559,
-      "grad_norm": 0.224609375,
+      "grad_norm": 0.283203125,
       "learning_rate": 0.00016969827797126555,
-      "loss": 0.0125,
+      "loss": 0.0137,
       "step": 1163
     },
     {
       "epoch": 2.561056105610561,
-      "grad_norm": 0.396484375,
+      "grad_norm": 0.2890625,
       "learning_rate": 0.00016964853057500778,
-      "loss": 0.0212,
+      "loss": 0.0195,
       "step": 1164
     },
     {
       "epoch": 2.563256325632563,
-      "grad_norm": 0.2392578125,
+      "grad_norm": 0.2578125,
       "learning_rate": 0.00016959874968102735,
-      "loss": 0.009,
+      "loss": 0.0079,
       "step": 1165
     },
     {
       "epoch": 2.5654565456545653,
-      "grad_norm": 0.23046875,
+      "grad_norm": 0.322265625,
       "learning_rate": 0.0001695489353132666,
-      "loss": 0.0167,
+      "loss": 0.0241,
       "step": 1166
     },
     {
       "epoch": 2.567656765676568,
-      "grad_norm": 0.2734375,
+      "grad_norm": 0.359375,
       "learning_rate": 0.00016949908749568397,
-      "loss": 0.0177,
+      "loss": 0.023,
       "step": 1167
     },
     {
       "epoch": 2.56985698569857,
-      "grad_norm": 0.220703125,
+      "grad_norm": 0.455078125,
       "learning_rate": 0.00016944920625225392,
-      "loss": 0.0135,
+      "loss": 0.0301,
       "step": 1168
     },
     {
       "epoch": 2.572057205720572,
-      "grad_norm": 0.3125,
+      "grad_norm": 0.5390625,
       "learning_rate": 0.000169399291606967,
-      "loss": 0.0163,
+      "loss": 0.0209,
       "step": 1169
     },
     {
       "epoch": 2.5742574257425743,
-      "grad_norm": 0.431640625,
+      "grad_norm": 0.30078125,
       "learning_rate": 0.00016934934358382985,
-      "loss": 0.0245,
+      "loss": 0.0181,
       "step": 1170
     },
     {
       "epoch": 2.5764576457645765,
-      "grad_norm": 0.41796875,
+      "grad_norm": 0.353515625,
       "learning_rate": 0.0001692993622068652,
-      "loss": 0.0287,
+      "loss": 0.0282,
       "step": 1171
     },
     {
       "epoch": 2.5786578657865786,
-      "grad_norm": 0.263671875,
+      "grad_norm": 0.287109375,
       "learning_rate": 0.00016924934750011175,
-      "loss": 0.0115,
+      "loss": 0.0167,
       "step": 1172
     },
     {
       "epoch": 2.580858085808581,
-      "grad_norm": 0.296875,
+      "grad_norm": 0.5078125,
       "learning_rate": 0.00016919929948762427,
-      "loss": 0.0117,
+      "loss": 0.024,
       "step": 1173
     },
     {
       "epoch": 2.583058305830583,
-      "grad_norm": 0.1767578125,
+      "grad_norm": 0.23828125,
       "learning_rate": 0.00016914921819347355,
-      "loss": 0.0112,
+      "loss": 0.0157,
       "step": 1174
     },
     {
       "epoch": 2.585258525852585,
-      "grad_norm": 0.439453125,
+      "grad_norm": 0.447265625,
       "learning_rate": 0.00016909910364174636,
-      "loss": 0.0236,
+      "loss": 0.0195,
       "step": 1175
     },
     {
       "epoch": 2.5874587458745877,
-      "grad_norm": 0.55078125,
+      "grad_norm": 0.29296875,
       "learning_rate": 0.00016904895585654548,
-      "loss": 0.0404,
+      "loss": 0.0213,
       "step": 1176
     },
     {
       "epoch": 2.5896589658965894,
-      "grad_norm": 0.1298828125,
+      "grad_norm": 0.31640625,
       "learning_rate": 0.0001689987748619897,
-      "loss": 0.0108,
+      "loss": 0.0128,
       "step": 1177
     },
     {
       "epoch": 2.591859185918592,
-      "grad_norm": 0.31640625,
+      "grad_norm": 0.30859375,
       "learning_rate": 0.00016894856068221372,
-      "loss": 0.0207,
+      "loss": 0.0263,
       "step": 1178
     },
     {
       "epoch": 2.594059405940594,
-      "grad_norm": 0.4453125,
+      "grad_norm": 0.263671875,
       "learning_rate": 0.00016889831334136827,
-      "loss": 0.0243,
+      "loss": 0.0188,
       "step": 1179
     },
     {
       "epoch": 2.5962596259625963,
-      "grad_norm": 0.197265625,
+      "grad_norm": 0.130859375,
       "learning_rate": 0.00016884803286362,
-      "loss": 0.0112,
+      "loss": 0.0125,
       "step": 1180
     },
     {
       "epoch": 2.5984598459845984,
-      "grad_norm": 0.400390625,
+      "grad_norm": 0.388671875,
       "learning_rate": 0.0001687977192731515,
-      "loss": 0.0199,
+      "loss": 0.0205,
       "step": 1181
     },
     {
       "epoch": 2.6006600660066006,
-      "grad_norm": 0.306640625,
+      "grad_norm": 0.283203125,
       "learning_rate": 0.00016874737259416122,
-      "loss": 0.018,
+      "loss": 0.0194,
       "step": 1182
     },
     {
       "epoch": 2.6028602860286028,
-      "grad_norm": 0.349609375,
+      "grad_norm": 0.1875,
       "learning_rate": 0.00016869699285086362,
-      "loss": 0.0177,
+      "loss": 0.0173,
       "step": 1183
     },
     {
       "epoch": 2.605060506050605,
-      "grad_norm": 0.26953125,
+      "grad_norm": 0.1552734375,
       "learning_rate": 0.00016864658006748905,
-      "loss": 0.0161,
+      "loss": 0.0149,
       "step": 1184
     },
     {
       "epoch": 2.6072607260726075,
-      "grad_norm": 0.287109375,
+      "grad_norm": 0.2431640625,
       "learning_rate": 0.00016859613426828374,
-      "loss": 0.0201,
+      "loss": 0.0214,
       "step": 1185
     },
     {
       "epoch": 2.609460946094609,
-      "grad_norm": 0.294921875,
+      "grad_norm": 0.47265625,
       "learning_rate": 0.00016854565547750973,
-      "loss": 0.0168,
+      "loss": 0.0243,
       "step": 1186
     },
     {
       "epoch": 2.611661166116612,
-      "grad_norm": 0.208984375,
+      "grad_norm": 0.279296875,
       "learning_rate": 0.00016849514371944504,
-      "loss": 0.0151,
+      "loss": 0.0171,
       "step": 1187
     },
     {
       "epoch": 2.613861386138614,
-      "grad_norm": 0.27734375,
+      "grad_norm": 0.455078125,
       "learning_rate": 0.00016844459901838344,
-      "loss": 0.0162,
+      "loss": 0.023,
       "step": 1188
     },
     {
       "epoch": 2.616061606160616,
-      "grad_norm": 0.173828125,
+      "grad_norm": 0.298828125,
       "learning_rate": 0.00016839402139863461,
-      "loss": 0.0185,
+      "loss": 0.0222,
       "step": 1189
     },
     {
       "epoch": 2.6182618261826183,
-      "grad_norm": 0.134765625,
+      "grad_norm": 0.2021484375,
       "learning_rate": 0.00016834341088452407,
-      "loss": 0.0092,
+      "loss": 0.0111,
       "step": 1190
     },
     {
       "epoch": 2.6204620462046204,
-      "grad_norm": 0.31640625,
+      "grad_norm": 0.349609375,
       "learning_rate": 0.00016829276750039314,
-      "loss": 0.0102,
+      "loss": 0.0175,
       "step": 1191
     },
     {
       "epoch": 2.6226622662266226,
-      "grad_norm": 0.55859375,
+      "grad_norm": 0.376953125,
       "learning_rate": 0.00016824209127059891,
-      "loss": 0.028,
+      "loss": 0.0214,
       "step": 1192
     },
     {
       "epoch": 2.6248624862486247,
-      "grad_norm": 0.365234375,
+      "grad_norm": 0.294921875,
       "learning_rate": 0.0001681913822195143,
-      "loss": 0.0133,
+      "loss": 0.0169,
       "step": 1193
     },
     {
       "epoch": 2.6270627062706273,
-      "grad_norm": 0.2578125,
+      "grad_norm": 0.29296875,
       "learning_rate": 0.00016814064037152805,
-      "loss": 0.0139,
+      "loss": 0.0164,
       "step": 1194
     },
     {
       "epoch": 2.629262926292629,
-      "grad_norm": 0.2578125,
+      "grad_norm": 0.3671875,
       "learning_rate": 0.00016808986575104465,
-      "loss": 0.0132,
+      "loss": 0.0195,
       "step": 1195
     },
     {
       "epoch": 2.6314631463146316,
-      "grad_norm": 0.41015625,
+      "grad_norm": 0.263671875,
       "learning_rate": 0.00016803905838248432,
-      "loss": 0.02,
+      "loss": 0.0183,
       "step": 1196
     },
     {
       "epoch": 2.633663366336634,
-      "grad_norm": 0.18359375,
+      "grad_norm": 0.283203125,
       "learning_rate": 0.00016798821829028303,
-      "loss": 0.0167,
+      "loss": 0.0195,
       "step": 1197
     },
     {
       "epoch": 2.635863586358636,
-      "grad_norm": 0.427734375,
+      "grad_norm": 0.33203125,
       "learning_rate": 0.00016793734549889257,
-      "loss": 0.0289,
+      "loss": 0.0219,
       "step": 1198
     },
     {
       "epoch": 2.638063806380638,
-      "grad_norm": 0.37890625,
+      "grad_norm": 0.318359375,
       "learning_rate": 0.00016788644003278038,
-      "loss": 0.0228,
+      "loss": 0.0206,
       "step": 1199
     },
     {
       "epoch": 2.6402640264026402,
-      "grad_norm": 0.2734375,
+      "grad_norm": 0.29296875,
       "learning_rate": 0.00016783550191642964,
-      "loss": 0.0136,
+      "loss": 0.0124,
       "step": 1200
     },
     {
       "epoch": 2.6424642464246424,
-      "grad_norm": 0.31640625,
+      "grad_norm": 0.34765625,
       "learning_rate": 0.00016778453117433916,
-      "loss": 0.0128,
+      "loss": 0.0169,
       "step": 1201
     },
     {
       "epoch": 2.6446644664466445,
-      "grad_norm": 0.408203125,
+      "grad_norm": 0.30078125,
       "learning_rate": 0.00016773352783102364,
-      "loss": 0.014,
+      "loss": 0.0194,
       "step": 1202
     },
     {
       "epoch": 2.6468646864686467,
-      "grad_norm": 0.361328125,
+      "grad_norm": 0.330078125,
       "learning_rate": 0.00016768249191101322,
-      "loss": 0.0125,
+      "loss": 0.0139,
       "step": 1203
     },
     {
       "epoch": 2.649064906490649,
-      "grad_norm": 0.388671875,
+      "grad_norm": 0.265625,
       "learning_rate": 0.00016763142343885384,
-      "loss": 0.0177,
+      "loss": 0.0113,
       "step": 1204
     },
     {
       "epoch": 2.6512651265126514,
-      "grad_norm": 0.314453125,
+      "grad_norm": 0.291015625,
       "learning_rate": 0.00016758032243910712,
-      "loss": 0.0216,
+      "loss": 0.0175,
       "step": 1205
     },
     {
       "epoch": 2.6534653465346536,
-      "grad_norm": 0.54296875,
+      "grad_norm": 0.349609375,
       "learning_rate": 0.00016752918893635024,
-      "loss": 0.024,
+      "loss": 0.0165,
       "step": 1206
     },
     {
       "epoch": 2.6556655665566558,
-      "grad_norm": 0.32421875,
+      "grad_norm": 0.27734375,
       "learning_rate": 0.00016747802295517605,
-      "loss": 0.0188,
+      "loss": 0.0122,
       "step": 1207
     },
     {
       "epoch": 2.657865786578658,
-      "grad_norm": 0.22265625,
+      "grad_norm": 0.373046875,
       "learning_rate": 0.00016742682452019304,
-      "loss": 0.017,
+      "loss": 0.0196,
       "step": 1208
     },
     {
       "epoch": 2.66006600660066,
-      "grad_norm": 0.208984375,
+      "grad_norm": 0.265625,
       "learning_rate": 0.0001673755936560253,
-      "loss": 0.0196,
+      "loss": 0.0243,
       "step": 1209
     },
     {
       "epoch": 2.662266226622662,
-      "grad_norm": 0.33203125,
+      "grad_norm": 0.25390625,
       "learning_rate": 0.00016732433038731242,
-      "loss": 0.0201,
+      "loss": 0.0112,
       "step": 1210
     },
     {
       "epoch": 2.6644664466446644,
-      "grad_norm": 0.34765625,
+      "grad_norm": 0.353515625,
       "learning_rate": 0.00016727303473870978,
-      "loss": 0.0122,
+      "loss": 0.0089,
       "step": 1211
     },
     {
       "epoch": 2.6666666666666665,
-      "grad_norm": 0.322265625,
+      "grad_norm": 0.240234375,
       "learning_rate": 0.00016722170673488814,
-      "loss": 0.0114,
+      "loss": 0.0098,
       "step": 1212
     },
     {
       "epoch": 2.6688668866886687,
-      "grad_norm": 0.3828125,
+      "grad_norm": 0.341796875,
       "learning_rate": 0.00016717034640053395,
-      "loss": 0.0232,
+      "loss": 0.0221,
       "step": 1213
     },
     {
       "epoch": 2.6710671067106713,
-      "grad_norm": 0.314453125,
+      "grad_norm": 0.404296875,
       "learning_rate": 0.0001671189537603491,
-      "loss": 0.0206,
+      "loss": 0.0213,
       "step": 1214
     },
     {
       "epoch": 2.6732673267326734,
-      "grad_norm": 0.326171875,
+      "grad_norm": 0.423828125,
       "learning_rate": 0.00016706752883905107,
-      "loss": 0.0136,
+      "loss": 0.0184,
       "step": 1215
     },
     {
       "epoch": 2.6754675467546756,
-      "grad_norm": 0.2001953125,
+      "grad_norm": 0.357421875,
       "learning_rate": 0.00016701607166137294,
-      "loss": 0.0176,
+      "loss": 0.0229,
       "step": 1216
     },
     {
       "epoch": 2.6776677667766777,
-      "grad_norm": 0.12255859375,
+      "grad_norm": 0.5,
       "learning_rate": 0.00016696458225206318,
-      "loss": 0.0122,
+      "loss": 0.0213,
       "step": 1217
     },
     {
       "epoch": 2.67986798679868,
-      "grad_norm": 0.26171875,
+      "grad_norm": 0.25390625,
       "learning_rate": 0.00016691306063588583,
-      "loss": 0.014,
+      "loss": 0.0097,
       "step": 1218
     },
     {
       "epoch": 2.682068206820682,
-      "grad_norm": 0.322265625,
+      "grad_norm": 0.357421875,
       "learning_rate": 0.00016686150683762038,
-      "loss": 0.0207,
+      "loss": 0.0201,
       "step": 1219
     },
     {
       "epoch": 2.684268426842684,
-      "grad_norm": 0.455078125,
+      "grad_norm": 0.275390625,
       "learning_rate": 0.0001668099208820619,
-      "loss": 0.0113,
+      "loss": 0.013,
       "step": 1220
     },
     {
       "epoch": 2.6864686468646863,
-      "grad_norm": 0.263671875,
+      "grad_norm": 0.3359375,
       "learning_rate": 0.0001667583027940207,
-      "loss": 0.0184,
+      "loss": 0.0182,
       "step": 1221
     },
     {
       "epoch": 2.6886688668866885,
-      "grad_norm": 0.271484375,
+      "grad_norm": 0.470703125,
       "learning_rate": 0.00016670665259832285,
-      "loss": 0.0171,
+      "loss": 0.0224,
       "step": 1222
     },
     {
       "epoch": 2.690869086908691,
-      "grad_norm": 0.404296875,
+      "grad_norm": 0.2578125,
       "learning_rate": 0.00016665497031980963,
-      "loss": 0.0136,
+      "loss": 0.0153,
       "step": 1223
     },
     {
       "epoch": 2.693069306930693,
-      "grad_norm": 0.251953125,
+      "grad_norm": 0.2119140625,
       "learning_rate": 0.00016660325598333783,
-      "loss": 0.0145,
+      "loss": 0.0183,
       "step": 1224
     },
     {
       "epoch": 2.6952695269526954,
-      "grad_norm": 0.390625,
+      "grad_norm": 0.29296875,
       "learning_rate": 0.0001665515096137797,
-      "loss": 0.0247,
+      "loss": 0.0222,
       "step": 1225
     },
     {
       "epoch": 2.6974697469746975,
-      "grad_norm": 0.2333984375,
+      "grad_norm": 0.361328125,
       "learning_rate": 0.00016649973123602276,
-      "loss": 0.0115,
+      "loss": 0.0148,
       "step": 1226
     },
     {
       "epoch": 2.6996699669966997,
-      "grad_norm": 0.248046875,
+      "grad_norm": 0.302734375,
       "learning_rate": 0.00016644792087497007,
-      "loss": 0.0199,
+      "loss": 0.0183,
       "step": 1227
     },
     {
       "epoch": 2.701870187018702,
-      "grad_norm": 0.189453125,
+      "grad_norm": 0.34765625,
       "learning_rate": 0.00016639607855554008,
-      "loss": 0.0122,
+      "loss": 0.0142,
       "step": 1228
     },
     {
       "epoch": 2.704070407040704,
-      "grad_norm": 0.419921875,
+      "grad_norm": 0.255859375,
       "learning_rate": 0.00016634420430266644,
-      "loss": 0.0287,
+      "loss": 0.0122,
       "step": 1229
     },
     {
       "epoch": 2.706270627062706,
-      "grad_norm": 0.3046875,
+      "grad_norm": 0.3828125,
       "learning_rate": 0.00016629229814129828,
-      "loss": 0.0258,
+      "loss": 0.0313,
       "step": 1230
     },
     {
       "epoch": 2.7084708470847083,
-      "grad_norm": 0.345703125,
+      "grad_norm": 0.376953125,
       "learning_rate": 0.00016624036009640016,
-      "loss": 0.0133,
+      "loss": 0.0163,
       "step": 1231
     },
     {
       "epoch": 2.710671067106711,
-      "grad_norm": 0.353515625,
+      "grad_norm": 0.314453125,
       "learning_rate": 0.0001661883901929518,
-      "loss": 0.0156,
+      "loss": 0.0123,
       "step": 1232
     },
     {
       "epoch": 2.7128712871287126,
-      "grad_norm": 0.2294921875,
+      "grad_norm": 0.306640625,
       "learning_rate": 0.0001661363884559483,
-      "loss": 0.0126,
+      "loss": 0.0172,
       "step": 1233
     },
     {
       "epoch": 2.715071507150715,
-      "grad_norm": 0.361328125,
+      "grad_norm": 0.474609375,
       "learning_rate": 0.00016608435491040016,
-      "loss": 0.0238,
+      "loss": 0.0264,
       "step": 1234
     },
     {
       "epoch": 2.7172717271727174,
-      "grad_norm": 0.30078125,
+      "grad_norm": 0.2197265625,
       "learning_rate": 0.00016603228958133305,
-      "loss": 0.0097,
+      "loss": 0.0091,
       "step": 1235
     },
     {
       "epoch": 2.7194719471947195,
-      "grad_norm": 0.318359375,
+      "grad_norm": 0.267578125,
       "learning_rate": 0.00016598019249378803,
-      "loss": 0.0152,
+      "loss": 0.011,
       "step": 1236
     },
     {
       "epoch": 2.7216721672167217,
-      "grad_norm": 0.54296875,
+      "grad_norm": 0.2578125,
       "learning_rate": 0.00016592806367282136,
-      "loss": 0.0292,
+      "loss": 0.0212,
       "step": 1237
     },
     {
       "epoch": 2.723872387238724,
-      "grad_norm": 0.302734375,
+      "grad_norm": 0.314453125,
       "learning_rate": 0.0001658759031435046,
-      "loss": 0.0252,
+      "loss": 0.0207,
       "step": 1238
     },
     {
       "epoch": 2.726072607260726,
-      "grad_norm": 0.40234375,
+      "grad_norm": 0.236328125,
       "learning_rate": 0.00016582371093092456,
-      "loss": 0.0123,
+      "loss": 0.0153,
       "step": 1239
     },
     {
       "epoch": 2.728272827282728,
-      "grad_norm": 0.314453125,
+      "grad_norm": 0.53125,
       "learning_rate": 0.00016577148706018328,
-      "loss": 0.0129,
+      "loss": 0.0178,
       "step": 1240
     },
     {
       "epoch": 2.7304730473047307,
-      "grad_norm": 0.34375,
+      "grad_norm": 0.361328125,
       "learning_rate": 0.00016571923155639802,
-      "loss": 0.0076,
+      "loss": 0.008,
       "step": 1241
     },
     {
       "epoch": 2.7326732673267324,
-      "grad_norm": 0.27734375,
+      "grad_norm": 0.275390625,
       "learning_rate": 0.00016566694444470126,
-      "loss": 0.0097,
+      "loss": 0.0107,
       "step": 1242
     },
     {
       "epoch": 2.734873487348735,
-      "grad_norm": 0.4609375,
+      "grad_norm": 0.283203125,
       "learning_rate": 0.0001656146257502407,
-      "loss": 0.0216,
+      "loss": 0.0165,
       "step": 1243
     },
     {
       "epoch": 2.737073707370737,
-      "grad_norm": 0.298828125,
+      "grad_norm": 0.318359375,
       "learning_rate": 0.00016556227549817919,
-      "loss": 0.0144,
+      "loss": 0.0156,
       "step": 1244
     },
     {
       "epoch": 2.7392739273927393,
-      "grad_norm": 0.33203125,
+      "grad_norm": 0.30859375,
       "learning_rate": 0.00016550989371369478,
-      "loss": 0.0189,
+      "loss": 0.0109,
       "step": 1245
     },
     {
       "epoch": 2.7414741474147415,
-      "grad_norm": 0.19921875,
+      "grad_norm": 0.5703125,
       "learning_rate": 0.00016545748042198072,
-      "loss": 0.011,
+      "loss": 0.0208,
       "step": 1246
     },
     {
       "epoch": 2.7436743674367436,
-      "grad_norm": 0.2890625,
+      "grad_norm": 0.478515625,
       "learning_rate": 0.00016540503564824538,
-      "loss": 0.0099,
+      "loss": 0.0138,
       "step": 1247
     },
     {
       "epoch": 2.745874587458746,
-      "grad_norm": 0.26953125,
+      "grad_norm": 0.349609375,
       "learning_rate": 0.0001653525594177122,
-      "loss": 0.0123,
+      "loss": 0.0206,
       "step": 1248
     },
     {
       "epoch": 2.748074807480748,
-      "grad_norm": 0.1962890625,
+      "grad_norm": 0.283203125,
       "learning_rate": 0.00016530005175561987,
-      "loss": 0.0127,
+      "loss": 0.0162,
       "step": 1249
     },
     {
       "epoch": 2.7502750275027505,
-      "grad_norm": 0.353515625,
+      "grad_norm": 0.240234375,
       "learning_rate": 0.00016524751268722217,
-      "loss": 0.0176,
+      "loss": 0.0148,
       "step": 1250
     },
     {
       "epoch": 2.7524752475247523,
-      "grad_norm": 0.40625,
+      "grad_norm": 0.478515625,
       "learning_rate": 0.00016519494223778792,
-      "loss": 0.0206,
+      "loss": 0.0194,
       "step": 1251
     },
     {
       "epoch": 2.754675467546755,
-      "grad_norm": 0.341796875,
+      "grad_norm": 0.275390625,
       "learning_rate": 0.0001651423404326011,
-      "loss": 0.0109,
+      "loss": 0.0173,
       "step": 1252
     },
     {
       "epoch": 2.756875687568757,
-      "grad_norm": 0.26953125,
+      "grad_norm": 0.34375,
       "learning_rate": 0.00016508970729696074,
-      "loss": 0.014,
+      "loss": 0.0191,
       "step": 1253
     },
     {
       "epoch": 2.759075907590759,
-      "grad_norm": 0.26953125,
+      "grad_norm": 0.44140625,
       "learning_rate": 0.00016503704285618094,
-      "loss": 0.0143,
+      "loss": 0.0317,
       "step": 1254
     },
     {
       "epoch": 2.759075907590759,
-      "eval_loss": 0.030484065413475037,
-      "eval_runtime": 10.9842,
-      "eval_samples_per_second": 34.868,
-      "eval_steps_per_second": 4.37,
+      "eval_loss": 0.034429945051670074,
+      "eval_runtime": 10.307,
+      "eval_samples_per_second": 37.159,
+      "eval_steps_per_second": 4.657,
       "step": 1254
     },
     {
       "epoch": 2.7612761276127613,
-      "grad_norm": 0.3515625,
+      "grad_norm": 0.322265625,
       "learning_rate": 0.00016498434713559088,
-      "loss": 0.0158,
+      "loss": 0.0155,
       "step": 1255
     },
     {
       "epoch": 2.7634763476347635,
-      "grad_norm": 0.345703125,
+      "grad_norm": 0.2490234375,
       "learning_rate": 0.00016493162016053474,
-      "loss": 0.0194,
+      "loss": 0.018,
       "step": 1256
     },
     {
       "epoch": 2.7656765676567656,
-      "grad_norm": 0.240234375,
+      "grad_norm": 0.36328125,
       "learning_rate": 0.00016487886195637173,
-      "loss": 0.0092,
+      "loss": 0.0183,
       "step": 1257
     },
     {
       "epoch": 2.7678767876787678,
-      "grad_norm": 0.291015625,
+      "grad_norm": 0.228515625,
       "learning_rate": 0.0001648260725484762,
-      "loss": 0.0121,
+      "loss": 0.0107,
       "step": 1258
     },
     {
       "epoch": 2.77007700770077,
-      "grad_norm": 0.158203125,
+      "grad_norm": 0.3046875,
       "learning_rate": 0.00016477325196223732,
-      "loss": 0.013,
+      "loss": 0.0163,
       "step": 1259
     },
     {
       "epoch": 2.772277227722772,
-      "grad_norm": 0.294921875,
+      "grad_norm": 0.353515625,
       "learning_rate": 0.0001647204002230594,
-      "loss": 0.022,
+      "loss": 0.0241,
       "step": 1260
     },
     {
       "epoch": 2.7744774477447747,
-      "grad_norm": 0.21875,
+      "grad_norm": 0.30859375,
       "learning_rate": 0.00016466751735636168,
-      "loss": 0.0159,
+      "loss": 0.0175,
       "step": 1261
     },
     {
       "epoch": 2.776677667766777,
-      "grad_norm": 0.59375,
+      "grad_norm": 0.203125,
       "learning_rate": 0.00016461460338757834,
-      "loss": 0.0295,
+      "loss": 0.012,
       "step": 1262
     },
     {
       "epoch": 2.778877887788779,
-      "grad_norm": 0.283203125,
+      "grad_norm": 0.314453125,
       "learning_rate": 0.0001645616583421586,
-      "loss": 0.0139,
+      "loss": 0.0203,
       "step": 1263
     },
     {
       "epoch": 2.781078107810781,
-      "grad_norm": 0.1767578125,
+      "grad_norm": 0.3359375,
       "learning_rate": 0.00016450868224556655,
-      "loss": 0.013,
+      "loss": 0.0185,
       "step": 1264
     },
     {
       "epoch": 2.7832783278327833,
-      "grad_norm": 0.53515625,
+      "grad_norm": 0.35546875,
       "learning_rate": 0.0001644556751232812,
-      "loss": 0.0299,
+      "loss": 0.0226,
       "step": 1265
     },
     {
       "epoch": 2.7854785478547854,
-      "grad_norm": 0.2138671875,
+      "grad_norm": 0.259765625,
       "learning_rate": 0.00016440263700079664,
-      "loss": 0.0114,
+      "loss": 0.0141,
       "step": 1266
     },
     {
       "epoch": 2.7876787678767876,
-      "grad_norm": 0.408203125,
+      "grad_norm": 0.376953125,
       "learning_rate": 0.00016434956790362167,
-      "loss": 0.0187,
+      "loss": 0.0205,
       "step": 1267
     },
     {
       "epoch": 2.7898789878987897,
-      "grad_norm": 0.396484375,
+      "grad_norm": 0.40625,
       "learning_rate": 0.00016429646785728004,
-      "loss": 0.0157,
+      "loss": 0.0209,
       "step": 1268
     },
     {
       "epoch": 2.792079207920792,
-      "grad_norm": 0.44140625,
+      "grad_norm": 0.392578125,
       "learning_rate": 0.0001642433368873105,
-      "loss": 0.0161,
+      "loss": 0.0156,
       "step": 1269
     },
     {
       "epoch": 2.7942794279427945,
-      "grad_norm": 0.337890625,
+      "grad_norm": 0.306640625,
       "learning_rate": 0.00016419017501926656,
-      "loss": 0.0163,
+      "loss": 0.0138,
       "step": 1270
     },
     {
       "epoch": 2.7964796479647966,
-      "grad_norm": 0.193359375,
+      "grad_norm": 0.25,
       "learning_rate": 0.00016413698227871662,
-      "loss": 0.009,
+      "loss": 0.0077,
       "step": 1271
     },
     {
       "epoch": 2.798679867986799,
-      "grad_norm": 0.1728515625,
+      "grad_norm": 0.35546875,
       "learning_rate": 0.0001640837586912439,
-      "loss": 0.0152,
+      "loss": 0.0271,
       "step": 1272
     },
     {
       "epoch": 2.800880088008801,
-      "grad_norm": 0.353515625,
+      "grad_norm": 0.40625,
       "learning_rate": 0.00016403050428244652,
-      "loss": 0.0191,
+      "loss": 0.0202,
       "step": 1273
     },
     {
       "epoch": 2.803080308030803,
-      "grad_norm": 0.271484375,
+      "grad_norm": 0.337890625,
       "learning_rate": 0.0001639772190779374,
-      "loss": 0.0173,
+      "loss": 0.0156,
       "step": 1274
     },
     {
       "epoch": 2.8052805280528053,
-      "grad_norm": 0.4453125,
+      "grad_norm": 0.30859375,
       "learning_rate": 0.0001639239031033442,
-      "loss": 0.0159,
+      "loss": 0.0186,
       "step": 1275
     },
     {
       "epoch": 2.8074807480748074,
-      "grad_norm": 0.224609375,
+      "grad_norm": 0.365234375,
       "learning_rate": 0.0001638705563843095,
-      "loss": 0.0109,
+      "loss": 0.0161,
       "step": 1276
     },
     {
       "epoch": 2.8096809680968096,
-      "grad_norm": 0.369140625,
+      "grad_norm": 0.267578125,
       "learning_rate": 0.00016381717894649054,
-      "loss": 0.0193,
+      "loss": 0.0137,
       "step": 1277
     },
     {
       "epoch": 2.8118811881188117,
-      "grad_norm": 0.279296875,
+      "grad_norm": 0.53125,
       "learning_rate": 0.00016376377081555945,
-      "loss": 0.0116,
+      "loss": 0.0295,
       "step": 1278
     },
     {
       "epoch": 2.8140814081408143,
-      "grad_norm": 0.423828125,
+      "grad_norm": 0.68359375,
       "learning_rate": 0.00016371033201720308,
-      "loss": 0.016,
+      "loss": 0.0318,
       "step": 1279
     },
     {
       "epoch": 2.816281628162816,
-      "grad_norm": 0.248046875,
+      "grad_norm": 0.16796875,
       "learning_rate": 0.00016365686257712302,
-      "loss": 0.0183,
+      "loss": 0.0169,
       "step": 1280
     },
     {
       "epoch": 2.8184818481848186,
-      "grad_norm": 0.30859375,
+      "grad_norm": 0.1767578125,
       "learning_rate": 0.00016360336252103553,
-      "loss": 0.0192,
+      "loss": 0.0127,
       "step": 1281
     },
     {
       "epoch": 2.8206820682068208,
-      "grad_norm": 0.62109375,
+      "grad_norm": 0.1953125,
       "learning_rate": 0.00016354983187467172,
-      "loss": 0.0135,
+      "loss": 0.0103,
       "step": 1282
     },
     {
       "epoch": 2.822882288228823,
-      "grad_norm": 0.1982421875,
+      "grad_norm": 0.40625,
       "learning_rate": 0.00016349627066377734,
-      "loss": 0.0155,
+      "loss": 0.0181,
       "step": 1283
     },
     {
       "epoch": 2.825082508250825,
-      "grad_norm": 0.208984375,
+      "grad_norm": 0.37890625,
       "learning_rate": 0.0001634426789141129,
-      "loss": 0.011,
+      "loss": 0.0184,
       "step": 1284
     },
     {
       "epoch": 2.8272827282728272,
-      "grad_norm": 0.41015625,
+      "grad_norm": 0.498046875,
       "learning_rate": 0.0001633890566514535,
-      "loss": 0.0228,
+      "loss": 0.0323,
       "step": 1285
     },
     {
       "epoch": 2.8294829482948294,
-      "grad_norm": 0.3515625,
+      "grad_norm": 0.275390625,
       "learning_rate": 0.00016333540390158898,
-      "loss": 0.0202,
+      "loss": 0.0243,
       "step": 1286
     },
     {
       "epoch": 2.8316831683168315,
-      "grad_norm": 0.10791015625,
+      "grad_norm": 0.171875,
       "learning_rate": 0.00016328172069032384,
-      "loss": 0.0076,
+      "loss": 0.0133,
       "step": 1287
     },
     {
       "epoch": 2.833883388338834,
-      "grad_norm": 0.380859375,
+      "grad_norm": 0.3046875,
       "learning_rate": 0.0001632280070434772,
-      "loss": 0.0137,
+      "loss": 0.0225,
       "step": 1288
     },
     {
       "epoch": 2.836083608360836,
-      "grad_norm": 0.294921875,
+      "grad_norm": 0.439453125,
       "learning_rate": 0.0001631742629868829,
-      "loss": 0.0142,
+      "loss": 0.0254,
       "step": 1289
     },
     {
       "epoch": 2.8382838283828384,
-      "grad_norm": 0.271484375,
+      "grad_norm": 0.318359375,
       "learning_rate": 0.0001631204885463893,
-      "loss": 0.0095,
+      "loss": 0.0157,
       "step": 1290
     },
     {
       "epoch": 2.8404840484048406,
-      "grad_norm": 0.1416015625,
+      "grad_norm": 0.287109375,
       "learning_rate": 0.0001630666837478594,
-      "loss": 0.0072,
+      "loss": 0.0132,
       "step": 1291
     },
     {
       "epoch": 2.8426842684268427,
-      "grad_norm": 0.376953125,
+      "grad_norm": 0.255859375,
       "learning_rate": 0.00016301284861717087,
-      "loss": 0.013,
+      "loss": 0.0117,
       "step": 1292
     },
     {
       "epoch": 2.844884488448845,
-      "grad_norm": 0.2255859375,
+      "grad_norm": 0.52734375,
       "learning_rate": 0.00016295898318021586,
-      "loss": 0.011,
+      "loss": 0.025,
       "step": 1293
     },
     {
       "epoch": 2.847084708470847,
-      "grad_norm": 0.470703125,
+      "grad_norm": 0.240234375,
       "learning_rate": 0.00016290508746290123,
-      "loss": 0.0104,
+      "loss": 0.0213,
       "step": 1294
     },
     {
       "epoch": 2.849284928492849,
-      "grad_norm": 0.35546875,
+      "grad_norm": 0.353515625,
       "learning_rate": 0.00016285116149114824,
-      "loss": 0.0083,
+      "loss": 0.0154,
       "step": 1295
     },
     {
       "epoch": 2.8514851485148514,
-      "grad_norm": 0.2333984375,
+      "grad_norm": 0.322265625,
       "learning_rate": 0.00016279720529089285,
-      "loss": 0.0127,
+      "loss": 0.0193,
       "step": 1296
     },
     {
       "epoch": 2.853685368536854,
-      "grad_norm": 0.361328125,
+      "grad_norm": 0.431640625,
       "learning_rate": 0.00016274321888808552,
-      "loss": 0.0168,
+      "loss": 0.0261,
       "step": 1297
     },
     {
       "epoch": 2.8558855885588557,
-      "grad_norm": 0.28515625,
+      "grad_norm": 0.31640625,
       "learning_rate": 0.00016268920230869116,
-      "loss": 0.0086,
+      "loss": 0.0146,
       "step": 1298
     },
     {
       "epoch": 2.8580858085808583,
-      "grad_norm": 0.455078125,
+      "grad_norm": 0.271484375,
       "learning_rate": 0.00016263515557868923,
-      "loss": 0.0157,
+      "loss": 0.0165,
       "step": 1299
     },
     {
       "epoch": 2.8602860286028604,
-      "grad_norm": 0.31640625,
+      "grad_norm": 0.451171875,
       "learning_rate": 0.00016258107872407375,
-      "loss": 0.0132,
+      "loss": 0.0146,
       "step": 1300
     },
     {
       "epoch": 2.8624862486248626,
-      "grad_norm": 0.3984375,
+      "grad_norm": 0.3515625,
       "learning_rate": 0.0001625269717708532,
-      "loss": 0.0145,
+      "loss": 0.0168,
       "step": 1301
     },
     {
       "epoch": 2.8646864686468647,
-      "grad_norm": 0.34375,
+      "grad_norm": 0.29296875,
       "learning_rate": 0.00016247283474505052,
-      "loss": 0.0164,
+      "loss": 0.0135,
       "step": 1302
     },
     {
       "epoch": 2.866886688668867,
-      "grad_norm": 0.2578125,
+      "grad_norm": 0.3828125,
       "learning_rate": 0.00016241866767270307,
-      "loss": 0.0111,
+      "loss": 0.0166,
       "step": 1303
     },
     {
       "epoch": 2.869086908690869,
-      "grad_norm": 0.59765625,
+      "grad_norm": 0.333984375,
       "learning_rate": 0.0001623644705798627,
-      "loss": 0.0212,
+      "loss": 0.0157,
       "step": 1304
     },
     {
       "epoch": 2.871287128712871,
-      "grad_norm": 0.283203125,
+      "grad_norm": 0.251953125,
       "learning_rate": 0.00016231024349259578,
-      "loss": 0.0128,
+      "loss": 0.0106,
       "step": 1305
     },
     {
       "epoch": 2.8734873487348733,
-      "grad_norm": 0.2470703125,
+      "grad_norm": 0.388671875,
       "learning_rate": 0.000162255986436983,
-      "loss": 0.0116,
+      "loss": 0.0199,
       "step": 1306
     },
     {
       "epoch": 2.8756875687568755,
-      "grad_norm": 0.380859375,
+      "grad_norm": 0.326171875,
       "learning_rate": 0.00016220169943911944,
-      "loss": 0.0267,
+      "loss": 0.0305,
       "step": 1307
     },
     {
       "epoch": 2.877887788778878,
-      "grad_norm": 0.3671875,
+      "grad_norm": 0.38671875,
       "learning_rate": 0.0001621473825251147,
-      "loss": 0.0206,
+      "loss": 0.0179,
       "step": 1308
     },
     {
       "epoch": 2.8800880088008802,
-      "grad_norm": 0.33984375,
+      "grad_norm": 0.396484375,
       "learning_rate": 0.0001620930357210927,
-      "loss": 0.0148,
+      "loss": 0.0172,
       "step": 1309
     },
     {
       "epoch": 2.8822882288228824,
-      "grad_norm": 0.2109375,
+      "grad_norm": 0.310546875,
       "learning_rate": 0.00016203865905319172,
-      "loss": 0.014,
+      "loss": 0.0159,
       "step": 1310
     },
     {
       "epoch": 2.8844884488448845,
-      "grad_norm": 0.259765625,
+      "grad_norm": 0.181640625,
       "learning_rate": 0.00016198425254756443,
-      "loss": 0.0151,
+      "loss": 0.0121,
       "step": 1311
     },
     {
       "epoch": 2.8866886688668867,
-      "grad_norm": 0.412109375,
+      "grad_norm": 0.423828125,
       "learning_rate": 0.00016192981623037783,
-      "loss": 0.0143,
+      "loss": 0.0169,
       "step": 1312
     },
     {
       "epoch": 2.888888888888889,
-      "grad_norm": 0.330078125,
+      "grad_norm": 0.345703125,
       "learning_rate": 0.00016187535012781326,
-      "loss": 0.0235,
+      "loss": 0.0305,
       "step": 1313
     },
     {
       "epoch": 2.891089108910891,
-      "grad_norm": 0.259765625,
+      "grad_norm": 0.21875,
       "learning_rate": 0.00016182085426606646,
-      "loss": 0.0102,
+      "loss": 0.0164,
       "step": 1314
     },
     {
       "epoch": 2.893289328932893,
-      "grad_norm": 0.37890625,
+      "grad_norm": 0.12060546875,
       "learning_rate": 0.0001617663286713474,
-      "loss": 0.0099,
+      "loss": 0.0056,
       "step": 1315
     },
     {
       "epoch": 2.8954895489548953,
-      "grad_norm": 0.337890625,
+      "grad_norm": 0.1796875,
       "learning_rate": 0.00016171177336988027,
-      "loss": 0.0126,
+      "loss": 0.0092,
       "step": 1316
     },
     {
       "epoch": 2.897689768976898,
-      "grad_norm": 0.1552734375,
+      "grad_norm": 0.291015625,
       "learning_rate": 0.00016165718838790375,
-      "loss": 0.0148,
+      "loss": 0.02,
       "step": 1317
     },
     {
       "epoch": 2.8998899889989,
-      "grad_norm": 0.2275390625,
+      "grad_norm": 0.11474609375,
       "learning_rate": 0.0001616025737516707,
-      "loss": 0.0128,
+      "loss": 0.0086,
       "step": 1318
     },
     {
       "epoch": 2.902090209020902,
-      "grad_norm": 0.2333984375,
+      "grad_norm": 0.25390625,
       "learning_rate": 0.0001615479294874482,
-      "loss": 0.0169,
+      "loss": 0.0168,
       "step": 1319
     },
     {
       "epoch": 2.9042904290429044,
-      "grad_norm": 0.34765625,
+      "grad_norm": 0.333984375,
       "learning_rate": 0.00016149325562151757,
-      "loss": 0.0191,
+      "loss": 0.0171,
       "step": 1320
     },
     {
       "epoch": 2.9064906490649065,
-      "grad_norm": 0.27734375,
+      "grad_norm": 0.2041015625,
       "learning_rate": 0.0001614385521801745,
-      "loss": 0.0126,
+      "loss": 0.016,
       "step": 1321
     },
     {
       "epoch": 2.9086908690869087,
-      "grad_norm": 0.330078125,
+      "grad_norm": 0.2421875,
       "learning_rate": 0.00016138381918972877,
-      "loss": 0.0126,
+      "loss": 0.0109,
       "step": 1322
     },
     {
       "epoch": 2.910891089108911,
-      "grad_norm": 0.21484375,
+      "grad_norm": 0.345703125,
       "learning_rate": 0.00016132905667650443,
-      "loss": 0.0087,
+      "loss": 0.0148,
       "step": 1323
     },
     {
       "epoch": 2.913091309130913,
-      "grad_norm": 0.375,
+      "grad_norm": 0.4140625,
       "learning_rate": 0.0001612742646668397,
-      "loss": 0.0148,
+      "loss": 0.0182,
       "step": 1324
     },
     {
       "epoch": 2.915291529152915,
-      "grad_norm": 0.2001953125,
+      "grad_norm": 0.31640625,
       "learning_rate": 0.00016121944318708698,
-      "loss": 0.012,
+      "loss": 0.0158,
       "step": 1325
     },
     {
       "epoch": 2.9174917491749177,
-      "grad_norm": 0.6171875,
+      "grad_norm": 0.48828125,
       "learning_rate": 0.00016116459226361295,
-      "loss": 0.0298,
+      "loss": 0.0195,
       "step": 1326
     },
     {
       "epoch": 2.9196919691969194,
-      "grad_norm": 0.29296875,
+      "grad_norm": 0.193359375,
       "learning_rate": 0.0001611097119227983,
-      "loss": 0.0075,
+      "loss": 0.0088,
       "step": 1327
     },
     {
       "epoch": 2.921892189218922,
-      "grad_norm": 0.212890625,
+      "grad_norm": 0.259765625,
       "learning_rate": 0.00016105480219103795,
-      "loss": 0.0136,
+      "loss": 0.0154,
       "step": 1328
     },
     {
       "epoch": 2.924092409240924,
-      "grad_norm": 0.2138671875,
+      "grad_norm": 0.2431640625,
       "learning_rate": 0.0001609998630947409,
-      "loss": 0.0067,
+      "loss": 0.0097,
       "step": 1329
     },
     {
       "epoch": 2.9262926292629263,
-      "grad_norm": 0.34765625,
+      "grad_norm": 0.3203125,
       "learning_rate": 0.00016094489466033043,
-      "loss": 0.0125,
+      "loss": 0.0081,
       "step": 1330
     },
     {
       "epoch": 2.9284928492849285,
-      "grad_norm": 0.1884765625,
+      "grad_norm": 0.3046875,
       "learning_rate": 0.0001608898969142437,
-      "loss": 0.0118,
+      "loss": 0.0111,
       "step": 1331
     },
     {
       "epoch": 2.9306930693069306,
-      "grad_norm": 0.298828125,
+      "grad_norm": 0.404296875,
       "learning_rate": 0.00016083486988293215,
-      "loss": 0.0095,
+      "loss": 0.0179,
       "step": 1332
     },
     {
       "epoch": 2.932893289328933,
-      "grad_norm": 0.28125,
+      "grad_norm": 0.470703125,
       "learning_rate": 0.00016077981359286119,
-      "loss": 0.0151,
+      "loss": 0.0249,
       "step": 1333
     },
     {
       "epoch": 2.935093509350935,
-      "grad_norm": 0.365234375,
+      "grad_norm": 0.22265625,
       "learning_rate": 0.0001607247280705104,
-      "loss": 0.0221,
+      "loss": 0.0138,
       "step": 1334
     },
     {
       "epoch": 2.9372937293729375,
-      "grad_norm": 0.26171875,
+      "grad_norm": 0.1416015625,
       "learning_rate": 0.00016066961334237334,
-      "loss": 0.0108,
+      "loss": 0.01,
       "step": 1335
     },
     {
       "epoch": 2.9394939493949392,
-      "grad_norm": 0.392578125,
+      "grad_norm": 0.279296875,
       "learning_rate": 0.00016061446943495763,
-      "loss": 0.0239,
+      "loss": 0.0234,
       "step": 1336
     },
     {
       "epoch": 2.941694169416942,
-      "grad_norm": 0.375,
+      "grad_norm": 0.44140625,
       "learning_rate": 0.00016055929637478496,
-      "loss": 0.0224,
+      "loss": 0.0169,
       "step": 1337
     },
     {
       "epoch": 2.943894389438944,
-      "grad_norm": 0.306640625,
+      "grad_norm": 0.251953125,
       "learning_rate": 0.00016050409418839103,
-      "loss": 0.0096,
+      "loss": 0.0081,
       "step": 1338
     },
     {
       "epoch": 2.946094609460946,
-      "grad_norm": 0.1943359375,
+      "grad_norm": 0.203125,
       "learning_rate": 0.00016044886290232551,
-      "loss": 0.0081,
+      "loss": 0.0104,
       "step": 1339
     },
     {
       "epoch": 2.9482948294829483,
-      "grad_norm": 0.453125,
+      "grad_norm": 0.39453125,
       "learning_rate": 0.00016039360254315214,
-      "loss": 0.0178,
+      "loss": 0.0125,
       "step": 1340
     },
     {
       "epoch": 2.9504950495049505,
-      "grad_norm": 0.1953125,
+      "grad_norm": 0.43359375,
       "learning_rate": 0.00016033831313744856,
-      "loss": 0.0178,
+      "loss": 0.0191,
       "step": 1341
     },
     {
       "epoch": 2.9526952695269526,
-      "grad_norm": 0.2392578125,
+      "grad_norm": 0.380859375,
       "learning_rate": 0.00016028299471180642,
-      "loss": 0.0097,
+      "loss": 0.0104,
       "step": 1342
     },
     {
       "epoch": 2.9548954895489548,
-      "grad_norm": 0.275390625,
+      "grad_norm": 0.263671875,
       "learning_rate": 0.00016022764729283135,
-      "loss": 0.0106,
+      "loss": 0.0111,
       "step": 1343
     },
     {
       "epoch": 2.9570957095709574,
-      "grad_norm": 0.37109375,
+      "grad_norm": 0.26953125,
       "learning_rate": 0.0001601722709071429,
-      "loss": 0.0109,
+      "loss": 0.0183,
       "step": 1344
     },
     {
       "epoch": 2.959295929592959,
-      "grad_norm": 0.4453125,
+      "grad_norm": 0.244140625,
       "learning_rate": 0.00016011686558137448,
-      "loss": 0.0124,
+      "loss": 0.0105,
       "step": 1345
     },
     {
       "epoch": 2.9614961496149617,
-      "grad_norm": 0.3359375,
+      "grad_norm": 0.263671875,
       "learning_rate": 0.00016006143134217357,
-      "loss": 0.0168,
+      "loss": 0.0174,
       "step": 1346
     },
     {
       "epoch": 2.963696369636964,
-      "grad_norm": 0.35546875,
+      "grad_norm": 0.396484375,
       "learning_rate": 0.00016000596821620145,
-      "loss": 0.0172,
+      "loss": 0.0195,
       "step": 1347
     },
     {
       "epoch": 2.965896589658966,
-      "grad_norm": 0.244140625,
+      "grad_norm": 0.31640625,
       "learning_rate": 0.00015995047623013333,
-      "loss": 0.0178,
+      "loss": 0.021,
       "step": 1348
     },
     {
       "epoch": 2.968096809680968,
-      "grad_norm": 0.34375,
+      "grad_norm": 0.3046875,
       "learning_rate": 0.00015989495541065825,
-      "loss": 0.0178,
+      "loss": 0.0197,
       "step": 1349
     },
     {
       "epoch": 2.9702970297029703,
-      "grad_norm": 0.1728515625,
+      "grad_norm": 0.4296875,
       "learning_rate": 0.0001598394057844792,
-      "loss": 0.0087,
+      "loss": 0.018,
       "step": 1350
     },
     {
       "epoch": 2.9724972497249724,
-      "grad_norm": 0.326171875,
+      "grad_norm": 0.28125,
       "learning_rate": 0.00015978382737831295,
-      "loss": 0.0139,
+      "loss": 0.0104,
       "step": 1351
     },
     {
       "epoch": 2.9746974697469746,
-      "grad_norm": 0.27734375,
+      "grad_norm": 0.39453125,
       "learning_rate": 0.00015972822021889017,
-      "loss": 0.0138,
+      "loss": 0.0154,
       "step": 1352
     },
     {
       "epoch": 2.976897689768977,
-      "grad_norm": 0.609375,
+      "grad_norm": 0.333984375,
       "learning_rate": 0.00015967258433295534,
-      "loss": 0.0212,
+      "loss": 0.0129,
       "step": 1353
     },
     {
       "epoch": 2.979097909790979,
-      "grad_norm": 0.169921875,
+      "grad_norm": 0.2314453125,
       "learning_rate": 0.0001596169197472667,
-      "loss": 0.0081,
+      "loss": 0.0087,
       "step": 1354
     },
     {
       "epoch": 2.9812981298129815,
-      "grad_norm": 0.43359375,
+      "grad_norm": 0.296875,
       "learning_rate": 0.0001595612264885964,
-      "loss": 0.0097,
+      "loss": 0.0077,
       "step": 1355
     },
     {
       "epoch": 2.9834983498349836,
-      "grad_norm": 0.33984375,
+      "grad_norm": 0.515625,
       "learning_rate": 0.0001595055045837303,
-      "loss": 0.0208,
+      "loss": 0.0248,
       "step": 1356
     },
     {
       "epoch": 2.985698569856986,
-      "grad_norm": 0.16015625,
+      "grad_norm": 0.28125,
       "learning_rate": 0.00015944975405946798,
-      "loss": 0.0111,
+      "loss": 0.0106,
       "step": 1357
     },
     {
       "epoch": 2.987898789878988,
-      "grad_norm": 0.515625,
+      "grad_norm": 0.337890625,
       "learning_rate": 0.000159393974942623,
-      "loss": 0.019,
+      "loss": 0.0112,
       "step": 1358
     },
     {
       "epoch": 2.99009900990099,
-      "grad_norm": 0.306640625,
+      "grad_norm": 0.6328125,
       "learning_rate": 0.00015933816726002245,
-      "loss": 0.0189,
+      "loss": 0.0191,
       "step": 1359
     },
     {
       "epoch": 2.9922992299229922,
-      "grad_norm": 0.515625,
+      "grad_norm": 0.279296875,
       "learning_rate": 0.0001592823310385073,
-      "loss": 0.0307,
+      "loss": 0.0183,
       "step": 1360
     },
     {
       "epoch": 2.9944994499449944,
-      "grad_norm": 0.30859375,
+      "grad_norm": 0.7265625,
       "learning_rate": 0.0001592264663049321,
-      "loss": 0.0164,
+      "loss": 0.0142,
       "step": 1361
     },
     {
       "epoch": 2.9966996699669965,
-      "grad_norm": 0.31640625,
+      "grad_norm": 0.353515625,
       "learning_rate": 0.0001591705730861653,
-      "loss": 0.0119,
+      "loss": 0.0135,
       "step": 1362
     },
     {
       "epoch": 2.9988998899889987,
-      "grad_norm": 0.412109375,
+      "grad_norm": 0.423828125,
       "learning_rate": 0.00015911465140908894,
-      "loss": 0.019,
+      "loss": 0.0177,
       "step": 1363
     },
     {
       "epoch": 3.0011001100110013,
-      "grad_norm": 0.33203125,
+      "grad_norm": 0.3046875,
       "learning_rate": 0.0001590587013005987,
-      "loss": 0.0143,
+      "loss": 0.0083,
       "step": 1364
     },
     {
       "epoch": 3.0033003300330035,
-      "grad_norm": 0.259765625,
+      "grad_norm": 0.56640625,
       "learning_rate": 0.00015900272278760407,
-      "loss": 0.012,
+      "loss": 0.0295,
       "step": 1365
     },
     {
       "epoch": 3.0055005500550056,
       "grad_norm": 0.2412109375,
       "learning_rate": 0.00015894671589702805,
-      "loss": 0.0064,
+      "loss": 0.0077,
       "step": 1366
     },
     {
       "epoch": 3.0077007700770078,
-      "grad_norm": 0.1279296875,
+      "grad_norm": 0.1884765625,
       "learning_rate": 0.00015889068065580742,
-      "loss": 0.0037,
+      "loss": 0.0028,
       "step": 1367
     },
     {
       "epoch": 3.00990099009901,
-      "grad_norm": 0.396484375,
+      "grad_norm": 0.244140625,
       "learning_rate": 0.00015883461709089252,
-      "loss": 0.0163,
+      "loss": 0.0086,
       "step": 1368
     },
     {
       "epoch": 3.00990099009901,
-      "eval_loss": 0.02657780796289444,
-      "eval_runtime": 10.4607,
-      "eval_samples_per_second": 36.613,
-      "eval_steps_per_second": 4.589,
+      "eval_loss": 0.02681020461022854,
+      "eval_runtime": 10.0714,
+      "eval_samples_per_second": 38.029,
+      "eval_steps_per_second": 4.766,
       "step": 1368
     },
     {
       "epoch": 3.012101210121012,
-      "grad_norm": 0.37890625,
+      "grad_norm": 0.38671875,
       "learning_rate": 0.00015877852522924732,
-      "loss": 0.0092,
+      "loss": 0.01,
       "step": 1369
     },
     {
       "epoch": 3.014301430143014,
-      "grad_norm": 0.1533203125,
+      "grad_norm": 0.1455078125,
       "learning_rate": 0.00015872240509784945,
-      "loss": 0.0074,
+      "loss": 0.0048,
       "step": 1370
     },
     {
       "epoch": 3.0165016501650164,
-      "grad_norm": 0.197265625,
+      "grad_norm": 0.337890625,
       "learning_rate": 0.00015866625672369006,
-      "loss": 0.0088,
+      "loss": 0.0134,
       "step": 1371
     },
     {
       "epoch": 3.0187018701870185,
       "grad_norm": 0.359375,
       "learning_rate": 0.00015861008013377394,
-      "loss": 0.0183,
+      "loss": 0.009,
       "step": 1372
     },
     {
       "epoch": 3.020902090209021,
-      "grad_norm": 0.1943359375,
+      "grad_norm": 0.283203125,
       "learning_rate": 0.00015855387535511944,
-      "loss": 0.0082,
+      "loss": 0.0084,
       "step": 1373
     },
     {
       "epoch": 3.0231023102310233,
-      "grad_norm": 0.73046875,
+      "grad_norm": 0.228515625,
       "learning_rate": 0.00015849764241475844,
-      "loss": 0.0418,
+      "loss": 0.007,
       "step": 1374
     },
     {
       "epoch": 3.0253025302530254,
-      "grad_norm": 0.095703125,
+      "grad_norm": 0.365234375,
       "learning_rate": 0.0001584413813397364,
-      "loss": 0.0059,
+      "loss": 0.0096,
       "step": 1375
     },
     {
       "epoch": 3.0275027502750276,
-      "grad_norm": 0.400390625,
+      "grad_norm": 0.3046875,
       "learning_rate": 0.00015838509215711225,
-      "loss": 0.0245,
+      "loss": 0.0155,
       "step": 1376
     },
     {
       "epoch": 3.0297029702970297,
-      "grad_norm": 0.2314453125,
+      "grad_norm": 0.2451171875,
       "learning_rate": 0.00015832877489395858,
-      "loss": 0.0065,
+      "loss": 0.0103,
       "step": 1377
     },
     {
       "epoch": 3.031903190319032,
-      "grad_norm": 0.25,
+      "grad_norm": 0.3203125,
       "learning_rate": 0.00015827242957736128,
-      "loss": 0.0119,
+      "loss": 0.0209,
       "step": 1378
     },
     {
       "epoch": 3.034103410341034,
-      "grad_norm": 0.1767578125,
+      "grad_norm": 0.365234375,
       "learning_rate": 0.00015821605623441993,
-      "loss": 0.0105,
+      "loss": 0.0259,
       "step": 1379
     },
     {
       "epoch": 3.036303630363036,
-      "grad_norm": 0.26171875,
+      "grad_norm": 0.359375,
       "learning_rate": 0.00015815965489224745,
-      "loss": 0.0119,
+      "loss": 0.0178,
       "step": 1380
     },
     {
       "epoch": 3.0385038503850383,
-      "grad_norm": 0.25,
+      "grad_norm": 0.33984375,
       "learning_rate": 0.00015810322557797027,
-      "loss": 0.0062,
+      "loss": 0.0077,
       "step": 1381
     },
     {
       "epoch": 3.0407040704070405,
-      "grad_norm": 0.220703125,
+      "grad_norm": 0.3203125,
       "learning_rate": 0.00015804676831872834,
-      "loss": 0.0067,
+      "loss": 0.0142,
       "step": 1382
     },
     {
       "epoch": 3.042904290429043,
-      "grad_norm": 0.314453125,
+      "grad_norm": 0.37109375,
       "learning_rate": 0.00015799028314167495,
-      "loss": 0.0076,
+      "loss": 0.0211,
       "step": 1383
     },
     {
       "epoch": 3.0451045104510452,
-      "grad_norm": 0.095703125,
+      "grad_norm": 0.267578125,
       "learning_rate": 0.00015793377007397683,
-      "loss": 0.0035,
+      "loss": 0.0075,
       "step": 1384
     },
     {
       "epoch": 3.0473047304730474,
-      "grad_norm": 0.189453125,
+      "grad_norm": 0.275390625,
       "learning_rate": 0.0001578772291428142,
-      "loss": 0.007,
+      "loss": 0.0105,
       "step": 1385
     },
     {
       "epoch": 3.0495049504950495,
-      "grad_norm": 0.30078125,
+      "grad_norm": 0.39453125,
       "learning_rate": 0.00015782066037538063,
-      "loss": 0.0112,
+      "loss": 0.0119,
       "step": 1386
     },
     {
       "epoch": 3.0517051705170517,
-      "grad_norm": 0.2451171875,
+      "grad_norm": 0.2353515625,
       "learning_rate": 0.00015776406379888308,
-      "loss": 0.0084,
+      "loss": 0.0077,
       "step": 1387
     },
     {
       "epoch": 3.053905390539054,
-      "grad_norm": 0.3125,
+      "grad_norm": 0.318359375,
       "learning_rate": 0.00015770743944054188,
-      "loss": 0.0184,
+      "loss": 0.0181,
       "step": 1388
     },
     {
       "epoch": 3.056105610561056,
-      "grad_norm": 0.283203125,
+      "grad_norm": 0.255859375,
       "learning_rate": 0.00015765078732759067,
-      "loss": 0.0104,
+      "loss": 0.0089,
       "step": 1389
     },
     {
       "epoch": 3.058305830583058,
-      "grad_norm": 0.2373046875,
+      "grad_norm": 0.349609375,
       "learning_rate": 0.00015759410748727662,
-      "loss": 0.0083,
+      "loss": 0.0117,
       "step": 1390
     },
     {
       "epoch": 3.0605060506050603,
-      "grad_norm": 0.408203125,
+      "grad_norm": 0.2734375,
       "learning_rate": 0.00015753739994686,
-      "loss": 0.0101,
+      "loss": 0.0068,
       "step": 1391
     },
     {
       "epoch": 3.062706270627063,
-      "grad_norm": 0.1630859375,
+      "grad_norm": 0.30078125,
       "learning_rate": 0.00015748066473361455,
-      "loss": 0.0042,
+      "loss": 0.0115,
       "step": 1392
     },
     {
       "epoch": 3.064906490649065,
-      "grad_norm": 0.20703125,
+      "grad_norm": 0.31640625,
       "learning_rate": 0.0001574239018748273,
-      "loss": 0.0064,
+      "loss": 0.0133,
       "step": 1393
     },
     {
       "epoch": 3.067106710671067,
-      "grad_norm": 0.251953125,
+      "grad_norm": 0.36328125,
       "learning_rate": 0.00015736711139779856,
-      "loss": 0.0074,
+      "loss": 0.0116,
       "step": 1394
     },
     {
       "epoch": 3.0693069306930694,
-      "grad_norm": 0.2060546875,
+      "grad_norm": 0.265625,
       "learning_rate": 0.0001573102933298419,
-      "loss": 0.0091,
+      "loss": 0.0156,
       "step": 1395
     },
     {
       "epoch": 3.0715071507150715,
-      "grad_norm": 0.31640625,
+      "grad_norm": 0.3125,
       "learning_rate": 0.00015725344769828423,
-      "loss": 0.0088,
+      "loss": 0.0137,
       "step": 1396
     },
     {
       "epoch": 3.0737073707370737,
-      "grad_norm": 0.2314453125,
+      "grad_norm": 0.5,
       "learning_rate": 0.00015719657453046565,
-      "loss": 0.0075,
+      "loss": 0.0119,
       "step": 1397
     },
     {
       "epoch": 3.075907590759076,
-      "grad_norm": 0.2265625,
+      "grad_norm": 0.1875,
       "learning_rate": 0.00015713967385373948,
-      "loss": 0.01,
+      "loss": 0.0117,
       "step": 1398
     },
     {
       "epoch": 3.078107810781078,
-      "grad_norm": 0.4296875,
+      "grad_norm": 0.35546875,
       "learning_rate": 0.00015708274569547231,
-      "loss": 0.019,
+      "loss": 0.0179,
       "step": 1399
     },
     {
       "epoch": 3.08030803080308,
-      "grad_norm": 0.431640625,
+      "grad_norm": 0.244140625,
       "learning_rate": 0.00015702579008304403,
-      "loss": 0.0179,
+      "loss": 0.011,
       "step": 1400
     },
     {
       "epoch": 3.0825082508250823,
-      "grad_norm": 0.302734375,
+      "grad_norm": 0.1591796875,
       "learning_rate": 0.0001569688070438476,
-      "loss": 0.0051,
+      "loss": 0.0054,
       "step": 1401
     },
     {
       "epoch": 3.084708470847085,
-      "grad_norm": 0.0859375,
+      "grad_norm": 0.154296875,
       "learning_rate": 0.0001569117966052893,
-      "loss": 0.002,
+      "loss": 0.005,
       "step": 1402
     },
     {
       "epoch": 3.086908690869087,
-      "grad_norm": 0.244140625,
+      "grad_norm": 0.267578125,
       "learning_rate": 0.00015685475879478842,
-      "loss": 0.0105,
+      "loss": 0.0129,
       "step": 1403
     },
     {
       "epoch": 3.089108910891089,
-      "grad_norm": 0.3515625,
+      "grad_norm": 0.2275390625,
       "learning_rate": 0.00015679769363977753,
-      "loss": 0.0088,
+      "loss": 0.0108,
       "step": 1404
     },
     {
       "epoch": 3.0913091309130913,
-      "grad_norm": 0.154296875,
+      "grad_norm": 0.232421875,
       "learning_rate": 0.00015674060116770236,
-      "loss": 0.0082,
+      "loss": 0.0097,
       "step": 1405
     },
     {
       "epoch": 3.0935093509350935,
-      "grad_norm": 0.2470703125,
+      "grad_norm": 0.2392578125,
       "learning_rate": 0.00015668348140602173,
-      "loss": 0.0085,
+      "loss": 0.0101,
       "step": 1406
     },
     {
       "epoch": 3.0957095709570956,
-      "grad_norm": 0.259765625,
+      "grad_norm": 0.291015625,
       "learning_rate": 0.00015662633438220762,
-      "loss": 0.0064,
+      "loss": 0.014,
       "step": 1407
     },
     {
       "epoch": 3.097909790979098,
-      "grad_norm": 0.1728515625,
+      "grad_norm": 0.2431640625,
       "learning_rate": 0.00015656916012374509,
-      "loss": 0.0069,
+      "loss": 0.0113,
       "step": 1408
     },
     {
       "epoch": 3.1001100110011,
-      "grad_norm": 0.35546875,
+      "grad_norm": 0.26953125,
       "learning_rate": 0.00015651195865813234,
-      "loss": 0.0162,
+      "loss": 0.0087,
       "step": 1409
     },
     {
       "epoch": 3.102310231023102,
-      "grad_norm": 0.1923828125,
+      "grad_norm": 0.25390625,
       "learning_rate": 0.00015645473001288057,
-      "loss": 0.0082,
+      "loss": 0.0089,
       "step": 1410
     },
     {
       "epoch": 3.1045104510451047,
-      "grad_norm": 0.2265625,
+      "grad_norm": 0.267578125,
       "learning_rate": 0.00015639747421551417,
-      "loss": 0.0126,
+      "loss": 0.011,
       "step": 1411
     },
     {
       "epoch": 3.106710671067107,
-      "grad_norm": 0.19140625,
+      "grad_norm": 0.2470703125,
       "learning_rate": 0.0001563401912935705,
-      "loss": 0.0032,
+      "loss": 0.0047,
       "step": 1412
     },
     {
       "epoch": 3.108910891089109,
-      "grad_norm": 0.21875,
+      "grad_norm": 0.13671875,
       "learning_rate": 0.00015628288127460005,
-      "loss": 0.0076,
+      "loss": 0.0049,
       "step": 1413
     },
     {
       "epoch": 3.111111111111111,
-      "grad_norm": 0.30078125,
+      "grad_norm": 0.1552734375,
       "learning_rate": 0.00015622554418616625,
-      "loss": 0.0061,
+      "loss": 0.0052,
       "step": 1414
     },
     {
       "epoch": 3.1133113311331133,
-      "grad_norm": 0.2578125,
+      "grad_norm": 0.2265625,
       "learning_rate": 0.00015616818005584554,
-      "loss": 0.0125,
+      "loss": 0.0039,
       "step": 1415
     },
     {
       "epoch": 3.1155115511551155,
-      "grad_norm": 0.1904296875,
+      "grad_norm": 0.28515625,
       "learning_rate": 0.0001561107889112275,
-      "loss": 0.008,
+      "loss": 0.0103,
       "step": 1416
     },
     {
       "epoch": 3.1177117711771176,
-      "grad_norm": 0.447265625,
+      "grad_norm": 0.318359375,
       "learning_rate": 0.00015605337077991455,
-      "loss": 0.0147,
+      "loss": 0.0154,
       "step": 1417
     },
     {
       "epoch": 3.1199119911991198,
-      "grad_norm": 0.162109375,
+      "grad_norm": 0.154296875,
       "learning_rate": 0.00015599592568952221,
-      "loss": 0.005,
+      "loss": 0.0054,
       "step": 1418
     },
     {
       "epoch": 3.122112211221122,
-      "grad_norm": 0.1328125,
+      "grad_norm": 0.126953125,
       "learning_rate": 0.0001559384536676789,
-      "loss": 0.0036,
+      "loss": 0.0049,
       "step": 1419
     },
     {
       "epoch": 3.1243124312431245,
-      "grad_norm": 0.1689453125,
+      "grad_norm": 0.3515625,
       "learning_rate": 0.00015588095474202595,
-      "loss": 0.0057,
+      "loss": 0.0125,
       "step": 1420
     },
     {
       "epoch": 3.1265126512651267,
-      "grad_norm": 0.140625,
+      "grad_norm": 0.26171875,
       "learning_rate": 0.00015582342894021774,
-      "loss": 0.0039,
+      "loss": 0.0048,
       "step": 1421
     },
     {
       "epoch": 3.128712871287129,
-      "grad_norm": 0.25,
+      "grad_norm": 0.2041015625,
       "learning_rate": 0.0001557658762899215,
-      "loss": 0.0045,
+      "loss": 0.0038,
       "step": 1422
     },
     {
       "epoch": 3.130913091309131,
-      "grad_norm": 0.294921875,
+      "grad_norm": 0.1630859375,
       "learning_rate": 0.00015570829681881742,
-      "loss": 0.0043,
+      "loss": 0.0047,
       "step": 1423
     },
     {
       "epoch": 3.133113311331133,
-      "grad_norm": 0.40625,
+      "grad_norm": 0.28515625,
       "learning_rate": 0.00015565069055459855,
-      "loss": 0.0257,
+      "loss": 0.0059,
       "step": 1424
     },
     {
       "epoch": 3.1353135313531353,
-      "grad_norm": 0.388671875,
+      "grad_norm": 0.1572265625,
       "learning_rate": 0.00015559305752497083,
-      "loss": 0.0161,
+      "loss": 0.0062,
       "step": 1425
     },
     {
       "epoch": 3.1375137513751374,
-      "grad_norm": 0.2490234375,
+      "grad_norm": 0.4140625,
       "learning_rate": 0.00015553539775765306,
-      "loss": 0.0082,
+      "loss": 0.0134,
       "step": 1426
     },
     {
       "epoch": 3.1397139713971396,
-      "grad_norm": 0.2734375,
+      "grad_norm": 0.326171875,
       "learning_rate": 0.00015547771128037697,
-      "loss": 0.0118,
+      "loss": 0.0033,
       "step": 1427
     },
     {
       "epoch": 3.1419141914191417,
-      "grad_norm": 0.310546875,
+      "grad_norm": 0.58984375,
       "learning_rate": 0.00015541999812088703,
-      "loss": 0.005,
+      "loss": 0.0131,
       "step": 1428
     },
     {
       "epoch": 3.1441144114411443,
-      "grad_norm": 0.1767578125,
+      "grad_norm": 0.25,
       "learning_rate": 0.00015536225830694068,
-      "loss": 0.0048,
+      "loss": 0.0064,
       "step": 1429
     },
     {
       "epoch": 3.1463146314631465,
-      "grad_norm": 0.271484375,
+      "grad_norm": 0.2265625,
       "learning_rate": 0.00015530449186630803,
-      "loss": 0.0075,
+      "loss": 0.0083,
       "step": 1430
     },
     {
       "epoch": 3.1485148514851486,
-      "grad_norm": 0.2216796875,
+      "grad_norm": 0.48828125,
       "learning_rate": 0.0001552466988267721,
-      "loss": 0.0087,
+      "loss": 0.0122,
       "step": 1431
     },
     {
       "epoch": 3.150715071507151,
-      "grad_norm": 0.154296875,
+      "grad_norm": 0.26171875,
       "learning_rate": 0.00015518887921612867,
-      "loss": 0.0049,
+      "loss": 0.0097,
       "step": 1432
     },
     {
       "epoch": 3.152915291529153,
-      "grad_norm": 0.140625,
+      "grad_norm": 0.29296875,
       "learning_rate": 0.0001551310330621863,
-      "loss": 0.0138,
+      "loss": 0.0156,
       "step": 1433
     },
     {
       "epoch": 3.155115511551155,
-      "grad_norm": 0.26953125,
+      "grad_norm": 0.46484375,
       "learning_rate": 0.0001550731603927663,
-      "loss": 0.0059,
+      "loss": 0.0208,
       "step": 1434
     },
     {
       "epoch": 3.1573157315731573,
-      "grad_norm": 0.255859375,
+      "grad_norm": 0.3203125,
       "learning_rate": 0.00015501526123570277,
-      "loss": 0.0062,
+      "loss": 0.0088,
       "step": 1435
     },
     {
       "epoch": 3.1595159515951594,
-      "grad_norm": 0.23046875,
+      "grad_norm": 0.1806640625,
       "learning_rate": 0.0001549573356188425,
-      "loss": 0.0064,
+      "loss": 0.0105,
       "step": 1436
     },
     {
       "epoch": 3.1617161716171616,
-      "grad_norm": 0.38671875,
+      "grad_norm": 0.30078125,
       "learning_rate": 0.00015489938357004504,
-      "loss": 0.0143,
+      "loss": 0.0164,
       "step": 1437
     },
     {
       "epoch": 3.1639163916391637,
-      "grad_norm": 0.263671875,
+      "grad_norm": 0.115234375,
       "learning_rate": 0.00015484140511718263,
-      "loss": 0.0057,
+      "loss": 0.0044,
       "step": 1438
     },
     {
       "epoch": 3.1661166116611663,
-      "grad_norm": 0.310546875,
+      "grad_norm": 0.52734375,
       "learning_rate": 0.00015478340028814028,
-      "loss": 0.0129,
+      "loss": 0.025,
       "step": 1439
     },
     {
       "epoch": 3.1683168316831685,
-      "grad_norm": 0.33203125,
+      "grad_norm": 0.1904296875,
       "learning_rate": 0.00015472536911081562,
-      "loss": 0.0074,
+      "loss": 0.0073,
       "step": 1440
     },
     {
       "epoch": 3.1705170517051706,
-      "grad_norm": 0.09912109375,
+      "grad_norm": 0.341796875,
       "learning_rate": 0.0001546673116131189,
-      "loss": 0.0013,
+      "loss": 0.0058,
       "step": 1441
     },
     {
       "epoch": 3.1727172717271728,
-      "grad_norm": 0.373046875,
+      "grad_norm": 0.232421875,
       "learning_rate": 0.00015460922782297313,
-      "loss": 0.0081,
+      "loss": 0.0039,
       "step": 1442
     },
     {
       "epoch": 3.174917491749175,
-      "grad_norm": 0.2734375,
+      "grad_norm": 0.255859375,
       "learning_rate": 0.00015455111776831396,
-      "loss": 0.0089,
+      "loss": 0.0077,
       "step": 1443
     },
     {
       "epoch": 3.177117711771177,
-      "grad_norm": 0.359375,
+      "grad_norm": 0.251953125,
       "learning_rate": 0.00015449298147708954,
-      "loss": 0.0125,
+      "loss": 0.0138,
       "step": 1444
     },
     {
       "epoch": 3.1793179317931792,
-      "grad_norm": 0.1572265625,
+      "grad_norm": 0.3984375,
       "learning_rate": 0.00015443481897726085,
-      "loss": 0.0037,
+      "loss": 0.0166,
       "step": 1445
     },
     {
       "epoch": 3.1815181518151814,
-      "grad_norm": 0.185546875,
+      "grad_norm": 0.333984375,
       "learning_rate": 0.00015437663029680137,
-      "loss": 0.0077,
+      "loss": 0.0101,
       "step": 1446
     },
     {
       "epoch": 3.1837183718371835,
-      "grad_norm": 0.279296875,
+      "grad_norm": 0.390625,
       "learning_rate": 0.00015431841546369705,
-      "loss": 0.0135,
+      "loss": 0.018,
       "step": 1447
     },
     {
       "epoch": 3.1859185918591857,
-      "grad_norm": 0.431640625,
+      "grad_norm": 0.54296875,
       "learning_rate": 0.00015426017450594664,
-      "loss": 0.0081,
+      "loss": 0.0251,
       "step": 1448
     },
     {
       "epoch": 3.1881188118811883,
-      "grad_norm": 0.306640625,
+      "grad_norm": 0.3828125,
       "learning_rate": 0.00015420190745156126,
-      "loss": 0.008,
+      "loss": 0.0162,
       "step": 1449
     },
     {
       "epoch": 3.1903190319031904,
-      "grad_norm": 0.44140625,
+      "grad_norm": 0.318359375,
       "learning_rate": 0.00015414361432856475,
-      "loss": 0.0108,
+      "loss": 0.0135,
       "step": 1450
     },
     {
       "epoch": 3.1925192519251926,
-      "grad_norm": 0.291015625,
+      "grad_norm": 0.390625,
       "learning_rate": 0.00015408529516499334,
-      "loss": 0.0065,
+      "loss": 0.0176,
       "step": 1451
     },
     {
       "epoch": 3.1947194719471947,
-      "grad_norm": 0.294921875,
+      "grad_norm": 0.2578125,
       "learning_rate": 0.00015402694998889588,
-      "loss": 0.0056,
+      "loss": 0.0085,
       "step": 1452
     },
     {
       "epoch": 3.196919691969197,
-      "grad_norm": 0.28125,
+      "grad_norm": 0.212890625,
       "learning_rate": 0.00015396857882833372,
-      "loss": 0.0098,
+      "loss": 0.0112,
       "step": 1453
     },
     {
       "epoch": 3.199119911991199,
-      "grad_norm": 0.287109375,
+      "grad_norm": 0.1875,
       "learning_rate": 0.0001539101817113807,
-      "loss": 0.01,
+      "loss": 0.0131,
       "step": 1454
     },
     {
       "epoch": 3.201320132013201,
-      "grad_norm": 0.14453125,
+      "grad_norm": 0.248046875,
       "learning_rate": 0.00015385175866612305,
-      "loss": 0.0032,
+      "loss": 0.0075,
       "step": 1455
     },
     {
       "epoch": 3.2035203520352034,
-      "grad_norm": 0.1533203125,
+      "grad_norm": 0.322265625,
       "learning_rate": 0.00015379330972065965,
-      "loss": 0.0065,
+      "loss": 0.0114,
       "step": 1456
     },
     {
       "epoch": 3.2057205720572055,
-      "grad_norm": 0.3046875,
+      "grad_norm": 0.349609375,
       "learning_rate": 0.00015373483490310167,
-      "loss": 0.0086,
+      "loss": 0.0133,
       "step": 1457
     },
     {
       "epoch": 3.207920792079208,
-      "grad_norm": 0.2353515625,
+      "grad_norm": 0.2412109375,
       "learning_rate": 0.00015367633424157284,
-      "loss": 0.0089,
+      "loss": 0.0144,
       "step": 1458
     },
     {
       "epoch": 3.2101210121012103,
-      "grad_norm": 0.1962890625,
+      "grad_norm": 0.1640625,
       "learning_rate": 0.00015361780776420924,
-      "loss": 0.0033,
+      "loss": 0.0049,
       "step": 1459
     },
     {
       "epoch": 3.2123212321232124,
-      "grad_norm": 0.390625,
+      "grad_norm": 0.1767578125,
       "learning_rate": 0.00015355925549915943,
-      "loss": 0.0234,
+      "loss": 0.0102,
       "step": 1460
     },
     {
       "epoch": 3.2145214521452146,
-      "grad_norm": 0.267578125,
+      "grad_norm": 0.1669921875,
       "learning_rate": 0.0001535006774745843,
-      "loss": 0.0103,
+      "loss": 0.0078,
       "step": 1461
     },
     {
       "epoch": 3.2167216721672167,
-      "grad_norm": 0.1689453125,
+      "grad_norm": 0.283203125,
       "learning_rate": 0.0001534420737186572,
-      "loss": 0.0073,
+      "loss": 0.0089,
       "step": 1462
     },
     {
       "epoch": 3.218921892189219,
-      "grad_norm": 0.283203125,
+      "grad_norm": 0.2275390625,
       "learning_rate": 0.00015338344425956386,
-      "loss": 0.011,
+      "loss": 0.0117,
       "step": 1463
     },
     {
       "epoch": 3.221122112211221,
-      "grad_norm": 0.34765625,
+      "grad_norm": 0.1591796875,
       "learning_rate": 0.00015332478912550229,
-      "loss": 0.0072,
+      "loss": 0.0063,
       "step": 1464
     },
     {
       "epoch": 3.223322332233223,
-      "grad_norm": 0.39453125,
+      "grad_norm": 0.193359375,
       "learning_rate": 0.0001532661083446829,
-      "loss": 0.0109,
+      "loss": 0.0088,
       "step": 1465
     },
     {
       "epoch": 3.2255225522552253,
-      "grad_norm": 0.28125,
+      "grad_norm": 0.2294921875,
       "learning_rate": 0.0001532074019453284,
-      "loss": 0.0109,
+      "loss": 0.0074,
       "step": 1466
     },
     {
       "epoch": 3.227722772277228,
-      "grad_norm": 0.158203125,
+      "grad_norm": 0.197265625,
       "learning_rate": 0.00015314866995567395,
-      "loss": 0.0047,
+      "loss": 0.0128,
       "step": 1467
     },
     {
       "epoch": 3.22992299229923,
-      "grad_norm": 0.248046875,
+      "grad_norm": 0.45703125,
       "learning_rate": 0.00015308991240396686,
-      "loss": 0.0073,
+      "loss": 0.0118,
       "step": 1468
     },
     {
       "epoch": 3.2321232123212322,
-      "grad_norm": 0.076171875,
+      "grad_norm": 0.341796875,
       "learning_rate": 0.0001530311293184668,
-      "loss": 0.0041,
+      "loss": 0.0084,
       "step": 1469
     },
     {
       "epoch": 3.2343234323432344,
-      "grad_norm": 0.23046875,
+      "grad_norm": 0.34765625,
       "learning_rate": 0.00015297232072744574,
-      "loss": 0.0052,
+      "loss": 0.0106,
       "step": 1470
     },
     {
       "epoch": 3.2365236523652365,
-      "grad_norm": 0.1650390625,
+      "grad_norm": 0.2041015625,
       "learning_rate": 0.00015291348665918782,
-      "loss": 0.0069,
+      "loss": 0.0075,
       "step": 1471
     },
     {
       "epoch": 3.2387238723872387,
-      "grad_norm": 0.1513671875,
+      "grad_norm": 0.2041015625,
       "learning_rate": 0.00015285462714198958,
-      "loss": 0.0071,
+      "loss": 0.007,
       "step": 1472
     },
     {
       "epoch": 3.240924092409241,
-      "grad_norm": 0.2431640625,
+      "grad_norm": 0.1884765625,
       "learning_rate": 0.0001527957422041597,
-      "loss": 0.01,
+      "loss": 0.0054,
       "step": 1473
     },
     {
       "epoch": 3.243124312431243,
-      "grad_norm": 0.2177734375,
+      "grad_norm": 0.181640625,
       "learning_rate": 0.00015273683187401913,
-      "loss": 0.0046,
+      "loss": 0.006,
       "step": 1474
     },
     {
       "epoch": 3.245324532453245,
-      "grad_norm": 0.353515625,
+      "grad_norm": 0.306640625,
       "learning_rate": 0.000152677896179901,
-      "loss": 0.0074,
+      "loss": 0.0066,
       "step": 1475
     },
     {
       "epoch": 3.2475247524752477,
-      "grad_norm": 0.171875,
+      "grad_norm": 0.353515625,
       "learning_rate": 0.00015261893515015066,
-      "loss": 0.0035,
+      "loss": 0.0084,
       "step": 1476
     },
     {
       "epoch": 3.24972497249725,
-      "grad_norm": 0.302734375,
+      "grad_norm": 0.349609375,
       "learning_rate": 0.0001525599488131256,
-      "loss": 0.0095,
+      "loss": 0.011,
       "step": 1477
     },
     {
       "epoch": 3.251925192519252,
-      "grad_norm": 0.2041015625,
+      "grad_norm": 0.37109375,
       "learning_rate": 0.00015250093719719555,
-      "loss": 0.0034,
+      "loss": 0.0084,
       "step": 1478
     },
     {
       "epoch": 3.254125412541254,
-      "grad_norm": 0.0986328125,
+      "grad_norm": 0.11083984375,
       "learning_rate": 0.00015244190033074243,
-      "loss": 0.0025,
+      "loss": 0.0022,
       "step": 1479
     },
     {
       "epoch": 3.2563256325632564,
-      "grad_norm": 0.1865234375,
+      "grad_norm": 0.5,
       "learning_rate": 0.00015238283824216015,
-      "loss": 0.0075,
+      "loss": 0.0138,
       "step": 1480
     },
     {
       "epoch": 3.2585258525852585,
-      "grad_norm": 0.2080078125,
+      "grad_norm": 0.21875,
       "learning_rate": 0.00015232375095985487,
-      "loss": 0.0048,
+      "loss": 0.0044,
       "step": 1481
     },
     {
       "epoch": 3.2607260726072607,
-      "grad_norm": 0.140625,
+      "grad_norm": 0.255859375,
       "learning_rate": 0.00015226463851224483,
-      "loss": 0.0057,
+      "loss": 0.0076,
       "step": 1482
     },
     {
       "epoch": 3.2607260726072607,
-      "eval_loss": 0.01804141327738762,
-      "eval_runtime": 10.6138,
-      "eval_samples_per_second": 36.085,
-      "eval_steps_per_second": 4.522,
+      "eval_loss": 0.0223908182233572,
+      "eval_runtime": 10.4248,
+      "eval_samples_per_second": 36.739,
+      "eval_steps_per_second": 4.604,
       "step": 1482
     },
     {
       "epoch": 3.262926292629263,
-      "grad_norm": 0.298828125,
+      "grad_norm": 0.10791015625,
       "learning_rate": 0.00015220550092776044,
-      "loss": 0.0057,
+      "loss": 0.0045,
       "step": 1483
     },
     {
       "epoch": 3.265126512651265,
-      "grad_norm": 0.248046875,
+      "grad_norm": 0.2041015625,
       "learning_rate": 0.0001521463382348441,
-      "loss": 0.0092,
+      "loss": 0.0049,
       "step": 1484
     },
     {
       "epoch": 3.2673267326732676,
-      "grad_norm": 0.244140625,
+      "grad_norm": 0.404296875,
       "learning_rate": 0.00015208715046195035,
-      "loss": 0.0093,
+      "loss": 0.0144,
       "step": 1485
     },
     {
       "epoch": 3.2695269526952697,
-      "grad_norm": 0.34375,
+      "grad_norm": 0.25390625,
       "learning_rate": 0.00015202793763754576,
-      "loss": 0.018,
+      "loss": 0.0145,
       "step": 1486
     },
     {
       "epoch": 3.271727172717272,
-      "grad_norm": 0.1904296875,
+      "grad_norm": 0.357421875,
       "learning_rate": 0.0001519686997901089,
-      "loss": 0.006,
+      "loss": 0.0085,
       "step": 1487
     },
     {
       "epoch": 3.273927392739274,
-      "grad_norm": 0.1767578125,
+      "grad_norm": 0.09765625,
       "learning_rate": 0.00015190943694813052,
-      "loss": 0.0095,
+      "loss": 0.0037,
       "step": 1488
     },
     {
       "epoch": 3.276127612761276,
-      "grad_norm": 0.154296875,
+      "grad_norm": 0.302734375,
       "learning_rate": 0.0001518501491401133,
-      "loss": 0.0046,
+      "loss": 0.0087,
       "step": 1489
     },
     {
       "epoch": 3.2783278327832783,
-      "grad_norm": 0.390625,
+      "grad_norm": 0.1826171875,
       "learning_rate": 0.0001517908363945719,
-      "loss": 0.0158,
+      "loss": 0.0066,
       "step": 1490
     },
     {
       "epoch": 3.2805280528052805,
-      "grad_norm": 0.515625,
+      "grad_norm": 0.3828125,
       "learning_rate": 0.00015173149874003308,
-      "loss": 0.0056,
+      "loss": 0.0084,
       "step": 1491
     },
     {
       "epoch": 3.2827282728272826,
-      "grad_norm": 0.216796875,
+      "grad_norm": 0.35546875,
       "learning_rate": 0.0001516721362050354,
-      "loss": 0.0062,
+      "loss": 0.0167,
       "step": 1492
     },
     {
       "epoch": 3.284928492849285,
-      "grad_norm": 0.2333984375,
+      "grad_norm": 0.33203125,
       "learning_rate": 0.00015161274881812954,
-      "loss": 0.0065,
+      "loss": 0.0094,
       "step": 1493
     },
     {
       "epoch": 3.287128712871287,
-      "grad_norm": 0.396484375,
+      "grad_norm": 0.2578125,
       "learning_rate": 0.00015155333660787806,
-      "loss": 0.0081,
+      "loss": 0.0039,
       "step": 1494
     },
     {
       "epoch": 3.289328932893289,
-      "grad_norm": 0.271484375,
+      "grad_norm": 0.466796875,
       "learning_rate": 0.00015149389960285558,
-      "loss": 0.008,
+      "loss": 0.0135,
       "step": 1495
     },
     {
       "epoch": 3.2915291529152917,
-      "grad_norm": 0.1591796875,
+      "grad_norm": 0.1826171875,
       "learning_rate": 0.00015143443783164847,
-      "loss": 0.0063,
+      "loss": 0.006,
       "step": 1496
     },
     {
       "epoch": 3.293729372937294,
-      "grad_norm": 0.306640625,
+      "grad_norm": 0.3046875,
       "learning_rate": 0.00015137495132285512,
-      "loss": 0.005,
+      "loss": 0.0054,
       "step": 1497
     },
     {
       "epoch": 3.295929592959296,
-      "grad_norm": 0.11083984375,
+      "grad_norm": 0.30078125,
       "learning_rate": 0.0001513154401050858,
-      "loss": 0.0027,
+      "loss": 0.0084,
       "step": 1498
     },
     {
       "epoch": 3.298129812981298,
-      "grad_norm": 0.287109375,
+      "grad_norm": 0.310546875,
       "learning_rate": 0.00015125590420696257,
-      "loss": 0.0048,
+      "loss": 0.0065,
       "step": 1499
     },
     {
       "epoch": 3.3003300330033003,
-      "grad_norm": 0.12158203125,
+      "grad_norm": 0.29296875,
       "learning_rate": 0.00015119634365711953,
-      "loss": 0.0125,
+      "loss": 0.0086,
       "step": 1500
     },
     {
       "epoch": 3.3025302530253025,
-      "grad_norm": 0.1845703125,
+      "grad_norm": 0.28515625,
       "learning_rate": 0.0001511367584842026,
-      "loss": 0.0077,
+      "loss": 0.0123,
       "step": 1501
     },
     {
       "epoch": 3.3047304730473046,
-      "grad_norm": 0.2333984375,
+      "grad_norm": 0.46484375,
       "learning_rate": 0.00015107714871686936,
-      "loss": 0.0072,
+      "loss": 0.0156,
       "step": 1502
     },
     {
       "epoch": 3.3069306930693068,
-      "grad_norm": 0.322265625,
+      "grad_norm": 0.25,
       "learning_rate": 0.00015101751438378942,
-      "loss": 0.0091,
+      "loss": 0.0132,
       "step": 1503
     },
     {
       "epoch": 3.309130913091309,
-      "grad_norm": 0.130859375,
+      "grad_norm": 0.443359375,
       "learning_rate": 0.00015095785551364412,
-      "loss": 0.0036,
+      "loss": 0.0303,
       "step": 1504
     },
     {
       "epoch": 3.3113311331133115,
-      "grad_norm": 0.251953125,
+      "grad_norm": 0.2451171875,
       "learning_rate": 0.00015089817213512663,
-      "loss": 0.0106,
+      "loss": 0.0151,
       "step": 1505
     },
     {
       "epoch": 3.3135313531353137,
-      "grad_norm": 0.30078125,
+      "grad_norm": 0.4921875,
       "learning_rate": 0.0001508384642769419,
-      "loss": 0.0048,
+      "loss": 0.0149,
       "step": 1506
     },
     {
       "epoch": 3.315731573157316,
-      "grad_norm": 0.267578125,
+      "grad_norm": 0.2021484375,
       "learning_rate": 0.00015077873196780666,
-      "loss": 0.0287,
+      "loss": 0.0065,
       "step": 1507
     },
     {
       "epoch": 3.317931793179318,
-      "grad_norm": 0.240234375,
+      "grad_norm": 0.2451171875,
       "learning_rate": 0.00015071897523644937,
-      "loss": 0.0069,
+      "loss": 0.0065,
       "step": 1508
     },
     {
       "epoch": 3.32013201320132,
-      "grad_norm": 0.2490234375,
+      "grad_norm": 0.349609375,
       "learning_rate": 0.00015065919411161026,
-      "loss": 0.0058,
+      "loss": 0.0127,
       "step": 1509
     },
     {
       "epoch": 3.3223322332233223,
-      "grad_norm": 0.2578125,
+      "grad_norm": 0.390625,
       "learning_rate": 0.00015059938862204127,
-      "loss": 0.007,
+      "loss": 0.0091,
       "step": 1510
     },
     {
       "epoch": 3.3245324532453244,
-      "grad_norm": 0.25,
+      "grad_norm": 0.330078125,
       "learning_rate": 0.00015053955879650607,
-      "loss": 0.006,
+      "loss": 0.0181,
       "step": 1511
     },
     {
       "epoch": 3.3267326732673266,
-      "grad_norm": 0.388671875,
+      "grad_norm": 0.24609375,
       "learning_rate": 0.00015047970466378012,
-      "loss": 0.0182,
+      "loss": 0.014,
       "step": 1512
     },
     {
       "epoch": 3.3289328932893287,
-      "grad_norm": 0.23828125,
+      "grad_norm": 0.57421875,
       "learning_rate": 0.0001504198262526504,
-      "loss": 0.0126,
+      "loss": 0.0202,
       "step": 1513
     },
     {
       "epoch": 3.3311331133113313,
-      "grad_norm": 0.16015625,
+      "grad_norm": 0.314453125,
       "learning_rate": 0.00015035992359191568,
-      "loss": 0.0061,
+      "loss": 0.0123,
       "step": 1514
     },
     {
       "epoch": 3.3333333333333335,
-      "grad_norm": 0.1796875,
+      "grad_norm": 0.19921875,
       "learning_rate": 0.00015029999671038635,
-      "loss": 0.0031,
+      "loss": 0.0096,
       "step": 1515
     },
     {
       "epoch": 3.3355335533553356,
-      "grad_norm": 0.201171875,
+      "grad_norm": 0.263671875,
       "learning_rate": 0.0001502400456368845,
-      "loss": 0.0025,
+      "loss": 0.0051,
       "step": 1516
     },
     {
       "epoch": 3.337733773377338,
-      "grad_norm": 0.1982421875,
+      "grad_norm": 0.326171875,
       "learning_rate": 0.00015018007040024382,
-      "loss": 0.0044,
+      "loss": 0.0161,
       "step": 1517
     },
     {
       "epoch": 3.33993399339934,
-      "grad_norm": 0.17578125,
+      "grad_norm": 0.1376953125,
       "learning_rate": 0.00015012007102930964,
-      "loss": 0.0059,
+      "loss": 0.0072,
       "step": 1518
     },
     {
       "epoch": 3.342134213421342,
-      "grad_norm": 0.384765625,
+      "grad_norm": 0.357421875,
       "learning_rate": 0.00015006004755293886,
-      "loss": 0.0101,
+      "loss": 0.0134,
       "step": 1519
     },
     {
       "epoch": 3.3443344334433442,
-      "grad_norm": 0.27734375,
+      "grad_norm": 0.1640625,
       "learning_rate": 0.00015000000000000001,
-      "loss": 0.0082,
+      "loss": 0.0076,
       "step": 1520
     },
     {
       "epoch": 3.3465346534653464,
-      "grad_norm": 0.1201171875,
+      "grad_norm": 0.41015625,
       "learning_rate": 0.00014993992839937317,
-      "loss": 0.003,
+      "loss": 0.0163,
       "step": 1521
     },
     {
       "epoch": 3.3487348734873486,
-      "grad_norm": 0.1123046875,
+      "grad_norm": 0.314453125,
       "learning_rate": 0.00014987983277995,
-      "loss": 0.0042,
+      "loss": 0.009,
       "step": 1522
     },
     {
       "epoch": 3.350935093509351,
-      "grad_norm": 0.1328125,
+      "grad_norm": 0.251953125,
       "learning_rate": 0.00014981971317063378,
-      "loss": 0.0049,
+      "loss": 0.0078,
       "step": 1523
     },
     {
       "epoch": 3.3531353135313533,
-      "grad_norm": 0.287109375,
+      "grad_norm": 0.2451171875,
       "learning_rate": 0.00014975956960033913,
-      "loss": 0.0034,
+      "loss": 0.0081,
       "step": 1524
     },
     {
       "epoch": 3.3553355335533555,
-      "grad_norm": 0.2080078125,
+      "grad_norm": 0.2421875,
       "learning_rate": 0.00014969940209799248,
-      "loss": 0.0119,
+      "loss": 0.0128,
       "step": 1525
     },
     {
       "epoch": 3.3575357535753576,
-      "grad_norm": 0.18359375,
+      "grad_norm": 0.244140625,
       "learning_rate": 0.0001496392106925315,
-      "loss": 0.0085,
+      "loss": 0.0105,
       "step": 1526
     },
     {
       "epoch": 3.3597359735973598,
-      "grad_norm": 0.23828125,
+      "grad_norm": 0.25,
       "learning_rate": 0.0001495789954129055,
-      "loss": 0.007,
+      "loss": 0.0097,
       "step": 1527
     },
     {
       "epoch": 3.361936193619362,
-      "grad_norm": 0.12109375,
+      "grad_norm": 0.27734375,
       "learning_rate": 0.00014951875628807528,
-      "loss": 0.0036,
+      "loss": 0.0078,
       "step": 1528
     },
     {
       "epoch": 3.364136413641364,
-      "grad_norm": 0.294921875,
+      "grad_norm": 0.251953125,
       "learning_rate": 0.00014945849334701308,
-      "loss": 0.0113,
+      "loss": 0.0102,
       "step": 1529
     },
     {
       "epoch": 3.366336633663366,
-      "grad_norm": 0.287109375,
+      "grad_norm": 0.1728515625,
       "learning_rate": 0.00014939820661870253,
-      "loss": 0.0079,
+      "loss": 0.0113,
       "step": 1530
     },
     {
       "epoch": 3.3685368536853684,
-      "grad_norm": 0.20703125,
+      "grad_norm": 0.220703125,
       "learning_rate": 0.00014933789613213877,
-      "loss": 0.0031,
+      "loss": 0.0085,
       "step": 1531
     },
     {
       "epoch": 3.370737073707371,
-      "grad_norm": 0.11572265625,
+      "grad_norm": 0.3203125,
       "learning_rate": 0.00014927756191632837,
-      "loss": 0.0055,
+      "loss": 0.0147,
       "step": 1532
     },
     {
       "epoch": 3.372937293729373,
-      "grad_norm": 0.185546875,
+      "grad_norm": 0.3671875,
       "learning_rate": 0.0001492172040002893,
-      "loss": 0.0047,
+      "loss": 0.0121,
       "step": 1533
     },
     {
       "epoch": 3.3751375137513753,
-      "grad_norm": 0.33203125,
+      "grad_norm": 0.166015625,
       "learning_rate": 0.000149156822413051,
-      "loss": 0.0101,
+      "loss": 0.0064,
       "step": 1534
     },
     {
       "epoch": 3.3773377337733774,
-      "grad_norm": 0.1435546875,
+      "grad_norm": 0.16796875,
       "learning_rate": 0.00014909641718365411,
-      "loss": 0.0058,
+      "loss": 0.0044,
       "step": 1535
     },
     {
       "epoch": 3.3795379537953796,
-      "grad_norm": 0.2734375,
+      "grad_norm": 0.1767578125,
       "learning_rate": 0.00014903598834115078,
-      "loss": 0.0058,
+      "loss": 0.0059,
       "step": 1536
     },
     {
       "epoch": 3.3817381738173817,
-      "grad_norm": 0.361328125,
+      "grad_norm": 0.267578125,
       "learning_rate": 0.00014897553591460453,
-      "loss": 0.0137,
+      "loss": 0.0095,
       "step": 1537
     },
     {
       "epoch": 3.383938393839384,
-      "grad_norm": 0.138671875,
+      "grad_norm": 0.271484375,
       "learning_rate": 0.0001489150599330901,
-      "loss": 0.0036,
+      "loss": 0.006,
       "step": 1538
     },
     {
       "epoch": 3.386138613861386,
-      "grad_norm": 0.21484375,
+      "grad_norm": 0.318359375,
       "learning_rate": 0.00014885456042569372,
-      "loss": 0.0061,
+      "loss": 0.0138,
       "step": 1539
     },
     {
       "epoch": 3.388338833883388,
-      "grad_norm": 0.384765625,
+      "grad_norm": 0.17578125,
       "learning_rate": 0.00014879403742151283,
-      "loss": 0.0103,
+      "loss": 0.0073,
       "step": 1540
     },
     {
       "epoch": 3.390539053905391,
-      "grad_norm": 0.2890625,
+      "grad_norm": 0.2041015625,
       "learning_rate": 0.00014873349094965621,
-      "loss": 0.0037,
+      "loss": 0.0045,
       "step": 1541
     },
     {
       "epoch": 3.3927392739273925,
-      "grad_norm": 0.1943359375,
+      "grad_norm": 0.23046875,
       "learning_rate": 0.00014867292103924387,
-      "loss": 0.0089,
+      "loss": 0.0114,
       "step": 1542
     },
     {
       "epoch": 3.394939493949395,
-      "grad_norm": 0.181640625,
+      "grad_norm": 0.2255859375,
       "learning_rate": 0.00014861232771940717,
-      "loss": 0.0066,
+      "loss": 0.0085,
       "step": 1543
     },
     {
       "epoch": 3.3971397139713972,
-      "grad_norm": 0.2138671875,
+      "grad_norm": 0.224609375,
       "learning_rate": 0.00014855171101928872,
-      "loss": 0.0043,
+      "loss": 0.0051,
       "step": 1544
     },
     {
       "epoch": 3.3993399339933994,
-      "grad_norm": 0.23046875,
+      "grad_norm": 0.291015625,
       "learning_rate": 0.00014849107096804231,
-      "loss": 0.0086,
+      "loss": 0.0122,
       "step": 1545
     },
     {
       "epoch": 3.4015401540154016,
-      "grad_norm": 0.2197265625,
+      "grad_norm": 0.3671875,
       "learning_rate": 0.000148430407594833,
-      "loss": 0.003,
+      "loss": 0.0074,
       "step": 1546
     },
     {
       "epoch": 3.4037403740374037,
-      "grad_norm": 0.421875,
+      "grad_norm": 0.3203125,
       "learning_rate": 0.0001483697209288371,
-      "loss": 0.0177,
+      "loss": 0.0068,
       "step": 1547
     },
     {
       "epoch": 3.405940594059406,
-      "grad_norm": 0.296875,
+      "grad_norm": 0.419921875,
       "learning_rate": 0.00014830901099924205,
-      "loss": 0.0042,
+      "loss": 0.0136,
       "step": 1548
     },
     {
       "epoch": 3.408140814081408,
-      "grad_norm": 0.234375,
+      "grad_norm": 0.12109375,
       "learning_rate": 0.0001482482778352465,
-      "loss": 0.0043,
+      "loss": 0.0044,
       "step": 1549
     },
     {
       "epoch": 3.41034103410341,
-      "grad_norm": 0.171875,
+      "grad_norm": 0.16015625,
       "learning_rate": 0.0001481875214660604,
-      "loss": 0.0066,
+      "loss": 0.0092,
       "step": 1550
     },
     {
       "epoch": 3.4125412541254123,
-      "grad_norm": 0.1220703125,
+      "grad_norm": 0.34375,
       "learning_rate": 0.00014812674192090465,
-      "loss": 0.0047,
+      "loss": 0.011,
       "step": 1551
     },
     {
       "epoch": 3.414741474147415,
-      "grad_norm": 0.17578125,
+      "grad_norm": 0.349609375,
       "learning_rate": 0.00014806593922901145,
-      "loss": 0.0077,
+      "loss": 0.0114,
       "step": 1552
     },
     {
       "epoch": 3.416941694169417,
-      "grad_norm": 0.224609375,
+      "grad_norm": 0.24609375,
       "learning_rate": 0.00014800511341962405,
-      "loss": 0.0036,
+      "loss": 0.005,
       "step": 1553
     },
     {
       "epoch": 3.419141914191419,
-      "grad_norm": 0.3359375,
+      "grad_norm": 0.1845703125,
       "learning_rate": 0.00014794426452199687,
       "loss": 0.0089,
       "step": 1554
     },
     {
       "epoch": 3.4213421342134214,
-      "grad_norm": 0.189453125,
+      "grad_norm": 0.30859375,
       "learning_rate": 0.00014788339256539544,
-      "loss": 0.0058,
+      "loss": 0.0072,
       "step": 1555
     },
     {
       "epoch": 3.4235423542354235,
-      "grad_norm": 0.130859375,
+      "grad_norm": 0.30859375,
       "learning_rate": 0.0001478224975790963,
-      "loss": 0.0021,
+      "loss": 0.0065,
       "step": 1556
     },
     {
       "epoch": 3.4257425742574257,
-      "grad_norm": 0.189453125,
+      "grad_norm": 0.11181640625,
       "learning_rate": 0.00014776157959238716,
-      "loss": 0.0089,
+      "loss": 0.0035,
       "step": 1557
     },
     {
       "epoch": 3.427942794279428,
-      "grad_norm": 0.1923828125,
+      "grad_norm": 0.462890625,
       "learning_rate": 0.00014770063863456676,
-      "loss": 0.0144,
+      "loss": 0.0108,
       "step": 1558
     },
     {
       "epoch": 3.43014301430143,
-      "grad_norm": 0.05810546875,
+      "grad_norm": 0.1611328125,
       "learning_rate": 0.0001476396747349449,
-      "loss": 0.0036,
+      "loss": 0.0066,
       "step": 1559
     },
     {
       "epoch": 3.432343234323432,
-      "grad_norm": 0.361328125,
+      "grad_norm": 0.1396484375,
       "learning_rate": 0.0001475786879228423,
-      "loss": 0.0052,
+      "loss": 0.0047,
       "step": 1560
     },
     {
       "epoch": 3.4345434543454347,
-      "grad_norm": 0.3828125,
+      "grad_norm": 0.380859375,
       "learning_rate": 0.00014751767822759093,
-      "loss": 0.007,
+      "loss": 0.0093,
       "step": 1561
     },
     {
       "epoch": 3.436743674367437,
-      "grad_norm": 0.123046875,
+      "grad_norm": 0.08447265625,
       "learning_rate": 0.00014745664567853355,
-      "loss": 0.0033,
+      "loss": 0.0034,
       "step": 1562
     },
     {
       "epoch": 3.438943894389439,
-      "grad_norm": 0.408203125,
+      "grad_norm": 0.2236328125,
       "learning_rate": 0.000147395590305024,
-      "loss": 0.0099,
+      "loss": 0.01,
       "step": 1563
     },
     {
       "epoch": 3.441144114411441,
-      "grad_norm": 0.1533203125,
+      "grad_norm": 0.21875,
       "learning_rate": 0.00014733451213642712,
-      "loss": 0.0036,
+      "loss": 0.0054,
       "step": 1564
     },
     {
       "epoch": 3.4433443344334433,
-      "grad_norm": 0.2216796875,
+      "grad_norm": 0.2294921875,
       "learning_rate": 0.00014727341120211867,
-      "loss": 0.003,
+      "loss": 0.0043,
       "step": 1565
     },
     {
       "epoch": 3.4455445544554455,
-      "grad_norm": 0.1806640625,
+      "grad_norm": 0.34375,
       "learning_rate": 0.00014721228753148538,
-      "loss": 0.0105,
+      "loss": 0.0138,
       "step": 1566
     },
     {
       "epoch": 3.4477447744774476,
-      "grad_norm": 0.12158203125,
+      "grad_norm": 0.435546875,
       "learning_rate": 0.0001471511411539249,
-      "loss": 0.0054,
+      "loss": 0.0113,
       "step": 1567
     },
     {
       "epoch": 3.44994499449945,
-      "grad_norm": 0.2041015625,
+      "grad_norm": 0.259765625,
       "learning_rate": 0.0001470899720988458,
-      "loss": 0.0063,
+      "loss": 0.009,
       "step": 1568
     },
     {
       "epoch": 3.452145214521452,
-      "grad_norm": 0.365234375,
+      "grad_norm": 0.1591796875,
       "learning_rate": 0.00014702878039566758,
-      "loss": 0.0082,
+      "loss": 0.0052,
       "step": 1569
     },
     {
       "epoch": 3.4543454345434546,
-      "grad_norm": 0.07763671875,
+      "grad_norm": 0.609375,
       "learning_rate": 0.0001469675660738206,
-      "loss": 0.0063,
+      "loss": 0.009,
       "step": 1570
     },
     {
       "epoch": 3.4565456545654567,
-      "grad_norm": 0.341796875,
+      "grad_norm": 0.27734375,
       "learning_rate": 0.00014690632916274617,
-      "loss": 0.0088,
+      "loss": 0.0056,
       "step": 1571
     },
     {
       "epoch": 3.458745874587459,
-      "grad_norm": 0.4609375,
+      "grad_norm": 0.2890625,
       "learning_rate": 0.00014684506969189637,
-      "loss": 0.0131,
+      "loss": 0.0077,
       "step": 1572
     },
     {
       "epoch": 3.460946094609461,
-      "grad_norm": 0.333984375,
+      "grad_norm": 0.328125,
       "learning_rate": 0.00014678378769073417,
-      "loss": 0.0075,
+      "loss": 0.0066,
       "step": 1573
     },
     {
       "epoch": 3.463146314631463,
-      "grad_norm": 0.26171875,
+      "grad_norm": 0.58984375,
       "learning_rate": 0.00014672248318873342,
-      "loss": 0.0053,
+      "loss": 0.0106,
       "step": 1574
     },
     {
       "epoch": 3.4653465346534653,
-      "grad_norm": 0.1435546875,
+      "grad_norm": 0.35546875,
       "learning_rate": 0.00014666115621537868,
-      "loss": 0.0048,
+      "loss": 0.0096,
       "step": 1575
     },
     {
       "epoch": 3.4675467546754675,
-      "grad_norm": 0.1142578125,
+      "grad_norm": 0.306640625,
       "learning_rate": 0.00014659980680016546,
-      "loss": 0.0029,
+      "loss": 0.0071,
       "step": 1576
     },
     {
       "epoch": 3.4697469746974696,
-      "grad_norm": 0.146484375,
+      "grad_norm": 0.251953125,
       "learning_rate": 0.00014653843497259997,
-      "loss": 0.0069,
+      "loss": 0.0052,
       "step": 1577
     },
     {
       "epoch": 3.4719471947194718,
-      "grad_norm": 0.1787109375,
+      "grad_norm": 0.52734375,
       "learning_rate": 0.0001464770407621992,
-      "loss": 0.0068,
+      "loss": 0.0165,
       "step": 1578
     },
     {
       "epoch": 3.4741474147414744,
-      "grad_norm": 0.1533203125,
+      "grad_norm": 0.1162109375,
       "learning_rate": 0.00014641562419849094,
-      "loss": 0.006,
+      "loss": 0.0052,
       "step": 1579
     },
     {
       "epoch": 3.4763476347634765,
-      "grad_norm": 0.1796875,
+      "grad_norm": 0.2333984375,
       "learning_rate": 0.0001463541853110137,
-      "loss": 0.0055,
+      "loss": 0.0095,
       "step": 1580
     },
     {
       "epoch": 3.4785478547854787,
-      "grad_norm": 0.291015625,
+      "grad_norm": 0.5,
       "learning_rate": 0.00014629272412931677,
-      "loss": 0.0102,
+      "loss": 0.014,
       "step": 1581
     },
     {
       "epoch": 3.480748074807481,
-      "grad_norm": 0.21484375,
+      "grad_norm": 0.44140625,
       "learning_rate": 0.00014623124068296014,
-      "loss": 0.0061,
+      "loss": 0.0105,
       "step": 1582
     },
     {
       "epoch": 3.482948294829483,
-      "grad_norm": 0.1650390625,
+      "grad_norm": 0.3125,
       "learning_rate": 0.0001461697350015145,
-      "loss": 0.0021,
+      "loss": 0.0089,
       "step": 1583
     },
     {
       "epoch": 3.485148514851485,
-      "grad_norm": 0.2001953125,
+      "grad_norm": 0.2294921875,
       "learning_rate": 0.00014610820711456122,
-      "loss": 0.0068,
+      "loss": 0.0092,
       "step": 1584
     },
     {
       "epoch": 3.4873487348734873,
-      "grad_norm": 0.341796875,
+      "grad_norm": 0.11669921875,
       "learning_rate": 0.00014604665705169237,
-      "loss": 0.0057,
+      "loss": 0.0049,
       "step": 1585
     },
     {
       "epoch": 3.4895489548954894,
-      "grad_norm": 0.212890625,
+      "grad_norm": 0.447265625,
       "learning_rate": 0.00014598508484251074,
-      "loss": 0.0055,
+      "loss": 0.0145,
       "step": 1586
     },
     {
       "epoch": 3.4917491749174916,
-      "grad_norm": 0.208984375,
+      "grad_norm": 0.46484375,
       "learning_rate": 0.00014592349051662963,
-      "loss": 0.0055,
+      "loss": 0.0116,
       "step": 1587
     },
     {
       "epoch": 3.493949394939494,
-      "grad_norm": 0.263671875,
+      "grad_norm": 0.427734375,
       "learning_rate": 0.00014586187410367307,
-      "loss": 0.0045,
+      "loss": 0.0136,
       "step": 1588
     },
     {
       "epoch": 3.4961496149614963,
-      "grad_norm": 0.1806640625,
+      "grad_norm": 0.318359375,
       "learning_rate": 0.0001458002356332758,
-      "loss": 0.0053,
+      "loss": 0.0087,
       "step": 1589
     },
     {
       "epoch": 3.4983498349834985,
-      "grad_norm": 0.09130859375,
+      "grad_norm": 0.46484375,
       "learning_rate": 0.00014573857513508294,
-      "loss": 0.0023,
+      "loss": 0.0157,
       "step": 1590
     },
     {
       "epoch": 3.5005500550055006,
-      "grad_norm": 0.263671875,
+      "grad_norm": 0.34375,
       "learning_rate": 0.00014567689263875048,
-      "loss": 0.0102,
+      "loss": 0.0153,
       "step": 1591
     },
     {
       "epoch": 3.502750275027503,
-      "grad_norm": 0.1748046875,
+      "grad_norm": 0.36328125,
       "learning_rate": 0.00014561518817394475,
-      "loss": 0.0114,
+      "loss": 0.008,
       "step": 1592
     },
     {
       "epoch": 3.504950495049505,
-      "grad_norm": 0.1396484375,
+      "grad_norm": 0.197265625,
       "learning_rate": 0.00014555346177034278,
-      "loss": 0.005,
+      "loss": 0.0047,
       "step": 1593
     },
     {
       "epoch": 3.507150715071507,
-      "grad_norm": 0.166015625,
+      "grad_norm": 0.359375,
       "learning_rate": 0.0001454917134576321,
-      "loss": 0.0079,
+      "loss": 0.0177,
       "step": 1594
     },
     {
       "epoch": 3.5093509350935093,
-      "grad_norm": 0.12890625,
+      "grad_norm": 0.251953125,
       "learning_rate": 0.0001454299432655108,
-      "loss": 0.0023,
+      "loss": 0.0055,
       "step": 1595
     },
     {
       "epoch": 3.5115511551155114,
-      "grad_norm": 0.40625,
+      "grad_norm": 0.09912109375,
       "learning_rate": 0.00014536815122368748,
-      "loss": 0.0087,
+      "loss": 0.0043,
       "step": 1596
     },
     {
       "epoch": 3.5115511551155114,
-      "eval_loss": 0.013583294115960598,
-      "eval_runtime": 10.5771,
-      "eval_samples_per_second": 36.21,
-      "eval_steps_per_second": 4.538,
+      "eval_loss": 0.020594414323568344,
+      "eval_runtime": 10.0909,
+      "eval_samples_per_second": 37.955,
+      "eval_steps_per_second": 4.757,
       "step": 1596
     },
     {
       "epoch": 3.513751375137514,
-      "grad_norm": 0.05224609375,
+      "grad_norm": 0.1787109375,
       "learning_rate": 0.00014530633736188126,
-      "loss": 0.0011,
+      "loss": 0.0041,
       "step": 1597
     },
     {
       "epoch": 3.5159515951595157,
-      "grad_norm": 0.1142578125,
+      "grad_norm": 0.330078125,
       "learning_rate": 0.00014524450170982177,
-      "loss": 0.0027,
+      "loss": 0.0111,
       "step": 1598
     },
     {
       "epoch": 3.5181518151815183,
-      "grad_norm": 0.1875,
+      "grad_norm": 0.283203125,
       "learning_rate": 0.0001451826442972491,
-      "loss": 0.0028,
+      "loss": 0.0056,
       "step": 1599
     },
     {
       "epoch": 3.5203520352035205,
-      "grad_norm": 0.1982421875,
+      "grad_norm": 0.2001953125,
       "learning_rate": 0.00014512076515391375,
-      "loss": 0.0049,
+      "loss": 0.0082,
       "step": 1600
     },
     {
       "epoch": 3.5225522552255226,
-      "grad_norm": 0.294921875,
+      "grad_norm": 0.234375,
       "learning_rate": 0.00014505886430957678,
-      "loss": 0.0079,
+      "loss": 0.0104,
       "step": 1601
     },
     {
       "epoch": 3.5247524752475248,
-      "grad_norm": 0.1240234375,
+      "grad_norm": 0.10498046875,
       "learning_rate": 0.0001449969417940096,
-      "loss": 0.0024,
+      "loss": 0.0037,
       "step": 1602
     },
     {
       "epoch": 3.526952695269527,
-      "grad_norm": 0.1318359375,
+      "grad_norm": 0.302734375,
       "learning_rate": 0.0001449349976369941,
-      "loss": 0.0052,
+      "loss": 0.0141,
       "step": 1603
     },
     {
       "epoch": 3.529152915291529,
-      "grad_norm": 0.380859375,
+      "grad_norm": 0.2099609375,
       "learning_rate": 0.00014487303186832255,
-      "loss": 0.0189,
+      "loss": 0.0135,
       "step": 1604
     },
     {
       "epoch": 3.5313531353135312,
-      "grad_norm": 0.193359375,
+      "grad_norm": 0.404296875,
       "learning_rate": 0.00014481104451779763,
-      "loss": 0.0044,
+      "loss": 0.0145,
       "step": 1605
     },
     {
       "epoch": 3.533553355335534,
-      "grad_norm": 0.2734375,
+      "grad_norm": 0.3515625,
       "learning_rate": 0.0001447490356152324,
-      "loss": 0.0054,
+      "loss": 0.0056,
       "step": 1606
     },
     {
       "epoch": 3.5357535753575355,
-      "grad_norm": 0.1279296875,
+      "grad_norm": 0.353515625,
       "learning_rate": 0.0001446870051904502,
-      "loss": 0.0047,
+      "loss": 0.0128,
       "step": 1607
     },
     {
       "epoch": 3.537953795379538,
-      "grad_norm": 0.21484375,
+      "grad_norm": 0.38671875,
       "learning_rate": 0.00014462495327328487,
-      "loss": 0.0034,
+      "loss": 0.01,
       "step": 1608
     },
     {
       "epoch": 3.5401540154015403,
-      "grad_norm": 0.095703125,
+      "grad_norm": 0.271484375,
       "learning_rate": 0.00014456287989358048,
-      "loss": 0.0026,
+      "loss": 0.0135,
       "step": 1609
     },
     {
       "epoch": 3.5423542354235424,
-      "grad_norm": 0.2734375,
+      "grad_norm": 0.326171875,
       "learning_rate": 0.00014450078508119148,
-      "loss": 0.0135,
+      "loss": 0.0049,
       "step": 1610
     },
     {
       "epoch": 3.5445544554455446,
-      "grad_norm": 0.31640625,
+      "grad_norm": 0.1513671875,
       "learning_rate": 0.00014443866886598258,
-      "loss": 0.0075,
+      "loss": 0.006,
       "step": 1611
     },
     {
       "epoch": 3.5467546754675467,
-      "grad_norm": 0.1826171875,
+      "grad_norm": 0.267578125,
       "learning_rate": 0.00014437653127782882,
-      "loss": 0.0079,
+      "loss": 0.0169,
       "step": 1612
     },
     {
       "epoch": 3.548954895489549,
-      "grad_norm": 0.2265625,
+      "grad_norm": 0.2421875,
       "learning_rate": 0.00014431437234661552,
-      "loss": 0.0075,
+      "loss": 0.0094,
       "step": 1613
     },
     {
       "epoch": 3.551155115511551,
-      "grad_norm": 0.3984375,
+      "grad_norm": 0.205078125,
       "learning_rate": 0.0001442521921022382,
-      "loss": 0.0064,
+      "loss": 0.0082,
       "step": 1614
     },
     {
       "epoch": 3.553355335533553,
-      "grad_norm": 0.2041015625,
+      "grad_norm": 0.328125,
       "learning_rate": 0.00014418999057460276,
-      "loss": 0.0057,
+      "loss": 0.0106,
       "step": 1615
     },
     {
       "epoch": 3.5555555555555554,
-      "grad_norm": 0.11279296875,
+      "grad_norm": 0.2236328125,
       "learning_rate": 0.0001441277677936252,
-      "loss": 0.0018,
+      "loss": 0.0036,
       "step": 1616
     },
     {
       "epoch": 3.557755775577558,
-      "grad_norm": 0.2490234375,
+      "grad_norm": 0.216796875,
       "learning_rate": 0.0001440655237892318,
-      "loss": 0.0092,
+      "loss": 0.0097,
       "step": 1617
     },
     {
       "epoch": 3.55995599559956,
-      "grad_norm": 0.197265625,
+      "grad_norm": 0.14453125,
       "learning_rate": 0.00014400325859135907,
-      "loss": 0.0031,
+      "loss": 0.0043,
       "step": 1618
     },
     {
       "epoch": 3.5621562156215623,
-      "grad_norm": 0.10546875,
+      "grad_norm": 0.26171875,
       "learning_rate": 0.0001439409722299537,
-      "loss": 0.0013,
+      "loss": 0.0049,
       "step": 1619
     },
     {
       "epoch": 3.5643564356435644,
-      "grad_norm": 0.06591796875,
+      "grad_norm": 0.2060546875,
       "learning_rate": 0.00014387866473497253,
-      "loss": 0.0011,
+      "loss": 0.0034,
       "step": 1620
     },
     {
       "epoch": 3.5665566556655666,
-      "grad_norm": 0.32421875,
+      "grad_norm": 0.64453125,
       "learning_rate": 0.0001438163361363826,
-      "loss": 0.0041,
+      "loss": 0.0085,
       "step": 1621
     },
     {
       "epoch": 3.5687568756875687,
-      "grad_norm": 0.15625,
+      "grad_norm": 0.2109375,
       "learning_rate": 0.00014375398646416105,
-      "loss": 0.0044,
+      "loss": 0.0079,
       "step": 1622
     },
     {
       "epoch": 3.570957095709571,
-      "grad_norm": 0.1005859375,
+      "grad_norm": 0.078125,
       "learning_rate": 0.00014369161574829522,
-      "loss": 0.0026,
+      "loss": 0.0033,
       "step": 1623
     },
     {
       "epoch": 3.573157315731573,
-      "grad_norm": 0.0986328125,
+      "grad_norm": 0.1826171875,
       "learning_rate": 0.00014362922401878254,
-      "loss": 0.0035,
+      "loss": 0.0046,
       "step": 1624
     },
     {
       "epoch": 3.575357535753575,
-      "grad_norm": 0.1171875,
+      "grad_norm": 0.2080078125,
       "learning_rate": 0.00014356681130563053,
-      "loss": 0.0024,
+      "loss": 0.0044,
       "step": 1625
     },
     {
       "epoch": 3.5775577557755778,
-      "grad_norm": 0.09423828125,
+      "grad_norm": 0.142578125,
       "learning_rate": 0.00014350437763885682,
-      "loss": 0.0025,
+      "loss": 0.0049,
       "step": 1626
     },
     {
       "epoch": 3.5797579757975795,
-      "grad_norm": 0.318359375,
+      "grad_norm": 0.2294921875,
       "learning_rate": 0.00014344192304848914,
-      "loss": 0.0066,
+      "loss": 0.0072,
       "step": 1627
     },
     {
       "epoch": 3.581958195819582,
-      "grad_norm": 0.146484375,
+      "grad_norm": 0.333984375,
       "learning_rate": 0.00014337944756456525,
-      "loss": 0.0041,
+      "loss": 0.0084,
       "step": 1628
     },
     {
       "epoch": 3.5841584158415842,
-      "grad_norm": 0.30078125,
+      "grad_norm": 0.25390625,
       "learning_rate": 0.00014331695121713297,
-      "loss": 0.0065,
+      "loss": 0.0106,
       "step": 1629
     },
     {
       "epoch": 3.5863586358635864,
-      "grad_norm": 0.34375,
+      "grad_norm": 0.345703125,
       "learning_rate": 0.0001432544340362501,
-      "loss": 0.0067,
+      "loss": 0.0142,
       "step": 1630
     },
     {
       "epoch": 3.5885588558855885,
-      "grad_norm": 0.318359375,
+      "grad_norm": 0.1806640625,
       "learning_rate": 0.0001431918960519846,
-      "loss": 0.0071,
+      "loss": 0.0083,
       "step": 1631
     },
     {
       "epoch": 3.5907590759075907,
-      "grad_norm": 0.33203125,
+      "grad_norm": 0.41796875,
       "learning_rate": 0.00014312933729441433,
-      "loss": 0.0058,
+      "loss": 0.0098,
       "step": 1632
     },
     {
       "epoch": 3.592959295929593,
-      "grad_norm": 0.291015625,
+      "grad_norm": 0.3671875,
       "learning_rate": 0.0001430667577936271,
-      "loss": 0.0118,
+      "loss": 0.014,
       "step": 1633
     },
     {
       "epoch": 3.595159515951595,
-      "grad_norm": 0.1357421875,
+      "grad_norm": 0.0771484375,
       "learning_rate": 0.0001430041575797208,
-      "loss": 0.0046,
+      "loss": 0.003,
       "step": 1634
     },
     {
       "epoch": 3.5973597359735976,
-      "grad_norm": 0.1328125,
+      "grad_norm": 0.35546875,
       "learning_rate": 0.00014294153668280326,
-      "loss": 0.0087,
+      "loss": 0.0138,
       "step": 1635
     },
     {
       "epoch": 3.5995599559955993,
-      "grad_norm": 0.2451171875,
+      "grad_norm": 0.35546875,
       "learning_rate": 0.00014287889513299218,
-      "loss": 0.0101,
+      "loss": 0.013,
       "step": 1636
     },
     {
       "epoch": 3.601760176017602,
-      "grad_norm": 0.1416015625,
+      "grad_norm": 0.2890625,
       "learning_rate": 0.0001428162329604153,
-      "loss": 0.0018,
+      "loss": 0.0044,
       "step": 1637
     },
     {
       "epoch": 3.603960396039604,
-      "grad_norm": 0.212890625,
+      "grad_norm": 0.34765625,
       "learning_rate": 0.00014275355019521019,
-      "loss": 0.0046,
+      "loss": 0.0088,
       "step": 1638
     },
     {
       "epoch": 3.606160616061606,
-      "grad_norm": 0.061767578125,
+      "grad_norm": 0.287109375,
       "learning_rate": 0.00014269084686752435,
-      "loss": 0.0017,
+      "loss": 0.0224,
       "step": 1639
     },
     {
       "epoch": 3.6083608360836084,
-      "grad_norm": 0.083984375,
+      "grad_norm": 0.30859375,
       "learning_rate": 0.00014262812300751526,
-      "loss": 0.0012,
+      "loss": 0.0091,
       "step": 1640
     },
     {
       "epoch": 3.6105610561056105,
-      "grad_norm": 0.296875,
+      "grad_norm": 0.2265625,
       "learning_rate": 0.00014256537864535012,
-      "loss": 0.0041,
+      "loss": 0.0035,
       "step": 1641
     },
     {
       "epoch": 3.6127612761276127,
-      "grad_norm": 0.330078125,
+      "grad_norm": 0.33203125,
       "learning_rate": 0.0001425026138112061,
-      "loss": 0.0026,
+      "loss": 0.014,
       "step": 1642
     },
     {
       "epoch": 3.614961496149615,
-      "grad_norm": 0.06591796875,
+      "grad_norm": 0.154296875,
       "learning_rate": 0.0001424398285352702,
-      "loss": 0.0009,
+      "loss": 0.0021,
       "step": 1643
     },
     {
       "epoch": 3.6171617161716174,
-      "grad_norm": 0.11962890625,
+      "grad_norm": 0.255859375,
       "learning_rate": 0.00014237702284773914,
-      "loss": 0.0072,
+      "loss": 0.0118,
       "step": 1644
     },
     {
       "epoch": 3.619361936193619,
-      "grad_norm": 0.212890625,
+      "grad_norm": 0.25,
       "learning_rate": 0.00014231419677881966,
-      "loss": 0.0142,
+      "loss": 0.0063,
       "step": 1645
     },
     {
       "epoch": 3.6215621562156217,
-      "grad_norm": 0.404296875,
+      "grad_norm": 0.3203125,
       "learning_rate": 0.00014225135035872817,
-      "loss": 0.004,
+      "loss": 0.0061,
       "step": 1646
     },
     {
       "epoch": 3.623762376237624,
-      "grad_norm": 0.2412109375,
+      "grad_norm": 0.3671875,
       "learning_rate": 0.00014218848361769086,
-      "loss": 0.0121,
+      "loss": 0.0173,
       "step": 1647
     },
     {
       "epoch": 3.625962596259626,
-      "grad_norm": 0.1708984375,
+      "grad_norm": 0.1806640625,
       "learning_rate": 0.00014212559658594373,
-      "loss": 0.0069,
+      "loss": 0.009,
       "step": 1648
     },
     {
       "epoch": 3.628162816281628,
-      "grad_norm": 0.0859375,
+      "grad_norm": 0.12890625,
       "learning_rate": 0.00014206268929373256,
-      "loss": 0.0024,
+      "loss": 0.0035,
       "step": 1649
     },
     {
       "epoch": 3.6303630363036303,
-      "grad_norm": 0.173828125,
+      "grad_norm": 0.296875,
       "learning_rate": 0.00014199976177131283,
-      "loss": 0.0039,
+      "loss": 0.0069,
       "step": 1650
     },
     {
       "epoch": 3.6325632563256325,
-      "grad_norm": 0.39453125,
+      "grad_norm": 0.2734375,
       "learning_rate": 0.00014193681404894978,
-      "loss": 0.014,
+      "loss": 0.0155,
       "step": 1651
     },
     {
       "epoch": 3.6347634763476346,
-      "grad_norm": 0.12109375,
+      "grad_norm": 0.1396484375,
       "learning_rate": 0.00014187384615691834,
-      "loss": 0.0034,
+      "loss": 0.0041,
       "step": 1652
     },
     {
       "epoch": 3.6369636963696372,
-      "grad_norm": 0.1748046875,
+      "grad_norm": 0.271484375,
       "learning_rate": 0.00014181085812550316,
-      "loss": 0.0035,
+      "loss": 0.0051,
       "step": 1653
     },
     {
       "epoch": 3.639163916391639,
-      "grad_norm": 0.1357421875,
+      "grad_norm": 0.228515625,
       "learning_rate": 0.0001417478499849986,
-      "loss": 0.0042,
+      "loss": 0.0094,
       "step": 1654
     },
     {
       "epoch": 3.6413641364136415,
-      "grad_norm": 0.27734375,
+      "grad_norm": 0.181640625,
       "learning_rate": 0.00014168482176570862,
-      "loss": 0.0041,
+      "loss": 0.0046,
       "step": 1655
     },
     {
       "epoch": 3.6435643564356437,
-      "grad_norm": 0.29296875,
+      "grad_norm": 0.255859375,
       "learning_rate": 0.00014162177349794685,
-      "loss": 0.0086,
+      "loss": 0.0077,
       "step": 1656
     },
     {
       "epoch": 3.645764576457646,
-      "grad_norm": 0.228515625,
+      "grad_norm": 0.263671875,
       "learning_rate": 0.00014155870521203668,
-      "loss": 0.0075,
+      "loss": 0.0098,
       "step": 1657
     },
     {
       "epoch": 3.647964796479648,
-      "grad_norm": 0.09716796875,
+      "grad_norm": 0.189453125,
       "learning_rate": 0.00014149561693831098,
-      "loss": 0.001,
+      "loss": 0.0032,
       "step": 1658
     },
     {
       "epoch": 3.65016501650165,
-      "grad_norm": 0.2451171875,
+      "grad_norm": 0.1923828125,
       "learning_rate": 0.00014143250870711233,
-      "loss": 0.0083,
+      "loss": 0.0074,
       "step": 1659
     },
     {
       "epoch": 3.6523652365236523,
-      "grad_norm": 0.1728515625,
+      "grad_norm": 0.1513671875,
       "learning_rate": 0.00014136938054879283,
-      "loss": 0.0029,
+      "loss": 0.0104,
       "step": 1660
     },
     {
       "epoch": 3.6545654565456545,
-      "grad_norm": 0.212890625,
+      "grad_norm": 0.16015625,
       "learning_rate": 0.0001413062324937142,
-      "loss": 0.0028,
+      "loss": 0.0024,
       "step": 1661
     },
     {
       "epoch": 3.6567656765676566,
-      "grad_norm": 0.1357421875,
+      "grad_norm": 0.197265625,
       "learning_rate": 0.0001412430645722478,
-      "loss": 0.0022,
+      "loss": 0.0083,
       "step": 1662
     },
     {
       "epoch": 3.6589658965896588,
-      "grad_norm": 0.40625,
+      "grad_norm": 0.38671875,
       "learning_rate": 0.0001411798768147744,
-      "loss": 0.0078,
+      "loss": 0.0108,
       "step": 1663
     },
     {
       "epoch": 3.6611661166116614,
-      "grad_norm": 0.1318359375,
+      "grad_norm": 0.2578125,
       "learning_rate": 0.00014111666925168442,
-      "loss": 0.0023,
+      "loss": 0.006,
       "step": 1664
     },
     {
       "epoch": 3.6633663366336635,
-      "grad_norm": 0.365234375,
+      "grad_norm": 0.2392578125,
       "learning_rate": 0.0001410534419133778,
-      "loss": 0.0123,
+      "loss": 0.0072,
       "step": 1665
     },
     {
       "epoch": 3.6655665566556657,
-      "grad_norm": 0.1650390625,
+      "grad_norm": 0.19921875,
       "learning_rate": 0.00014099019483026395,
-      "loss": 0.0061,
+      "loss": 0.0073,
       "step": 1666
     },
     {
       "epoch": 3.667766776677668,
-      "grad_norm": 0.43359375,
+      "grad_norm": 0.1376953125,
       "learning_rate": 0.0001409269280327618,
-      "loss": 0.0083,
+      "loss": 0.0029,
       "step": 1667
     },
     {
       "epoch": 3.66996699669967,
-      "grad_norm": 0.07958984375,
+      "grad_norm": 0.2236328125,
       "learning_rate": 0.0001408636415512997,
-      "loss": 0.0014,
+      "loss": 0.005,
       "step": 1668
     },
     {
       "epoch": 3.672167216721672,
-      "grad_norm": 0.10498046875,
+      "grad_norm": 0.1923828125,
       "learning_rate": 0.0001408003354163156,
-      "loss": 0.0028,
+      "loss": 0.0049,
       "step": 1669
     },
     {
       "epoch": 3.6743674367436743,
-      "grad_norm": 0.11767578125,
+      "grad_norm": 0.306640625,
       "learning_rate": 0.0001407370096582568,
-      "loss": 0.0049,
+      "loss": 0.0056,
       "step": 1670
     },
     {
       "epoch": 3.6765676567656764,
-      "grad_norm": 0.189453125,
+      "grad_norm": 0.11669921875,
       "learning_rate": 0.00014067366430758004,
-      "loss": 0.0045,
+      "loss": 0.0017,
       "step": 1671
     },
     {
       "epoch": 3.6787678767876786,
-      "grad_norm": 0.138671875,
+      "grad_norm": 0.1435546875,
       "learning_rate": 0.0001406102993947515,
-      "loss": 0.0018,
+      "loss": 0.0033,
       "step": 1672
     },
     {
       "epoch": 3.680968096809681,
-      "grad_norm": 0.1884765625,
+      "grad_norm": 0.302734375,
       "learning_rate": 0.0001405469149502468,
-      "loss": 0.0045,
+      "loss": 0.0061,
       "step": 1673
     },
     {
       "epoch": 3.6831683168316833,
-      "grad_norm": 0.431640625,
+      "grad_norm": 0.2119140625,
       "learning_rate": 0.0001404835110045509,
-      "loss": 0.0128,
+      "loss": 0.0079,
       "step": 1674
     },
     {
       "epoch": 3.6853685368536855,
-      "grad_norm": 0.1572265625,
+      "grad_norm": 0.314453125,
       "learning_rate": 0.00014042008758815818,
-      "loss": 0.005,
+      "loss": 0.0075,
       "step": 1675
     },
     {
       "epoch": 3.6875687568756876,
-      "grad_norm": 0.0859375,
+      "grad_norm": 0.29296875,
       "learning_rate": 0.0001403566447315724,
-      "loss": 0.0028,
+      "loss": 0.0123,
       "step": 1676
     },
     {
       "epoch": 3.68976897689769,
-      "grad_norm": 0.28125,
+      "grad_norm": 0.3203125,
       "learning_rate": 0.00014029318246530656,
-      "loss": 0.0123,
+      "loss": 0.0111,
       "step": 1677
     },
     {
       "epoch": 3.691969196919692,
-      "grad_norm": 0.05908203125,
+      "grad_norm": 0.2275390625,
       "learning_rate": 0.00014022970081988315,
-      "loss": 0.0019,
+      "loss": 0.0066,
       "step": 1678
     },
     {
       "epoch": 3.694169416941694,
-      "grad_norm": 0.15625,
+      "grad_norm": 0.353515625,
       "learning_rate": 0.0001401661998258339,
-      "loss": 0.0082,
+      "loss": 0.0069,
       "step": 1679
     },
     {
       "epoch": 3.6963696369636962,
-      "grad_norm": 0.078125,
+      "grad_norm": 0.177734375,
       "learning_rate": 0.00014010267951369983,
-      "loss": 0.003,
+      "loss": 0.0085,
       "step": 1680
     },
     {
       "epoch": 3.6985698569856984,
-      "grad_norm": 0.1748046875,
+      "grad_norm": 0.0478515625,
       "learning_rate": 0.0001400391399140313,
-      "loss": 0.0154,
+      "loss": 0.0022,
       "step": 1681
     },
     {
       "epoch": 3.700770077007701,
-      "grad_norm": 0.3125,
+      "grad_norm": 0.34765625,
       "learning_rate": 0.00013997558105738786,
-      "loss": 0.0051,
+      "loss": 0.0058,
       "step": 1682
     },
     {
       "epoch": 3.7029702970297027,
-      "grad_norm": 0.150390625,
+      "grad_norm": 0.18359375,
       "learning_rate": 0.00013991200297433842,
-      "loss": 0.0028,
+      "loss": 0.0039,
       "step": 1683
     },
     {
       "epoch": 3.7051705170517053,
-      "grad_norm": 0.09619140625,
+      "grad_norm": 0.30078125,
       "learning_rate": 0.0001398484056954611,
-      "loss": 0.0015,
+      "loss": 0.0061,
       "step": 1684
     },
     {
       "epoch": 3.7073707370737075,
-      "grad_norm": 0.2080078125,
+      "grad_norm": 0.2890625,
       "learning_rate": 0.00013978478925134324,
-      "loss": 0.0069,
+      "loss": 0.0059,
       "step": 1685
     },
     {
       "epoch": 3.7095709570957096,
-      "grad_norm": 0.08935546875,
+      "grad_norm": 0.251953125,
       "learning_rate": 0.00013972115367258145,
-      "loss": 0.0017,
+      "loss": 0.0064,
       "step": 1686
     },
     {
       "epoch": 3.7117711771177118,
-      "grad_norm": 0.0546875,
+      "grad_norm": 0.330078125,
       "learning_rate": 0.00013965749898978141,
-      "loss": 0.0017,
+      "loss": 0.0028,
       "step": 1687
     },
     {
       "epoch": 3.713971397139714,
-      "grad_norm": 0.1337890625,
+      "grad_norm": 0.208984375,
       "learning_rate": 0.00013959382523355814,
-      "loss": 0.0031,
+      "loss": 0.0061,
       "step": 1688
     },
     {
       "epoch": 3.716171617161716,
-      "grad_norm": 0.337890625,
+      "grad_norm": 0.1767578125,
       "learning_rate": 0.00013953013243453582,
-      "loss": 0.0042,
+      "loss": 0.0067,
       "step": 1689
     },
     {
       "epoch": 3.718371837183718,
-      "grad_norm": 0.361328125,
+      "grad_norm": 0.2294921875,
       "learning_rate": 0.00013946642062334766,
-      "loss": 0.0092,
+      "loss": 0.0085,
       "step": 1690
     },
     {
       "epoch": 3.720572057205721,
-      "grad_norm": 0.2177734375,
+      "grad_norm": 0.1611328125,
       "learning_rate": 0.0001394026898306361,
-      "loss": 0.0043,
+      "loss": 0.0059,
       "step": 1691
     },
     {
       "epoch": 3.7227722772277225,
-      "grad_norm": 0.373046875,
+      "grad_norm": 0.3359375,
       "learning_rate": 0.00013933894008705277,
-      "loss": 0.0077,
+      "loss": 0.0076,
       "step": 1692
     },
     {
       "epoch": 3.724972497249725,
-      "grad_norm": 0.09814453125,
+      "grad_norm": 0.1328125,
       "learning_rate": 0.0001392751714232583,
       "loss": 0.0037,
       "step": 1693
     },
     {
       "epoch": 3.7271727172717273,
-      "grad_norm": 0.057861328125,
+      "grad_norm": 0.193359375,
       "learning_rate": 0.00013921138386992243,
-      "loss": 0.001,
+      "loss": 0.0116,
       "step": 1694
     },
     {
       "epoch": 3.7293729372937294,
-      "grad_norm": 0.11474609375,
+      "grad_norm": 0.451171875,
       "learning_rate": 0.00013914757745772408,
-      "loss": 0.0026,
+      "loss": 0.0066,
       "step": 1695
     },
     {
       "epoch": 3.7315731573157316,
-      "grad_norm": 0.1640625,
+      "grad_norm": 0.1796875,
       "learning_rate": 0.0001390837522173512,
-      "loss": 0.0071,
+      "loss": 0.0104,
       "step": 1696
     },
     {
       "epoch": 3.7337733773377337,
-      "grad_norm": 0.240234375,
+      "grad_norm": 0.1923828125,
       "learning_rate": 0.00013901990817950078,
-      "loss": 0.0049,
+      "loss": 0.0085,
       "step": 1697
     },
     {
       "epoch": 3.735973597359736,
-      "grad_norm": 0.318359375,
+      "grad_norm": 0.09130859375,
       "learning_rate": 0.0001389560453748788,
-      "loss": 0.008,
+      "loss": 0.0009,
       "step": 1698
     },
     {
       "epoch": 3.738173817381738,
-      "grad_norm": 0.12353515625,
+      "grad_norm": 0.2236328125,
       "learning_rate": 0.0001388921638342003,
-      "loss": 0.0035,
+      "loss": 0.006,
       "step": 1699
     },
     {
       "epoch": 3.7403740374037406,
-      "grad_norm": 0.12890625,
+      "grad_norm": 0.193359375,
       "learning_rate": 0.00013882826358818935,
-      "loss": 0.0052,
+      "loss": 0.0076,
       "step": 1700
     },
     {
       "epoch": 3.7425742574257423,
-      "grad_norm": 0.236328125,
+      "grad_norm": 0.1748046875,
       "learning_rate": 0.00013876434466757913,
-      "loss": 0.0063,
+      "loss": 0.0083,
       "step": 1701
     },
     {
       "epoch": 3.744774477447745,
-      "grad_norm": 0.1337890625,
+      "grad_norm": 0.1240234375,
       "learning_rate": 0.0001387004071031116,
-      "loss": 0.0031,
+      "loss": 0.0023,
       "step": 1702
     },
     {
       "epoch": 3.746974697469747,
-      "grad_norm": 0.051025390625,
+      "grad_norm": 0.11279296875,
       "learning_rate": 0.00013863645092553775,
-      "loss": 0.001,
+      "loss": 0.0031,
       "step": 1703
     },
     {
       "epoch": 3.7491749174917492,
-      "grad_norm": 0.1396484375,
+      "grad_norm": 0.2353515625,
       "learning_rate": 0.00013857247616561757,
-      "loss": 0.0025,
+      "loss": 0.0048,
       "step": 1704
     },
     {
       "epoch": 3.7513751375137514,
-      "grad_norm": 0.177734375,
+      "grad_norm": 0.21875,
       "learning_rate": 0.00013850848285411994,
-      "loss": 0.0034,
+      "loss": 0.0055,
       "step": 1705
     },
     {
       "epoch": 3.7535753575357536,
-      "grad_norm": 0.158203125,
+      "grad_norm": 0.12255859375,
       "learning_rate": 0.00013844447102182272,
-      "loss": 0.0031,
+      "loss": 0.005,
       "step": 1706
     },
     {
       "epoch": 3.7557755775577557,
-      "grad_norm": 0.111328125,
+      "grad_norm": 0.28125,
       "learning_rate": 0.00013838044069951265,
-      "loss": 0.0025,
+      "loss": 0.0062,
       "step": 1707
     },
     {
       "epoch": 3.757975797579758,
-      "grad_norm": 0.2021484375,
+      "grad_norm": 0.2392578125,
       "learning_rate": 0.00013831639191798532,
-      "loss": 0.0057,
+      "loss": 0.0046,
       "step": 1708
     },
     {
       "epoch": 3.76017601760176,
-      "grad_norm": 0.171875,
+      "grad_norm": 0.1318359375,
       "learning_rate": 0.00013825232470804523,
-      "loss": 0.0057,
+      "loss": 0.0056,
       "step": 1709
     },
     {
       "epoch": 3.762376237623762,
-      "grad_norm": 0.259765625,
+      "grad_norm": 0.23828125,
       "learning_rate": 0.00013818823910050578,
-      "loss": 0.0045,
+      "loss": 0.0085,
       "step": 1710
     },
     {
       "epoch": 3.762376237623762,
-      "eval_loss": 0.007744603790342808,
-      "eval_runtime": 10.1911,
-      "eval_samples_per_second": 37.582,
-      "eval_steps_per_second": 4.71,
+      "eval_loss": 0.012671160511672497,
+      "eval_runtime": 10.7843,
+      "eval_samples_per_second": 35.515,
+      "eval_steps_per_second": 4.451,
       "step": 1710
     },
     {
       "epoch": 3.7645764576457648,
-      "grad_norm": 0.103515625,
+      "grad_norm": 0.1376953125,
       "learning_rate": 0.00013812413512618912,
-      "loss": 0.002,
+      "loss": 0.0043,
       "step": 1711
     },
     {
       "epoch": 3.766776677667767,
-      "grad_norm": 0.32421875,
+      "grad_norm": 0.232421875,
       "learning_rate": 0.00013806001281592634,
-      "loss": 0.0123,
+      "loss": 0.009,
       "step": 1712
     },
     {
       "epoch": 3.768976897689769,
-      "grad_norm": 0.2294921875,
+      "grad_norm": 0.150390625,
       "learning_rate": 0.00013799587220055735,
-      "loss": 0.0024,
+      "loss": 0.0018,
       "step": 1713
     },
     {
       "epoch": 3.771177117711771,
-      "grad_norm": 0.45703125,
+      "grad_norm": 0.208984375,
       "learning_rate": 0.00013793171331093077,
-      "loss": 0.0085,
+      "loss": 0.0029,
       "step": 1714
     },
     {
       "epoch": 3.7733773377337734,
-      "grad_norm": 0.3203125,
+      "grad_norm": 0.2138671875,
       "learning_rate": 0.00013786753617790404,
-      "loss": 0.0026,
+      "loss": 0.0048,
       "step": 1715
     },
     {
       "epoch": 3.7755775577557755,
-      "grad_norm": 0.10009765625,
+      "grad_norm": 0.1416015625,
       "learning_rate": 0.0001378033408323434,
-      "loss": 0.0031,
+      "loss": 0.0027,
       "step": 1716
     },
     {
       "epoch": 3.7777777777777777,
-      "grad_norm": 0.1123046875,
+      "grad_norm": 0.40234375,
       "learning_rate": 0.0001377391273051239,
-      "loss": 0.0053,
+      "loss": 0.0057,
       "step": 1717
     },
     {
       "epoch": 3.77997799779978,
-      "grad_norm": 0.255859375,
+      "grad_norm": 0.1201171875,
       "learning_rate": 0.00013767489562712922,
-      "loss": 0.0059,
+      "loss": 0.0036,
       "step": 1718
     },
     {
       "epoch": 3.782178217821782,
-      "grad_norm": 0.11279296875,
+      "grad_norm": 0.23828125,
       "learning_rate": 0.0001376106458292519,
-      "loss": 0.0018,
+      "loss": 0.0043,
       "step": 1719
     },
     {
       "epoch": 3.7843784378437846,
-      "grad_norm": 0.05810546875,
+      "grad_norm": 0.1806640625,
       "learning_rate": 0.000137546377942393,
-      "loss": 0.0017,
+      "loss": 0.0043,
       "step": 1720
     },
     {
       "epoch": 3.7865786578657867,
-      "grad_norm": 0.06884765625,
+      "grad_norm": 0.10302734375,
       "learning_rate": 0.00013748209199746251,
-      "loss": 0.0007,
+      "loss": 0.0013,
       "step": 1721
     },
     {
       "epoch": 3.788778877887789,
-      "grad_norm": 0.08642578125,
+      "grad_norm": 0.10498046875,
       "learning_rate": 0.0001374177880253789,
-      "loss": 0.0015,
+      "loss": 0.0013,
       "step": 1722
     },
     {
       "epoch": 3.790979097909791,
-      "grad_norm": 0.10986328125,
+      "grad_norm": 0.2890625,
       "learning_rate": 0.00013735346605706948,
-      "loss": 0.0044,
+      "loss": 0.0106,
       "step": 1723
     },
     {
       "epoch": 3.793179317931793,
-      "grad_norm": 0.126953125,
+      "grad_norm": 0.310546875,
       "learning_rate": 0.00013728912612347017,
-      "loss": 0.0044,
+      "loss": 0.0105,
       "step": 1724
     },
     {
       "epoch": 3.7953795379537953,
-      "grad_norm": 0.1103515625,
+      "grad_norm": 0.140625,
       "learning_rate": 0.00013722476825552544,
-      "loss": 0.0015,
+      "loss": 0.0038,
       "step": 1725
     },
     {
       "epoch": 3.7975797579757975,
-      "grad_norm": 0.11669921875,
+      "grad_norm": 0.33984375,
       "learning_rate": 0.00013716039248418845,
-      "loss": 0.0048,
+      "loss": 0.0112,
       "step": 1726
     },
     {
       "epoch": 3.7997799779977997,
-      "grad_norm": 0.2421875,
+      "grad_norm": 0.10693359375,
       "learning_rate": 0.000137095998840421,
-      "loss": 0.0029,
+      "loss": 0.002,
       "step": 1727
     },
     {
       "epoch": 3.801980198019802,
-      "grad_norm": 0.1123046875,
+      "grad_norm": 0.08740234375,
       "learning_rate": 0.00013703158735519342,
-      "loss": 0.0034,
+      "loss": 0.0049,
       "step": 1728
     },
     {
       "epoch": 3.8041804180418044,
-      "grad_norm": 0.28515625,
+      "grad_norm": 0.361328125,
       "learning_rate": 0.00013696715805948474,
-      "loss": 0.0089,
+      "loss": 0.0062,
       "step": 1729
     },
     {
       "epoch": 3.806380638063806,
-      "grad_norm": 0.2236328125,
+      "grad_norm": 0.32421875,
       "learning_rate": 0.00013690271098428232,
-      "loss": 0.0044,
+      "loss": 0.0046,
       "step": 1730
     },
     {
       "epoch": 3.8085808580858087,
-      "grad_norm": 0.19140625,
+      "grad_norm": 0.064453125,
       "learning_rate": 0.0001368382461605824,
-      "loss": 0.0043,
+      "loss": 0.0014,
       "step": 1731
     },
     {
       "epoch": 3.810781078107811,
-      "grad_norm": 0.1845703125,
+      "grad_norm": 0.234375,
       "learning_rate": 0.00013677376361938943,
-      "loss": 0.0051,
+      "loss": 0.0108,
       "step": 1732
     },
     {
       "epoch": 3.812981298129813,
-      "grad_norm": 0.08154296875,
+      "grad_norm": 0.1865234375,
       "learning_rate": 0.00013670926339171658,
-      "loss": 0.0033,
+      "loss": 0.0066,
       "step": 1733
     },
     {
       "epoch": 3.815181518151815,
-      "grad_norm": 0.234375,
+      "grad_norm": 0.19921875,
       "learning_rate": 0.00013664474550858553,
-      "loss": 0.0065,
+      "loss": 0.004,
       "step": 1734
     },
     {
       "epoch": 3.8173817381738173,
-      "grad_norm": 0.2041015625,
+      "grad_norm": 0.1669921875,
       "learning_rate": 0.00013658021000102636,
-      "loss": 0.0038,
+      "loss": 0.0019,
       "step": 1735
     },
     {
       "epoch": 3.8195819581958195,
-      "grad_norm": 0.271484375,
+      "grad_norm": 0.330078125,
       "learning_rate": 0.00013651565690007766,
-      "loss": 0.0063,
+      "loss": 0.0054,
       "step": 1736
     },
     {
       "epoch": 3.8217821782178216,
-      "grad_norm": 0.185546875,
+      "grad_norm": 0.169921875,
       "learning_rate": 0.00013645108623678648,
-      "loss": 0.0037,
+      "loss": 0.0039,
       "step": 1737
     },
     {
       "epoch": 3.823982398239824,
-      "grad_norm": 0.1865234375,
+      "grad_norm": 0.30078125,
       "learning_rate": 0.0001363864980422083,
-      "loss": 0.0027,
+      "loss": 0.0036,
       "step": 1738
     },
     {
       "epoch": 3.826182618261826,
-      "grad_norm": 0.06396484375,
+      "grad_norm": 0.279296875,
       "learning_rate": 0.00013632189234740713,
-      "loss": 0.0009,
+      "loss": 0.0061,
       "step": 1739
     },
     {
       "epoch": 3.8283828382838285,
-      "grad_norm": 0.18359375,
+      "grad_norm": 0.88671875,
       "learning_rate": 0.0001362572691834553,
-      "loss": 0.0026,
+      "loss": 0.0062,
       "step": 1740
     },
     {
       "epoch": 3.8305830583058307,
-      "grad_norm": 0.1640625,
+      "grad_norm": 0.2421875,
       "learning_rate": 0.0001361926285814335,
-      "loss": 0.0034,
+      "loss": 0.0083,
       "step": 1741
     },
     {
       "epoch": 3.832783278327833,
-      "grad_norm": 0.076171875,
+      "grad_norm": 0.1416015625,
       "learning_rate": 0.00013612797057243093,
-      "loss": 0.0021,
+      "loss": 0.0029,
       "step": 1742
     },
     {
       "epoch": 3.834983498349835,
-      "grad_norm": 0.169921875,
+      "grad_norm": 0.1318359375,
       "learning_rate": 0.00013606329518754508,
-      "loss": 0.002,
+      "loss": 0.005,
       "step": 1743
     },
     {
       "epoch": 3.837183718371837,
-      "grad_norm": 0.09375,
+      "grad_norm": 0.263671875,
       "learning_rate": 0.00013599860245788178,
-      "loss": 0.0024,
+      "loss": 0.0062,
       "step": 1744
     },
     {
       "epoch": 3.8393839383938393,
-      "grad_norm": 0.296875,
+      "grad_norm": 0.1884765625,
       "learning_rate": 0.00013593389241455537,
-      "loss": 0.0041,
+      "loss": 0.0054,
       "step": 1745
     },
     {
       "epoch": 3.8415841584158414,
-      "grad_norm": 0.134765625,
+      "grad_norm": 0.376953125,
       "learning_rate": 0.00013586916508868827,
-      "loss": 0.0034,
+      "loss": 0.0046,
       "step": 1746
     },
     {
       "epoch": 3.843784378437844,
-      "grad_norm": 0.055908203125,
+      "grad_norm": 0.30859375,
       "learning_rate": 0.00013580442051141137,
-      "loss": 0.0011,
+      "loss": 0.0051,
       "step": 1747
     },
     {
       "epoch": 3.8459845984598457,
-      "grad_norm": 0.26953125,
+      "grad_norm": 0.318359375,
       "learning_rate": 0.0001357396587138638,
-      "loss": 0.0077,
+      "loss": 0.0065,
       "step": 1748
     },
     {
       "epoch": 3.8481848184818483,
-      "grad_norm": 0.1455078125,
+      "grad_norm": 0.126953125,
       "learning_rate": 0.00013567487972719305,
-      "loss": 0.0049,
+      "loss": 0.0065,
       "step": 1749
     },
     {
       "epoch": 3.8503850385038505,
-      "grad_norm": 0.21484375,
+      "grad_norm": 0.1669921875,
       "learning_rate": 0.00013561008358255468,
-      "loss": 0.0043,
+      "loss": 0.0066,
       "step": 1750
     },
     {
       "epoch": 3.8525852585258527,
-      "grad_norm": 0.1787109375,
+      "grad_norm": 0.275390625,
       "learning_rate": 0.0001355452703111128,
-      "loss": 0.0058,
+      "loss": 0.0125,
       "step": 1751
     },
     {
       "epoch": 3.854785478547855,
-      "grad_norm": 0.1416015625,
+      "grad_norm": 0.296875,
       "learning_rate": 0.00013548043994403953,
-      "loss": 0.0029,
+      "loss": 0.0092,
       "step": 1752
     },
     {
       "epoch": 3.856985698569857,
-      "grad_norm": 0.2197265625,
+      "grad_norm": 0.248046875,
       "learning_rate": 0.00013541559251251527,
-      "loss": 0.0038,
+      "loss": 0.0158,
       "step": 1753
     },
     {
       "epoch": 3.859185918591859,
-      "grad_norm": 0.10498046875,
+      "grad_norm": 0.25,
       "learning_rate": 0.00013535072804772864,
-      "loss": 0.0023,
+      "loss": 0.0083,
       "step": 1754
     },
     {
       "epoch": 3.8613861386138613,
-      "grad_norm": 0.291015625,
+      "grad_norm": 0.2353515625,
       "learning_rate": 0.0001352858465808764,
-      "loss": 0.0182,
+      "loss": 0.009,
       "step": 1755
     },
     {
       "epoch": 3.863586358635864,
-      "grad_norm": 0.32421875,
+      "grad_norm": 0.3203125,
       "learning_rate": 0.0001352209481431637,
-      "loss": 0.0032,
+      "loss": 0.0052,
       "step": 1756
     },
     {
       "epoch": 3.8657865786578656,
-      "grad_norm": 0.130859375,
+      "grad_norm": 0.353515625,
       "learning_rate": 0.0001351560327658035,
-      "loss": 0.0076,
+      "loss": 0.0109,
       "step": 1757
     },
     {
       "epoch": 3.867986798679868,
-      "grad_norm": 0.193359375,
+      "grad_norm": 0.16015625,
       "learning_rate": 0.00013509110048001715,
-      "loss": 0.0051,
+      "loss": 0.0042,
       "step": 1758
     },
     {
       "epoch": 3.8701870187018703,
-      "grad_norm": 0.1845703125,
+      "grad_norm": 0.09619140625,
       "learning_rate": 0.00013502615131703413,
-      "loss": 0.0035,
+      "loss": 0.002,
       "step": 1759
     },
     {
       "epoch": 3.8723872387238725,
-      "grad_norm": 0.08251953125,
+      "grad_norm": 0.2470703125,
       "learning_rate": 0.00013496118530809193,
-      "loss": 0.0027,
+      "loss": 0.0042,
       "step": 1760
     },
     {
       "epoch": 3.8745874587458746,
-      "grad_norm": 0.404296875,
+      "grad_norm": 0.12353515625,
       "learning_rate": 0.00013489620248443622,
-      "loss": 0.01,
+      "loss": 0.0029,
       "step": 1761
     },
     {
       "epoch": 3.8767876787678768,
-      "grad_norm": 0.1865234375,
+      "grad_norm": 0.1533203125,
       "learning_rate": 0.00013483120287732073,
-      "loss": 0.0062,
+      "loss": 0.0053,
       "step": 1762
     },
     {
       "epoch": 3.878987898789879,
-      "grad_norm": 0.1357421875,
+      "grad_norm": 0.1396484375,
       "learning_rate": 0.0001347661865180073,
-      "loss": 0.003,
+      "loss": 0.0034,
       "step": 1763
     },
     {
       "epoch": 3.881188118811881,
-      "grad_norm": 0.06201171875,
+      "grad_norm": 0.0966796875,
       "learning_rate": 0.0001347011534377657,
-      "loss": 0.0009,
+      "loss": 0.0028,
       "step": 1764
     },
     {
       "epoch": 3.8833883388338832,
-      "grad_norm": 0.1630859375,
+      "grad_norm": 0.5546875,
       "learning_rate": 0.00013463610366787392,
-      "loss": 0.0036,
+      "loss": 0.0073,
       "step": 1765
     },
     {
       "epoch": 3.8855885588558854,
-      "grad_norm": 0.134765625,
+      "grad_norm": 0.1923828125,
       "learning_rate": 0.00013457103723961783,
-      "loss": 0.0037,
+      "loss": 0.0064,
       "step": 1766
     },
     {
       "epoch": 3.887788778877888,
-      "grad_norm": 0.302734375,
+      "grad_norm": 0.54296875,
       "learning_rate": 0.00013450595418429146,
-      "loss": 0.0046,
+      "loss": 0.016,
       "step": 1767
     },
     {
       "epoch": 3.88998899889989,
-      "grad_norm": 0.1611328125,
+      "grad_norm": 0.06005859375,
       "learning_rate": 0.0001344408545331967,
-      "loss": 0.0026,
+      "loss": 0.0017,
       "step": 1768
     },
     {
       "epoch": 3.8921892189218923,
-      "grad_norm": 0.1279296875,
+      "grad_norm": 0.169921875,
       "learning_rate": 0.00013437573831764343,
-      "loss": 0.0044,
+      "loss": 0.0061,
       "step": 1769
     },
     {
       "epoch": 3.8943894389438944,
-      "grad_norm": 0.1474609375,
+      "grad_norm": 0.08447265625,
       "learning_rate": 0.0001343106055689496,
-      "loss": 0.0045,
+      "loss": 0.0032,
       "step": 1770
     },
     {
       "epoch": 3.8965896589658966,
-      "grad_norm": 0.08203125,
+      "grad_norm": 0.087890625,
       "learning_rate": 0.000134245456318441,
-      "loss": 0.0027,
+      "loss": 0.0029,
       "step": 1771
     },
     {
       "epoch": 3.8987898789878987,
-      "grad_norm": 0.04345703125,
+      "grad_norm": 0.1181640625,
       "learning_rate": 0.0001341802905974514,
-      "loss": 0.001,
+      "loss": 0.0065,
       "step": 1772
     },
     {
       "epoch": 3.900990099009901,
-      "grad_norm": 0.1552734375,
+      "grad_norm": 0.15234375,
       "learning_rate": 0.00013411510843732262,
-      "loss": 0.0042,
+      "loss": 0.0053,
       "step": 1773
     },
     {
       "epoch": 3.903190319031903,
-      "grad_norm": 0.06201171875,
+      "grad_norm": 0.07470703125,
       "learning_rate": 0.00013404990986940412,
-      "loss": 0.0013,
+      "loss": 0.0011,
       "step": 1774
     },
     {
       "epoch": 3.905390539053905,
-      "grad_norm": 0.04541015625,
+      "grad_norm": 0.2421875,
       "learning_rate": 0.0001339846949250535,
-      "loss": 0.0015,
+      "loss": 0.0062,
       "step": 1775
     },
     {
       "epoch": 3.907590759075908,
-      "grad_norm": 0.06884765625,
+      "grad_norm": 0.08935546875,
       "learning_rate": 0.00013391946363563604,
-      "loss": 0.0023,
+      "loss": 0.0017,
       "step": 1776
     },
     {
       "epoch": 3.9097909790979095,
-      "grad_norm": 0.408203125,
+      "grad_norm": 0.189453125,
       "learning_rate": 0.00013385421603252505,
-      "loss": 0.0097,
+      "loss": 0.0048,
       "step": 1777
     },
     {
       "epoch": 3.911991199119912,
-      "grad_norm": 0.08544921875,
+      "grad_norm": 0.1416015625,
       "learning_rate": 0.00013378895214710156,
-      "loss": 0.0039,
+      "loss": 0.0069,
       "step": 1778
     },
     {
       "epoch": 3.9141914191419143,
-      "grad_norm": 0.04931640625,
+      "grad_norm": 0.2470703125,
       "learning_rate": 0.00013372367201075453,
-      "loss": 0.0012,
+      "loss": 0.0035,
       "step": 1779
     },
     {
       "epoch": 3.9163916391639164,
-      "grad_norm": 0.13671875,
+      "grad_norm": 0.24609375,
       "learning_rate": 0.00013365837565488064,
-      "loss": 0.0024,
+      "loss": 0.0045,
       "step": 1780
     },
     {
       "epoch": 3.9185918591859186,
-      "grad_norm": 0.076171875,
+      "grad_norm": 0.1748046875,
       "learning_rate": 0.00013359306311088443,
-      "loss": 0.0007,
+      "loss": 0.0019,
       "step": 1781
     },
     {
       "epoch": 3.9207920792079207,
-      "grad_norm": 0.10009765625,
+      "grad_norm": 0.228515625,
       "learning_rate": 0.00013352773441017822,
-      "loss": 0.002,
+      "loss": 0.0042,
       "step": 1782
     },
     {
       "epoch": 3.922992299229923,
-      "grad_norm": 0.1318359375,
+      "grad_norm": 0.236328125,
       "learning_rate": 0.00013346238958418212,
-      "loss": 0.0015,
+      "loss": 0.002,
       "step": 1783
     },
     {
       "epoch": 3.925192519251925,
-      "grad_norm": 0.310546875,
+      "grad_norm": 0.212890625,
       "learning_rate": 0.00013339702866432392,
-      "loss": 0.0133,
+      "loss": 0.0099,
       "step": 1784
     },
     {
       "epoch": 3.9273927392739276,
-      "grad_norm": 0.11669921875,
+      "grad_norm": 0.357421875,
       "learning_rate": 0.00013333165168203924,
-      "loss": 0.005,
+      "loss": 0.0088,
       "step": 1785
     },
     {
       "epoch": 3.9295929592959293,
-      "grad_norm": 0.2490234375,
+      "grad_norm": 0.1943359375,
       "learning_rate": 0.00013326625866877135,
-      "loss": 0.0111,
+      "loss": 0.0048,
       "step": 1786
     },
     {
       "epoch": 3.931793179317932,
-      "grad_norm": 0.1533203125,
+      "grad_norm": 0.345703125,
       "learning_rate": 0.0001332008496559713,
-      "loss": 0.0012,
+      "loss": 0.0123,
       "step": 1787
     },
     {
       "epoch": 3.933993399339934,
-      "grad_norm": 0.09423828125,
+      "grad_norm": 0.1728515625,
       "learning_rate": 0.00013313542467509776,
-      "loss": 0.0016,
+      "loss": 0.0033,
       "step": 1788
     },
     {
       "epoch": 3.9361936193619362,
-      "grad_norm": 0.1962890625,
+      "grad_norm": 0.44140625,
       "learning_rate": 0.00013306998375761718,
-      "loss": 0.0043,
+      "loss": 0.0136,
       "step": 1789
     },
     {
       "epoch": 3.9383938393839384,
-      "grad_norm": 0.2275390625,
+      "grad_norm": 0.1689453125,
       "learning_rate": 0.0001330045269350036,
-      "loss": 0.0042,
+      "loss": 0.0028,
       "step": 1790
     },
     {
       "epoch": 3.9405940594059405,
-      "grad_norm": 0.310546875,
+      "grad_norm": 0.2138671875,
       "learning_rate": 0.00013293905423873866,
-      "loss": 0.0041,
+      "loss": 0.006,
       "step": 1791
     },
     {
       "epoch": 3.9427942794279427,
-      "grad_norm": 0.220703125,
+      "grad_norm": 0.11572265625,
       "learning_rate": 0.00013287356570031175,
-      "loss": 0.0036,
+      "loss": 0.0027,
       "step": 1792
     },
     {
       "epoch": 3.944994499449945,
-      "grad_norm": 0.244140625,
+      "grad_norm": 0.1591796875,
       "learning_rate": 0.00013280806135121983,
       "loss": 0.0043,
       "step": 1793
     },
     {
       "epoch": 3.9471947194719474,
-      "grad_norm": 0.0966796875,
+      "grad_norm": 0.173828125,
       "learning_rate": 0.00013274254122296747,
-      "loss": 0.0023,
+      "loss": 0.0036,
       "step": 1794
     },
     {
       "epoch": 3.949394939493949,
-      "grad_norm": 0.0908203125,
+      "grad_norm": 0.28515625,
       "learning_rate": 0.0001326770053470668,
-      "loss": 0.0021,
+      "loss": 0.0134,
       "step": 1795
     },
     {
       "epoch": 3.9515951595159517,
-      "grad_norm": 0.08349609375,
+      "grad_norm": 0.298828125,
       "learning_rate": 0.00013261145375503757,
-      "loss": 0.0012,
+      "loss": 0.0045,
       "step": 1796
     },
     {
       "epoch": 3.953795379537954,
-      "grad_norm": 0.0218505859375,
+      "grad_norm": 0.18359375,
       "learning_rate": 0.000132545886478407,
-      "loss": 0.0004,
+      "loss": 0.0068,
       "step": 1797
     },
     {
       "epoch": 3.955995599559956,
-      "grad_norm": 0.31640625,
+      "grad_norm": 0.25,
       "learning_rate": 0.00013248030354871004,
-      "loss": 0.0037,
+      "loss": 0.0027,
       "step": 1798
     },
     {
       "epoch": 3.958195819581958,
-      "grad_norm": 0.244140625,
+      "grad_norm": 0.2109375,
       "learning_rate": 0.00013241470499748893,
-      "loss": 0.0056,
+      "loss": 0.0082,
       "step": 1799
     },
     {
       "epoch": 3.9603960396039604,
-      "grad_norm": 0.1591796875,
+      "grad_norm": 0.2216796875,
       "learning_rate": 0.0001323490908562936,
-      "loss": 0.0028,
+      "loss": 0.0051,
       "step": 1800
     },
     {
       "epoch": 3.9625962596259625,
-      "grad_norm": 0.08203125,
+      "grad_norm": 0.255859375,
       "learning_rate": 0.0001322834611566814,
-      "loss": 0.0019,
+      "loss": 0.0067,
       "step": 1801
     },
     {
       "epoch": 3.9647964796479647,
-      "grad_norm": 0.337890625,
+      "grad_norm": 0.1259765625,
       "learning_rate": 0.00013221781593021717,
-      "loss": 0.004,
+      "loss": 0.0032,
       "step": 1802
     },
     {
       "epoch": 3.9669966996699673,
-      "grad_norm": 0.08447265625,
+      "grad_norm": 0.21484375,
       "learning_rate": 0.0001321521552084733,
-      "loss": 0.0009,
+      "loss": 0.0041,
       "step": 1803
     },
     {
       "epoch": 3.969196919691969,
-      "grad_norm": 0.201171875,
+      "grad_norm": 0.11083984375,
       "learning_rate": 0.00013208647902302945,
-      "loss": 0.0023,
+      "loss": 0.0039,
       "step": 1804
     },
     {
       "epoch": 3.9713971397139716,
-      "grad_norm": 0.1806640625,
+      "grad_norm": 0.15234375,
       "learning_rate": 0.00013202078740547288,
-      "loss": 0.0061,
+      "loss": 0.0075,
       "step": 1805
     },
     {
       "epoch": 3.9735973597359737,
-      "grad_norm": 0.07763671875,
+      "grad_norm": 0.25,
       "learning_rate": 0.00013195508038739824,
-      "loss": 0.0017,
+      "loss": 0.0054,
       "step": 1806
     },
     {
       "epoch": 3.975797579757976,
-      "grad_norm": 0.056884765625,
+      "grad_norm": 0.07763671875,
       "learning_rate": 0.00013188935800040753,
-      "loss": 0.0009,
+      "loss": 0.0029,
       "step": 1807
     },
     {
       "epoch": 3.977997799779978,
-      "grad_norm": 0.04638671875,
+      "grad_norm": 0.1474609375,
       "learning_rate": 0.0001318236202761102,
-      "loss": 0.0007,
+      "loss": 0.0028,
       "step": 1808
     },
     {
       "epoch": 3.98019801980198,
-      "grad_norm": 0.11181640625,
+      "grad_norm": 0.06689453125,
       "learning_rate": 0.00013175786724612307,
-      "loss": 0.0023,
+      "loss": 0.0012,
       "step": 1809
     },
     {
       "epoch": 3.9823982398239823,
-      "grad_norm": 0.0234375,
+      "grad_norm": 0.049560546875,
       "learning_rate": 0.0001316920989420703,
-      "loss": 0.0005,
+      "loss": 0.0011,
       "step": 1810
     },
     {
       "epoch": 3.9845984598459845,
-      "grad_norm": 0.12255859375,
+      "grad_norm": 0.12890625,
       "learning_rate": 0.00013162631539558338,
-      "loss": 0.0022,
+      "loss": 0.0024,
       "step": 1811
     },
     {
       "epoch": 3.9867986798679866,
-      "grad_norm": 0.244140625,
+      "grad_norm": 0.1357421875,
       "learning_rate": 0.00013156051663830114,
-      "loss": 0.0057,
+      "loss": 0.0043,
       "step": 1812
     },
     {
       "epoch": 3.988998899889989,
-      "grad_norm": 0.201171875,
+      "grad_norm": 0.283203125,
       "learning_rate": 0.00013149470270186982,
-      "loss": 0.0042,
+      "loss": 0.0101,
       "step": 1813
     },
     {
       "epoch": 3.9911991199119914,
-      "grad_norm": 0.0390625,
+      "grad_norm": 0.1083984375,
       "learning_rate": 0.00013142887361794277,
-      "loss": 0.0006,
+      "loss": 0.0029,
       "step": 1814
     },
     {
       "epoch": 3.9933993399339935,
-      "grad_norm": 0.1494140625,
+      "grad_norm": 0.1298828125,
       "learning_rate": 0.00013136302941818085,
-      "loss": 0.0064,
+      "loss": 0.006,
       "step": 1815
     },
     {
       "epoch": 3.9955995599559957,
-      "grad_norm": 0.1494140625,
+      "grad_norm": 0.302734375,
       "learning_rate": 0.00013129717013425198,
-      "loss": 0.0007,
+      "loss": 0.0093,
       "step": 1816
     },
     {
       "epoch": 3.997799779977998,
-      "grad_norm": 0.236328125,
+      "grad_norm": 0.31640625,
       "learning_rate": 0.00013123129579783148,
-      "loss": 0.0048,
+      "loss": 0.0089,
       "step": 1817
     },
     {
       "epoch": 4.0,
-      "grad_norm": 0.16015625,
+      "grad_norm": 0.2109375,
       "learning_rate": 0.00013116540644060186,
-      "loss": 0.0018,
+      "loss": 0.003,
       "step": 1818
     },
     {
       "epoch": 4.002200220022003,
-      "grad_norm": 0.400390625,
+      "grad_norm": 0.07568359375,
       "learning_rate": 0.00013109950209425284,
-      "loss": 0.0083,
+      "loss": 0.001,
       "step": 1819
     },
     {
       "epoch": 4.004400440044004,
-      "grad_norm": 0.0301513671875,
+      "grad_norm": 0.3203125,
       "learning_rate": 0.00013103358279048134,
-      "loss": 0.0006,
+      "loss": 0.0013,
       "step": 1820
     },
     {
       "epoch": 4.006600660066007,
-      "grad_norm": 0.1904296875,
+      "grad_norm": 0.1767578125,
       "learning_rate": 0.00013096764856099154,
-      "loss": 0.0027,
+      "loss": 0.0024,
       "step": 1821
     },
     {
       "epoch": 4.008800880088009,
-      "grad_norm": 0.10107421875,
+      "grad_norm": 0.0673828125,
       "learning_rate": 0.00013090169943749476,
-      "loss": 0.0009,
+      "loss": 0.001,
       "step": 1822
     },
     {
       "epoch": 4.011001100110011,
-      "grad_norm": 0.048828125,
+      "grad_norm": 0.44921875,
       "learning_rate": 0.00013083573545170945,
-      "loss": 0.0005,
+      "loss": 0.0038,
       "step": 1823
     },
     {
       "epoch": 4.013201320132013,
-      "grad_norm": 0.021484375,
+      "grad_norm": 0.259765625,
       "learning_rate": 0.00013076975663536123,
-      "loss": 0.0008,
+      "loss": 0.0071,
       "step": 1824
     },
     {
       "epoch": 4.013201320132013,
-      "eval_loss": 0.00523857306689024,
-      "eval_runtime": 10.85,
-      "eval_samples_per_second": 35.299,
-      "eval_steps_per_second": 4.424,
+      "eval_loss": 0.008145586587488651,
+      "eval_runtime": 11.0938,
+      "eval_samples_per_second": 34.524,
+      "eval_steps_per_second": 4.327,
       "step": 1824
     },
     {
       "epoch": 4.0154015401540155,
-      "grad_norm": 0.1845703125,
+      "grad_norm": 0.091796875,
       "learning_rate": 0.00013070376302018287,
       "loss": 0.002,
       "step": 1825
     },
     {
       "epoch": 4.017601760176017,
-      "grad_norm": 0.08154296875,
+      "grad_norm": 0.02490234375,
       "learning_rate": 0.00013063775463791427,
-      "loss": 0.0017,
+      "loss": 0.0009,
       "step": 1826
     },
     {
       "epoch": 4.01980198019802,
-      "grad_norm": 0.08154296875,
+      "grad_norm": 0.02001953125,
       "learning_rate": 0.00013057173152030235,
-      "loss": 0.0009,
+      "loss": 0.0006,
       "step": 1827
     },
     {
       "epoch": 4.022002200220022,
-      "grad_norm": 0.1552734375,
+      "grad_norm": 0.26953125,
       "learning_rate": 0.00013050569369910125,
-      "loss": 0.002,
+      "loss": 0.0186,
       "step": 1828
     },
     {
       "epoch": 4.024202420242024,
-      "grad_norm": 0.158203125,
+      "grad_norm": 0.0478515625,
       "learning_rate": 0.0001304396412060721,
-      "loss": 0.0045,
+      "loss": 0.001,
       "step": 1829
     },
     {
       "epoch": 4.026402640264027,
-      "grad_norm": 0.1806640625,
+      "grad_norm": 0.1298828125,
       "learning_rate": 0.000130373574072983,
-      "loss": 0.0033,
+      "loss": 0.003,
       "step": 1830
     },
     {
       "epoch": 4.028602860286028,
-      "grad_norm": 0.140625,
+      "grad_norm": 0.28125,
       "learning_rate": 0.00013030749233160932,
-      "loss": 0.0013,
+      "loss": 0.004,
       "step": 1831
     },
     {
       "epoch": 4.030803080308031,
-      "grad_norm": 0.08984375,
+      "grad_norm": 0.13671875,
       "learning_rate": 0.00013024139601373324,
-      "loss": 0.0014,
+      "loss": 0.0021,
       "step": 1832
     },
     {
       "epoch": 4.033003300330033,
-      "grad_norm": 0.203125,
+      "grad_norm": 0.2177734375,
       "learning_rate": 0.00013017528515114405,
-      "loss": 0.0028,
+      "loss": 0.0065,
       "step": 1833
     },
     {
       "epoch": 4.035203520352035,
-      "grad_norm": 0.11083984375,
+      "grad_norm": 0.2138671875,
       "learning_rate": 0.00013010915977563803,
-      "loss": 0.0011,
+      "loss": 0.0018,
       "step": 1834
     },
     {
       "epoch": 4.037403740374037,
-      "grad_norm": 0.150390625,
+      "grad_norm": 0.2236328125,
       "learning_rate": 0.00013004301991901838,
-      "loss": 0.0031,
+      "loss": 0.004,
       "step": 1835
     },
     {
       "epoch": 4.03960396039604,
-      "grad_norm": 0.26953125,
+      "grad_norm": 0.05322265625,
       "learning_rate": 0.00012997686561309538,
-      "loss": 0.0064,
+      "loss": 0.0016,
       "step": 1836
     },
     {
       "epoch": 4.041804180418042,
-      "grad_norm": 0.01318359375,
+      "grad_norm": 0.19140625,
       "learning_rate": 0.00012991069688968618,
-      "loss": 0.0005,
+      "loss": 0.0033,
       "step": 1837
     },
     {
       "epoch": 4.044004400440044,
-      "grad_norm": 0.16796875,
+      "grad_norm": 0.08544921875,
       "learning_rate": 0.00012984451378061484,
-      "loss": 0.003,
+      "loss": 0.0026,
       "step": 1838
     },
     {
       "epoch": 4.0462046204620465,
-      "grad_norm": 0.032958984375,
+      "grad_norm": 0.271484375,
       "learning_rate": 0.00012977831631771238,
-      "loss": 0.0007,
+      "loss": 0.0046,
       "step": 1839
     },
     {
       "epoch": 4.048404840484048,
-      "grad_norm": 0.01409912109375,
+      "grad_norm": 0.2041015625,
       "learning_rate": 0.00012971210453281674,
-      "loss": 0.0004,
+      "loss": 0.0018,
       "step": 1840
     },
     {
       "epoch": 4.050605060506051,
-      "grad_norm": 0.07275390625,
+      "grad_norm": 0.10595703125,
       "learning_rate": 0.0001296458784577727,
-      "loss": 0.0018,
+      "loss": 0.0032,
       "step": 1841
     },
     {
       "epoch": 4.052805280528053,
-      "grad_norm": 0.10693359375,
+      "grad_norm": 0.1015625,
       "learning_rate": 0.000129579638124432,
-      "loss": 0.0019,
+      "loss": 0.0023,
       "step": 1842
     },
     {
       "epoch": 4.055005500550055,
-      "grad_norm": 0.29296875,
+      "grad_norm": 0.0625,
       "learning_rate": 0.00012951338356465312,
-      "loss": 0.0085,
+      "loss": 0.0009,
       "step": 1843
     },
     {
       "epoch": 4.057205720572057,
-      "grad_norm": 0.0078125,
+      "grad_norm": 0.26953125,
       "learning_rate": 0.00012944711481030144,
-      "loss": 0.0003,
+      "loss": 0.0035,
       "step": 1844
     },
     {
       "epoch": 4.0594059405940595,
-      "grad_norm": 0.047119140625,
+      "grad_norm": 0.04248046875,
       "learning_rate": 0.0001293808318932492,
-      "loss": 0.0005,
+      "loss": 0.0008,
       "step": 1845
     },
     {
       "epoch": 4.061606160616061,
-      "grad_norm": 0.048828125,
+      "grad_norm": 0.11328125,
       "learning_rate": 0.0001293145348453754,
-      "loss": 0.0009,
+      "loss": 0.0014,
       "step": 1846
     },
     {
       "epoch": 4.063806380638064,
-      "grad_norm": 0.057373046875,
+      "grad_norm": 0.28125,
       "learning_rate": 0.00012924822369856585,
-      "loss": 0.0006,
+      "loss": 0.0079,
       "step": 1847
     },
     {
       "epoch": 4.066006600660066,
-      "grad_norm": 0.240234375,
+      "grad_norm": 0.1298828125,
       "learning_rate": 0.00012918189848471317,
-      "loss": 0.005,
+      "loss": 0.0017,
       "step": 1848
     },
     {
       "epoch": 4.068206820682068,
-      "grad_norm": 0.1630859375,
+      "grad_norm": 0.23046875,
       "learning_rate": 0.00012911555923571667,
-      "loss": 0.0081,
+      "loss": 0.0074,
       "step": 1849
     },
     {
       "epoch": 4.070407040704071,
-      "grad_norm": 0.16796875,
+      "grad_norm": 0.279296875,
       "learning_rate": 0.00012904920598348253,
-      "loss": 0.0023,
+      "loss": 0.0031,
       "step": 1850
     },
     {
       "epoch": 4.072607260726072,
-      "grad_norm": 0.041015625,
+      "grad_norm": 0.13671875,
       "learning_rate": 0.00012898283875992353,
-      "loss": 0.0012,
+      "loss": 0.0021,
       "step": 1851
     },
     {
       "epoch": 4.074807480748075,
-      "grad_norm": 0.1767578125,
+      "grad_norm": 0.431640625,
       "learning_rate": 0.00012891645759695926,
-      "loss": 0.0023,
+      "loss": 0.0038,
       "step": 1852
     },
     {
       "epoch": 4.077007700770077,
-      "grad_norm": 0.09326171875,
+      "grad_norm": 0.1982421875,
       "learning_rate": 0.00012885006252651602,
-      "loss": 0.0013,
+      "loss": 0.0016,
       "step": 1853
     },
     {
       "epoch": 4.079207920792079,
-      "grad_norm": 0.0166015625,
+      "grad_norm": 0.049072265625,
       "learning_rate": 0.0001287836535805267,
       "loss": 0.0004,
       "step": 1854
     },
     {
       "epoch": 4.081408140814081,
-      "grad_norm": 0.0245361328125,
+      "grad_norm": 0.2109375,
       "learning_rate": 0.000128717230790931,
-      "loss": 0.0007,
+      "loss": 0.0033,
       "step": 1855
     },
     {
       "epoch": 4.083608360836084,
-      "grad_norm": 0.024169921875,
+      "grad_norm": 0.0693359375,
       "learning_rate": 0.00012865079418967522,
-      "loss": 0.0005,
+      "loss": 0.0007,
       "step": 1856
     },
     {
       "epoch": 4.085808580858086,
-      "grad_norm": 0.279296875,
+      "grad_norm": 0.353515625,
       "learning_rate": 0.0001285843438087122,
-      "loss": 0.0078,
+      "loss": 0.0064,
       "step": 1857
     },
     {
       "epoch": 4.088008800880088,
-      "grad_norm": 0.03662109375,
+      "grad_norm": 0.1689453125,
       "learning_rate": 0.0001285178796800016,
-      "loss": 0.001,
+      "loss": 0.0034,
       "step": 1858
     },
     {
       "epoch": 4.0902090209020905,
-      "grad_norm": 0.06494140625,
+      "grad_norm": 0.1103515625,
       "learning_rate": 0.00012845140183550952,
-      "loss": 0.0021,
+      "loss": 0.004,
       "step": 1859
     },
     {
       "epoch": 4.092409240924092,
-      "grad_norm": 0.09814453125,
+      "grad_norm": 0.12158203125,
       "learning_rate": 0.0001283849103072088,
-      "loss": 0.0016,
+      "loss": 0.0029,
       "step": 1860
     },
     {
       "epoch": 4.094609460946095,
-      "grad_norm": 0.037841796875,
+      "grad_norm": 0.1513671875,
       "learning_rate": 0.0001283184051270787,
-      "loss": 0.0008,
+      "loss": 0.002,
       "step": 1861
     },
     {
       "epoch": 4.0968096809680965,
-      "grad_norm": 0.0091552734375,
+      "grad_norm": 0.244140625,
       "learning_rate": 0.00012825188632710523,
-      "loss": 0.0004,
+      "loss": 0.0058,
       "step": 1862
     },
     {
       "epoch": 4.099009900990099,
-      "grad_norm": 0.390625,
+      "grad_norm": 0.0283203125,
       "learning_rate": 0.0001281853539392808,
-      "loss": 0.0129,
+      "loss": 0.0005,
       "step": 1863
     },
     {
       "epoch": 4.101210121012101,
-      "grad_norm": 0.11962890625,
+      "grad_norm": 0.1787109375,
       "learning_rate": 0.00012811880799560443,
-      "loss": 0.0018,
+      "loss": 0.0037,
       "step": 1864
     },
     {
       "epoch": 4.103410341034103,
-      "grad_norm": 0.10205078125,
+      "grad_norm": 0.14453125,
       "learning_rate": 0.00012805224852808163,
       "loss": 0.0016,
       "step": 1865
     },
     {
       "epoch": 4.105610561056106,
-      "grad_norm": 0.11962890625,
+      "grad_norm": 0.06298828125,
       "learning_rate": 0.00012798567556872444,
-      "loss": 0.0007,
+      "loss": 0.0009,
       "step": 1866
     },
     {
       "epoch": 4.107810781078108,
-      "grad_norm": 0.11083984375,
+      "grad_norm": 0.40234375,
       "learning_rate": 0.00012791908914955137,
-      "loss": 0.0006,
+      "loss": 0.0022,
       "step": 1867
     },
     {
       "epoch": 4.11001100110011,
-      "grad_norm": 0.056884765625,
+      "grad_norm": 0.21484375,
       "learning_rate": 0.00012785248930258743,
-      "loss": 0.0011,
+      "loss": 0.0015,
       "step": 1868
     },
     {
       "epoch": 4.112211221122112,
-      "grad_norm": 0.2255859375,
+      "grad_norm": 0.2001953125,
       "learning_rate": 0.00012778587605986403,
-      "loss": 0.0019,
+      "loss": 0.0023,
       "step": 1869
     },
     {
       "epoch": 4.114411441144115,
-      "grad_norm": 0.0146484375,
+      "grad_norm": 0.035888671875,
       "learning_rate": 0.00012771924945341906,
-      "loss": 0.0004,
+      "loss": 0.0008,
       "step": 1870
     },
     {
       "epoch": 4.116611661166116,
-      "grad_norm": 0.06298828125,
+      "grad_norm": 0.1474609375,
       "learning_rate": 0.0001276526095152969,
-      "loss": 0.0009,
+      "loss": 0.0018,
       "step": 1871
     },
     {
       "epoch": 4.118811881188119,
-      "grad_norm": 0.07421875,
+      "grad_norm": 0.115234375,
       "learning_rate": 0.00012758595627754816,
-      "loss": 0.0009,
+      "loss": 0.0016,
       "step": 1872
     },
     {
       "epoch": 4.121012101210121,
-      "grad_norm": 0.0791015625,
+      "grad_norm": 0.031982421875,
       "learning_rate": 0.00012751928977223007,
-      "loss": 0.0009,
+      "loss": 0.0005,
       "step": 1873
     },
     {
       "epoch": 4.123212321232123,
-      "grad_norm": 0.0167236328125,
+      "grad_norm": 0.150390625,
       "learning_rate": 0.0001274526100314061,
-      "loss": 0.0005,
+      "loss": 0.0023,
       "step": 1874
     },
     {
       "epoch": 4.125412541254126,
-      "grad_norm": 0.2470703125,
+      "grad_norm": 0.220703125,
       "learning_rate": 0.00012738591708714615,
-      "loss": 0.0023,
+      "loss": 0.0035,
       "step": 1875
     },
     {
       "epoch": 4.1276127612761275,
-      "grad_norm": 0.099609375,
+      "grad_norm": 0.220703125,
       "learning_rate": 0.0001273192109715264,
-      "loss": 0.0012,
+      "loss": 0.0061,
       "step": 1876
     },
     {
       "epoch": 4.12981298129813,
-      "grad_norm": 0.2021484375,
+      "grad_norm": 0.06396484375,
       "learning_rate": 0.00012725249171662942,
-      "loss": 0.0012,
+      "loss": 0.0007,
       "step": 1877
     },
     {
       "epoch": 4.132013201320132,
-      "grad_norm": 0.045166015625,
+      "grad_norm": 0.1796875,
       "learning_rate": 0.00012718575935454407,
-      "loss": 0.0005,
+      "loss": 0.0015,
       "step": 1878
     },
     {
       "epoch": 4.134213421342134,
-      "grad_norm": 0.318359375,
+      "grad_norm": 0.07470703125,
       "learning_rate": 0.00012711901391736555,
-      "loss": 0.0097,
+      "loss": 0.0012,
       "step": 1879
     },
     {
       "epoch": 4.136413641364136,
-      "grad_norm": 0.0296630859375,
+      "grad_norm": 0.251953125,
       "learning_rate": 0.00012705225543719538,
-      "loss": 0.0005,
+      "loss": 0.0041,
       "step": 1880
     },
     {
       "epoch": 4.138613861386139,
-      "grad_norm": 0.07666015625,
+      "grad_norm": 0.0498046875,
       "learning_rate": 0.0001269854839461412,
-      "loss": 0.0008,
+      "loss": 0.0011,
       "step": 1881
     },
     {
       "epoch": 4.1408140814081404,
-      "grad_norm": 0.25390625,
+      "grad_norm": 0.2734375,
       "learning_rate": 0.00012691869947631713,
-      "loss": 0.0024,
+      "loss": 0.0021,
       "step": 1882
     },
     {
       "epoch": 4.143014301430143,
-      "grad_norm": 0.0947265625,
+      "grad_norm": 0.2275390625,
       "learning_rate": 0.00012685190205984325,
-      "loss": 0.0009,
+      "loss": 0.0021,
       "step": 1883
     },
     {
       "epoch": 4.145214521452146,
-      "grad_norm": 0.044921875,
+      "grad_norm": 0.1279296875,
       "learning_rate": 0.00012678509172884617,
-      "loss": 0.001,
+      "loss": 0.0026,
       "step": 1884
     },
     {
       "epoch": 4.147414741474147,
-      "grad_norm": 0.302734375,
+      "grad_norm": 0.123046875,
       "learning_rate": 0.00012671826851545851,
-      "loss": 0.0023,
+      "loss": 0.0022,
       "step": 1885
     },
     {
       "epoch": 4.14961496149615,
-      "grad_norm": 0.041015625,
+      "grad_norm": 0.3828125,
       "learning_rate": 0.00012665143245181918,
-      "loss": 0.0006,
+      "loss": 0.0137,
       "step": 1886
     },
     {
       "epoch": 4.151815181518152,
-      "grad_norm": 0.01611328125,
+      "grad_norm": 0.302734375,
       "learning_rate": 0.0001265845835700732,
-      "loss": 0.0004,
+      "loss": 0.0059,
       "step": 1887
     },
     {
       "epoch": 4.154015401540154,
-      "grad_norm": 0.05517578125,
+      "grad_norm": 0.251953125,
       "learning_rate": 0.00012651772190237182,
-      "loss": 0.0009,
+      "loss": 0.004,
       "step": 1888
     },
     {
       "epoch": 4.156215621562156,
-      "grad_norm": 0.01434326171875,
+      "grad_norm": 0.1962890625,
       "learning_rate": 0.00012645084748087236,
-      "loss": 0.0003,
+      "loss": 0.0058,
       "step": 1889
     },
     {
       "epoch": 4.158415841584159,
-      "grad_norm": 0.025634765625,
+      "grad_norm": 0.173828125,
       "learning_rate": 0.00012638396033773835,
-      "loss": 0.0005,
+      "loss": 0.0038,
       "step": 1890
     },
     {
       "epoch": 4.16061606160616,
-      "grad_norm": 0.1591796875,
+      "grad_norm": 0.2001953125,
       "learning_rate": 0.00012631706050513948,
-      "loss": 0.0008,
+      "loss": 0.0016,
       "step": 1891
     },
     {
       "epoch": 4.162816281628163,
-      "grad_norm": 0.33203125,
+      "grad_norm": 0.2099609375,
       "learning_rate": 0.00012625014801525137,
-      "loss": 0.0047,
+      "loss": 0.0015,
       "step": 1892
     },
     {
       "epoch": 4.165016501650165,
-      "grad_norm": 0.0130615234375,
+      "grad_norm": 0.09912109375,
       "learning_rate": 0.00012618322290025593,
-      "loss": 0.0003,
+      "loss": 0.0012,
       "step": 1893
     },
     {
       "epoch": 4.167216721672167,
-      "grad_norm": 0.283203125,
+      "grad_norm": 0.1865234375,
       "learning_rate": 0.00012611628519234094,
-      "loss": 0.003,
+      "loss": 0.011,
       "step": 1894
     },
     {
       "epoch": 4.16941694169417,
-      "grad_norm": 0.1865234375,
+      "grad_norm": 0.09375,
       "learning_rate": 0.00012604933492370044,
-      "loss": 0.0024,
+      "loss": 0.0023,
       "step": 1895
     },
     {
       "epoch": 4.1716171617161715,
-      "grad_norm": 0.0184326171875,
+      "grad_norm": 0.2734375,
       "learning_rate": 0.00012598237212653435,
-      "loss": 0.0005,
+      "loss": 0.0028,
       "step": 1896
     },
     {
       "epoch": 4.173817381738174,
-      "grad_norm": 0.0186767578125,
+      "grad_norm": 0.173828125,
       "learning_rate": 0.00012591539683304872,
-      "loss": 0.0004,
+      "loss": 0.0024,
       "step": 1897
     },
     {
       "epoch": 4.176017601760176,
-      "grad_norm": 0.1474609375,
+      "grad_norm": 0.1162109375,
       "learning_rate": 0.00012584840907545558,
-      "loss": 0.0026,
+      "loss": 0.0019,
       "step": 1898
     },
     {
       "epoch": 4.178217821782178,
-      "grad_norm": 0.026123046875,
+      "grad_norm": 0.03125,
       "learning_rate": 0.00012578140888597284,
-      "loss": 0.0005,
+      "loss": 0.0011,
       "step": 1899
     },
     {
       "epoch": 4.18041804180418,
-      "grad_norm": 0.015380859375,
+      "grad_norm": 0.058349609375,
       "learning_rate": 0.0001257143962968246,
-      "loss": 0.0004,
+      "loss": 0.0013,
       "step": 1900
     },
     {
       "epoch": 4.182618261826183,
-      "grad_norm": 0.0869140625,
+      "grad_norm": 0.052490234375,
       "learning_rate": 0.0001256473713402408,
-      "loss": 0.0006,
+      "loss": 0.0004,
       "step": 1901
     },
     {
       "epoch": 4.184818481848184,
-      "grad_norm": 0.044921875,
+      "grad_norm": 0.10107421875,
       "learning_rate": 0.00012558033404845733,
-      "loss": 0.0006,
+      "loss": 0.002,
       "step": 1902
     },
     {
       "epoch": 4.187018701870187,
-      "grad_norm": 0.0771484375,
+      "grad_norm": 0.06982421875,
       "learning_rate": 0.00012551328445371598,
-      "loss": 0.0086,
+      "loss": 0.0015,
       "step": 1903
     },
     {
       "epoch": 4.18921892189219,
-      "grad_norm": 0.0272216796875,
+      "grad_norm": 0.07666015625,
       "learning_rate": 0.00012544622258826464,
-      "loss": 0.0004,
+      "loss": 0.0009,
       "step": 1904
     },
     {
       "epoch": 4.191419141914191,
-      "grad_norm": 0.09326171875,
+      "grad_norm": 0.1474609375,
       "learning_rate": 0.00012537914848435682,
-      "loss": 0.0024,
+      "loss": 0.0028,
       "step": 1905
     },
     {
       "epoch": 4.193619361936194,
-      "grad_norm": 0.12255859375,
+      "grad_norm": 0.056884765625,
       "learning_rate": 0.0001253120621742521,
       "loss": 0.0011,
       "step": 1906
     },
     {
       "epoch": 4.195819581958196,
-      "grad_norm": 0.25,
+      "grad_norm": 0.01806640625,
       "learning_rate": 0.00012524496369021597,
-      "loss": 0.0044,
+      "loss": 0.0003,
       "step": 1907
     },
     {
       "epoch": 4.198019801980198,
-      "grad_norm": 0.197265625,
+      "grad_norm": 0.109375,
       "learning_rate": 0.0001251778530645196,
-      "loss": 0.0021,
+      "loss": 0.0015,
       "step": 1908
     },
     {
       "epoch": 4.2002200220022,
-      "grad_norm": 0.034423828125,
+      "grad_norm": 0.333984375,
       "learning_rate": 0.00012511073032944018,
-      "loss": 0.0007,
+      "loss": 0.0051,
       "step": 1909
     },
     {
       "epoch": 4.2024202420242025,
-      "grad_norm": 0.306640625,
+      "grad_norm": 0.142578125,
       "learning_rate": 0.0001250435955172606,
-      "loss": 0.0033,
+      "loss": 0.0014,
       "step": 1910
     },
     {
       "epoch": 4.204620462046204,
-      "grad_norm": 0.01068115234375,
+      "grad_norm": 0.255859375,
       "learning_rate": 0.00012497644866026953,
-      "loss": 0.0003,
+      "loss": 0.0034,
       "step": 1911
     },
     {
       "epoch": 4.206820682068207,
-      "grad_norm": 0.0172119140625,
+      "grad_norm": 0.06884765625,
       "learning_rate": 0.00012490928979076165,
-      "loss": 0.0003,
+      "loss": 0.0006,
       "step": 1912
     },
     {
       "epoch": 4.209020902090209,
-      "grad_norm": 0.0302734375,
+      "grad_norm": 0.291015625,
       "learning_rate": 0.00012484211894103718,
-      "loss": 0.0004,
+      "loss": 0.003,
       "step": 1913
     },
     {
       "epoch": 4.211221122112211,
-      "grad_norm": 0.27734375,
+      "grad_norm": 0.07177734375,
       "learning_rate": 0.0001247749361434022,
-      "loss": 0.0069,
+      "loss": 0.0007,
       "step": 1914
     },
     {
       "epoch": 4.213421342134214,
-      "grad_norm": 0.076171875,
+      "grad_norm": 0.08154296875,
       "learning_rate": 0.00012470774143016853,
-      "loss": 0.0006,
+      "loss": 0.0009,
       "step": 1915
     },
     {
       "epoch": 4.215621562156215,
-      "grad_norm": 0.1552734375,
+      "grad_norm": 0.3125,
       "learning_rate": 0.00012464053483365374,
-      "loss": 0.001,
+      "loss": 0.0031,
       "step": 1916
     },
     {
       "epoch": 4.217821782178218,
-      "grad_norm": 0.2060546875,
+      "grad_norm": 0.0615234375,
       "learning_rate": 0.00012457331638618106,
-      "loss": 0.0062,
+      "loss": 0.0009,
       "step": 1917
     },
     {
       "epoch": 4.22002200220022,
-      "grad_norm": 0.058837890625,
+      "grad_norm": 0.103515625,
       "learning_rate": 0.0001245060861200795,
-      "loss": 0.0008,
+      "loss": 0.002,
       "step": 1918
     },
     {
       "epoch": 4.222222222222222,
-      "grad_norm": 0.039306640625,
+      "grad_norm": 0.1376953125,
       "learning_rate": 0.00012443884406768368,
-      "loss": 0.0005,
+      "loss": 0.0024,
       "step": 1919
     },
     {
       "epoch": 4.224422442244224,
-      "grad_norm": 0.1455078125,
+      "grad_norm": 0.1953125,
       "learning_rate": 0.00012437159026133397,
-      "loss": 0.0016,
+      "loss": 0.0019,
       "step": 1920
     },
     {
       "epoch": 4.226622662266227,
-      "grad_norm": 0.0283203125,
+      "grad_norm": 0.11767578125,
       "learning_rate": 0.00012430432473337622,
-      "loss": 0.0007,
+      "loss": 0.0018,
       "step": 1921
     },
     {
       "epoch": 4.228822882288229,
-      "grad_norm": 0.416015625,
+      "grad_norm": 0.0908203125,
       "learning_rate": 0.00012423704751616214,
-      "loss": 0.0043,
+      "loss": 0.0012,
       "step": 1922
     },
     {
       "epoch": 4.231023102310231,
-      "grad_norm": 0.10791015625,
+      "grad_norm": 0.055419921875,
       "learning_rate": 0.00012416975864204896,
-      "loss": 0.0012,
+      "loss": 0.0008,
       "step": 1923
     },
     {
       "epoch": 4.2332233223322335,
-      "grad_norm": 0.38671875,
+      "grad_norm": 0.06787109375,
       "learning_rate": 0.00012410245814339948,
-      "loss": 0.0054,
+      "loss": 0.0013,
       "step": 1924
     },
     {
       "epoch": 4.235423542354235,
-      "grad_norm": 0.07958984375,
+      "grad_norm": 0.150390625,
       "learning_rate": 0.00012403514605258215,
-      "loss": 0.0005,
+      "loss": 0.0013,
       "step": 1925
     },
     {
       "epoch": 4.237623762376238,
-      "grad_norm": 0.3046875,
+      "grad_norm": 0.1865234375,
       "learning_rate": 0.00012396782240197096,
-      "loss": 0.0091,
+      "loss": 0.0017,
       "step": 1926
     },
     {
       "epoch": 4.2398239823982395,
-      "grad_norm": 0.048095703125,
+      "grad_norm": 0.26171875,
       "learning_rate": 0.00012390048722394544,
-      "loss": 0.0006,
+      "loss": 0.0074,
       "step": 1927
     },
     {
       "epoch": 4.242024202420242,
-      "grad_norm": 0.2138671875,
+      "grad_norm": 0.177734375,
       "learning_rate": 0.00012383314055089076,
-      "loss": 0.0034,
+      "loss": 0.0023,
       "step": 1928
     },
     {
       "epoch": 4.244224422442244,
-      "grad_norm": 0.062255859375,
+      "grad_norm": 0.09765625,
       "learning_rate": 0.0001237657824151975,
-      "loss": 0.0008,
+      "loss": 0.0017,
       "step": 1929
     },
     {
       "epoch": 4.2464246424642464,
-      "grad_norm": 0.1494140625,
+      "grad_norm": 0.2138671875,
       "learning_rate": 0.00012369841284926188,
-      "loss": 0.0011,
+      "loss": 0.0021,
       "step": 1930
     },
     {
       "epoch": 4.248624862486249,
-      "grad_norm": 0.2373046875,
+      "grad_norm": 0.0810546875,
       "learning_rate": 0.0001236310318854855,
-      "loss": 0.0021,
+      "loss": 0.001,
       "step": 1931
     },
     {
       "epoch": 4.250825082508251,
-      "grad_norm": 0.01104736328125,
+      "grad_norm": 0.0213623046875,
       "learning_rate": 0.0001235636395562755,
-      "loss": 0.0003,
+      "loss": 0.0004,
       "step": 1932
     },
     {
       "epoch": 4.253025302530253,
-      "grad_norm": 0.0234375,
+      "grad_norm": 0.236328125,
       "learning_rate": 0.00012349623589404444,
-      "loss": 0.0005,
+      "loss": 0.002,
       "step": 1933
     },
     {
       "epoch": 4.255225522552255,
-      "grad_norm": 0.10546875,
+      "grad_norm": 0.01275634765625,
       "learning_rate": 0.0001234288209312104,
-      "loss": 0.0006,
+      "loss": 0.0003,
       "step": 1934
     },
     {
       "epoch": 4.257425742574258,
-      "grad_norm": 0.259765625,
+      "grad_norm": 0.0751953125,
       "learning_rate": 0.00012336139470019684,
-      "loss": 0.0054,
+      "loss": 0.0007,
       "step": 1935
     },
     {
       "epoch": 4.259625962596259,
-      "grad_norm": 0.2734375,
+      "grad_norm": 0.2060546875,
       "learning_rate": 0.0001232939572334327,
-      "loss": 0.0029,
+      "loss": 0.0012,
       "step": 1936
     },
     {
       "epoch": 4.261826182618262,
-      "grad_norm": 0.37109375,
+      "grad_norm": 0.052001953125,
       "learning_rate": 0.00012322650856335223,
-      "loss": 0.004,
+      "loss": 0.0006,
       "step": 1937
     },
     {
       "epoch": 4.264026402640264,
-      "grad_norm": 0.10546875,
+      "grad_norm": 0.11474609375,
       "learning_rate": 0.00012315904872239514,
-      "loss": 0.001,
+      "loss": 0.002,
       "step": 1938
     },
     {
       "epoch": 4.264026402640264,
-      "eval_loss": 0.003189276671037078,
-      "eval_runtime": 11.0869,
-      "eval_samples_per_second": 34.545,
-      "eval_steps_per_second": 4.329,
+      "eval_loss": 0.005318178329616785,
+      "eval_runtime": 10.2331,
+      "eval_samples_per_second": 37.427,
+      "eval_steps_per_second": 4.691,
       "step": 1938
     },
     {
       "epoch": 4.266226622662266,
-      "grad_norm": 0.025146484375,
+      "grad_norm": 0.1123046875,
       "learning_rate": 0.0001230915777430065,
-      "loss": 0.0005,
+      "loss": 0.0012,
       "step": 1939
     },
     {
       "epoch": 4.268426842684269,
-      "grad_norm": 0.058837890625,
+      "grad_norm": 0.1318359375,
       "learning_rate": 0.0001230240956576367,
-      "loss": 0.0006,
+      "loss": 0.0011,
       "step": 1940
     },
     {
       "epoch": 4.270627062706271,
-      "grad_norm": 0.07275390625,
+      "grad_norm": 0.0361328125,
       "learning_rate": 0.0001229566024987416,
-      "loss": 0.002,
+      "loss": 0.0007,
       "step": 1941
     },
     {
       "epoch": 4.272827282728273,
-      "grad_norm": 0.0947265625,
+      "grad_norm": 0.05029296875,
       "learning_rate": 0.00012288909829878217,
       "loss": 0.0008,
       "step": 1942
     },
     {
       "epoch": 4.275027502750275,
-      "grad_norm": 0.1376953125,
+      "grad_norm": 0.23046875,
       "learning_rate": 0.00012282158309022488,
-      "loss": 0.003,
+      "loss": 0.0064,
       "step": 1943
     },
     {
       "epoch": 4.2772277227722775,
-      "grad_norm": 0.2109375,
+      "grad_norm": 0.263671875,
       "learning_rate": 0.00012275405690554135,
-      "loss": 0.0067,
+      "loss": 0.0037,
       "step": 1944
     },
     {
       "epoch": 4.279427942794279,
-      "grad_norm": 0.020751953125,
+      "grad_norm": 0.1708984375,
       "learning_rate": 0.00012268651977720866,
-      "loss": 0.0004,
+      "loss": 0.0041,
       "step": 1945
     },
     {
       "epoch": 4.281628162816282,
-      "grad_norm": 0.2177734375,
+      "grad_norm": 0.1064453125,
       "learning_rate": 0.00012261897173770897,
-      "loss": 0.0022,
+      "loss": 0.0009,
       "step": 1946
     },
     {
       "epoch": 4.2838283828382835,
-      "grad_norm": 0.02099609375,
+      "grad_norm": 0.058837890625,
       "learning_rate": 0.00012255141281952982,
-      "loss": 0.0004,
+      "loss": 0.0006,
       "step": 1947
     },
     {
       "epoch": 4.286028602860286,
-      "grad_norm": 0.0546875,
+      "grad_norm": 0.1923828125,
       "learning_rate": 0.00012248384305516387,
-      "loss": 0.0007,
+      "loss": 0.0039,
       "step": 1948
     },
     {
       "epoch": 4.288228822882289,
-      "grad_norm": 0.1630859375,
+      "grad_norm": 0.2099609375,
       "learning_rate": 0.00012241626247710906,
-      "loss": 0.0009,
+      "loss": 0.0028,
       "step": 1949
     },
     {
       "epoch": 4.29042904290429,
-      "grad_norm": 0.04443359375,
+      "grad_norm": 0.021728515625,
       "learning_rate": 0.00012234867111786851,
-      "loss": 0.0005,
+      "loss": 0.0004,
       "step": 1950
     },
     {
       "epoch": 4.292629262926293,
-      "grad_norm": 0.162109375,
+      "grad_norm": 0.0086669921875,
       "learning_rate": 0.0001222810690099506,
-      "loss": 0.0013,
+      "loss": 0.0002,
       "step": 1951
     },
     {
       "epoch": 4.294829482948295,
-      "grad_norm": 0.1474609375,
+      "grad_norm": 0.251953125,
       "learning_rate": 0.00012221345618586881,
-      "loss": 0.0037,
+      "loss": 0.0007,
       "step": 1952
     },
     {
       "epoch": 4.297029702970297,
-      "grad_norm": 0.0225830078125,
+      "grad_norm": 0.26953125,
       "learning_rate": 0.00012214583267814173,
-      "loss": 0.0003,
+      "loss": 0.0034,
       "step": 1953
     },
     {
       "epoch": 4.299229922992299,
-      "grad_norm": 0.01556396484375,
+      "grad_norm": 0.01025390625,
       "learning_rate": 0.00012207819851929315,
-      "loss": 0.0004,
+      "loss": 0.0003,
       "step": 1954
     },
     {
       "epoch": 4.301430143014302,
-      "grad_norm": 0.2041015625,
+      "grad_norm": 0.1982421875,
       "learning_rate": 0.00012201055374185198,
-      "loss": 0.0035,
+      "loss": 0.0017,
       "step": 1955
     },
     {
       "epoch": 4.303630363036303,
-      "grad_norm": 0.0322265625,
+      "grad_norm": 0.030517578125,
       "learning_rate": 0.00012194289837835224,
-      "loss": 0.0006,
+      "loss": 0.0005,
       "step": 1956
     },
     {
       "epoch": 4.305830583058306,
-      "grad_norm": 0.0712890625,
+      "grad_norm": 0.27734375,
       "learning_rate": 0.00012187523246133303,
-      "loss": 0.0006,
+      "loss": 0.0065,
       "step": 1957
     },
     {
       "epoch": 4.3080308030803085,
-      "grad_norm": 0.068359375,
+      "grad_norm": 0.416015625,
       "learning_rate": 0.00012180755602333856,
-      "loss": 0.0007,
+      "loss": 0.0044,
       "step": 1958
     },
     {
       "epoch": 4.31023102310231,
-      "grad_norm": 0.061279296875,
+      "grad_norm": 0.2392578125,
       "learning_rate": 0.00012173986909691799,
-      "loss": 0.0006,
+      "loss": 0.0024,
       "step": 1959
     },
     {
       "epoch": 4.312431243124313,
-      "grad_norm": 0.06298828125,
+      "grad_norm": 0.1982421875,
       "learning_rate": 0.00012167217171462566,
-      "loss": 0.0008,
+      "loss": 0.0043,
       "step": 1960
     },
     {
       "epoch": 4.3146314631463145,
-      "grad_norm": 0.0322265625,
+      "grad_norm": 0.1162109375,
       "learning_rate": 0.00012160446390902085,
-      "loss": 0.0004,
+      "loss": 0.001,
       "step": 1961
     },
     {
       "epoch": 4.316831683168317,
-      "grad_norm": 0.271484375,
+      "grad_norm": 0.16796875,
       "learning_rate": 0.00012153674571266791,
-      "loss": 0.008,
+      "loss": 0.0025,
       "step": 1962
     },
     {
       "epoch": 4.319031903190319,
-      "grad_norm": 0.07373046875,
+      "grad_norm": 0.0252685546875,
       "learning_rate": 0.00012146901715813617,
-      "loss": 0.0005,
+      "loss": 0.0006,
       "step": 1963
     },
     {
       "epoch": 4.321232123212321,
-      "grad_norm": 0.20703125,
+      "grad_norm": 0.1689453125,
       "learning_rate": 0.0001214012782779999,
-      "loss": 0.0017,
+      "loss": 0.0069,
       "step": 1964
     },
     {
       "epoch": 4.323432343234323,
-      "grad_norm": 0.1396484375,
+      "grad_norm": 0.032470703125,
       "learning_rate": 0.00012133352910483838,
-      "loss": 0.0008,
+      "loss": 0.0004,
       "step": 1965
     },
     {
       "epoch": 4.325632563256326,
-      "grad_norm": 0.2041015625,
+      "grad_norm": 0.2138671875,
       "learning_rate": 0.00012126576967123581,
-      "loss": 0.005,
+      "loss": 0.0031,
       "step": 1966
     },
     {
       "epoch": 4.327832783278327,
-      "grad_norm": 0.1357421875,
+      "grad_norm": 0.392578125,
       "learning_rate": 0.00012119800000978136,
-      "loss": 0.0025,
+      "loss": 0.0066,
       "step": 1967
     },
     {
       "epoch": 4.33003300330033,
-      "grad_norm": 0.018798828125,
+      "grad_norm": 0.1787109375,
       "learning_rate": 0.00012113022015306916,
-      "loss": 0.0003,
+      "loss": 0.0018,
       "step": 1968
     },
     {
       "epoch": 4.332233223322333,
-      "grad_norm": 0.0257568359375,
+      "grad_norm": 0.185546875,
       "learning_rate": 0.00012106243013369811,
-      "loss": 0.0005,
+      "loss": 0.004,
       "step": 1969
     },
     {
       "epoch": 4.334433443344334,
-      "grad_norm": 0.08837890625,
+      "grad_norm": 0.07421875,
       "learning_rate": 0.0001209946299842721,
-      "loss": 0.0006,
+      "loss": 0.0012,
       "step": 1970
     },
     {
       "epoch": 4.336633663366337,
-      "grad_norm": 0.041259765625,
+      "grad_norm": 0.173828125,
       "learning_rate": 0.00012092681973739988,
-      "loss": 0.0006,
+      "loss": 0.0036,
       "step": 1971
     },
     {
       "epoch": 4.338833883388339,
-      "grad_norm": 0.259765625,
+      "grad_norm": 0.1103515625,
       "learning_rate": 0.00012085899942569498,
-      "loss": 0.0038,
+      "loss": 0.0025,
       "step": 1972
     },
     {
       "epoch": 4.341034103410341,
-      "grad_norm": 0.06689453125,
+      "grad_norm": 0.2041015625,
       "learning_rate": 0.00012079116908177593,
-      "loss": 0.0005,
+      "loss": 0.0059,
       "step": 1973
     },
     {
       "epoch": 4.343234323432343,
-      "grad_norm": 0.01043701171875,
+      "grad_norm": 0.421875,
       "learning_rate": 0.00012072332873826595,
-      "loss": 0.0003,
+      "loss": 0.0034,
       "step": 1974
     },
     {
       "epoch": 4.3454345434543455,
-      "grad_norm": 0.1943359375,
+      "grad_norm": 0.2578125,
       "learning_rate": 0.0001206554784277931,
-      "loss": 0.0019,
+      "loss": 0.002,
       "step": 1975
     },
     {
       "epoch": 4.347634763476347,
-      "grad_norm": 0.00958251953125,
+      "grad_norm": 0.10302734375,
       "learning_rate": 0.00012058761818299022,
-      "loss": 0.0003,
+      "loss": 0.0012,
       "step": 1976
     },
     {
       "epoch": 4.34983498349835,
-      "grad_norm": 0.2109375,
+      "grad_norm": 0.158203125,
       "learning_rate": 0.000120519748036495,
-      "loss": 0.0021,
+      "loss": 0.0015,
       "step": 1977
     },
     {
       "epoch": 4.3520352035203524,
-      "grad_norm": 0.048583984375,
+      "grad_norm": 0.248046875,
       "learning_rate": 0.00012045186802094977,
-      "loss": 0.001,
+      "loss": 0.0047,
       "step": 1978
     },
     {
       "epoch": 4.354235423542354,
-      "grad_norm": 0.1669921875,
+      "grad_norm": 0.10693359375,
       "learning_rate": 0.00012038397816900177,
-      "loss": 0.0028,
+      "loss": 0.0016,
       "step": 1979
     },
     {
       "epoch": 4.356435643564357,
-      "grad_norm": 0.251953125,
+      "grad_norm": 0.1484375,
       "learning_rate": 0.00012031607851330282,
-      "loss": 0.0017,
+      "loss": 0.0029,
       "step": 1980
     },
     {
       "epoch": 4.3586358635863585,
-      "grad_norm": 0.03955078125,
+      "grad_norm": 0.255859375,
       "learning_rate": 0.00012024816908650953,
-      "loss": 0.0007,
+      "loss": 0.0069,
       "step": 1981
     },
     {
       "epoch": 4.360836083608361,
-      "grad_norm": 0.1650390625,
+      "grad_norm": 0.09130859375,
       "learning_rate": 0.00012018024992128318,
-      "loss": 0.0033,
+      "loss": 0.0017,
       "step": 1982
     },
     {
       "epoch": 4.363036303630363,
-      "grad_norm": 0.0625,
+      "grad_norm": 0.1533203125,
       "learning_rate": 0.00012011232105028975,
-      "loss": 0.0013,
+      "loss": 0.0087,
       "step": 1983
     },
     {
       "epoch": 4.365236523652365,
-      "grad_norm": 0.0281982421875,
+      "grad_norm": 0.01397705078125,
       "learning_rate": 0.00012004438250619991,
       "loss": 0.0003,
       "step": 1984
     },
     {
       "epoch": 4.367436743674367,
-      "grad_norm": 0.05419921875,
+      "grad_norm": 0.07861328125,
       "learning_rate": 0.000119976434321689,
-      "loss": 0.0004,
+      "loss": 0.0007,
       "step": 1985
     },
     {
       "epoch": 4.36963696369637,
-      "grad_norm": 0.00848388671875,
+      "grad_norm": 0.068359375,
       "learning_rate": 0.00011990847652943685,
-      "loss": 0.0003,
+      "loss": 0.001,
       "step": 1986
     },
     {
       "epoch": 4.371837183718371,
-      "grad_norm": 0.0380859375,
+      "grad_norm": 0.0634765625,
       "learning_rate": 0.00011984050916212806,
-      "loss": 0.0008,
+      "loss": 0.0011,
       "step": 1987
     },
     {
       "epoch": 4.374037403740374,
-      "grad_norm": 0.0125732421875,
+      "grad_norm": 0.06591796875,
       "learning_rate": 0.00011977253225245179,
-      "loss": 0.0004,
+      "loss": 0.0016,
       "step": 1988
     },
     {
       "epoch": 4.376237623762377,
-      "grad_norm": 0.022216796875,
+      "grad_norm": 0.060791015625,
       "learning_rate": 0.0001197045458331018,
-      "loss": 0.0003,
+      "loss": 0.0011,
       "step": 1989
     },
     {
       "epoch": 4.378437843784378,
-      "grad_norm": 0.006317138671875,
+      "grad_norm": 0.0289306640625,
       "learning_rate": 0.00011963654993677645,
-      "loss": 0.0003,
+      "loss": 0.0005,
       "step": 1990
     },
     {
       "epoch": 4.380638063806381,
-      "grad_norm": 0.06884765625,
+      "grad_norm": 0.1708984375,
       "learning_rate": 0.00011956854459617854,
-      "loss": 0.0023,
+      "loss": 0.0042,
       "step": 1991
     },
     {
       "epoch": 4.382838283828383,
-      "grad_norm": 0.04931640625,
+      "grad_norm": 0.044189453125,
       "learning_rate": 0.00011950052984401554,
-      "loss": 0.001,
+      "loss": 0.0012,
       "step": 1992
     },
     {
       "epoch": 4.385038503850385,
-      "grad_norm": 0.173828125,
+      "grad_norm": 0.0244140625,
       "learning_rate": 0.0001194325057129994,
-      "loss": 0.0027,
+      "loss": 0.0004,
       "step": 1993
     },
     {
       "epoch": 4.387238723872387,
-      "grad_norm": 0.017333984375,
+      "grad_norm": 0.2373046875,
       "learning_rate": 0.00011936447223584657,
-      "loss": 0.0004,
+      "loss": 0.0019,
       "step": 1994
     },
     {
       "epoch": 4.3894389438943895,
-      "grad_norm": 0.0306396484375,
+      "grad_norm": 0.03125,
       "learning_rate": 0.00011929642944527806,
-      "loss": 0.0004,
+      "loss": 0.0005,
       "step": 1995
     },
     {
       "epoch": 4.391639163916391,
-      "grad_norm": 0.054931640625,
+      "grad_norm": 0.07958984375,
       "learning_rate": 0.00011922837737401922,
-      "loss": 0.0005,
+      "loss": 0.001,
       "step": 1996
     },
     {
       "epoch": 4.393839383938394,
-      "grad_norm": 0.0771484375,
+      "grad_norm": 0.2578125,
       "learning_rate": 0.0001191603160548,
-      "loss": 0.0008,
+      "loss": 0.0141,
       "step": 1997
     },
     {
       "epoch": 4.396039603960396,
-      "grad_norm": 0.2021484375,
+      "grad_norm": 0.455078125,
       "learning_rate": 0.00011909224552035474,
-      "loss": 0.0016,
+      "loss": 0.0063,
       "step": 1998
     },
     {
       "epoch": 4.398239823982398,
-      "grad_norm": 0.224609375,
+      "grad_norm": 0.05322265625,
       "learning_rate": 0.00011902416580342221,
-      "loss": 0.0014,
+      "loss": 0.0007,
       "step": 1999
     },
     {
       "epoch": 4.400440044004401,
-      "grad_norm": 0.01409912109375,
+      "grad_norm": 0.05224609375,
       "learning_rate": 0.0001189560769367456,
-      "loss": 0.0003,
+      "loss": 0.0007,
       "step": 2000
     },
     {
       "epoch": 4.402640264026402,
-      "grad_norm": 0.019775390625,
+      "grad_norm": 0.291015625,
       "learning_rate": 0.00011888797895307258,
-      "loss": 0.0003,
+      "loss": 0.0044,
       "step": 2001
     },
     {
       "epoch": 4.404840484048405,
-      "grad_norm": 0.00848388671875,
+      "grad_norm": 0.1533203125,
       "learning_rate": 0.00011881987188515502,
-      "loss": 0.0002,
+      "loss": 0.002,
       "step": 2002
     },
     {
       "epoch": 4.407040704070407,
-      "grad_norm": 0.06201171875,
+      "grad_norm": 0.04638671875,
       "learning_rate": 0.00011875175576574933,
-      "loss": 0.001,
+      "loss": 0.0009,
       "step": 2003
     },
     {
       "epoch": 4.409240924092409,
-      "grad_norm": 0.06982421875,
+      "grad_norm": 0.0303955078125,
       "learning_rate": 0.00011868363062761621,
-      "loss": 0.001,
+      "loss": 0.0007,
       "step": 2004
     },
     {
       "epoch": 4.411441144114411,
-      "grad_norm": 0.0361328125,
+      "grad_norm": 0.181640625,
       "learning_rate": 0.00011861549650352069,
-      "loss": 0.0006,
+      "loss": 0.0029,
       "step": 2005
     },
     {
       "epoch": 4.413641364136414,
-      "grad_norm": 0.1474609375,
+      "grad_norm": 0.3359375,
       "learning_rate": 0.00011854735342623213,
-      "loss": 0.0016,
+      "loss": 0.0084,
       "step": 2006
     },
     {
       "epoch": 4.415841584158416,
-      "grad_norm": 0.046875,
+      "grad_norm": 0.10498046875,
       "learning_rate": 0.00011847920142852424,
-      "loss": 0.0012,
+      "loss": 0.0018,
       "step": 2007
     },
     {
       "epoch": 4.418041804180418,
-      "grad_norm": 0.050048828125,
+      "grad_norm": 0.2109375,
       "learning_rate": 0.00011841104054317494,
-      "loss": 0.0006,
+      "loss": 0.0035,
       "step": 2008
     },
     {
       "epoch": 4.4202420242024205,
-      "grad_norm": 0.0120849609375,
+      "grad_norm": 0.07177734375,
       "learning_rate": 0.00011834287080296644,
-      "loss": 0.0003,
+      "loss": 0.0015,
       "step": 2009
     },
     {
       "epoch": 4.422442244224422,
-      "grad_norm": 0.08642578125,
+      "grad_norm": 0.28125,
       "learning_rate": 0.0001182746922406853,
-      "loss": 0.001,
+      "loss": 0.0018,
       "step": 2010
     },
     {
       "epoch": 4.424642464246425,
-      "grad_norm": 0.138671875,
+      "grad_norm": 0.515625,
       "learning_rate": 0.0001182065048891222,
-      "loss": 0.0013,
+      "loss": 0.0026,
       "step": 2011
     },
     {
       "epoch": 4.4268426842684265,
-      "grad_norm": 0.01416015625,
+      "grad_norm": 0.26953125,
       "learning_rate": 0.00011813830878107215,
-      "loss": 0.0004,
+      "loss": 0.0045,
       "step": 2012
     },
     {
       "epoch": 4.429042904290429,
-      "grad_norm": 0.0228271484375,
+      "grad_norm": 0.033447265625,
       "learning_rate": 0.00011807010394933429,
-      "loss": 0.0004,
+      "loss": 0.0007,
       "step": 2013
     },
     {
       "epoch": 4.431243124312431,
-      "grad_norm": 0.10693359375,
+      "grad_norm": 0.031982421875,
       "learning_rate": 0.00011800189042671198,
       "loss": 0.0007,
       "step": 2014
     },
     {
       "epoch": 4.433443344334433,
-      "grad_norm": 0.06396484375,
+      "grad_norm": 0.1279296875,
       "learning_rate": 0.0001179336682460128,
-      "loss": 0.001,
+      "loss": 0.0012,
       "step": 2015
     },
     {
       "epoch": 4.435643564356436,
-      "grad_norm": 0.009033203125,
+      "grad_norm": 0.1875,
       "learning_rate": 0.00011786543744004846,
-      "loss": 0.0002,
+      "loss": 0.0042,
       "step": 2016
     },
     {
       "epoch": 4.437843784378438,
-      "grad_norm": 0.08935546875,
+      "grad_norm": 0.041015625,
       "learning_rate": 0.0001177971980416348,
       "loss": 0.0007,
       "step": 2017
     },
     {
       "epoch": 4.44004400440044,
-      "grad_norm": 0.07763671875,
+      "grad_norm": 0.1875,
       "learning_rate": 0.00011772895008359186,
-      "loss": 0.0012,
+      "loss": 0.0016,
       "step": 2018
     },
     {
       "epoch": 4.442244224422442,
       "grad_norm": 0.043701171875,
       "learning_rate": 0.0001176606935987437,
-      "loss": 0.0005,
+      "loss": 0.0006,
       "step": 2019
     },
     {
       "epoch": 4.444444444444445,
-      "grad_norm": 0.016357421875,
+      "grad_norm": 0.11669921875,
       "learning_rate": 0.00011759242861991855,
-      "loss": 0.0003,
+      "loss": 0.0014,
       "step": 2020
     },
     {
       "epoch": 4.446644664466446,
-      "grad_norm": 0.322265625,
+      "grad_norm": 0.0830078125,
       "learning_rate": 0.00011752415517994872,
-      "loss": 0.0012,
+      "loss": 0.0015,
       "step": 2021
     },
     {
       "epoch": 4.448844884488449,
-      "grad_norm": 0.05859375,
+      "grad_norm": 0.040771484375,
       "learning_rate": 0.00011745587331167058,
-      "loss": 0.0007,
+      "loss": 0.0008,
       "step": 2022
     },
     {
       "epoch": 4.451045104510451,
-      "grad_norm": 0.0054931640625,
+      "grad_norm": 0.158203125,
       "learning_rate": 0.00011738758304792456,
-      "loss": 0.0002,
+      "loss": 0.0031,
       "step": 2023
     },
     {
       "epoch": 4.453245324532453,
-      "grad_norm": 0.462890625,
+      "grad_norm": 0.0927734375,
       "learning_rate": 0.00011731928442155508,
-      "loss": 0.0073,
+      "loss": 0.0016,
       "step": 2024
     },
     {
       "epoch": 4.455445544554456,
-      "grad_norm": 0.0096435546875,
+      "grad_norm": 0.1279296875,
       "learning_rate": 0.00011725097746541065,
-      "loss": 0.0002,
+      "loss": 0.0008,
       "step": 2025
     },
     {
       "epoch": 4.457645764576458,
-      "grad_norm": 0.0986328125,
+      "grad_norm": 0.033447265625,
       "learning_rate": 0.00011718266221234375,
-      "loss": 0.0023,
+      "loss": 0.0007,
       "step": 2026
     },
     {
       "epoch": 4.45984598459846,
-      "grad_norm": 0.006378173828125,
+      "grad_norm": 0.1728515625,
       "learning_rate": 0.00011711433869521087,
-      "loss": 0.0002,
+      "loss": 0.0035,
       "step": 2027
     },
     {
       "epoch": 4.462046204620462,
-      "grad_norm": 0.19140625,
+      "grad_norm": 0.1064453125,
       "learning_rate": 0.00011704600694687242,
-      "loss": 0.0054,
+      "loss": 0.0023,
       "step": 2028
     },
     {
       "epoch": 4.4642464246424645,
-      "grad_norm": 0.03759765625,
+      "grad_norm": 0.2490234375,
       "learning_rate": 0.00011697766700019289,
-      "loss": 0.0009,
+      "loss": 0.005,
       "step": 2029
     },
     {
       "epoch": 4.466446644664466,
-      "grad_norm": 0.2216796875,
+      "grad_norm": 0.047119140625,
       "learning_rate": 0.00011690931888804055,
-      "loss": 0.0036,
+      "loss": 0.0009,
       "step": 2030
     },
     {
       "epoch": 4.468646864686469,
-      "grad_norm": 0.04443359375,
+      "grad_norm": 0.04296875,
       "learning_rate": 0.00011684096264328773,
-      "loss": 0.0005,
+      "loss": 0.0007,
       "step": 2031
     },
     {
       "epoch": 4.4708470847084705,
-      "grad_norm": 0.009521484375,
+      "grad_norm": 0.01495361328125,
       "learning_rate": 0.0001167725982988106,
-      "loss": 0.0002,
+      "loss": 0.0004,
       "step": 2032
     },
     {
       "epoch": 4.473047304730473,
-      "grad_norm": 0.359375,
+      "grad_norm": 0.2119140625,
       "learning_rate": 0.00011670422588748925,
-      "loss": 0.0026,
+      "loss": 0.0012,
       "step": 2033
     },
     {
       "epoch": 4.475247524752476,
-      "grad_norm": 0.05712890625,
+      "grad_norm": 0.02490234375,
       "learning_rate": 0.0001166358454422077,
       "loss": 0.0005,
       "step": 2034
     },
     {
       "epoch": 4.477447744774477,
-      "grad_norm": 0.09716796875,
+      "grad_norm": 0.09033203125,
       "learning_rate": 0.00011656745699585371,
-      "loss": 0.0013,
+      "loss": 0.0021,
       "step": 2035
     },
     {
       "epoch": 4.47964796479648,
-      "grad_norm": 0.12255859375,
+      "grad_norm": 0.11474609375,
       "learning_rate": 0.00011649906058131901,
-      "loss": 0.001,
+      "loss": 0.0014,
       "step": 2036
     },
     {
       "epoch": 4.481848184818482,
-      "grad_norm": 0.0142822265625,
+      "grad_norm": 0.19921875,
       "learning_rate": 0.00011643065623149907,
-      "loss": 0.0003,
+      "loss": 0.0025,
       "step": 2037
     },
     {
       "epoch": 4.484048404840484,
-      "grad_norm": 0.042236328125,
+      "grad_norm": 0.318359375,
       "learning_rate": 0.00011636224397929328,
-      "loss": 0.0009,
+      "loss": 0.0038,
       "step": 2038
     },
     {
       "epoch": 4.486248624862486,
-      "grad_norm": 0.138671875,
+      "grad_norm": 0.1025390625,
       "learning_rate": 0.0001162938238576047,
-      "loss": 0.0006,
+      "loss": 0.0012,
       "step": 2039
     },
     {
       "epoch": 4.488448844884489,
-      "grad_norm": 0.0166015625,
+      "grad_norm": 0.2041015625,
       "learning_rate": 0.00011622539589934027,
-      "loss": 0.0004,
+      "loss": 0.0025,
       "step": 2040
     },
     {
       "epoch": 4.49064906490649,
-      "grad_norm": 0.10546875,
+      "grad_norm": 0.158203125,
       "learning_rate": 0.0001161569601374107,
-      "loss": 0.0015,
+      "loss": 0.0013,
       "step": 2041
     },
     {
       "epoch": 4.492849284928493,
-      "grad_norm": 0.01611328125,
+      "grad_norm": 0.224609375,
       "learning_rate": 0.0001160885166047304,
-      "loss": 0.0004,
+      "loss": 0.0021,
       "step": 2042
     },
     {
       "epoch": 4.4950495049504955,
-      "grad_norm": 0.150390625,
+      "grad_norm": 0.171875,
       "learning_rate": 0.00011602006533421756,
-      "loss": 0.0032,
+      "loss": 0.0017,
       "step": 2043
     },
     {
       "epoch": 4.497249724972497,
-      "grad_norm": 0.011474609375,
+      "grad_norm": 0.1279296875,
       "learning_rate": 0.00011595160635879407,
-      "loss": 0.0002,
+      "loss": 0.0006,
       "step": 2044
     },
     {
       "epoch": 4.4994499449945,
-      "grad_norm": 0.04541015625,
+      "grad_norm": 0.07421875,
       "learning_rate": 0.00011588313971138552,
-      "loss": 0.0006,
+      "loss": 0.0021,
       "step": 2045
     },
     {
       "epoch": 4.5016501650165015,
-      "grad_norm": 0.048828125,
+      "grad_norm": 0.076171875,
       "learning_rate": 0.0001158146654249212,
-      "loss": 0.0006,
+      "loss": 0.0015,
       "step": 2046
     },
     {
       "epoch": 4.503850385038504,
-      "grad_norm": 0.337890625,
+      "grad_norm": 0.150390625,
       "learning_rate": 0.00011574618353233413,
-      "loss": 0.0026,
+      "loss": 0.0084,
       "step": 2047
     },
     {
       "epoch": 4.506050605060506,
-      "grad_norm": 0.07373046875,
+      "grad_norm": 0.0196533203125,
       "learning_rate": 0.00011567769406656085,
-      "loss": 0.0006,
+      "loss": 0.0005,
       "step": 2048
     },
     {
       "epoch": 4.508250825082508,
-      "grad_norm": 0.06689453125,
+      "grad_norm": 0.064453125,
       "learning_rate": 0.00011560919706054167,
-      "loss": 0.0006,
+      "loss": 0.0012,
       "step": 2049
     },
     {
       "epoch": 4.51045104510451,
-      "grad_norm": 0.16015625,
+      "grad_norm": 0.08349609375,
       "learning_rate": 0.00011554069254722051,
-      "loss": 0.0024,
+      "loss": 0.0007,
       "step": 2050
     },
     {
       "epoch": 4.512651265126513,
-      "grad_norm": 0.0166015625,
+      "grad_norm": 0.10986328125,
       "learning_rate": 0.00011547218055954481,
-      "loss": 0.0003,
+      "loss": 0.0011,
       "step": 2051
     },
     {
       "epoch": 4.514851485148515,
-      "grad_norm": 0.01129150390625,
+      "grad_norm": 0.18359375,
       "learning_rate": 0.00011540366113046573,
-      "loss": 0.0002,
+      "loss": 0.0028,
       "step": 2052
     },
     {
       "epoch": 4.514851485148515,
-      "eval_loss": 0.002514768624678254,
-      "eval_runtime": 10.3209,
-      "eval_samples_per_second": 37.109,
-      "eval_steps_per_second": 4.651,
+      "eval_loss": 0.0034243022091686726,
+      "eval_runtime": 10.4077,
+      "eval_samples_per_second": 36.8,
+      "eval_steps_per_second": 4.612,
       "step": 2052
     },
     {
       "epoch": 4.517051705170517,
-      "grad_norm": 0.03369140625,
+      "grad_norm": 0.11328125,
       "learning_rate": 0.0001153351342929379,
-      "loss": 0.0004,
+      "loss": 0.001,
       "step": 2053
     },
     {
       "epoch": 4.51925192519252,
-      "grad_norm": 0.328125,
+      "grad_norm": 0.0137939453125,
       "learning_rate": 0.00011526660007991956,
-      "loss": 0.0033,
+      "loss": 0.0003,
       "step": 2054
     },
     {
       "epoch": 4.521452145214521,
-      "grad_norm": 0.1787109375,
+      "grad_norm": 0.189453125,
       "learning_rate": 0.00011519805852437253,
-      "loss": 0.0012,
+      "loss": 0.0046,
       "step": 2055
     },
     {
       "epoch": 4.523652365236524,
-      "grad_norm": 0.09912109375,
+      "grad_norm": 0.10107421875,
       "learning_rate": 0.00011512950965926213,
-      "loss": 0.0015,
+      "loss": 0.0017,
       "step": 2056
     },
     {
       "epoch": 4.525852585258526,
-      "grad_norm": 0.1005859375,
+      "grad_norm": 0.318359375,
       "learning_rate": 0.00011506095351755717,
-      "loss": 0.0006,
+      "loss": 0.0033,
       "step": 2057
     },
     {
       "epoch": 4.528052805280528,
-      "grad_norm": 0.01495361328125,
+      "grad_norm": 0.111328125,
       "learning_rate": 0.00011499239013222997,
-      "loss": 0.0004,
+      "loss": 0.0011,
       "step": 2058
     },
     {
       "epoch": 4.53025302530253,
-      "grad_norm": 0.130859375,
+      "grad_norm": 0.060546875,
       "learning_rate": 0.0001149238195362564,
-      "loss": 0.0014,
+      "loss": 0.0007,
       "step": 2059
     },
     {
       "epoch": 4.5324532453245325,
-      "grad_norm": 0.01458740234375,
+      "grad_norm": 0.02783203125,
       "learning_rate": 0.0001148552417626157,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2060
     },
     {
       "epoch": 4.534653465346535,
-      "grad_norm": 0.03369140625,
+      "grad_norm": 0.054931640625,
       "learning_rate": 0.00011478665684429064,
-      "loss": 0.0007,
+      "loss": 0.001,
       "step": 2061
     },
     {
       "epoch": 4.536853685368537,
-      "grad_norm": 0.07958984375,
+      "grad_norm": 0.0150146484375,
       "learning_rate": 0.00011471806481426737,
-      "loss": 0.0006,
+      "loss": 0.0004,
       "step": 2062
     },
     {
       "epoch": 4.539053905390539,
-      "grad_norm": 0.0206298828125,
+      "grad_norm": 0.16015625,
       "learning_rate": 0.00011464946570553547,
-      "loss": 0.0003,
+      "loss": 0.0024,
       "step": 2063
     },
     {
       "epoch": 4.541254125412541,
-      "grad_norm": 0.17578125,
+      "grad_norm": 0.10400390625,
       "learning_rate": 0.000114580859551088,
-      "loss": 0.0023,
+      "loss": 0.0012,
       "step": 2064
     },
     {
       "epoch": 4.543454345434544,
-      "grad_norm": 0.01397705078125,
+      "grad_norm": 0.017333984375,
       "learning_rate": 0.00011451224638392129,
-      "loss": 0.0003,
+      "loss": 0.0004,
       "step": 2065
     },
     {
       "epoch": 4.5456545654565454,
-      "grad_norm": 0.0186767578125,
+      "grad_norm": 0.11328125,
       "learning_rate": 0.00011444362623703513,
-      "loss": 0.0003,
+      "loss": 0.0024,
       "step": 2066
     },
     {
       "epoch": 4.547854785478548,
-      "grad_norm": 0.29296875,
+      "grad_norm": 0.130859375,
       "learning_rate": 0.00011437499914343262,
-      "loss": 0.0014,
+      "loss": 0.0008,
       "step": 2067
     },
     {
       "epoch": 4.55005500550055,
-      "grad_norm": 0.205078125,
+      "grad_norm": 0.1630859375,
       "learning_rate": 0.00011430636513612023,
-      "loss": 0.0018,
+      "loss": 0.0023,
       "step": 2068
     },
     {
       "epoch": 4.552255225522552,
-      "grad_norm": 0.10302734375,
+      "grad_norm": 0.01055908203125,
       "learning_rate": 0.00011423772424810775,
-      "loss": 0.0006,
+      "loss": 0.0003,
       "step": 2069
     },
     {
       "epoch": 4.554455445544555,
-      "grad_norm": 0.08837890625,
+      "grad_norm": 0.09130859375,
       "learning_rate": 0.00011416907651240825,
-      "loss": 0.0009,
+      "loss": 0.0012,
       "step": 2070
     },
     {
       "epoch": 4.556655665566557,
-      "grad_norm": 0.055419921875,
+      "grad_norm": 0.049560546875,
       "learning_rate": 0.00011410042196203814,
-      "loss": 0.0017,
+      "loss": 0.0011,
       "step": 2071
     },
     {
       "epoch": 4.558855885588558,
-      "grad_norm": 0.130859375,
+      "grad_norm": 0.09326171875,
       "learning_rate": 0.00011403176063001708,
-      "loss": 0.0018,
+      "loss": 0.0015,
       "step": 2072
     },
     {
       "epoch": 4.561056105610561,
-      "grad_norm": 0.057861328125,
+      "grad_norm": 0.05712890625,
       "learning_rate": 0.00011396309254936798,
-      "loss": 0.0008,
+      "loss": 0.0004,
       "step": 2073
     },
     {
       "epoch": 4.563256325632564,
-      "grad_norm": 0.044189453125,
+      "grad_norm": 0.078125,
       "learning_rate": 0.00011389441775311704,
-      "loss": 0.0011,
+      "loss": 0.0012,
       "step": 2074
     },
     {
       "epoch": 4.565456545654565,
-      "grad_norm": 0.0218505859375,
+      "grad_norm": 0.12060546875,
       "learning_rate": 0.00011382573627429362,
-      "loss": 0.0003,
+      "loss": 0.0009,
       "step": 2075
     },
     {
       "epoch": 4.567656765676568,
-      "grad_norm": 0.0400390625,
+      "grad_norm": 0.036376953125,
       "learning_rate": 0.00011375704814593037,
-      "loss": 0.0004,
+      "loss": 0.0007,
       "step": 2076
     },
     {
       "epoch": 4.56985698569857,
-      "grad_norm": 0.039794921875,
+      "grad_norm": 0.0235595703125,
       "learning_rate": 0.00011368835340106309,
-      "loss": 0.0006,
+      "loss": 0.0005,
       "step": 2077
     },
     {
       "epoch": 4.572057205720572,
-      "grad_norm": 0.031494140625,
+      "grad_norm": 0.1630859375,
       "learning_rate": 0.0001136196520727308,
-      "loss": 0.0004,
+      "loss": 0.0012,
       "step": 2078
     },
     {
       "epoch": 4.574257425742574,
-      "grad_norm": 0.0419921875,
+      "grad_norm": 0.11376953125,
       "learning_rate": 0.00011355094419397563,
-      "loss": 0.0004,
+      "loss": 0.0009,
       "step": 2079
     },
     {
       "epoch": 4.5764576457645765,
-      "grad_norm": 0.050048828125,
+      "grad_norm": 0.10302734375,
       "learning_rate": 0.00011348222979784289,
-      "loss": 0.0005,
+      "loss": 0.0014,
       "step": 2080
     },
     {
       "epoch": 4.578657865786578,
-      "grad_norm": 0.08837890625,
+      "grad_norm": 0.08935546875,
       "learning_rate": 0.00011341350891738105,
-      "loss": 0.0014,
+      "loss": 0.0018,
       "step": 2081
     },
     {
       "epoch": 4.580858085808581,
-      "grad_norm": 0.0732421875,
+      "grad_norm": 0.1689453125,
       "learning_rate": 0.00011334478158564168,
-      "loss": 0.0006,
+      "loss": 0.0012,
       "step": 2082
     },
     {
       "epoch": 4.583058305830583,
-      "grad_norm": 0.1552734375,
+      "grad_norm": 0.1826171875,
       "learning_rate": 0.00011327604783567945,
-      "loss": 0.0018,
+      "loss": 0.0041,
       "step": 2083
     },
     {
       "epoch": 4.585258525852585,
-      "grad_norm": 0.04296875,
+      "grad_norm": 0.1484375,
       "learning_rate": 0.00011320730770055204,
-      "loss": 0.0004,
+      "loss": 0.0027,
       "step": 2084
     },
     {
       "epoch": 4.587458745874588,
-      "grad_norm": 0.00860595703125,
+      "grad_norm": 0.0234375,
       "learning_rate": 0.00011313856121332036,
-      "loss": 0.0002,
+      "loss": 0.0004,
       "step": 2085
     },
     {
       "epoch": 4.589658965896589,
-      "grad_norm": 0.142578125,
+      "grad_norm": 0.00714111328125,
       "learning_rate": 0.00011306980840704828,
-      "loss": 0.0007,
+      "loss": 0.0002,
       "step": 2086
     },
     {
       "epoch": 4.591859185918592,
-      "grad_norm": 0.08984375,
+      "grad_norm": 0.06494140625,
       "learning_rate": 0.00011300104931480271,
-      "loss": 0.0081,
+      "loss": 0.001,
       "step": 2087
     },
     {
       "epoch": 4.594059405940594,
-      "grad_norm": 0.0301513671875,
+      "grad_norm": 0.1318359375,
       "learning_rate": 0.00011293228396965361,
-      "loss": 0.0004,
+      "loss": 0.0007,
       "step": 2088
     },
     {
       "epoch": 4.596259625962596,
-      "grad_norm": 0.0196533203125,
+      "grad_norm": 0.193359375,
       "learning_rate": 0.00011286351240467387,
-      "loss": 0.0006,
+      "loss": 0.0029,
       "step": 2089
     },
     {
       "epoch": 4.598459845984598,
-      "grad_norm": 0.0184326171875,
+      "grad_norm": 0.0172119140625,
       "learning_rate": 0.00011279473465293952,
-      "loss": 0.0003,
+      "loss": 0.0004,
       "step": 2090
     },
     {
       "epoch": 4.600660066006601,
-      "grad_norm": 0.130859375,
+      "grad_norm": 0.02685546875,
       "learning_rate": 0.00011272595074752941,
-      "loss": 0.0006,
+      "loss": 0.0004,
       "step": 2091
     },
     {
       "epoch": 4.602860286028603,
-      "grad_norm": 0.06201171875,
+      "grad_norm": 0.025390625,
       "learning_rate": 0.00011265716072152546,
-      "loss": 0.0007,
+      "loss": 0.0005,
       "step": 2092
     },
     {
       "epoch": 4.605060506050605,
-      "grad_norm": 0.103515625,
+      "grad_norm": 0.07568359375,
       "learning_rate": 0.00011258836460801246,
-      "loss": 0.0006,
+      "loss": 0.0008,
       "step": 2093
     },
     {
       "epoch": 4.6072607260726075,
-      "grad_norm": 0.02197265625,
+      "grad_norm": 0.0947265625,
       "learning_rate": 0.00011251956244007819,
-      "loss": 0.0003,
+      "loss": 0.0011,
       "step": 2094
     },
     {
       "epoch": 4.609460946094609,
-      "grad_norm": 0.01324462890625,
+      "grad_norm": 0.048095703125,
       "learning_rate": 0.00011245075425081328,
-      "loss": 0.0003,
+      "loss": 0.0009,
       "step": 2095
     },
     {
       "epoch": 4.611661166116612,
-      "grad_norm": 0.0203857421875,
+      "grad_norm": 0.034423828125,
       "learning_rate": 0.0001123819400733113,
-      "loss": 0.0004,
+      "loss": 0.0006,
       "step": 2096
     },
     {
       "epoch": 4.6138613861386135,
-      "grad_norm": 0.00836181640625,
+      "grad_norm": 0.05712890625,
       "learning_rate": 0.0001123131199406687,
-      "loss": 0.0002,
+      "loss": 0.0006,
       "step": 2097
     },
     {
       "epoch": 4.616061606160616,
-      "grad_norm": 0.0096435546875,
+      "grad_norm": 0.06201171875,
       "learning_rate": 0.00011224429388598478,
-      "loss": 0.0003,
+      "loss": 0.0013,
       "step": 2098
     },
     {
       "epoch": 4.618261826182618,
-      "grad_norm": 0.0263671875,
+      "grad_norm": 0.0830078125,
       "learning_rate": 0.0001121754619423617,
-      "loss": 0.0004,
+      "loss": 0.0014,
       "step": 2099
     },
     {
       "epoch": 4.62046204620462,
-      "grad_norm": 0.09814453125,
+      "grad_norm": 0.041748046875,
       "learning_rate": 0.00011210662414290438,
-      "loss": 0.0008,
+      "loss": 0.0006,
       "step": 2100
     },
     {
       "epoch": 4.622662266226623,
-      "grad_norm": 0.008056640625,
+      "grad_norm": 0.02880859375,
       "learning_rate": 0.00011203778052072067,
-      "loss": 0.0002,
+      "loss": 0.0006,
       "step": 2101
     },
     {
       "epoch": 4.624862486248625,
-      "grad_norm": 0.0159912109375,
+      "grad_norm": 0.03271484375,
       "learning_rate": 0.00011196893110892118,
-      "loss": 0.0004,
+      "loss": 0.0005,
       "step": 2102
     },
     {
       "epoch": 4.627062706270627,
-      "grad_norm": 0.003875732421875,
+      "grad_norm": 0.01416015625,
       "learning_rate": 0.00011190007594061927,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2103
     },
     {
       "epoch": 4.629262926292629,
-      "grad_norm": 0.024169921875,
+      "grad_norm": 0.0419921875,
       "learning_rate": 0.00011183121504893108,
-      "loss": 0.0003,
+      "loss": 0.0004,
       "step": 2104
     },
     {
       "epoch": 4.631463146314632,
-      "grad_norm": 0.0247802734375,
+      "grad_norm": 0.03271484375,
       "learning_rate": 0.00011176234846697558,
-      "loss": 0.0003,
+      "loss": 0.0004,
       "step": 2105
     },
     {
       "epoch": 4.633663366336633,
-      "grad_norm": 0.27734375,
+      "grad_norm": 0.09130859375,
       "learning_rate": 0.00011169347622787429,
-      "loss": 0.0015,
+      "loss": 0.0004,
       "step": 2106
     },
     {
       "epoch": 4.635863586358636,
-      "grad_norm": 0.046875,
+      "grad_norm": 0.06689453125,
       "learning_rate": 0.0001116245983647517,
-      "loss": 0.0013,
+      "loss": 0.0017,
       "step": 2107
     },
     {
       "epoch": 4.638063806380638,
-      "grad_norm": 0.055908203125,
+      "grad_norm": 0.07958984375,
       "learning_rate": 0.0001115557149107348,
-      "loss": 0.0005,
+      "loss": 0.0018,
       "step": 2108
     },
     {
       "epoch": 4.64026402640264,
-      "grad_norm": 0.25,
+      "grad_norm": 0.06298828125,
       "learning_rate": 0.00011148682589895339,
-      "loss": 0.0027,
+      "loss": 0.0014,
       "step": 2109
     },
     {
       "epoch": 4.642464246424643,
-      "grad_norm": 0.054443359375,
+      "grad_norm": 0.09130859375,
       "learning_rate": 0.00011141793136253986,
-      "loss": 0.0008,
+      "loss": 0.001,
       "step": 2110
     },
     {
       "epoch": 4.6446644664466445,
-      "grad_norm": 0.0230712890625,
+      "grad_norm": 0.0576171875,
       "learning_rate": 0.0001113490313346293,
-      "loss": 0.0005,
+      "loss": 0.0009,
       "step": 2111
     },
     {
       "epoch": 4.646864686468647,
-      "grad_norm": 0.01312255859375,
+      "grad_norm": 0.1552734375,
       "learning_rate": 0.00011128012584835948,
-      "loss": 0.0002,
+      "loss": 0.0011,
       "step": 2112
     },
     {
       "epoch": 4.649064906490649,
-      "grad_norm": 0.1376953125,
+      "grad_norm": 0.162109375,
       "learning_rate": 0.0001112112149368707,
-      "loss": 0.0009,
+      "loss": 0.0036,
       "step": 2113
     },
     {
       "epoch": 4.6512651265126514,
-      "grad_norm": 0.015869140625,
+      "grad_norm": 0.0123291015625,
       "learning_rate": 0.00011114229863330596,
       "loss": 0.0002,
       "step": 2114
     },
     {
       "epoch": 4.653465346534653,
-      "grad_norm": 0.034912109375,
+      "grad_norm": 0.058349609375,
       "learning_rate": 0.0001110733769708108,
-      "loss": 0.0004,
+      "loss": 0.001,
       "step": 2115
     },
     {
       "epoch": 4.655665566556656,
-      "grad_norm": 0.083984375,
+      "grad_norm": 0.07470703125,
       "learning_rate": 0.00011100444998253338,
-      "loss": 0.0019,
+      "loss": 0.0007,
       "step": 2116
     },
     {
       "epoch": 4.6578657865786575,
-      "grad_norm": 0.0111083984375,
+      "grad_norm": 0.0177001953125,
       "learning_rate": 0.00011093551770162431,
       "loss": 0.0003,
       "step": 2117
     },
     {
       "epoch": 4.66006600660066,
-      "grad_norm": 0.047119140625,
+      "grad_norm": 0.431640625,
       "learning_rate": 0.00011086658016123693,
-      "loss": 0.0005,
+      "loss": 0.0017,
       "step": 2118
     },
     {
       "epoch": 4.662266226622663,
-      "grad_norm": 0.05712890625,
+      "grad_norm": 0.0927734375,
       "learning_rate": 0.00011079763739452696,
-      "loss": 0.0008,
+      "loss": 0.0015,
       "step": 2119
     },
     {
       "epoch": 4.664466446644664,
-      "grad_norm": 0.0078125,
+      "grad_norm": 0.052978515625,
       "learning_rate": 0.00011072868943465269,
-      "loss": 0.0002,
+      "loss": 0.0007,
       "step": 2120
     },
     {
       "epoch": 4.666666666666667,
-      "grad_norm": 0.09033203125,
+      "grad_norm": 0.01611328125,
       "learning_rate": 0.00011065973631477492,
-      "loss": 0.001,
+      "loss": 0.0003,
       "step": 2121
     },
     {
       "epoch": 4.668866886688669,
-      "grad_norm": 0.09814453125,
+      "grad_norm": 0.0279541015625,
       "learning_rate": 0.00011059077806805684,
-      "loss": 0.0007,
+      "loss": 0.0004,
       "step": 2122
     },
     {
       "epoch": 4.671067106710671,
-      "grad_norm": 0.0517578125,
+      "grad_norm": 0.2294921875,
       "learning_rate": 0.00011052181472766427,
-      "loss": 0.0004,
+      "loss": 0.0043,
       "step": 2123
     },
     {
       "epoch": 4.673267326732673,
-      "grad_norm": 0.00836181640625,
+      "grad_norm": 0.1875,
       "learning_rate": 0.00011045284632676536,
-      "loss": 0.0002,
+      "loss": 0.0018,
       "step": 2124
     },
     {
       "epoch": 4.675467546754676,
-      "grad_norm": 0.099609375,
+      "grad_norm": 0.130859375,
       "learning_rate": 0.0001103838728985307,
-      "loss": 0.0011,
+      "loss": 0.0021,
       "step": 2125
     },
     {
       "epoch": 4.677667766776677,
-      "grad_norm": 0.037841796875,
+      "grad_norm": 0.055908203125,
       "learning_rate": 0.00011031489447613333,
       "loss": 0.0004,
       "step": 2126
     },
     {
       "epoch": 4.67986798679868,
-      "grad_norm": 0.045166015625,
+      "grad_norm": 0.0177001953125,
       "learning_rate": 0.00011024591109274866,
-      "loss": 0.0005,
+      "loss": 0.0004,
       "step": 2127
     },
     {
       "epoch": 4.6820682068206825,
-      "grad_norm": 0.012939453125,
+      "grad_norm": 0.045166015625,
       "learning_rate": 0.00011017692278155453,
-      "loss": 0.0003,
+      "loss": 0.0007,
       "step": 2128
     },
     {
       "epoch": 4.684268426842684,
-      "grad_norm": 0.0908203125,
+      "grad_norm": 0.036865234375,
       "learning_rate": 0.00011010792957573115,
-      "loss": 0.0007,
+      "loss": 0.0004,
       "step": 2129
     },
     {
       "epoch": 4.686468646864687,
-      "grad_norm": 0.203125,
+      "grad_norm": 0.034423828125,
       "learning_rate": 0.00011003893150846103,
-      "loss": 0.0013,
+      "loss": 0.0008,
       "step": 2130
     },
     {
       "epoch": 4.6886688668866885,
-      "grad_norm": 0.006866455078125,
+      "grad_norm": 0.09521484375,
       "learning_rate": 0.00010996992861292906,
-      "loss": 0.0002,
+      "loss": 0.0005,
       "step": 2131
     },
     {
       "epoch": 4.690869086908691,
-      "grad_norm": 0.2412109375,
+      "grad_norm": 0.08837890625,
       "learning_rate": 0.00010990092092232246,
-      "loss": 0.0008,
+      "loss": 0.0006,
       "step": 2132
     },
     {
       "epoch": 4.693069306930693,
-      "grad_norm": 0.13671875,
+      "grad_norm": 0.0189208984375,
       "learning_rate": 0.0001098319084698307,
-      "loss": 0.0019,
+      "loss": 0.0003,
       "step": 2133
     },
     {
       "epoch": 4.695269526952695,
-      "grad_norm": 0.24609375,
+      "grad_norm": 0.07080078125,
       "learning_rate": 0.00010976289128864556,
-      "loss": 0.0048,
+      "loss": 0.0011,
       "step": 2134
     },
     {
       "epoch": 4.697469746974697,
-      "grad_norm": 0.01190185546875,
+      "grad_norm": 0.038818359375,
       "learning_rate": 0.0001096938694119612,
-      "loss": 0.0003,
+      "loss": 0.0006,
       "step": 2135
     },
     {
       "epoch": 4.6996699669967,
-      "grad_norm": 0.00714111328125,
+      "grad_norm": 0.021484375,
       "learning_rate": 0.00010962484287297388,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2136
     },
     {
       "epoch": 4.701870187018702,
-      "grad_norm": 0.048828125,
+      "grad_norm": 0.0458984375,
       "learning_rate": 0.00010955581170488223,
-      "loss": 0.0006,
+      "loss": 0.0007,
       "step": 2137
     },
     {
       "epoch": 4.704070407040704,
-      "grad_norm": 0.0201416015625,
+      "grad_norm": 0.04833984375,
       "learning_rate": 0.00010948677594088698,
-      "loss": 0.0004,
+      "loss": 0.0008,
       "step": 2138
     },
     {
       "epoch": 4.706270627062707,
-      "grad_norm": 0.00872802734375,
+      "grad_norm": 0.00921630859375,
       "learning_rate": 0.00010941773561419117,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2139
     },
     {
       "epoch": 4.708470847084708,
-      "grad_norm": 0.076171875,
+      "grad_norm": 0.006439208984375,
       "learning_rate": 0.000109348690758,
-      "loss": 0.001,
+      "loss": 0.0002,
       "step": 2140
     },
     {
       "epoch": 4.710671067106711,
-      "grad_norm": 0.08935546875,
+      "grad_norm": 0.01220703125,
       "learning_rate": 0.0001092796414055209,
-      "loss": 0.0005,
+      "loss": 0.0003,
       "step": 2141
     },
     {
       "epoch": 4.712871287128713,
-      "grad_norm": 0.0361328125,
+      "grad_norm": 0.173828125,
       "learning_rate": 0.00010921058758996336,
-      "loss": 0.0006,
+      "loss": 0.002,
       "step": 2142
     },
     {
       "epoch": 4.715071507150715,
-      "grad_norm": 0.0625,
+      "grad_norm": 0.1328125,
       "learning_rate": 0.00010914152934453911,
-      "loss": 0.001,
+      "loss": 0.0015,
       "step": 2143
     },
     {
       "epoch": 4.717271727172717,
-      "grad_norm": 0.02587890625,
+      "grad_norm": 0.01287841796875,
       "learning_rate": 0.00010907246670246194,
-      "loss": 0.0004,
+      "loss": 0.0002,
       "step": 2144
     },
     {
       "epoch": 4.7194719471947195,
-      "grad_norm": 0.02490234375,
+      "grad_norm": 0.00994873046875,
       "learning_rate": 0.00010900339969694777,
       "loss": 0.0003,
       "step": 2145
     },
     {
       "epoch": 4.721672167216722,
-      "grad_norm": 0.016357421875,
+      "grad_norm": 0.154296875,
       "learning_rate": 0.0001089343283612147,
-      "loss": 0.0004,
+      "loss": 0.0018,
       "step": 2146
     },
     {
       "epoch": 4.723872387238724,
-      "grad_norm": 0.00762939453125,
+      "grad_norm": 0.2080078125,
       "learning_rate": 0.0001088652527284828,
-      "loss": 0.0002,
+      "loss": 0.0027,
       "step": 2147
     },
     {
       "epoch": 4.726072607260726,
-      "grad_norm": 0.00933837890625,
+      "grad_norm": 0.00738525390625,
       "learning_rate": 0.0001087961728319743,
       "loss": 0.0002,
       "step": 2148
     },
     {
       "epoch": 4.728272827282728,
-      "grad_norm": 0.064453125,
+      "grad_norm": 0.00970458984375,
       "learning_rate": 0.00010872708870491337,
-      "loss": 0.0004,
+      "loss": 0.0002,
       "step": 2149
     },
     {
       "epoch": 4.730473047304731,
-      "grad_norm": 0.04541015625,
+      "grad_norm": 0.0233154296875,
       "learning_rate": 0.00010865800038052631,
-      "loss": 0.0004,
+      "loss": 0.0002,
       "step": 2150
     },
     {
       "epoch": 4.732673267326732,
-      "grad_norm": 0.03466796875,
+      "grad_norm": 0.07275390625,
       "learning_rate": 0.00010858890789204144,
-      "loss": 0.0003,
+      "loss": 0.0008,
       "step": 2151
     },
     {
       "epoch": 4.734873487348735,
-      "grad_norm": 0.232421875,
+      "grad_norm": 0.00494384765625,
       "learning_rate": 0.00010851981127268902,
-      "loss": 0.0053,
+      "loss": 0.0002,
       "step": 2152
     },
     {
       "epoch": 4.737073707370737,
-      "grad_norm": 0.1669921875,
+      "grad_norm": 0.0888671875,
       "learning_rate": 0.00010845071055570137,
-      "loss": 0.0018,
+      "loss": 0.001,
       "step": 2153
     },
     {
       "epoch": 4.739273927392739,
-      "grad_norm": 0.049560546875,
+      "grad_norm": 0.01239013671875,
       "learning_rate": 0.00010838160577431269,
-      "loss": 0.0006,
+      "loss": 0.0003,
       "step": 2154
     },
     {
       "epoch": 4.741474147414742,
-      "grad_norm": 0.0257568359375,
+      "grad_norm": 0.0296630859375,
       "learning_rate": 0.00010831249696175918,
       "loss": 0.0005,
       "step": 2155
     },
     {
       "epoch": 4.743674367436744,
-      "grad_norm": 0.0439453125,
+      "grad_norm": 0.0086669921875,
       "learning_rate": 0.000108243384151279,
       "loss": 0.0003,
       "step": 2156
     },
     {
       "epoch": 4.745874587458746,
-      "grad_norm": 0.017578125,
+      "grad_norm": 0.0390625,
       "learning_rate": 0.00010817426737611223,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2157
     },
     {
       "epoch": 4.748074807480748,
-      "grad_norm": 0.01531982421875,
+      "grad_norm": 0.091796875,
       "learning_rate": 0.00010810514666950084,
-      "loss": 0.0003,
+      "loss": 0.0005,
       "step": 2158
     },
     {
       "epoch": 4.7502750275027505,
-      "grad_norm": 0.2294921875,
+      "grad_norm": 0.050537109375,
       "learning_rate": 0.0001080360220646887,
-      "loss": 0.0013,
+      "loss": 0.0007,
       "step": 2159
     },
     {
       "epoch": 4.752475247524752,
-      "grad_norm": 0.0291748046875,
+      "grad_norm": 0.00885009765625,
       "learning_rate": 0.00010796689359492153,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2160
     },
     {
       "epoch": 4.754675467546755,
-      "grad_norm": 0.0032196044921875,
+      "grad_norm": 0.00494384765625,
       "learning_rate": 0.0001078977612934469,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2161
     },
     {
       "epoch": 4.756875687568757,
-      "grad_norm": 0.016357421875,
+      "grad_norm": 0.25390625,
       "learning_rate": 0.00010782862519351431,
-      "loss": 0.0002,
+      "loss": 0.0018,
       "step": 2162
     },
     {
       "epoch": 4.759075907590759,
-      "grad_norm": 0.00946044921875,
+      "grad_norm": 0.00579833984375,
       "learning_rate": 0.00010775948532837499,
       "loss": 0.0002,
       "step": 2163
     },
     {
       "epoch": 4.761276127612762,
-      "grad_norm": 0.003204345703125,
+      "grad_norm": 0.06005859375,
       "learning_rate": 0.00010769034173128207,
-      "loss": 0.0001,
+      "loss": 0.0004,
       "step": 2164
     },
     {
       "epoch": 4.7634763476347635,
-      "grad_norm": 0.08837890625,
+      "grad_norm": 0.1171875,
       "learning_rate": 0.00010762119443549035,
-      "loss": 0.0005,
+      "loss": 0.0007,
       "step": 2165
     },
     {
       "epoch": 4.765676567656766,
-      "grad_norm": 0.02294921875,
+      "grad_norm": 0.037353515625,
       "learning_rate": 0.00010755204347425652,
-      "loss": 0.0003,
+      "loss": 0.0007,
       "step": 2166
     },
     {
       "epoch": 4.765676567656766,
-      "eval_loss": 0.0013566253473982215,
-      "eval_runtime": 11.2021,
-      "eval_samples_per_second": 34.19,
-      "eval_steps_per_second": 4.285,
+      "eval_loss": 0.001631063874810934,
+      "eval_runtime": 10.3171,
+      "eval_samples_per_second": 37.123,
+      "eval_steps_per_second": 4.652,
       "step": 2166
     },
     {
       "epoch": 4.767876787678768,
-      "grad_norm": 0.048583984375,
+      "grad_norm": 0.037841796875,
       "learning_rate": 0.00010748288888083895,
-      "loss": 0.0005,
+      "loss": 0.0004,
       "step": 2167
     },
     {
       "epoch": 4.77007700770077,
-      "grad_norm": 0.006500244140625,
+      "grad_norm": 0.004425048828125,
       "learning_rate": 0.00010741373068849788,
       "loss": 0.0002,
       "step": 2168
     },
     {
       "epoch": 4.772277227722772,
-      "grad_norm": 0.009521484375,
+      "grad_norm": 0.060546875,
       "learning_rate": 0.00010734456893049514,
-      "loss": 0.0003,
+      "loss": 0.001,
       "step": 2169
     },
     {
       "epoch": 4.774477447744775,
-      "grad_norm": 0.01025390625,
+      "grad_norm": 0.00823974609375,
       "learning_rate": 0.0001072754036400944,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2170
     },
     {
       "epoch": 4.776677667766776,
-      "grad_norm": 0.01263427734375,
+      "grad_norm": 0.017578125,
       "learning_rate": 0.00010720623485056089,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2171
     },
     {
       "epoch": 4.778877887788779,
-      "grad_norm": 0.0296630859375,
+      "grad_norm": 0.02099609375,
       "learning_rate": 0.00010713706259516165,
-      "loss": 0.0003,
+      "loss": 0.0004,
       "step": 2172
     },
     {
       "epoch": 4.781078107810782,
-      "grad_norm": 0.02587890625,
+      "grad_norm": 0.006195068359375,
       "learning_rate": 0.00010706788690716529,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2173
     },
     {
       "epoch": 4.783278327832783,
-      "grad_norm": 0.1337890625,
+      "grad_norm": 0.0869140625,
       "learning_rate": 0.00010699870781984218,
-      "loss": 0.0019,
+      "loss": 0.0016,
       "step": 2174
     },
     {
       "epoch": 4.785478547854785,
-      "grad_norm": 0.158203125,
+      "grad_norm": 0.087890625,
       "learning_rate": 0.00010692952536646426,
-      "loss": 0.0015,
+      "loss": 0.001,
       "step": 2175
     },
     {
       "epoch": 4.787678767876788,
-      "grad_norm": 0.162109375,
+      "grad_norm": 0.00811767578125,
       "learning_rate": 0.00010686033958030504,
-      "loss": 0.0014,
+      "loss": 0.0002,
       "step": 2176
     },
     {
       "epoch": 4.78987898789879,
-      "grad_norm": 0.091796875,
+      "grad_norm": 0.0869140625,
       "learning_rate": 0.0001067911504946397,
-      "loss": 0.0004,
+      "loss": 0.001,
       "step": 2177
     },
     {
       "epoch": 4.792079207920792,
-      "grad_norm": 0.365234375,
+      "grad_norm": 0.003997802734375,
       "learning_rate": 0.00010672195814274496,
-      "loss": 0.0021,
+      "loss": 0.0001,
       "step": 2178
     },
     {
       "epoch": 4.7942794279427945,
-      "grad_norm": 0.00762939453125,
+      "grad_norm": 0.0123291015625,
       "learning_rate": 0.00010665276255789923,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2179
     },
     {
       "epoch": 4.796479647964796,
-      "grad_norm": 0.028076171875,
+      "grad_norm": 0.09619140625,
       "learning_rate": 0.00010658356377338234,
-      "loss": 0.0003,
+      "loss": 0.0011,
       "step": 2180
     },
     {
       "epoch": 4.798679867986799,
-      "grad_norm": 0.1708984375,
+      "grad_norm": 0.058349609375,
       "learning_rate": 0.0001065143618224757,
       "loss": 0.001,
       "step": 2181
     },
     {
       "epoch": 4.8008800880088005,
-      "grad_norm": 0.30859375,
+      "grad_norm": 0.007659912109375,
       "learning_rate": 0.00010644515673846223,
-      "loss": 0.0011,
+      "loss": 0.0002,
       "step": 2182
     },
     {
       "epoch": 4.803080308030803,
-      "grad_norm": 0.078125,
+      "grad_norm": 0.005584716796875,
       "learning_rate": 0.00010637594855462638,
-      "loss": 0.0007,
+      "loss": 0.0002,
       "step": 2183
     },
     {
       "epoch": 4.805280528052805,
-      "grad_norm": 0.29296875,
+      "grad_norm": 0.0211181640625,
       "learning_rate": 0.00010630673730425412,
-      "loss": 0.0029,
+      "loss": 0.0003,
       "step": 2184
     },
     {
       "epoch": 4.807480748074807,
-      "grad_norm": 0.01409912109375,
+      "grad_norm": 0.041259765625,
       "learning_rate": 0.00010623752302063283,
-      "loss": 0.0002,
+      "loss": 0.0004,
       "step": 2185
     },
     {
       "epoch": 4.80968096809681,
-      "grad_norm": 0.0037994384765625,
+      "grad_norm": 0.064453125,
       "learning_rate": 0.00010616830573705143,
-      "loss": 0.0002,
+      "loss": 0.0009,
       "step": 2186
     },
     {
       "epoch": 4.811881188118812,
-      "grad_norm": 0.00384521484375,
+      "grad_norm": 0.004608154296875,
       "learning_rate": 0.00010609908548680016,
       "loss": 0.0001,
       "step": 2187
     },
     {
       "epoch": 4.814081408140814,
-      "grad_norm": 0.263671875,
+      "grad_norm": 0.059326171875,
       "learning_rate": 0.00010602986230317083,
-      "loss": 0.0034,
+      "loss": 0.001,
       "step": 2188
     },
     {
       "epoch": 4.816281628162816,
-      "grad_norm": 0.009765625,
+      "grad_norm": 0.0089111328125,
       "learning_rate": 0.0001059606362194565,
       "loss": 0.0002,
       "step": 2189
     },
     {
       "epoch": 4.818481848184819,
-      "grad_norm": 0.0225830078125,
+      "grad_norm": 0.0213623046875,
       "learning_rate": 0.00010589140726895179,
       "loss": 0.0003,
       "step": 2190
     },
     {
       "epoch": 4.82068206820682,
-      "grad_norm": 0.0076904296875,
+      "grad_norm": 0.12255859375,
       "learning_rate": 0.00010582217548495265,
-      "loss": 0.0002,
+      "loss": 0.0011,
       "step": 2191
     },
     {
       "epoch": 4.822882288228823,
-      "grad_norm": 0.047607421875,
+      "grad_norm": 0.01708984375,
       "learning_rate": 0.00010575294090075629,
-      "loss": 0.0004,
+      "loss": 0.0003,
       "step": 2192
     },
     {
       "epoch": 4.825082508250825,
-      "grad_norm": 0.197265625,
+      "grad_norm": 0.00567626953125,
       "learning_rate": 0.00010568370354966137,
-      "loss": 0.0009,
+      "loss": 0.0002,
       "step": 2193
     },
     {
       "epoch": 4.827282728272827,
-      "grad_norm": 0.015380859375,
+      "grad_norm": 0.1923828125,
       "learning_rate": 0.00010561446346496786,
-      "loss": 0.0003,
+      "loss": 0.0029,
       "step": 2194
     },
     {
       "epoch": 4.82948294829483,
-      "grad_norm": 0.0888671875,
+      "grad_norm": 0.022216796875,
       "learning_rate": 0.00010554522067997703,
-      "loss": 0.002,
+      "loss": 0.0006,
       "step": 2195
     },
     {
       "epoch": 4.8316831683168315,
-      "grad_norm": 0.240234375,
+      "grad_norm": 0.013671875,
       "learning_rate": 0.00010547597522799146,
-      "loss": 0.0019,
+      "loss": 0.0002,
       "step": 2196
     },
     {
       "epoch": 4.833883388338834,
-      "grad_norm": 0.035888671875,
+      "grad_norm": 0.08251953125,
       "learning_rate": 0.00010540672714231507,
       "loss": 0.0005,
       "step": 2197
     },
     {
       "epoch": 4.836083608360836,
-      "grad_norm": 0.265625,
+      "grad_norm": 0.2158203125,
       "learning_rate": 0.0001053374764562529,
-      "loss": 0.0022,
+      "loss": 0.0014,
       "step": 2198
     },
     {
       "epoch": 4.838283828382838,
-      "grad_norm": 0.1044921875,
+      "grad_norm": 0.01055908203125,
       "learning_rate": 0.00010526822320311136,
-      "loss": 0.0019,
+      "loss": 0.0002,
       "step": 2199
     },
     {
       "epoch": 4.84048404840484,
-      "grad_norm": 0.0634765625,
+      "grad_norm": 0.005645751953125,
       "learning_rate": 0.00010519896741619803,
-      "loss": 0.0005,
+      "loss": 0.0002,
       "step": 2200
     },
     {
       "epoch": 4.842684268426843,
-      "grad_norm": 0.0130615234375,
+      "grad_norm": 0.1025390625,
       "learning_rate": 0.0001051297091288218,
-      "loss": 0.0003,
+      "loss": 0.0005,
       "step": 2201
     },
     {
       "epoch": 4.8448844884488445,
-      "grad_norm": 0.064453125,
+      "grad_norm": 0.142578125,
       "learning_rate": 0.00010506044837429272,
-      "loss": 0.0006,
+      "loss": 0.001,
       "step": 2202
     },
     {
       "epoch": 4.847084708470847,
-      "grad_norm": 0.09423828125,
+      "grad_norm": 0.030517578125,
       "learning_rate": 0.00010499118518592195,
-      "loss": 0.0012,
+      "loss": 0.0004,
       "step": 2203
     },
     {
       "epoch": 4.84928492849285,
-      "grad_norm": 0.00616455078125,
+      "grad_norm": 0.004486083984375,
       "learning_rate": 0.00010492191959702187,
       "loss": 0.0002,
       "step": 2204
     },
     {
       "epoch": 4.851485148514851,
-      "grad_norm": 0.06640625,
+      "grad_norm": 0.01953125,
       "learning_rate": 0.00010485265164090608,
-      "loss": 0.0007,
+      "loss": 0.0003,
       "step": 2205
     },
     {
       "epoch": 4.853685368536854,
-      "grad_norm": 0.0093994140625,
+      "grad_norm": 0.0101318359375,
       "learning_rate": 0.00010478338135088921,
       "loss": 0.0002,
       "step": 2206
     },
     {
       "epoch": 4.855885588558856,
-      "grad_norm": 0.0732421875,
+      "grad_norm": 0.0177001953125,
       "learning_rate": 0.00010471410876028713,
-      "loss": 0.0006,
+      "loss": 0.0003,
       "step": 2207
     },
     {
       "epoch": 4.858085808580858,
-      "grad_norm": 0.39453125,
+      "grad_norm": 0.0203857421875,
       "learning_rate": 0.0001046448339024167,
-      "loss": 0.0029,
+      "loss": 0.0003,
       "step": 2208
     },
     {
       "epoch": 4.86028602860286,
-      "grad_norm": 0.08642578125,
+      "grad_norm": 0.03076171875,
       "learning_rate": 0.00010457555681059597,
-      "loss": 0.0013,
+      "loss": 0.0005,
       "step": 2209
     },
     {
       "epoch": 4.862486248624863,
-      "grad_norm": 0.271484375,
+      "grad_norm": 0.00445556640625,
       "learning_rate": 0.00010450627751814394,
-      "loss": 0.0043,
+      "loss": 0.0002,
       "step": 2210
     },
     {
       "epoch": 4.864686468646864,
-      "grad_norm": 0.05517578125,
+      "grad_norm": 0.00775146484375,
       "learning_rate": 0.00010443699605838084,
-      "loss": 0.0004,
+      "loss": 0.0002,
       "step": 2211
     },
     {
       "epoch": 4.866886688668867,
-      "grad_norm": 0.0252685546875,
+      "grad_norm": 0.01336669921875,
       "learning_rate": 0.00010436771246462775,
-      "loss": 0.0004,
+      "loss": 0.0002,
       "step": 2212
     },
     {
       "epoch": 4.8690869086908695,
-      "grad_norm": 0.02783203125,
+      "grad_norm": 0.01300048828125,
       "learning_rate": 0.000104298426770207,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2213
     },
     {
       "epoch": 4.871287128712871,
-      "grad_norm": 0.0849609375,
+      "grad_norm": 0.033203125,
       "learning_rate": 0.00010422913900844169,
-      "loss": 0.0011,
+      "loss": 0.0004,
       "step": 2214
     },
     {
       "epoch": 4.873487348734874,
-      "grad_norm": 0.01904296875,
+      "grad_norm": 0.006103515625,
       "learning_rate": 0.00010415984921265609,
       "loss": 0.0002,
       "step": 2215
     },
     {
       "epoch": 4.8756875687568755,
-      "grad_norm": 0.006317138671875,
+      "grad_norm": 0.0079345703125,
       "learning_rate": 0.00010409055741617537,
       "loss": 0.0002,
       "step": 2216
     },
     {
       "epoch": 4.877887788778878,
-      "grad_norm": 0.0341796875,
+      "grad_norm": 0.06689453125,
       "learning_rate": 0.00010402126365232565,
-      "loss": 0.0003,
+      "loss": 0.0006,
       "step": 2217
     },
     {
       "epoch": 4.88008800880088,
-      "grad_norm": 0.00714111328125,
+      "grad_norm": 0.0091552734375,
       "learning_rate": 0.00010395196795443412,
       "loss": 0.0002,
       "step": 2218
     },
     {
       "epoch": 4.882288228822882,
-      "grad_norm": 0.04443359375,
+      "grad_norm": 0.0203857421875,
       "learning_rate": 0.0001038826703558287,
-      "loss": 0.0004,
+      "loss": 0.0002,
       "step": 2219
     },
     {
       "epoch": 4.884488448844884,
-      "grad_norm": 0.56640625,
+      "grad_norm": 0.0184326171875,
       "learning_rate": 0.00010381337088983838,
-      "loss": 0.0017,
+      "loss": 0.0003,
       "step": 2220
     },
     {
       "epoch": 4.886688668866887,
-      "grad_norm": 0.41015625,
+      "grad_norm": 0.01068115234375,
       "learning_rate": 0.00010374406958979301,
-      "loss": 0.0033,
+      "loss": 0.0002,
       "step": 2221
     },
     {
       "epoch": 4.888888888888889,
-      "grad_norm": 0.01373291015625,
+      "grad_norm": 0.0081787109375,
       "learning_rate": 0.00010367476648902328,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2222
     },
     {
       "epoch": 4.891089108910891,
-      "grad_norm": 0.0101318359375,
+      "grad_norm": 0.010498046875,
       "learning_rate": 0.0001036054616208608,
       "loss": 0.0002,
       "step": 2223
     },
     {
       "epoch": 4.893289328932894,
-      "grad_norm": 0.053955078125,
+      "grad_norm": 0.01043701171875,
       "learning_rate": 0.00010353615501863799,
-      "loss": 0.0005,
+      "loss": 0.0003,
       "step": 2224
     },
     {
       "epoch": 4.895489548954895,
-      "grad_norm": 0.0125732421875,
+      "grad_norm": 0.060302734375,
       "learning_rate": 0.00010346684671568814,
-      "loss": 0.0002,
+      "loss": 0.0004,
       "step": 2225
     },
     {
       "epoch": 4.897689768976898,
-      "grad_norm": 0.018798828125,
+      "grad_norm": 0.150390625,
       "learning_rate": 0.00010339753674534531,
-      "loss": 0.0003,
+      "loss": 0.0014,
       "step": 2226
     },
     {
       "epoch": 4.8998899889989,
-      "grad_norm": 0.05029296875,
+      "grad_norm": 0.0281982421875,
       "learning_rate": 0.00010332822514094442,
-      "loss": 0.0005,
+      "loss": 0.0004,
       "step": 2227
     },
     {
       "epoch": 4.902090209020902,
-      "grad_norm": 0.01495361328125,
+      "grad_norm": 0.0308837890625,
       "learning_rate": 0.00010325891193582111,
       "loss": 0.0003,
       "step": 2228
     },
     {
       "epoch": 4.904290429042904,
-      "grad_norm": 0.00872802734375,
+      "grad_norm": 0.12890625,
       "learning_rate": 0.00010318959716331191,
-      "loss": 0.0003,
+      "loss": 0.0008,
       "step": 2229
     },
     {
       "epoch": 4.9064906490649065,
-      "grad_norm": 0.1904296875,
+      "grad_norm": 0.01434326171875,
       "learning_rate": 0.00010312028085675391,
-      "loss": 0.0021,
+      "loss": 0.0002,
       "step": 2230
     },
     {
       "epoch": 4.908690869086909,
-      "grad_norm": 0.0128173828125,
+      "grad_norm": 0.12109375,
       "learning_rate": 0.00010305096304948514,
-      "loss": 0.0002,
+      "loss": 0.0005,
       "step": 2231
     },
     {
       "epoch": 4.910891089108911,
-      "grad_norm": 0.0322265625,
+      "grad_norm": 0.25390625,
       "learning_rate": 0.0001029816437748442,
-      "loss": 0.0006,
+      "loss": 0.0049,
       "step": 2232
     },
     {
       "epoch": 4.913091309130913,
-      "grad_norm": 0.046875,
+      "grad_norm": 0.0234375,
       "learning_rate": 0.00010291232306617047,
-      "loss": 0.0006,
+      "loss": 0.0003,
       "step": 2233
     },
     {
       "epoch": 4.915291529152915,
-      "grad_norm": 0.0322265625,
+      "grad_norm": 0.044921875,
       "learning_rate": 0.00010284300095680403,
       "loss": 0.0005,
       "step": 2234
     },
     {
       "epoch": 4.917491749174918,
-      "grad_norm": 0.431640625,
+      "grad_norm": 0.044189453125,
       "learning_rate": 0.00010277367748008557,
-      "loss": 0.0044,
+      "loss": 0.0004,
       "step": 2235
     },
     {
       "epoch": 4.919691969196919,
-      "grad_norm": 0.059326171875,
+      "grad_norm": 0.158203125,
       "learning_rate": 0.00010270435266935651,
-      "loss": 0.0007,
+      "loss": 0.0008,
       "step": 2236
     },
     {
       "epoch": 4.921892189218922,
-      "grad_norm": 0.00970458984375,
+      "grad_norm": 0.0177001953125,
       "learning_rate": 0.00010263502655795887,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2237
     },
     {
       "epoch": 4.924092409240924,
-      "grad_norm": 0.0186767578125,
+      "grad_norm": 0.0120849609375,
       "learning_rate": 0.0001025656991792353,
-      "loss": 0.0004,
+      "loss": 0.0002,
       "step": 2238
     },
     {
       "epoch": 4.926292629262926,
-      "grad_norm": 0.0234375,
+      "grad_norm": 0.0113525390625,
       "learning_rate": 0.00010249637056652906,
-      "loss": 0.0005,
+      "loss": 0.0003,
       "step": 2239
     },
     {
       "epoch": 4.928492849284929,
-      "grad_norm": 0.169921875,
+      "grad_norm": 0.015625,
       "learning_rate": 0.00010242704075318402,
-      "loss": 0.0012,
+      "loss": 0.0002,
       "step": 2240
     },
     {
       "epoch": 4.930693069306931,
-      "grad_norm": 0.018798828125,
+      "grad_norm": 0.0247802734375,
       "learning_rate": 0.0001023577097725446,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2241
     },
     {
       "epoch": 4.932893289328933,
-      "grad_norm": 0.1728515625,
+      "grad_norm": 0.0159912109375,
       "learning_rate": 0.00010228837765795578,
-      "loss": 0.0025,
+      "loss": 0.0003,
       "step": 2242
     },
     {
       "epoch": 4.935093509350935,
-      "grad_norm": 0.00634765625,
+      "grad_norm": 0.06640625,
       "learning_rate": 0.00010221904444276315,
-      "loss": 0.0002,
+      "loss": 0.001,
       "step": 2243
     },
     {
       "epoch": 4.9372937293729375,
-      "grad_norm": 0.1650390625,
+      "grad_norm": 0.007415771484375,
       "learning_rate": 0.00010214971016031274,
-      "loss": 0.001,
+      "loss": 0.0002,
       "step": 2244
     },
     {
       "epoch": 4.939493949394939,
-      "grad_norm": 0.0281982421875,
+      "grad_norm": 0.025146484375,
       "learning_rate": 0.00010208037484395114,
-      "loss": 0.0005,
+      "loss": 0.0003,
       "step": 2245
     },
     {
       "epoch": 4.941694169416942,
-      "grad_norm": 0.028564453125,
+      "grad_norm": 0.029296875,
       "learning_rate": 0.00010201103852702545,
-      "loss": 0.0004,
+      "loss": 0.0005,
       "step": 2246
     },
     {
       "epoch": 4.9438943894389435,
-      "grad_norm": 0.0257568359375,
+      "grad_norm": 0.02197265625,
       "learning_rate": 0.00010194170124288322,
-      "loss": 0.0004,
+      "loss": 0.0003,
       "step": 2247
     },
     {
       "epoch": 4.946094609460946,
-      "grad_norm": 0.38671875,
+      "grad_norm": 0.01611328125,
       "learning_rate": 0.00010187236302487247,
-      "loss": 0.0019,
+      "loss": 0.0003,
       "step": 2248
     },
     {
       "epoch": 4.948294829482949,
-      "grad_norm": 0.1455078125,
+      "grad_norm": 0.0203857421875,
       "learning_rate": 0.00010180302390634168,
-      "loss": 0.0011,
+      "loss": 0.0004,
       "step": 2249
     },
     {
       "epoch": 4.9504950495049505,
-      "grad_norm": 0.11669921875,
+      "grad_norm": 0.09033203125,
       "learning_rate": 0.00010173368392063978,
-      "loss": 0.001,
+      "loss": 0.0008,
       "step": 2250
     },
     {
       "epoch": 4.952695269526953,
-      "grad_norm": 0.11474609375,
+      "grad_norm": 0.1630859375,
       "learning_rate": 0.00010166434310111608,
       "loss": 0.0014,
       "step": 2251
     },
     {
       "epoch": 4.954895489548955,
-      "grad_norm": 0.0162353515625,
+      "grad_norm": 0.005859375,
       "learning_rate": 0.00010159500148112029,
       "loss": 0.0002,
       "step": 2252
     },
     {
       "epoch": 4.957095709570957,
-      "grad_norm": 0.034423828125,
+      "grad_norm": 0.06494140625,
       "learning_rate": 0.00010152565909400256,
-      "loss": 0.0007,
+      "loss": 0.0021,
       "step": 2253
     },
     {
       "epoch": 4.959295929592959,
-      "grad_norm": 0.029296875,
+      "grad_norm": 0.01226806640625,
       "learning_rate": 0.00010145631597311334,
-      "loss": 0.0004,
+      "loss": 0.0002,
       "step": 2254
     },
     {
       "epoch": 4.961496149614962,
-      "grad_norm": 0.0274658203125,
+      "grad_norm": 0.10693359375,
       "learning_rate": 0.00010138697215180346,
-      "loss": 0.0003,
+      "loss": 0.001,
       "step": 2255
     },
     {
       "epoch": 4.963696369636963,
-      "grad_norm": 0.30859375,
+      "grad_norm": 0.06005859375,
       "learning_rate": 0.00010131762766342414,
-      "loss": 0.0023,
+      "loss": 0.0008,
       "step": 2256
     },
     {
       "epoch": 4.965896589658966,
-      "grad_norm": 0.07421875,
+      "grad_norm": 0.02734375,
       "learning_rate": 0.00010124828254132683,
-      "loss": 0.0006,
+      "loss": 0.0004,
       "step": 2257
     },
     {
       "epoch": 4.968096809680969,
-      "grad_norm": 0.0159912109375,
+      "grad_norm": 0.0050048828125,
       "learning_rate": 0.00010117893681886334,
-      "loss": 0.0003,
+      "loss": 0.0001,
       "step": 2258
     },
     {
       "epoch": 4.97029702970297,
-      "grad_norm": 0.055419921875,
+      "grad_norm": 0.322265625,
       "learning_rate": 0.00010110959052938575,
-      "loss": 0.0006,
+      "loss": 0.0013,
       "step": 2259
     },
     {
       "epoch": 4.972497249724973,
-      "grad_norm": 0.08447265625,
+      "grad_norm": 0.138671875,
       "learning_rate": 0.00010104024370624644,
-      "loss": 0.0003,
+      "loss": 0.0009,
       "step": 2260
     },
     {
       "epoch": 4.974697469746975,
-      "grad_norm": 0.2333984375,
+      "grad_norm": 0.01434326171875,
       "learning_rate": 0.00010097089638279798,
-      "loss": 0.0013,
+      "loss": 0.0003,
       "step": 2261
     },
     {
       "epoch": 4.976897689768977,
-      "grad_norm": 0.01904296875,
+      "grad_norm": 0.0185546875,
       "learning_rate": 0.00010090154859239328,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2262
     },
     {
       "epoch": 4.979097909790979,
-      "grad_norm": 0.00714111328125,
+      "grad_norm": 0.1328125,
       "learning_rate": 0.00010083220036838539,
-      "loss": 0.0002,
+      "loss": 0.0004,
       "step": 2263
     },
     {
       "epoch": 4.9812981298129815,
-      "grad_norm": 0.01068115234375,
+      "grad_norm": 0.00506591796875,
       "learning_rate": 0.00010076285174412759,
       "loss": 0.0002,
       "step": 2264
     },
     {
       "epoch": 4.983498349834983,
-      "grad_norm": 0.0235595703125,
+      "grad_norm": 0.03125,
       "learning_rate": 0.00010069350275297337,
       "loss": 0.0004,
       "step": 2265
     },
     {
       "epoch": 4.985698569856986,
-      "grad_norm": 0.011474609375,
+      "grad_norm": 0.0096435546875,
       "learning_rate": 0.00010062415342827642,
       "loss": 0.0002,
       "step": 2266
     },
     {
       "epoch": 4.987898789878988,
-      "grad_norm": 0.039306640625,
+      "grad_norm": 0.02783203125,
       "learning_rate": 0.00010055480380339053,
-      "loss": 0.0006,
+      "loss": 0.0004,
       "step": 2267
     },
     {
       "epoch": 4.99009900990099,
-      "grad_norm": 0.3515625,
+      "grad_norm": 0.171875,
       "learning_rate": 0.00010048545391166966,
-      "loss": 0.0031,
+      "loss": 0.0022,
       "step": 2268
     },
     {
       "epoch": 4.992299229922993,
-      "grad_norm": 0.045166015625,
+      "grad_norm": 0.0224609375,
       "learning_rate": 0.00010041610378646789,
-      "loss": 0.0004,
+      "loss": 0.0003,
       "step": 2269
     },
     {
       "epoch": 4.994499449944994,
-      "grad_norm": 0.00726318359375,
+      "grad_norm": 0.013671875,
       "learning_rate": 0.00010034675346113945,
       "loss": 0.0002,
       "step": 2270
     },
     {
       "epoch": 4.996699669966997,
-      "grad_norm": 0.2353515625,
+      "grad_norm": 0.010986328125,
       "learning_rate": 0.00010027740296903862,
-      "loss": 0.0022,
+      "loss": 0.0002,
       "step": 2271
     },
     {
       "epoch": 4.998899889988999,
-      "grad_norm": 0.051025390625,
+      "grad_norm": 0.01318359375,
       "learning_rate": 0.00010020805234351982,
-      "loss": 0.0006,
+      "loss": 0.0002,
       "step": 2272
     },
     {
       "epoch": 5.001100110011001,
-      "grad_norm": 0.00726318359375,
+      "grad_norm": 0.01214599609375,
       "learning_rate": 0.00010013870161793744,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2273
     },
     {
       "epoch": 5.003300330033003,
-      "grad_norm": 0.012451171875,
+      "grad_norm": 0.005462646484375,
       "learning_rate": 0.00010006935082564599,
       "loss": 0.0002,
       "step": 2274
     },
     {
       "epoch": 5.005500550055006,
-      "grad_norm": 0.068359375,
+      "grad_norm": 0.0140380859375,
       "learning_rate": 0.0001,
-      "loss": 0.0005,
+      "loss": 0.0002,
       "step": 2275
     },
     {
       "epoch": 5.007700770077007,
-      "grad_norm": 0.240234375,
+      "grad_norm": 0.01544189453125,
       "learning_rate": 9.9930649174354e-05,
-      "loss": 0.0014,
+      "loss": 0.0003,
       "step": 2276
     },
     {
       "epoch": 5.00990099009901,
-      "grad_norm": 0.00457763671875,
+      "grad_norm": 0.002655029296875,
       "learning_rate": 9.986129838206259e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2277
     },
     {
       "epoch": 5.0121012101210125,
-      "grad_norm": 0.1474609375,
+      "grad_norm": 0.026123046875,
       "learning_rate": 9.97919476564802e-05,
-      "loss": 0.0007,
+      "loss": 0.0002,
       "step": 2278
     },
     {
       "epoch": 5.014301430143014,
-      "grad_norm": 0.010498046875,
+      "grad_norm": 0.052978515625,
       "learning_rate": 9.97225970309614e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2279
     },
     {
       "epoch": 5.016501650165017,
-      "grad_norm": 0.00640869140625,
+      "grad_norm": 0.06298828125,
       "learning_rate": 9.965324653886056e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2280
     },
     {
       "epoch": 5.016501650165017,
-      "eval_loss": 0.0008992512011900544,
-      "eval_runtime": 10.2748,
-      "eval_samples_per_second": 37.276,
-      "eval_steps_per_second": 4.672,
+      "eval_loss": 0.0007948374841362238,
+      "eval_runtime": 10.9525,
+      "eval_samples_per_second": 34.969,
+      "eval_steps_per_second": 4.383,
       "step": 2280
     },
     {
       "epoch": 5.0187018701870185,
-      "grad_norm": 0.09619140625,
+      "grad_norm": 0.0673828125,
       "learning_rate": 9.958389621353211e-05,
-      "loss": 0.0006,
+      "loss": 0.0005,
       "step": 2281
     },
     {
       "epoch": 5.020902090209021,
-      "grad_norm": 0.01708984375,
+      "grad_norm": 0.0927734375,
       "learning_rate": 9.951454608833037e-05,
-      "loss": 0.0003,
+      "loss": 0.0009,
       "step": 2282
     },
     {
       "epoch": 5.023102310231023,
-      "grad_norm": 0.0125732421875,
+      "grad_norm": 0.0048828125,
       "learning_rate": 9.94451961966095e-05,
       "loss": 0.0002,
       "step": 2283
     },
     {
       "epoch": 5.025302530253025,
-      "grad_norm": 0.00518798828125,
+      "grad_norm": 0.0205078125,
       "learning_rate": 9.937584657172361e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2284
     },
     {
       "epoch": 5.027502750275027,
-      "grad_norm": 0.005401611328125,
+      "grad_norm": 0.0076904296875,
       "learning_rate": 9.930649724702664e-05,
       "loss": 0.0002,
       "step": 2285
     },
     {
       "epoch": 5.02970297029703,
-      "grad_norm": 0.0208740234375,
+      "grad_norm": 0.0030059814453125,
       "learning_rate": 9.923714825587246e-05,
-      "loss": 0.0003,
+      "loss": 0.0001,
       "step": 2286
     },
     {
       "epoch": 5.031903190319032,
-      "grad_norm": 0.0155029296875,
+      "grad_norm": 0.0086669921875,
       "learning_rate": 9.916779963161465e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2287
     },
     {
       "epoch": 5.034103410341034,
-      "grad_norm": 0.00592041015625,
+      "grad_norm": 0.003143310546875,
       "learning_rate": 9.909845140760675e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2288
     },
     {
       "epoch": 5.036303630363037,
-      "grad_norm": 0.103515625,
+      "grad_norm": 0.006988525390625,
       "learning_rate": 9.902910361720203e-05,
-      "loss": 0.0005,
+      "loss": 0.0002,
       "step": 2289
     },
     {
       "epoch": 5.038503850385038,
-      "grad_norm": 0.0133056640625,
+      "grad_norm": 0.0184326171875,
       "learning_rate": 9.895975629375359e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2290
     },
     {
       "epoch": 5.040704070407041,
-      "grad_norm": 0.04248046875,
+      "grad_norm": 0.023193359375,
       "learning_rate": 9.889040947061429e-05,
-      "loss": 0.0004,
+      "loss": 0.0003,
       "step": 2291
     },
     {
       "epoch": 5.042904290429043,
-      "grad_norm": 0.1005859375,
+      "grad_norm": 0.01153564453125,
       "learning_rate": 9.882106318113668e-05,
-      "loss": 0.0011,
+      "loss": 0.0002,
       "step": 2292
     },
     {
       "epoch": 5.045104510451045,
-      "grad_norm": 0.00799560546875,
+      "grad_norm": 0.00360107421875,
       "learning_rate": 9.875171745867318e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2293
     },
     {
       "epoch": 5.047304730473047,
-      "grad_norm": 0.00604248046875,
+      "grad_norm": 0.0028839111328125,
       "learning_rate": 9.868237233657588e-05,
       "loss": 0.0001,
       "step": 2294
     },
     {
       "epoch": 5.0495049504950495,
-      "grad_norm": 0.01129150390625,
+      "grad_norm": 0.00738525390625,
       "learning_rate": 9.861302784819655e-05,
       "loss": 0.0002,
       "step": 2295
     },
     {
       "epoch": 5.051705170517051,
-      "grad_norm": 0.04345703125,
+      "grad_norm": 0.003631591796875,
       "learning_rate": 9.854368402688672e-05,
-      "loss": 0.0003,
+      "loss": 0.0001,
       "step": 2296
     },
     {
       "epoch": 5.053905390539054,
-      "grad_norm": 0.006988525390625,
+      "grad_norm": 0.1416015625,
       "learning_rate": 9.847434090599748e-05,
-      "loss": 0.0002,
+      "loss": 0.0005,
       "step": 2297
     },
     {
       "epoch": 5.0561056105610565,
-      "grad_norm": 0.01141357421875,
+      "grad_norm": 0.01422119140625,
       "learning_rate": 9.840499851887971e-05,
       "loss": 0.0002,
       "step": 2298
     },
     {
       "epoch": 5.058305830583058,
-      "grad_norm": 0.0159912109375,
+      "grad_norm": 0.006988525390625,
       "learning_rate": 9.833565689888395e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2299
     },
     {
       "epoch": 5.060506050605061,
-      "grad_norm": 0.15625,
+      "grad_norm": 0.029541015625,
       "learning_rate": 9.826631607936025e-05,
-      "loss": 0.0005,
+      "loss": 0.0002,
       "step": 2300
     },
     {
       "epoch": 5.0627062706270625,
-      "grad_norm": 0.3671875,
+      "grad_norm": 0.37890625,
       "learning_rate": 9.819697609365835e-05,
-      "loss": 0.004,
+      "loss": 0.003,
       "step": 2301
     },
     {
       "epoch": 5.064906490649065,
-      "grad_norm": 0.0037384033203125,
+      "grad_norm": 0.00408935546875,
       "learning_rate": 9.812763697512754e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2302
     },
     {
       "epoch": 5.067106710671067,
-      "grad_norm": 0.01043701171875,
+      "grad_norm": 0.00543212890625,
       "learning_rate": 9.805829875711679e-05,
-      "loss": 0.0003,
+      "loss": 0.0001,
       "step": 2303
     },
     {
       "epoch": 5.069306930693069,
-      "grad_norm": 0.00823974609375,
+      "grad_norm": 0.011962890625,
       "learning_rate": 9.798896147297457e-05,
       "loss": 0.0002,
       "step": 2304
     },
     {
       "epoch": 5.071507150715071,
-      "grad_norm": 0.028076171875,
+      "grad_norm": 0.0167236328125,
       "learning_rate": 9.791962515604887e-05,
       "loss": 0.0003,
       "step": 2305
     },
     {
       "epoch": 5.073707370737074,
-      "grad_norm": 0.003448486328125,
+      "grad_norm": 0.008056640625,
       "learning_rate": 9.78502898396873e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2306
     },
     {
       "epoch": 5.075907590759076,
-      "grad_norm": 0.00555419921875,
+      "grad_norm": 0.004486083984375,
       "learning_rate": 9.778095555723687e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2307
     },
     {
       "epoch": 5.078107810781078,
-      "grad_norm": 0.0172119140625,
+      "grad_norm": 0.0037384033203125,
       "learning_rate": 9.771162234204425e-05,
-      "loss": 0.0003,
+      "loss": 0.0001,
       "step": 2308
     },
     {
       "epoch": 5.080308030803081,
-      "grad_norm": 0.021484375,
+      "grad_norm": 0.01806640625,
       "learning_rate": 9.764229022745543e-05,
       "loss": 0.0002,
       "step": 2309
     },
     {
       "epoch": 5.082508250825082,
-      "grad_norm": 0.01708984375,
+      "grad_norm": 0.039794921875,
       "learning_rate": 9.7572959246816e-05,
       "loss": 0.0002,
       "step": 2310
     },
     {
       "epoch": 5.084708470847085,
-      "grad_norm": 0.0038909912109375,
+      "grad_norm": 0.008056640625,
       "learning_rate": 9.750362943347098e-05,
       "loss": 0.0002,
       "step": 2311
     },
     {
       "epoch": 5.086908690869087,
-      "grad_norm": 0.00848388671875,
+      "grad_norm": 0.0230712890625,
       "learning_rate": 9.743430082076472e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2312
     },
     {
       "epoch": 5.089108910891089,
-      "grad_norm": 0.00982666015625,
+      "grad_norm": 0.01507568359375,
       "learning_rate": 9.736497344204117e-05,
       "loss": 0.0003,
       "step": 2313
     },
     {
       "epoch": 5.091309130913091,
-      "grad_norm": 0.029541015625,
+      "grad_norm": 0.0234375,
       "learning_rate": 9.72956473306435e-05,
-      "loss": 0.0006,
+      "loss": 0.0005,
       "step": 2314
     },
     {
       "epoch": 5.0935093509350935,
-      "grad_norm": 0.010009765625,
+      "grad_norm": 0.006256103515625,
       "learning_rate": 9.722632251991444e-05,
       "loss": 0.0002,
       "step": 2315
     },
     {
       "epoch": 5.095709570957096,
-      "grad_norm": 0.04638671875,
+      "grad_norm": 0.00732421875,
       "learning_rate": 9.715699904319599e-05,
-      "loss": 0.0005,
+      "loss": 0.0002,
       "step": 2316
     },
     {
       "epoch": 5.097909790979098,
-      "grad_norm": 0.01806640625,
+      "grad_norm": 0.00823974609375,
       "learning_rate": 9.708767693382953e-05,
-      "loss": 0.0004,
+      "loss": 0.0003,
       "step": 2317
     },
     {
       "epoch": 5.1001100110011,
-      "grad_norm": 0.0203857421875,
+      "grad_norm": 0.03466796875,
       "learning_rate": 9.701835622515584e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2318
     },
     {
       "epoch": 5.102310231023102,
-      "grad_norm": 0.0123291015625,
+      "grad_norm": 0.0037841796875,
       "learning_rate": 9.694903695051488e-05,
       "loss": 0.0002,
       "step": 2319
     },
     {
       "epoch": 5.104510451045105,
-      "grad_norm": 0.00732421875,
+      "grad_norm": 0.008544921875,
       "learning_rate": 9.687971914324607e-05,
       "loss": 0.0002,
       "step": 2320
     },
     {
       "epoch": 5.106710671067106,
-      "grad_norm": 0.08349609375,
+      "grad_norm": 0.0206298828125,
       "learning_rate": 9.681040283668812e-05,
-      "loss": 0.0021,
+      "loss": 0.0004,
       "step": 2321
     },
     {
       "epoch": 5.108910891089109,
-      "grad_norm": 0.015869140625,
+      "grad_norm": 0.00543212890625,
       "learning_rate": 9.674108806417889e-05,
       "loss": 0.0002,
       "step": 2322
     },
     {
       "epoch": 5.111111111111111,
-      "grad_norm": 0.01165771484375,
+      "grad_norm": 0.01556396484375,
       "learning_rate": 9.667177485905562e-05,
       "loss": 0.0002,
       "step": 2323
     },
     {
       "epoch": 5.113311331133113,
-      "grad_norm": 0.004058837890625,
+      "grad_norm": 0.0059814453125,
       "learning_rate": 9.660246325465471e-05,
       "loss": 0.0002,
       "step": 2324
     },
     {
       "epoch": 5.115511551155116,
-      "grad_norm": 0.01434326171875,
+      "grad_norm": 0.005279541015625,
       "learning_rate": 9.653315328431191e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2325
     },
     {
       "epoch": 5.117711771177118,
-      "grad_norm": 0.59765625,
+      "grad_norm": 0.004150390625,
       "learning_rate": 9.646384498136202e-05,
-      "loss": 0.0092,
+      "loss": 0.0002,
       "step": 2326
     },
     {
       "epoch": 5.11991199119912,
-      "grad_norm": 0.0086669921875,
+      "grad_norm": 0.01068115234375,
       "learning_rate": 9.639453837913923e-05,
       "loss": 0.0003,
       "step": 2327
     },
     {
       "epoch": 5.122112211221122,
-      "grad_norm": 0.0098876953125,
+      "grad_norm": 0.0247802734375,
       "learning_rate": 9.632523351097676e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2328
     },
     {
       "epoch": 5.1243124312431245,
-      "grad_norm": 0.05517578125,
+      "grad_norm": 0.00848388671875,
       "learning_rate": 9.625593041020701e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2329
     },
     {
       "epoch": 5.126512651265126,
-      "grad_norm": 0.047119140625,
+      "grad_norm": 0.02734375,
       "learning_rate": 9.618662911016165e-05,
-      "loss": 0.0003,
+      "loss": 0.0004,
       "step": 2330
     },
     {
       "epoch": 5.128712871287129,
-      "grad_norm": 0.005859375,
+      "grad_norm": 0.00482177734375,
       "learning_rate": 9.611732964417132e-05,
       "loss": 0.0002,
       "step": 2331
     },
     {
       "epoch": 5.1309130913091305,
-      "grad_norm": 0.014404296875,
+      "grad_norm": 0.0184326171875,
       "learning_rate": 9.604803204556592e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2332
     },
     {
       "epoch": 5.133113311331133,
-      "grad_norm": 0.0186767578125,
+      "grad_norm": 0.01007080078125,
       "learning_rate": 9.597873634767437e-05,
       "loss": 0.0002,
       "step": 2333
     },
     {
       "epoch": 5.135313531353136,
-      "grad_norm": 0.0205078125,
+      "grad_norm": 0.01019287109375,
       "learning_rate": 9.590944258382466e-05,
       "loss": 0.0002,
       "step": 2334
     },
     {
       "epoch": 5.137513751375137,
-      "grad_norm": 0.00787353515625,
+      "grad_norm": 0.0155029296875,
       "learning_rate": 9.584015078734395e-05,
       "loss": 0.0002,
       "step": 2335
     },
     {
       "epoch": 5.13971397139714,
-      "grad_norm": 0.00445556640625,
+      "grad_norm": 0.008056640625,
       "learning_rate": 9.577086099155832e-05,
       "loss": 0.0002,
       "step": 2336
     },
     {
       "epoch": 5.141914191419142,
-      "grad_norm": 0.0751953125,
+      "grad_norm": 0.0101318359375,
       "learning_rate": 9.570157322979303e-05,
-      "loss": 0.0005,
+      "loss": 0.0002,
       "step": 2337
     },
     {
       "epoch": 5.144114411441144,
-      "grad_norm": 0.0068359375,
+      "grad_norm": 0.0263671875,
       "learning_rate": 9.563228753537227e-05,
       "loss": 0.0002,
       "step": 2338
     },
     {
       "epoch": 5.146314631463146,
-      "grad_norm": 0.04345703125,
+      "grad_norm": 0.003662109375,
       "learning_rate": 9.556300394161919e-05,
-      "loss": 0.0004,
+      "loss": 0.0002,
       "step": 2339
     },
     {
       "epoch": 5.148514851485149,
-      "grad_norm": 0.006591796875,
+      "grad_norm": 0.0157470703125,
       "learning_rate": 9.54937224818561e-05,
       "loss": 0.0002,
       "step": 2340
     },
     {
       "epoch": 5.15071507150715,
-      "grad_norm": 0.038330078125,
+      "grad_norm": 0.012451171875,
       "learning_rate": 9.542444318940407e-05,
-      "loss": 0.0004,
+      "loss": 0.0002,
       "step": 2341
     },
     {
       "epoch": 5.152915291529153,
-      "grad_norm": 0.0157470703125,
+      "grad_norm": 0.0174560546875,
       "learning_rate": 9.535516609758332e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2342
     },
     {
       "epoch": 5.1551155115511555,
-      "grad_norm": 0.00860595703125,
+      "grad_norm": 0.007720947265625,
       "learning_rate": 9.528589123971288e-05,
       "loss": 0.0002,
       "step": 2343
     },
     {
       "epoch": 5.157315731573157,
-      "grad_norm": 0.0086669921875,
+      "grad_norm": 0.003326416015625,
       "learning_rate": 9.52166186491108e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2344
     },
     {
       "epoch": 5.15951595159516,
-      "grad_norm": 0.0162353515625,
+      "grad_norm": 0.019775390625,
       "learning_rate": 9.514734835909397e-05,
       "loss": 0.0003,
       "step": 2345
     },
     {
       "epoch": 5.161716171617162,
-      "grad_norm": 0.02392578125,
+      "grad_norm": 0.01483154296875,
       "learning_rate": 9.507808040297814e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2346
     },
     {
       "epoch": 5.163916391639164,
-      "grad_norm": 0.0125732421875,
+      "grad_norm": 0.01397705078125,
       "learning_rate": 9.50088148140781e-05,
       "loss": 0.0002,
       "step": 2347
     },
     {
       "epoch": 5.166116611661166,
-      "grad_norm": 0.008056640625,
+      "grad_norm": 0.0230712890625,
       "learning_rate": 9.49395516257073e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2348
     },
     {
       "epoch": 5.1683168316831685,
-      "grad_norm": 0.008544921875,
+      "grad_norm": 0.0040283203125,
       "learning_rate": 9.48702908711782e-05,
       "loss": 0.0002,
       "step": 2349
     },
     {
       "epoch": 5.17051705170517,
-      "grad_norm": 0.045166015625,
+      "grad_norm": 0.009765625,
       "learning_rate": 9.480103258380198e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2350
     },
     {
       "epoch": 5.172717271727173,
-      "grad_norm": 0.05712890625,
+      "grad_norm": 0.011962890625,
       "learning_rate": 9.473177679688867e-05,
-      "loss": 0.0006,
+      "loss": 0.0003,
       "step": 2351
     },
     {
       "epoch": 5.174917491749175,
-      "grad_norm": 0.1416015625,
+      "grad_norm": 0.0303955078125,
       "learning_rate": 9.466252354374716e-05,
-      "loss": 0.001,
+      "loss": 0.0002,
       "step": 2352
     },
     {
       "epoch": 5.177117711771177,
-      "grad_norm": 0.01397705078125,
+      "grad_norm": 0.0089111328125,
       "learning_rate": 9.459327285768495e-05,
       "loss": 0.0002,
       "step": 2353
     },
     {
       "epoch": 5.17931793179318,
-      "grad_norm": 0.0093994140625,
+      "grad_norm": 0.0084228515625,
       "learning_rate": 9.452402477200852e-05,
       "loss": 0.0002,
       "step": 2354
     },
     {
       "epoch": 5.181518151815181,
-      "grad_norm": 0.00787353515625,
+      "grad_norm": 0.02001953125,
       "learning_rate": 9.4454779320023e-05,
       "loss": 0.0003,
       "step": 2355
     },
     {
       "epoch": 5.183718371837184,
-      "grad_norm": 0.01055908203125,
+      "grad_norm": 0.01141357421875,
       "learning_rate": 9.438553653503215e-05,
       "loss": 0.0002,
       "step": 2356
     },
     {
       "epoch": 5.185918591859186,
-      "grad_norm": 0.0107421875,
+      "grad_norm": 0.005035400390625,
       "learning_rate": 9.431629645033867e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2357
     },
     {
       "epoch": 5.188118811881188,
-      "grad_norm": 0.00970458984375,
+      "grad_norm": 0.01055908203125,
       "learning_rate": 9.424705909924372e-05,
       "loss": 0.0002,
       "step": 2358
     },
     {
       "epoch": 5.19031903190319,
-      "grad_norm": 0.005950927734375,
+      "grad_norm": 0.00482177734375,
       "learning_rate": 9.417782451504737e-05,
       "loss": 0.0001,
       "step": 2359
     },
     {
       "epoch": 5.192519251925193,
-      "grad_norm": 0.0037841796875,
+      "grad_norm": 0.0029144287109375,
       "learning_rate": 9.410859273104822e-05,
       "loss": 0.0001,
       "step": 2360
     },
     {
       "epoch": 5.194719471947194,
-      "grad_norm": 0.00872802734375,
+      "grad_norm": 0.0093994140625,
       "learning_rate": 9.40393637805435e-05,
       "loss": 0.0002,
       "step": 2361
     },
     {
       "epoch": 5.196919691969197,
-      "grad_norm": 0.236328125,
+      "grad_norm": 0.00958251953125,
       "learning_rate": 9.397013769682924e-05,
-      "loss": 0.0011,
+      "loss": 0.0002,
       "step": 2362
     },
     {
       "epoch": 5.1991199119911995,
-      "grad_norm": 0.017822265625,
+      "grad_norm": 0.03466796875,
       "learning_rate": 9.390091451319985e-05,
-      "loss": 0.0004,
+      "loss": 0.0003,
       "step": 2363
     },
     {
       "epoch": 5.201320132013201,
-      "grad_norm": 0.0181884765625,
+      "grad_norm": 0.00347900390625,
       "learning_rate": 9.383169426294861e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2364
     },
     {
       "epoch": 5.203520352035204,
-      "grad_norm": 0.034423828125,
+      "grad_norm": 0.06591796875,
       "learning_rate": 9.376247697936719e-05,
-      "loss": 0.0004,
+      "loss": 0.0008,
       "step": 2365
     },
     {
       "epoch": 5.2057205720572055,
-      "grad_norm": 0.0194091796875,
+      "grad_norm": 0.0263671875,
       "learning_rate": 9.369326269574589e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2366
     },
     {
       "epoch": 5.207920792079208,
-      "grad_norm": 0.0162353515625,
+      "grad_norm": 0.0086669921875,
       "learning_rate": 9.362405144537365e-05,
       "loss": 0.0002,
       "step": 2367
     },
     {
       "epoch": 5.21012101210121,
-      "grad_norm": 0.01287841796875,
+      "grad_norm": 0.00518798828125,
       "learning_rate": 9.35548432615378e-05,
       "loss": 0.0002,
       "step": 2368
     },
     {
       "epoch": 5.212321232123212,
-      "grad_norm": 0.012939453125,
+      "grad_norm": 0.00274658203125,
       "learning_rate": 9.348563817752437e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2369
     },
     {
       "epoch": 5.214521452145214,
-      "grad_norm": 0.0169677734375,
+      "grad_norm": 0.0140380859375,
       "learning_rate": 9.341643622661768e-05,
       "loss": 0.0002,
       "step": 2370
     },
     {
       "epoch": 5.216721672167217,
-      "grad_norm": 0.004791259765625,
+      "grad_norm": 0.00927734375,
       "learning_rate": 9.334723744210077e-05,
       "loss": 0.0002,
       "step": 2371
     },
     {
       "epoch": 5.218921892189219,
-      "grad_norm": 0.02001953125,
+      "grad_norm": 0.0185546875,
       "learning_rate": 9.327804185725505e-05,
       "loss": 0.0003,
       "step": 2372
     },
     {
       "epoch": 5.221122112211221,
-      "grad_norm": 0.146484375,
+      "grad_norm": 0.00677490234375,
       "learning_rate": 9.320884950536034e-05,
-      "loss": 0.0005,
+      "loss": 0.0002,
       "step": 2373
     },
     {
       "epoch": 5.223322332233224,
-      "grad_norm": 0.0045166015625,
+      "grad_norm": 0.007598876953125,
       "learning_rate": 9.313966041969501e-05,
       "loss": 0.0002,
       "step": 2374
     },
     {
       "epoch": 5.225522552255225,
-      "grad_norm": 0.2216796875,
+      "grad_norm": 0.0042724609375,
       "learning_rate": 9.307047463353575e-05,
-      "loss": 0.0018,
+      "loss": 0.0001,
       "step": 2375
     },
     {
       "epoch": 5.227722772277228,
-      "grad_norm": 0.005889892578125,
+      "grad_norm": 0.01019287109375,
       "learning_rate": 9.300129218015781e-05,
       "loss": 0.0002,
       "step": 2376
     },
     {
       "epoch": 5.22992299229923,
-      "grad_norm": 0.01446533203125,
+      "grad_norm": 0.05029296875,
       "learning_rate": 9.293211309283472e-05,
-      "loss": 0.0003,
+      "loss": 0.0008,
       "step": 2377
     },
     {
       "epoch": 5.232123212321232,
-      "grad_norm": 0.1318359375,
+      "grad_norm": 0.0478515625,
       "learning_rate": 9.286293740483837e-05,
-      "loss": 0.0008,
+      "loss": 0.0003,
       "step": 2378
     },
     {
       "epoch": 5.234323432343234,
-      "grad_norm": 0.00836181640625,
+      "grad_norm": 0.007171630859375,
       "learning_rate": 9.279376514943915e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2379
     },
     {
       "epoch": 5.2365236523652365,
-      "grad_norm": 0.09375,
+      "grad_norm": 0.0498046875,
       "learning_rate": 9.272459635990562e-05,
-      "loss": 0.0004,
+      "loss": 0.0005,
       "step": 2380
     },
     {
       "epoch": 5.238723872387239,
-      "grad_norm": 0.009521484375,
+      "grad_norm": 0.0096435546875,
       "learning_rate": 9.265543106950487e-05,
       "loss": 0.0002,
       "step": 2381
     },
     {
       "epoch": 5.240924092409241,
-      "grad_norm": 0.033203125,
+      "grad_norm": 0.004547119140625,
       "learning_rate": 9.258626931150215e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2382
     },
     {
       "epoch": 5.243124312431243,
-      "grad_norm": 0.01251220703125,
+      "grad_norm": 0.00909423828125,
       "learning_rate": 9.251711111916105e-05,
-      "loss": 0.0004,
+      "loss": 0.0003,
       "step": 2383
     },
     {
       "epoch": 5.245324532453245,
-      "grad_norm": 0.01165771484375,
+      "grad_norm": 0.045654296875,
       "learning_rate": 9.244795652574354e-05,
-      "loss": 0.0002,
+      "loss": 0.001,
       "step": 2384
     },
     {
       "epoch": 5.247524752475248,
-      "grad_norm": 0.013671875,
+      "grad_norm": 0.0032501220703125,
       "learning_rate": 9.237880556450967e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2385
     },
     {
       "epoch": 5.2497249724972495,
-      "grad_norm": 0.034912109375,
+      "grad_norm": 0.004486083984375,
       "learning_rate": 9.230965826871798e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2386
     },
     {
       "epoch": 5.251925192519252,
-      "grad_norm": 0.396484375,
+      "grad_norm": 0.01165771484375,
       "learning_rate": 9.224051467162502e-05,
-      "loss": 0.0093,
+      "loss": 0.0002,
       "step": 2387
     },
     {
       "epoch": 5.254125412541254,
-      "grad_norm": 0.006591796875,
+      "grad_norm": 0.019775390625,
       "learning_rate": 9.217137480648569e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2388
     },
     {
       "epoch": 5.256325632563256,
-      "grad_norm": 0.004180908203125,
+      "grad_norm": 0.00872802734375,
       "learning_rate": 9.210223870655312e-05,
       "loss": 0.0002,
       "step": 2389
     },
     {
       "epoch": 5.258525852585258,
-      "grad_norm": 0.012939453125,
+      "grad_norm": 0.02099609375,
       "learning_rate": 9.203310640507849e-05,
       "loss": 0.0002,
       "step": 2390
     },
     {
       "epoch": 5.260726072607261,
-      "grad_norm": 0.0084228515625,
+      "grad_norm": 0.0078125,
       "learning_rate": 9.196397793531134e-05,
       "loss": 0.0002,
       "step": 2391
     },
     {
       "epoch": 5.262926292629263,
-      "grad_norm": 0.0223388671875,
+      "grad_norm": 0.00360107421875,
       "learning_rate": 9.189485333049918e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2392
     },
     {
       "epoch": 5.265126512651265,
-      "grad_norm": 0.0030059814453125,
+      "grad_norm": 0.00341796875,
       "learning_rate": 9.182573262388777e-05,
       "loss": 0.0001,
       "step": 2393
     },
     {
       "epoch": 5.267326732673268,
-      "grad_norm": 0.0274658203125,
+      "grad_norm": 0.005462646484375,
       "learning_rate": 9.175661584872103e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2394
     },
     {
       "epoch": 5.267326732673268,
-      "eval_loss": 0.0005059956456534564,
-      "eval_runtime": 10.3657,
-      "eval_samples_per_second": 36.949,
-      "eval_steps_per_second": 4.631,
+      "eval_loss": 0.0005059099639765918,
+      "eval_runtime": 10.5852,
+      "eval_samples_per_second": 36.183,
+      "eval_steps_per_second": 4.535,
       "step": 2394
     },
     {
       "epoch": 5.269526952695269,
-      "grad_norm": 0.01611328125,
+      "grad_norm": 0.00665283203125,
       "learning_rate": 9.168750303824084e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2395
     },
     {
       "epoch": 5.271727172717272,
-      "grad_norm": 0.018798828125,
+      "grad_norm": 0.003326416015625,
       "learning_rate": 9.161839422568736e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2396
     },
     {
       "epoch": 5.273927392739274,
-      "grad_norm": 0.07861328125,
+      "grad_norm": 0.00433349609375,
       "learning_rate": 9.154928944429865e-05,
-      "loss": 0.0003,
+      "loss": 0.0001,
       "step": 2397
     },
     {
       "epoch": 5.276127612761276,
-      "grad_norm": 0.0181884765625,
+      "grad_norm": 0.0196533203125,
       "learning_rate": 9.148018872731097e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2398
     },
     {
       "epoch": 5.278327832783278,
-      "grad_norm": 0.0048828125,
+      "grad_norm": 0.0185546875,
       "learning_rate": 9.141109210795859e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2399
     },
     {
       "epoch": 5.2805280528052805,
-      "grad_norm": 0.057373046875,
+      "grad_norm": 0.00634765625,
       "learning_rate": 9.134199961947368e-05,
-      "loss": 0.0004,
+      "loss": 0.0001,
       "step": 2400
     },
     {
       "epoch": 5.282728272827283,
-      "grad_norm": 0.0238037109375,
+      "grad_norm": 0.007598876953125,
       "learning_rate": 9.127291129508666e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2401
     },
     {
       "epoch": 5.284928492849285,
-      "grad_norm": 0.01397705078125,
+      "grad_norm": 0.008544921875,
       "learning_rate": 9.120382716802573e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2402
     },
     {
       "epoch": 5.287128712871287,
-      "grad_norm": 0.00494384765625,
+      "grad_norm": 0.0034027099609375,
       "learning_rate": 9.11347472715172e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2403
     },
     {
       "epoch": 5.289328932893289,
-      "grad_norm": 0.005950927734375,
+      "grad_norm": 0.004730224609375,
       "learning_rate": 9.106567163878533e-05,
       "loss": 0.0002,
       "step": 2404
     },
     {
       "epoch": 5.291529152915292,
-      "grad_norm": 0.003997802734375,
+      "grad_norm": 0.00396728515625,
       "learning_rate": 9.099660030305224e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2405
     },
     {
       "epoch": 5.293729372937293,
-      "grad_norm": 0.017822265625,
+      "grad_norm": 0.0059814453125,
       "learning_rate": 9.092753329753812e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2406
     },
     {
       "epoch": 5.295929592959296,
-      "grad_norm": 0.0026092529296875,
+      "grad_norm": 0.00543212890625,
       "learning_rate": 9.085847065546091e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2407
     },
     {
       "epoch": 5.298129812981298,
-      "grad_norm": 0.0101318359375,
+      "grad_norm": 0.0042724609375,
       "learning_rate": 9.078941241003666e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2408
     },
     {
       "epoch": 5.3003300330033,
-      "grad_norm": 0.0198974609375,
+      "grad_norm": 0.00286865234375,
       "learning_rate": 9.072035859447913e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2409
     },
     {
       "epoch": 5.302530253025303,
-      "grad_norm": 0.0157470703125,
+      "grad_norm": 0.0086669921875,
       "learning_rate": 9.065130924199998e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2410
     },
     {
       "epoch": 5.304730473047305,
-      "grad_norm": 0.00830078125,
+      "grad_norm": 0.00372314453125,
       "learning_rate": 9.058226438580885e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2411
     },
     {
       "epoch": 5.306930693069307,
-      "grad_norm": 0.004180908203125,
+      "grad_norm": 0.00494384765625,
       "learning_rate": 9.051322405911304e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2412
     },
     {
       "epoch": 5.309130913091309,
-      "grad_norm": 0.00347900390625,
+      "grad_norm": 0.0034027099609375,
       "learning_rate": 9.044418829511781e-05,
       "loss": 0.0002,
       "step": 2413
     },
     {
       "epoch": 5.3113311331133115,
-      "grad_norm": 0.005218505859375,
+      "grad_norm": 0.004974365234375,
       "learning_rate": 9.037515712702613e-05,
       "loss": 0.0002,
       "step": 2414
     },
     {
       "epoch": 5.313531353135313,
-      "grad_norm": 0.0068359375,
+      "grad_norm": 0.00592041015625,
       "learning_rate": 9.030613058803881e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2415
     },
     {
       "epoch": 5.315731573157316,
-      "grad_norm": 0.02734375,
+      "grad_norm": 0.0038604736328125,
       "learning_rate": 9.023710871135445e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2416
     },
     {
       "epoch": 5.3179317931793175,
-      "grad_norm": 0.01141357421875,
+      "grad_norm": 0.0050048828125,
       "learning_rate": 9.016809153016932e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2417
     },
     {
       "epoch": 5.32013201320132,
-      "grad_norm": 0.00799560546875,
+      "grad_norm": 0.0027008056640625,
       "learning_rate": 9.009907907767759e-05,
       "loss": 0.0001,
       "step": 2418
     },
     {
       "epoch": 5.322332233223323,
-      "grad_norm": 0.01019287109375,
+      "grad_norm": 0.005706787109375,
       "learning_rate": 9.003007138707095e-05,
       "loss": 0.0002,
       "step": 2419
     },
     {
       "epoch": 5.324532453245324,
-      "grad_norm": 0.00579833984375,
+      "grad_norm": 0.00604248046875,
       "learning_rate": 8.996106849153897e-05,
       "loss": 0.0002,
       "step": 2420
     },
     {
       "epoch": 5.326732673267327,
-      "grad_norm": 0.004302978515625,
+      "grad_norm": 0.047119140625,
       "learning_rate": 8.989207042426887e-05,
-      "loss": 0.0002,
+      "loss": 0.0005,
       "step": 2421
     },
     {
       "epoch": 5.328932893289329,
-      "grad_norm": 0.003509521484375,
+      "grad_norm": 0.00775146484375,
       "learning_rate": 8.982307721844546e-05,
       "loss": 0.0002,
       "step": 2422
     },
     {
       "epoch": 5.331133113311331,
-      "grad_norm": 0.01470947265625,
+      "grad_norm": 0.0081787109375,
       "learning_rate": 8.975408890725136e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2423
     },
     {
       "epoch": 5.333333333333333,
-      "grad_norm": 0.00946044921875,
+      "grad_norm": 0.00927734375,
       "learning_rate": 8.968510552386668e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2424
     },
     {
       "epoch": 5.335533553355336,
-      "grad_norm": 0.0028076171875,
+      "grad_norm": 0.004547119140625,
       "learning_rate": 8.961612710146934e-05,
       "loss": 0.0001,
       "step": 2425
     },
     {
       "epoch": 5.337733773377337,
-      "grad_norm": 0.060546875,
+      "grad_norm": 0.0093994140625,
       "learning_rate": 8.954715367323468e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2426
     },
     {
       "epoch": 5.33993399339934,
-      "grad_norm": 0.060791015625,
+      "grad_norm": 0.042724609375,
       "learning_rate": 8.947818527233572e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2427
     },
     {
       "epoch": 5.3421342134213425,
-      "grad_norm": 0.00390625,
+      "grad_norm": 0.07373046875,
       "learning_rate": 8.940922193194317e-05,
-      "loss": 0.0002,
+      "loss": 0.0005,
       "step": 2428
     },
     {
       "epoch": 5.344334433443344,
-      "grad_norm": 0.0751953125,
+      "grad_norm": 0.00286865234375,
       "learning_rate": 8.93402636852251e-05,
-      "loss": 0.0007,
+      "loss": 0.0001,
       "step": 2429
     },
     {
       "epoch": 5.346534653465347,
-      "grad_norm": 0.00823974609375,
+      "grad_norm": 0.0028839111328125,
       "learning_rate": 8.927131056534733e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2430
     },
     {
       "epoch": 5.3487348734873486,
-      "grad_norm": 0.00799560546875,
+      "grad_norm": 0.0035858154296875,
       "learning_rate": 8.920236260547305e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2431
     },
     {
       "epoch": 5.350935093509351,
-      "grad_norm": 0.01318359375,
+      "grad_norm": 0.033203125,
       "learning_rate": 8.913341983876307e-05,
-      "loss": 0.0002,
+      "loss": 0.0004,
       "step": 2432
     },
     {
       "epoch": 5.353135313531353,
-      "grad_norm": 0.1953125,
+      "grad_norm": 0.0174560546875,
       "learning_rate": 8.90644822983757e-05,
-      "loss": 0.001,
+      "loss": 0.0002,
       "step": 2433
     },
     {
       "epoch": 5.3553355335533555,
-      "grad_norm": 0.004486083984375,
+      "grad_norm": 0.004669189453125,
       "learning_rate": 8.899555001746664e-05,
       "loss": 0.0002,
       "step": 2434
@@ -17226,432 +17226,432 @@
       "epoch": 5.357535753575357,
       "grad_norm": 0.00433349609375,
       "learning_rate": 8.892662302918922e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2435
     },
     {
       "epoch": 5.35973597359736,
-      "grad_norm": 0.00775146484375,
+      "grad_norm": 0.004150390625,
       "learning_rate": 8.885770136669406e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2436
     },
     {
       "epoch": 5.361936193619362,
-      "grad_norm": 0.0079345703125,
+      "grad_norm": 0.037841796875,
       "learning_rate": 8.87887850631293e-05,
       "loss": 0.0002,
       "step": 2437
     },
     {
       "epoch": 5.364136413641364,
-      "grad_norm": 0.0135498046875,
+      "grad_norm": 0.0032958984375,
       "learning_rate": 8.871987415164055e-05,
-      "loss": 0.0003,
+      "loss": 0.0001,
       "step": 2438
     },
     {
       "epoch": 5.366336633663367,
-      "grad_norm": 0.01068115234375,
+      "grad_norm": 0.0091552734375,
       "learning_rate": 8.865096866537071e-05,
       "loss": 0.0002,
       "step": 2439
     },
     {
       "epoch": 5.368536853685368,
-      "grad_norm": 0.0072021484375,
+      "grad_norm": 0.00286865234375,
       "learning_rate": 8.858206863746018e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2440
     },
     {
       "epoch": 5.370737073707371,
-      "grad_norm": 0.0047607421875,
+      "grad_norm": 0.00823974609375,
       "learning_rate": 8.851317410104665e-05,
       "loss": 0.0002,
       "step": 2441
     },
     {
       "epoch": 5.372937293729373,
-      "grad_norm": 0.00732421875,
+      "grad_norm": 0.0037078857421875,
       "learning_rate": 8.84442850892652e-05,
       "loss": 0.0002,
       "step": 2442
     },
     {
       "epoch": 5.375137513751375,
-      "grad_norm": 0.00787353515625,
+      "grad_norm": 0.00616455078125,
       "learning_rate": 8.837540163524833e-05,
       "loss": 0.0002,
       "step": 2443
     },
     {
       "epoch": 5.377337733773377,
-      "grad_norm": 0.0279541015625,
+      "grad_norm": 0.0040283203125,
       "learning_rate": 8.83065237721257e-05,
-      "loss": 0.0003,
+      "loss": 0.0001,
       "step": 2444
     },
     {
       "epoch": 5.37953795379538,
-      "grad_norm": 0.042724609375,
+      "grad_norm": 0.00982666015625,
       "learning_rate": 8.823765153302447e-05,
-      "loss": 0.0004,
+      "loss": 0.0003,
       "step": 2445
     },
     {
       "epoch": 5.381738173817382,
-      "grad_norm": 0.0029144287109375,
+      "grad_norm": 0.0030670166015625,
       "learning_rate": 8.816878495106893e-05,
       "loss": 0.0001,
       "step": 2446
     },
     {
       "epoch": 5.383938393839384,
-      "grad_norm": 0.2314453125,
+      "grad_norm": 0.00531005859375,
       "learning_rate": 8.809992405938076e-05,
-      "loss": 0.0046,
+      "loss": 0.0002,
       "step": 2447
     },
     {
       "epoch": 5.3861386138613865,
-      "grad_norm": 0.00860595703125,
+      "grad_norm": 0.041015625,
       "learning_rate": 8.803106889107883e-05,
-      "loss": 0.0002,
+      "loss": 0.0005,
       "step": 2448
     },
     {
       "epoch": 5.388338833883388,
-      "grad_norm": 0.00396728515625,
+      "grad_norm": 0.0054931640625,
       "learning_rate": 8.796221947927932e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2449
     },
     {
       "epoch": 5.390539053905391,
-      "grad_norm": 0.004638671875,
+      "grad_norm": 0.004180908203125,
       "learning_rate": 8.789337585709564e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2450
     },
     {
       "epoch": 5.3927392739273925,
-      "grad_norm": 0.009765625,
+      "grad_norm": 0.0101318359375,
       "learning_rate": 8.782453805763833e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2451
     },
     {
       "epoch": 5.394939493949395,
-      "grad_norm": 0.0169677734375,
+      "grad_norm": 0.0213623046875,
       "learning_rate": 8.775570611401523e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2452
     },
     {
       "epoch": 5.397139713971397,
-      "grad_norm": 0.007049560546875,
+      "grad_norm": 0.005859375,
       "learning_rate": 8.76868800593313e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2453
     },
     {
       "epoch": 5.399339933993399,
-      "grad_norm": 0.003692626953125,
+      "grad_norm": 0.00616455078125,
       "learning_rate": 8.761805992668869e-05,
       "loss": 0.0002,
       "step": 2454
     },
     {
       "epoch": 5.401540154015402,
-      "grad_norm": 0.0037384033203125,
+      "grad_norm": 0.0025482177734375,
       "learning_rate": 8.754924574918675e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2455
     },
     {
       "epoch": 5.403740374037404,
-      "grad_norm": 0.01495361328125,
+      "grad_norm": 0.002410888671875,
       "learning_rate": 8.748043755992182e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2456
     },
     {
       "epoch": 5.405940594059406,
-      "grad_norm": 0.049072265625,
+      "grad_norm": 0.007720947265625,
       "learning_rate": 8.741163539198755e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2457
     },
     {
       "epoch": 5.408140814081408,
-      "grad_norm": 0.01141357421875,
+      "grad_norm": 0.0140380859375,
       "learning_rate": 8.734283927847456e-05,
       "loss": 0.0002,
       "step": 2458
     },
     {
       "epoch": 5.410341034103411,
-      "grad_norm": 0.03662109375,
+      "grad_norm": 0.00311279296875,
       "learning_rate": 8.727404925247058e-05,
-      "loss": 0.0003,
+      "loss": 0.0001,
       "step": 2459
     },
     {
       "epoch": 5.412541254125412,
-      "grad_norm": 0.2890625,
+      "grad_norm": 0.006103515625,
       "learning_rate": 8.72052653470605e-05,
-      "loss": 0.0016,
+      "loss": 0.0002,
       "step": 2460
     },
     {
       "epoch": 5.414741474147415,
-      "grad_norm": 0.006195068359375,
+      "grad_norm": 0.00469970703125,
       "learning_rate": 8.713648759532611e-05,
       "loss": 0.0002,
       "step": 2461
     },
     {
       "epoch": 5.416941694169417,
-      "grad_norm": 0.08251953125,
+      "grad_norm": 0.0037078857421875,
       "learning_rate": 8.706771603034642e-05,
-      "loss": 0.0008,
+      "loss": 0.0001,
       "step": 2462
     },
     {
       "epoch": 5.419141914191419,
-      "grad_norm": 0.00616455078125,
+      "grad_norm": 0.014892578125,
       "learning_rate": 8.69989506851973e-05,
       "loss": 0.0001,
       "step": 2463
     },
     {
       "epoch": 5.421342134213421,
-      "grad_norm": 0.0072021484375,
+      "grad_norm": 0.004302978515625,
       "learning_rate": 8.693019159295176e-05,
       "loss": 0.0001,
       "step": 2464
     },
     {
       "epoch": 5.4235423542354235,
-      "grad_norm": 0.03662109375,
+      "grad_norm": 0.007415771484375,
       "learning_rate": 8.686143878667965e-05,
-      "loss": 0.0004,
+      "loss": 0.0002,
       "step": 2465
     },
     {
       "epoch": 5.425742574257426,
-      "grad_norm": 0.01495361328125,
+      "grad_norm": 0.0036468505859375,
       "learning_rate": 8.679269229944796e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2466
     },
     {
       "epoch": 5.427942794279428,
-      "grad_norm": 0.1552734375,
+      "grad_norm": 0.00274658203125,
       "learning_rate": 8.67239521643206e-05,
-      "loss": 0.0007,
+      "loss": 0.0001,
       "step": 2467
     },
     {
       "epoch": 5.43014301430143,
-      "grad_norm": 0.01263427734375,
+      "grad_norm": 0.00421142578125,
       "learning_rate": 8.665521841435835e-05,
       "loss": 0.0001,
       "step": 2468
     },
     {
       "epoch": 5.432343234323432,
-      "grad_norm": 0.004241943359375,
+      "grad_norm": 0.004486083984375,
       "learning_rate": 8.658649108261899e-05,
       "loss": 0.0002,
       "step": 2469
     },
     {
       "epoch": 5.434543454345435,
-      "grad_norm": 0.035400390625,
+      "grad_norm": 0.011962890625,
       "learning_rate": 8.651777020215712e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2470
     },
     {
       "epoch": 5.436743674367436,
-      "grad_norm": 0.00311279296875,
+      "grad_norm": 0.0032806396484375,
       "learning_rate": 8.644905580602436e-05,
       "loss": 0.0001,
       "step": 2471
     },
     {
       "epoch": 5.438943894389439,
-      "grad_norm": 0.00860595703125,
+      "grad_norm": 0.004180908203125,
       "learning_rate": 8.638034792726923e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2472
     },
     {
       "epoch": 5.441144114411441,
-      "grad_norm": 0.0142822265625,
+      "grad_norm": 0.003662109375,
       "learning_rate": 8.631164659893692e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2473
     },
     {
       "epoch": 5.443344334433443,
-      "grad_norm": 0.022216796875,
+      "grad_norm": 0.006744384765625,
       "learning_rate": 8.624295185406964e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2474
     },
     {
       "epoch": 5.445544554455446,
-      "grad_norm": 0.028564453125,
+      "grad_norm": 0.005279541015625,
       "learning_rate": 8.61742637257064e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2475
     },
     {
       "epoch": 5.447744774477448,
-      "grad_norm": 0.21875,
+      "grad_norm": 0.020751953125,
       "learning_rate": 8.610558224688297e-05,
-      "loss": 0.0037,
+      "loss": 0.0002,
       "step": 2476
     },
     {
       "epoch": 5.44994499449945,
-      "grad_norm": 0.022705078125,
+      "grad_norm": 0.0032806396484375,
       "learning_rate": 8.603690745063203e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2477
     },
     {
       "epoch": 5.452145214521452,
-      "grad_norm": 0.04345703125,
+      "grad_norm": 0.004852294921875,
       "learning_rate": 8.596823936998292e-05,
-      "loss": 0.0003,
+      "loss": 0.0001,
       "step": 2478
     },
     {
       "epoch": 5.4543454345434546,
-      "grad_norm": 0.0034027099609375,
+      "grad_norm": 0.0033721923828125,
       "learning_rate": 8.589957803796187e-05,
       "loss": 0.0001,
       "step": 2479
     },
     {
       "epoch": 5.456545654565456,
-      "grad_norm": 0.021728515625,
+      "grad_norm": 0.004730224609375,
       "learning_rate": 8.583092348759176e-05,
       "loss": 0.0002,
       "step": 2480
     },
     {
       "epoch": 5.458745874587459,
-      "grad_norm": 0.019287109375,
+      "grad_norm": 0.003936767578125,
       "learning_rate": 8.576227575189225e-05,
-      "loss": 0.0004,
+      "loss": 0.0001,
       "step": 2481
     },
     {
       "epoch": 5.460946094609461,
-      "grad_norm": 0.005767822265625,
+      "grad_norm": 0.0191650390625,
       "learning_rate": 8.569363486387979e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2482
     },
     {
       "epoch": 5.463146314631463,
-      "grad_norm": 0.00921630859375,
+      "grad_norm": 0.00469970703125,
       "learning_rate": 8.562500085656737e-05,
       "loss": 0.0002,
       "step": 2483
     },
     {
       "epoch": 5.465346534653466,
-      "grad_norm": 0.00628662109375,
+      "grad_norm": 0.005828857421875,
       "learning_rate": 8.555637376296489e-05,
       "loss": 0.0002,
       "step": 2484
     },
     {
       "epoch": 5.4675467546754675,
-      "grad_norm": 0.025390625,
+      "grad_norm": 0.0033416748046875,
       "learning_rate": 8.548775361607872e-05,
-      "loss": 0.0003,
+      "loss": 0.0001,
       "step": 2485
     },
     {
       "epoch": 5.46974697469747,
-      "grad_norm": 0.013671875,
+      "grad_norm": 0.026123046875,
       "learning_rate": 8.541914044891205e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2486
     },
     {
       "epoch": 5.471947194719472,
-      "grad_norm": 0.004119873046875,
+      "grad_norm": 0.004974365234375,
       "learning_rate": 8.535053429446453e-05,
       "loss": 0.0002,
       "step": 2487
     },
     {
       "epoch": 5.474147414741474,
-      "grad_norm": 0.004425048828125,
+      "grad_norm": 0.01373291015625,
       "learning_rate": 8.528193518573264e-05,
       "loss": 0.0002,
       "step": 2488
     },
     {
       "epoch": 5.476347634763476,
-      "grad_norm": 0.003021240234375,
+      "grad_norm": 0.00567626953125,
       "learning_rate": 8.521334315570939e-05,
       "loss": 0.0001,
       "step": 2489
     },
     {
       "epoch": 5.478547854785479,
-      "grad_norm": 0.01123046875,
+      "grad_norm": 0.008544921875,
       "learning_rate": 8.514475823738432e-05,
       "loss": 0.0002,
       "step": 2490
     },
     {
       "epoch": 5.48074807480748,
-      "grad_norm": 0.01025390625,
+      "grad_norm": 0.002532958984375,
       "learning_rate": 8.507618046374364e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2491
     },
     {
       "epoch": 5.482948294829483,
-      "grad_norm": 0.004791259765625,
+      "grad_norm": 0.01043701171875,
       "learning_rate": 8.500760986777005e-05,
       "loss": 0.0002,
       "step": 2492
     },
     {
       "epoch": 5.485148514851485,
-      "grad_norm": 0.0169677734375,
+      "grad_norm": 0.003509521484375,
       "learning_rate": 8.493904648244283e-05,
       "loss": 0.0002,
       "step": 2493
     },
     {
       "epoch": 5.487348734873487,
-      "grad_norm": 0.0062255859375,
+      "grad_norm": 0.006744384765625,
       "learning_rate": 8.48704903407379e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2494
     },
     {
       "epoch": 5.48954895489549,
-      "grad_norm": 0.004119873046875,
+      "grad_norm": 0.005401611328125,
       "learning_rate": 8.480194147562748e-05,
       "loss": 0.0001,
       "step": 2495
     },
     {
       "epoch": 5.491749174917492,
-      "grad_norm": 0.0194091796875,
+      "grad_norm": 0.004913330078125,
       "learning_rate": 8.473339992008048e-05,
       "loss": 0.0002,
       "step": 2496
@@ -17660,2676 +17660,2676 @@
       "epoch": 5.493949394939494,
       "grad_norm": 0.004608154296875,
       "learning_rate": 8.466486570706214e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2497
     },
     {
       "epoch": 5.496149614961496,
-      "grad_norm": 0.0033721923828125,
+      "grad_norm": 0.007110595703125,
       "learning_rate": 8.459633886953428e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2498
     },
     {
       "epoch": 5.4983498349834985,
-      "grad_norm": 0.049560546875,
+      "grad_norm": 0.0029449462890625,
       "learning_rate": 8.45278194404552e-05,
-      "loss": 0.0004,
+      "loss": 0.0001,
       "step": 2499
     },
     {
       "epoch": 5.5005500550055,
-      "grad_norm": 0.0115966796875,
+      "grad_norm": 0.006622314453125,
       "learning_rate": 8.445930745277953e-05,
       "loss": 0.0002,
       "step": 2500
     },
     {
       "epoch": 5.502750275027503,
-      "grad_norm": 0.019287109375,
+      "grad_norm": 0.0029449462890625,
       "learning_rate": 8.439080293945834e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2501
     },
     {
       "epoch": 5.5049504950495045,
-      "grad_norm": 0.01806640625,
+      "grad_norm": 0.0031280517578125,
       "learning_rate": 8.432230593343917e-05,
-      "loss": 0.0003,
+      "loss": 0.0001,
       "step": 2502
     },
     {
       "epoch": 5.507150715071507,
-      "grad_norm": 0.07958984375,
+      "grad_norm": 0.0025177001953125,
       "learning_rate": 8.425381646766592e-05,
-      "loss": 0.0003,
+      "loss": 0.0001,
       "step": 2503
     },
     {
       "epoch": 5.50935093509351,
-      "grad_norm": 0.03955078125,
+      "grad_norm": 0.007080078125,
       "learning_rate": 8.41853345750788e-05,
-      "loss": 0.0005,
+      "loss": 0.0002,
       "step": 2504
     },
     {
       "epoch": 5.511551155115511,
-      "grad_norm": 0.00555419921875,
+      "grad_norm": 0.004974365234375,
       "learning_rate": 8.411686028861447e-05,
       "loss": 0.0001,
       "step": 2505
     },
     {
       "epoch": 5.513751375137514,
-      "grad_norm": 0.0341796875,
+      "grad_norm": 0.007659912109375,
       "learning_rate": 8.404839364120595e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2506
     },
     {
       "epoch": 5.515951595159516,
-      "grad_norm": 0.006011962890625,
+      "grad_norm": 0.00994873046875,
       "learning_rate": 8.397993466578246e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2507
     },
     {
       "epoch": 5.518151815181518,
-      "grad_norm": 0.22265625,
+      "grad_norm": 0.00347900390625,
       "learning_rate": 8.391148339526964e-05,
-      "loss": 0.0013,
+      "loss": 0.0002,
       "step": 2508
     },
     {
       "epoch": 5.518151815181518,
-      "eval_loss": 0.0003648173005785793,
-      "eval_runtime": 10.2871,
-      "eval_samples_per_second": 37.231,
-      "eval_steps_per_second": 4.666,
+      "eval_loss": 0.00043490156531333923,
+      "eval_runtime": 10.1854,
+      "eval_samples_per_second": 37.603,
+      "eval_steps_per_second": 4.713,
       "step": 2508
     },
     {
       "epoch": 5.52035203520352,
-      "grad_norm": 0.0054931640625,
+      "grad_norm": 0.004425048828125,
       "learning_rate": 8.384303986258932e-05,
       "loss": 0.0002,
       "step": 2509
     },
     {
       "epoch": 5.522552255225523,
-      "grad_norm": 0.08935546875,
+      "grad_norm": 0.0035400390625,
       "learning_rate": 8.377460410065973e-05,
-      "loss": 0.0005,
+      "loss": 0.0001,
       "step": 2510
     },
     {
       "epoch": 5.524752475247524,
-      "grad_norm": 0.0086669921875,
+      "grad_norm": 0.0057373046875,
       "learning_rate": 8.370617614239533e-05,
       "loss": 0.0002,
       "step": 2511
     },
     {
       "epoch": 5.526952695269527,
-      "grad_norm": 0.00848388671875,
+      "grad_norm": 0.0191650390625,
       "learning_rate": 8.363775602070676e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2512
     },
     {
       "epoch": 5.5291529152915295,
-      "grad_norm": 0.0033111572265625,
+      "grad_norm": 0.005889892578125,
       "learning_rate": 8.356934376850097e-05,
       "loss": 0.0001,
       "step": 2513
     },
     {
       "epoch": 5.531353135313531,
-      "grad_norm": 0.0038604736328125,
+      "grad_norm": 0.0031585693359375,
       "learning_rate": 8.350093941868101e-05,
       "loss": 0.0001,
       "step": 2514
     },
     {
       "epoch": 5.533553355335534,
-      "grad_norm": 0.0159912109375,
+      "grad_norm": 0.0034942626953125,
       "learning_rate": 8.343254300414628e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2515
     },
     {
       "epoch": 5.5357535753575355,
-      "grad_norm": 0.1669921875,
+      "grad_norm": 0.005279541015625,
       "learning_rate": 8.336415455779232e-05,
-      "loss": 0.0016,
+      "loss": 0.0001,
       "step": 2516
     },
     {
       "epoch": 5.537953795379538,
-      "grad_norm": 0.004547119140625,
+      "grad_norm": 0.006622314453125,
       "learning_rate": 8.329577411251075e-05,
       "loss": 0.0002,
       "step": 2517
     },
     {
       "epoch": 5.54015401540154,
-      "grad_norm": 0.005035400390625,
+      "grad_norm": 0.0166015625,
       "learning_rate": 8.322740170118945e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2518
     },
     {
       "epoch": 5.542354235423542,
-      "grad_norm": 0.004791259765625,
+      "grad_norm": 0.0029144287109375,
       "learning_rate": 8.31590373567123e-05,
       "loss": 0.0002,
       "step": 2519
     },
     {
       "epoch": 5.544554455445544,
-      "grad_norm": 0.0059814453125,
+      "grad_norm": 0.00250244140625,
       "learning_rate": 8.309068111195945e-05,
       "loss": 0.0001,
       "step": 2520
     },
     {
       "epoch": 5.546754675467547,
-      "grad_norm": 0.018798828125,
+      "grad_norm": 0.0067138671875,
       "learning_rate": 8.302233299980715e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2521
     },
     {
       "epoch": 5.548954895489549,
-      "grad_norm": 0.0050048828125,
+      "grad_norm": 0.0029296875,
       "learning_rate": 8.29539930531276e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2522
     },
     {
       "epoch": 5.551155115511551,
-      "grad_norm": 0.007354736328125,
+      "grad_norm": 0.0033416748046875,
       "learning_rate": 8.288566130478919e-05,
       "loss": 0.0002,
       "step": 2523
     },
     {
       "epoch": 5.553355335533554,
-      "grad_norm": 0.310546875,
+      "grad_norm": 0.00604248046875,
       "learning_rate": 8.281733778765628e-05,
-      "loss": 0.0015,
+      "loss": 0.0001,
       "step": 2524
     },
     {
       "epoch": 5.555555555555555,
-      "grad_norm": 0.005950927734375,
+      "grad_norm": 0.008544921875,
       "learning_rate": 8.274902253458939e-05,
       "loss": 0.0002,
       "step": 2525
     },
     {
       "epoch": 5.557755775577558,
-      "grad_norm": 0.0098876953125,
+      "grad_norm": 0.002838134765625,
       "learning_rate": 8.268071557844494e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2526
     },
     {
       "epoch": 5.55995599559956,
-      "grad_norm": 0.053955078125,
+      "grad_norm": 0.01519775390625,
       "learning_rate": 8.261241695207546e-05,
-      "loss": 0.0008,
+      "loss": 0.0004,
       "step": 2527
     },
     {
       "epoch": 5.562156215621562,
-      "grad_norm": 0.00433349609375,
+      "grad_norm": 0.00799560546875,
       "learning_rate": 8.254412668832942e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2528
     },
     {
       "epoch": 5.564356435643564,
-      "grad_norm": 0.0133056640625,
+      "grad_norm": 0.0031890869140625,
       "learning_rate": 8.247584482005129e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2529
     },
     {
       "epoch": 5.566556655665567,
-      "grad_norm": 0.003509521484375,
+      "grad_norm": 0.01483154296875,
       "learning_rate": 8.240757138008149e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2530
     },
     {
       "epoch": 5.568756875687569,
-      "grad_norm": 0.0030670166015625,
+      "grad_norm": 0.01190185546875,
       "learning_rate": 8.233930640125634e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2531
     },
     {
       "epoch": 5.570957095709571,
-      "grad_norm": 0.0081787109375,
+      "grad_norm": 0.028076171875,
       "learning_rate": 8.227104991640815e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2532
     },
     {
       "epoch": 5.5731573157315735,
-      "grad_norm": 0.00469970703125,
+      "grad_norm": 0.004913330078125,
       "learning_rate": 8.220280195836522e-05,
       "loss": 0.0001,
       "step": 2533
     },
     {
       "epoch": 5.575357535753575,
-      "grad_norm": 0.01202392578125,
+      "grad_norm": 0.01214599609375,
       "learning_rate": 8.213456255995157e-05,
       "loss": 0.0002,
       "step": 2534
     },
     {
       "epoch": 5.577557755775578,
-      "grad_norm": 0.00909423828125,
+      "grad_norm": 0.002410888671875,
       "learning_rate": 8.206633175398725e-05,
       "loss": 0.0001,
       "step": 2535
     },
     {
       "epoch": 5.5797579757975795,
-      "grad_norm": 0.004669189453125,
+      "grad_norm": 0.003662109375,
       "learning_rate": 8.199810957328803e-05,
       "loss": 0.0001,
       "step": 2536
     },
     {
       "epoch": 5.581958195819582,
-      "grad_norm": 0.007781982421875,
+      "grad_norm": 0.00830078125,
       "learning_rate": 8.192989605066572e-05,
       "loss": 0.0002,
       "step": 2537
     },
     {
       "epoch": 5.584158415841584,
-      "grad_norm": 0.0140380859375,
+      "grad_norm": 0.005401611328125,
       "learning_rate": 8.186169121892789e-05,
       "loss": 0.0002,
       "step": 2538
     },
     {
       "epoch": 5.586358635863586,
-      "grad_norm": 0.0269775390625,
+      "grad_norm": 0.006011962890625,
       "learning_rate": 8.179349511087781e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2539
     },
     {
       "epoch": 5.588558855885589,
-      "grad_norm": 0.302734375,
+      "grad_norm": 0.007110595703125,
       "learning_rate": 8.172530775931474e-05,
-      "loss": 0.0006,
+      "loss": 0.0002,
       "step": 2540
     },
     {
       "epoch": 5.590759075907591,
-      "grad_norm": 0.006561279296875,
+      "grad_norm": 0.005523681640625,
       "learning_rate": 8.165712919703357e-05,
       "loss": 0.0002,
       "step": 2541
     },
     {
       "epoch": 5.592959295929593,
-      "grad_norm": 0.0177001953125,
+      "grad_norm": 0.0037841796875,
       "learning_rate": 8.158895945682512e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2542
     },
     {
       "epoch": 5.595159515951595,
-      "grad_norm": 0.010009765625,
+      "grad_norm": 0.00457763671875,
       "learning_rate": 8.152079857147578e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2543
     },
     {
       "epoch": 5.597359735973598,
-      "grad_norm": 0.004119873046875,
+      "grad_norm": 0.00634765625,
       "learning_rate": 8.145264657376788e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2544
     },
     {
       "epoch": 5.599559955995599,
-      "grad_norm": 0.0029449462890625,
+      "grad_norm": 0.003387451171875,
       "learning_rate": 8.138450349647936e-05,
       "loss": 0.0001,
       "step": 2545
     },
     {
       "epoch": 5.601760176017602,
-      "grad_norm": 0.00433349609375,
+      "grad_norm": 0.004425048828125,
       "learning_rate": 8.131636937238381e-05,
       "loss": 0.0001,
       "step": 2546
     },
     {
       "epoch": 5.603960396039604,
-      "grad_norm": 0.01275634765625,
+      "grad_norm": 0.00518798828125,
       "learning_rate": 8.124824423425071e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2547
     },
     {
       "epoch": 5.606160616061606,
-      "grad_norm": 0.00811767578125,
+      "grad_norm": 0.00701904296875,
       "learning_rate": 8.1180128114845e-05,
       "loss": 0.0001,
       "step": 2548
     },
     {
       "epoch": 5.608360836083609,
-      "grad_norm": 0.02880859375,
+      "grad_norm": 0.0174560546875,
       "learning_rate": 8.111202104692746e-05,
       "loss": 0.0002,
       "step": 2549
     },
     {
       "epoch": 5.6105610561056105,
-      "grad_norm": 0.0042724609375,
+      "grad_norm": 0.00457763671875,
       "learning_rate": 8.104392306325442e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2550
     },
     {
       "epoch": 5.612761276127613,
-      "grad_norm": 0.003204345703125,
+      "grad_norm": 0.0036773681640625,
       "learning_rate": 8.09758341965778e-05,
       "loss": 0.0001,
       "step": 2551
     },
     {
       "epoch": 5.614961496149615,
-      "grad_norm": 0.044189453125,
+      "grad_norm": 0.006866455078125,
       "learning_rate": 8.090775447964531e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2552
     },
     {
       "epoch": 5.617161716171617,
-      "grad_norm": 0.004638671875,
+      "grad_norm": 0.00701904296875,
       "learning_rate": 8.083968394520003e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2553
     },
     {
       "epoch": 5.619361936193619,
-      "grad_norm": 0.00579833984375,
+      "grad_norm": 0.026123046875,
       "learning_rate": 8.07716226259808e-05,
       "loss": 0.0002,
       "step": 2554
     },
     {
       "epoch": 5.621562156215622,
-      "grad_norm": 0.00457763671875,
+      "grad_norm": 0.0159912109375,
       "learning_rate": 8.070357055472197e-05,
-      "loss": 0.0001,
+      "loss": 0.0003,
       "step": 2555
     },
     {
       "epoch": 5.623762376237623,
-      "grad_norm": 0.01348876953125,
+      "grad_norm": 0.003082275390625,
       "learning_rate": 8.063552776415343e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2556
     },
     {
       "epoch": 5.625962596259626,
-      "grad_norm": 0.0294189453125,
+      "grad_norm": 0.00933837890625,
       "learning_rate": 8.056749428700063e-05,
       "loss": 0.0002,
       "step": 2557
     },
     {
       "epoch": 5.628162816281629,
-      "grad_norm": 0.2158203125,
+      "grad_norm": 0.00299072265625,
       "learning_rate": 8.049947015598447e-05,
-      "loss": 0.0006,
+      "loss": 0.0001,
       "step": 2558
     },
     {
       "epoch": 5.63036303630363,
-      "grad_norm": 0.002716064453125,
+      "grad_norm": 0.002960205078125,
       "learning_rate": 8.043145540382147e-05,
       "loss": 0.0001,
       "step": 2559
     },
     {
       "epoch": 5.632563256325633,
-      "grad_norm": 0.005584716796875,
+      "grad_norm": 0.043212890625,
       "learning_rate": 8.036345006322359e-05,
       "loss": 0.0002,
       "step": 2560
     },
     {
       "epoch": 5.634763476347635,
-      "grad_norm": 0.0185546875,
+      "grad_norm": 0.00396728515625,
       "learning_rate": 8.029545416689821e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2561
     },
     {
       "epoch": 5.636963696369637,
-      "grad_norm": 0.019775390625,
+      "grad_norm": 0.005584716796875,
       "learning_rate": 8.022746774754824e-05,
       "loss": 0.0002,
       "step": 2562
     },
     {
       "epoch": 5.639163916391639,
-      "grad_norm": 0.00360107421875,
+      "grad_norm": 0.01153564453125,
       "learning_rate": 8.015949083787196e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2563
     },
     {
       "epoch": 5.6413641364136415,
-      "grad_norm": 0.00408935546875,
+      "grad_norm": 0.003387451171875,
       "learning_rate": 8.00915234705632e-05,
       "loss": 0.0001,
       "step": 2564
     },
     {
       "epoch": 5.643564356435643,
-      "grad_norm": 0.009765625,
+      "grad_norm": 0.0030670166015625,
       "learning_rate": 8.002356567831103e-05,
       "loss": 0.0001,
       "step": 2565
     },
     {
       "epoch": 5.645764576457646,
-      "grad_norm": 0.0034027099609375,
+      "grad_norm": 0.04443359375,
       "learning_rate": 7.995561749380007e-05,
-      "loss": 0.0002,
+      "loss": 0.0006,
       "step": 2566
     },
     {
       "epoch": 5.647964796479648,
-      "grad_norm": 0.0035858154296875,
+      "grad_norm": 0.03857421875,
       "learning_rate": 7.988767894971027e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2567
     },
     {
       "epoch": 5.65016501650165,
-      "grad_norm": 0.005340576171875,
+      "grad_norm": 0.00433349609375,
       "learning_rate": 7.981975007871683e-05,
       "loss": 0.0002,
       "step": 2568
     },
     {
       "epoch": 5.652365236523653,
-      "grad_norm": 0.015869140625,
+      "grad_norm": 0.03125,
       "learning_rate": 7.975183091349051e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2569
     },
     {
       "epoch": 5.6545654565456545,
-      "grad_norm": 0.1298828125,
+      "grad_norm": 0.0029144287109375,
       "learning_rate": 7.96839214866972e-05,
-      "loss": 0.0005,
+      "loss": 0.0001,
       "step": 2570
     },
     {
       "epoch": 5.656765676567657,
-      "grad_norm": 0.02783203125,
+      "grad_norm": 0.0093994140625,
       "learning_rate": 7.961602183099825e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2571
     },
     {
       "epoch": 5.658965896589659,
-      "grad_norm": 0.003509521484375,
+      "grad_norm": 0.01544189453125,
       "learning_rate": 7.954813197905025e-05,
-      "loss": 0.0001,
+      "loss": 0.0003,
       "step": 2572
     },
     {
       "epoch": 5.661166116611661,
-      "grad_norm": 0.00506591796875,
+      "grad_norm": 0.00714111328125,
       "learning_rate": 7.948025196350504e-05,
       "loss": 0.0002,
       "step": 2573
     },
     {
       "epoch": 5.663366336633663,
-      "grad_norm": 0.00335693359375,
+      "grad_norm": 0.00592041015625,
       "learning_rate": 7.941238181700982e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2574
     },
     {
       "epoch": 5.665566556655666,
-      "grad_norm": 0.0050048828125,
+      "grad_norm": 0.004058837890625,
       "learning_rate": 7.934452157220694e-05,
       "loss": 0.0001,
       "step": 2575
     },
     {
       "epoch": 5.667766776677668,
-      "grad_norm": 0.00238037109375,
+      "grad_norm": 0.01177978515625,
       "learning_rate": 7.927667126173405e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2576
     },
     {
       "epoch": 5.66996699669967,
-      "grad_norm": 0.041748046875,
+      "grad_norm": 0.0103759765625,
       "learning_rate": 7.920883091822408e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2577
     },
     {
       "epoch": 5.672167216721672,
-      "grad_norm": 0.002960205078125,
+      "grad_norm": 0.00555419921875,
       "learning_rate": 7.914100057430501e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2578
     },
     {
       "epoch": 5.674367436743674,
-      "grad_norm": 0.005889892578125,
+      "grad_norm": 0.0096435546875,
       "learning_rate": 7.907318026260018e-05,
       "loss": 0.0002,
       "step": 2579
     },
     {
       "epoch": 5.676567656765677,
-      "grad_norm": 0.00567626953125,
+      "grad_norm": 0.042236328125,
       "learning_rate": 7.900537001572792e-05,
-      "loss": 0.0002,
+      "loss": 0.0006,
       "step": 2580
     },
     {
       "epoch": 5.678767876787679,
-      "grad_norm": 0.0029449462890625,
+      "grad_norm": 0.003204345703125,
       "learning_rate": 7.893756986630193e-05,
       "loss": 0.0001,
       "step": 2581
     },
     {
       "epoch": 5.680968096809681,
-      "grad_norm": 0.005828857421875,
+      "grad_norm": 0.007110595703125,
       "learning_rate": 7.886977984693085e-05,
       "loss": 0.0002,
       "step": 2582
     },
     {
       "epoch": 5.683168316831683,
-      "grad_norm": 0.0033416748046875,
+      "grad_norm": 0.00408935546875,
       "learning_rate": 7.880199999021862e-05,
       "loss": 0.0001,
       "step": 2583
     },
     {
       "epoch": 5.6853685368536855,
-      "grad_norm": 0.0028228759765625,
+      "grad_norm": 0.006103515625,
       "learning_rate": 7.873423032876422e-05,
       "loss": 0.0001,
       "step": 2584
     },
     {
       "epoch": 5.687568756875687,
-      "grad_norm": 0.004119873046875,
+      "grad_norm": 0.005279541015625,
       "learning_rate": 7.866647089516164e-05,
       "loss": 0.0002,
       "step": 2585
     },
     {
       "epoch": 5.68976897689769,
-      "grad_norm": 0.0038299560546875,
+      "grad_norm": 0.00604248046875,
       "learning_rate": 7.859872172200015e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2586
     },
     {
       "epoch": 5.6919691969196915,
-      "grad_norm": 0.0084228515625,
+      "grad_norm": 0.052001953125,
       "learning_rate": 7.853098284186385e-05,
-      "loss": 0.0001,
+      "loss": 0.0006,
       "step": 2587
     },
     {
       "epoch": 5.694169416941694,
-      "grad_norm": 0.01116943359375,
+      "grad_norm": 0.007598876953125,
       "learning_rate": 7.846325428733209e-05,
       "loss": 0.0002,
       "step": 2588
     },
     {
       "epoch": 5.696369636963697,
-      "grad_norm": 0.003143310546875,
+      "grad_norm": 0.00299072265625,
       "learning_rate": 7.839553609097918e-05,
       "loss": 0.0001,
       "step": 2589
     },
     {
       "epoch": 5.698569856985698,
-      "grad_norm": 0.004425048828125,
+      "grad_norm": 0.0164794921875,
       "learning_rate": 7.832782828537437e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2590
     },
     {
       "epoch": 5.700770077007701,
-      "grad_norm": 0.002960205078125,
+      "grad_norm": 0.007537841796875,
       "learning_rate": 7.826013090308205e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2591
     },
     {
       "epoch": 5.702970297029703,
-      "grad_norm": 0.006683349609375,
+      "grad_norm": 0.006072998046875,
       "learning_rate": 7.819244397666148e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2592
     },
     {
       "epoch": 5.705170517051705,
-      "grad_norm": 0.00225830078125,
+      "grad_norm": 0.003692626953125,
       "learning_rate": 7.812476753866698e-05,
       "loss": 0.0001,
       "step": 2593
     },
     {
       "epoch": 5.707370737073707,
-      "grad_norm": 0.00628662109375,
+      "grad_norm": 0.004302978515625,
       "learning_rate": 7.805710162164778e-05,
       "loss": 0.0001,
       "step": 2594
     },
     {
       "epoch": 5.70957095709571,
-      "grad_norm": 0.005767822265625,
+      "grad_norm": 0.00726318359375,
       "learning_rate": 7.798944625814803e-05,
       "loss": 0.0001,
       "step": 2595
     },
     {
       "epoch": 5.711771177117711,
-      "grad_norm": 0.039306640625,
+      "grad_norm": 0.005401611328125,
       "learning_rate": 7.79218014807069e-05,
-      "loss": 0.0003,
+      "loss": 0.0001,
       "step": 2596
     },
     {
       "epoch": 5.713971397139714,
-      "grad_norm": 0.0035552978515625,
+      "grad_norm": 0.007171630859375,
       "learning_rate": 7.78541673218583e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2597
     },
     {
       "epoch": 5.7161716171617165,
-      "grad_norm": 0.0033416748046875,
+      "grad_norm": 0.007415771484375,
       "learning_rate": 7.778654381413121e-05,
       "loss": 0.0001,
       "step": 2598
     },
     {
       "epoch": 5.718371837183718,
-      "grad_norm": 0.01519775390625,
+      "grad_norm": 0.01531982421875,
       "learning_rate": 7.77189309900494e-05,
       "loss": 0.0002,
       "step": 2599
     },
     {
       "epoch": 5.720572057205721,
-      "grad_norm": 0.01043701171875,
+      "grad_norm": 0.0067138671875,
       "learning_rate": 7.765132888213147e-05,
       "loss": 0.0002,
       "step": 2600
     },
     {
       "epoch": 5.7227722772277225,
-      "grad_norm": 0.0059814453125,
+      "grad_norm": 0.003753662109375,
       "learning_rate": 7.758373752289098e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2601
     },
     {
       "epoch": 5.724972497249725,
-      "grad_norm": 0.01153564453125,
+      "grad_norm": 0.0111083984375,
       "learning_rate": 7.751615694483615e-05,
       "loss": 0.0002,
       "step": 2602
     },
     {
       "epoch": 5.727172717271727,
-      "grad_norm": 0.0037078857421875,
+      "grad_norm": 0.005828857421875,
       "learning_rate": 7.744858718047022e-05,
       "loss": 0.0002,
       "step": 2603
     },
     {
       "epoch": 5.729372937293729,
-      "grad_norm": 0.0040283203125,
+      "grad_norm": 0.0230712890625,
       "learning_rate": 7.738102826229105e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2604
     },
     {
       "epoch": 5.731573157315731,
-      "grad_norm": 0.0107421875,
+      "grad_norm": 0.0033721923828125,
       "learning_rate": 7.731348022279134e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2605
     },
     {
       "epoch": 5.733773377337734,
-      "grad_norm": 0.00732421875,
+      "grad_norm": 0.00360107421875,
       "learning_rate": 7.724594309445866e-05,
       "loss": 0.0002,
       "step": 2606
     },
     {
       "epoch": 5.735973597359736,
-      "grad_norm": 0.058349609375,
+      "grad_norm": 0.0022430419921875,
       "learning_rate": 7.717841690977515e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2607
     },
     {
       "epoch": 5.738173817381738,
-      "grad_norm": 0.0033111572265625,
+      "grad_norm": 0.006011962890625,
       "learning_rate": 7.711090170121787e-05,
       "loss": 0.0001,
       "step": 2608
     },
     {
       "epoch": 5.740374037403741,
-      "grad_norm": 0.01324462890625,
+      "grad_norm": 0.003631591796875,
       "learning_rate": 7.704339750125845e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2609
     },
     {
       "epoch": 5.742574257425742,
-      "grad_norm": 0.0032958984375,
+      "grad_norm": 0.00274658203125,
       "learning_rate": 7.69759043423633e-05,
       "loss": 0.0001,
       "step": 2610
     },
     {
       "epoch": 5.744774477447745,
-      "grad_norm": 0.01007080078125,
+      "grad_norm": 0.002197265625,
       "learning_rate": 7.690842225699355e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2611
     },
     {
       "epoch": 5.746974697469747,
-      "grad_norm": 0.00555419921875,
+      "grad_norm": 0.004486083984375,
       "learning_rate": 7.68409512776049e-05,
       "loss": 0.0002,
       "step": 2612
     },
     {
       "epoch": 5.749174917491749,
-      "grad_norm": 0.0034027099609375,
+      "grad_norm": 0.034423828125,
       "learning_rate": 7.677349143664782e-05,
-      "loss": 0.0001,
+      "loss": 0.0003,
       "step": 2613
     },
     {
       "epoch": 5.751375137513751,
-      "grad_norm": 0.0029296875,
+      "grad_norm": 0.004241943359375,
       "learning_rate": 7.670604276656733e-05,
       "loss": 0.0002,
       "step": 2614
     },
     {
       "epoch": 5.7535753575357536,
-      "grad_norm": 0.003387451171875,
+      "grad_norm": 0.00628662109375,
       "learning_rate": 7.663860529980317e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2615
     },
     {
       "epoch": 5.755775577557756,
-      "grad_norm": 0.00531005859375,
+      "grad_norm": 0.00421142578125,
       "learning_rate": 7.657117906878963e-05,
       "loss": 0.0001,
       "step": 2616
     },
     {
       "epoch": 5.757975797579758,
-      "grad_norm": 0.010009765625,
+      "grad_norm": 0.0067138671875,
       "learning_rate": 7.650376410595556e-05,
       "loss": 0.0002,
       "step": 2617
     },
     {
       "epoch": 5.7601760176017605,
-      "grad_norm": 0.003387451171875,
+      "grad_norm": 0.00341796875,
       "learning_rate": 7.643636044372453e-05,
       "loss": 0.0002,
       "step": 2618
     },
     {
       "epoch": 5.762376237623762,
-      "grad_norm": 0.0213623046875,
+      "grad_norm": 0.0059814453125,
       "learning_rate": 7.63689681145145e-05,
       "loss": 0.0002,
       "step": 2619
     },
     {
       "epoch": 5.764576457645765,
-      "grad_norm": 0.020263671875,
+      "grad_norm": 0.0224609375,
       "learning_rate": 7.630158715073813e-05,
       "loss": 0.0003,
       "step": 2620
     },
     {
       "epoch": 5.7667766776677665,
-      "grad_norm": 0.011962890625,
+      "grad_norm": 0.00555419921875,
       "learning_rate": 7.623421758480249e-05,
       "loss": 0.0001,
       "step": 2621
     },
     {
       "epoch": 5.768976897689769,
-      "grad_norm": 0.0023040771484375,
+      "grad_norm": 0.003570556640625,
       "learning_rate": 7.616685944910924e-05,
       "loss": 0.0001,
       "step": 2622
     },
     {
       "epoch": 5.768976897689769,
-      "eval_loss": 0.0002637207508087158,
-      "eval_runtime": 10.168,
-      "eval_samples_per_second": 37.667,
-      "eval_steps_per_second": 4.721,
+      "eval_loss": 0.0003930090751964599,
+      "eval_runtime": 10.135,
+      "eval_samples_per_second": 37.79,
+      "eval_steps_per_second": 4.736,
       "step": 2622
     },
     {
       "epoch": 5.771177117711771,
-      "grad_norm": 0.002716064453125,
+      "grad_norm": 0.003997802734375,
       "learning_rate": 7.609951277605459e-05,
       "loss": 0.0001,
       "step": 2623
     },
     {
       "epoch": 5.773377337733773,
-      "grad_norm": 0.007080078125,
+      "grad_norm": 0.00628662109375,
       "learning_rate": 7.603217759802908e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2624
     },
     {
       "epoch": 5.775577557755776,
-      "grad_norm": 0.009521484375,
+      "grad_norm": 0.00640869140625,
       "learning_rate": 7.596485394741788e-05,
       "loss": 0.0002,
       "step": 2625
     },
     {
       "epoch": 5.777777777777778,
-      "grad_norm": 0.0032196044921875,
+      "grad_norm": 0.0036773681640625,
       "learning_rate": 7.589754185660055e-05,
       "loss": 0.0001,
       "step": 2626
     },
     {
       "epoch": 5.77997799779978,
-      "grad_norm": 0.013427734375,
+      "grad_norm": 0.0040283203125,
       "learning_rate": 7.583024135795105e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2627
     },
     {
       "epoch": 5.782178217821782,
-      "grad_norm": 0.0078125,
+      "grad_norm": 0.00885009765625,
       "learning_rate": 7.576295248383787e-05,
       "loss": 0.0002,
       "step": 2628
     },
     {
       "epoch": 5.784378437843785,
-      "grad_norm": 0.0031890869140625,
+      "grad_norm": 0.00439453125,
       "learning_rate": 7.569567526662377e-05,
       "loss": 0.0002,
       "step": 2629
     },
     {
       "epoch": 5.786578657865786,
-      "grad_norm": 0.01177978515625,
+      "grad_norm": 0.016357421875,
       "learning_rate": 7.56284097386661e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2630
     },
     {
       "epoch": 5.788778877887789,
-      "grad_norm": 0.00579833984375,
+      "grad_norm": 0.003936767578125,
       "learning_rate": 7.556115593231634e-05,
       "loss": 0.0002,
       "step": 2631
     },
     {
       "epoch": 5.790979097909791,
-      "grad_norm": 0.006683349609375,
+      "grad_norm": 0.00482177734375,
       "learning_rate": 7.54939138799205e-05,
       "loss": 0.0002,
       "step": 2632
     },
     {
       "epoch": 5.793179317931793,
-      "grad_norm": 0.0033721923828125,
+      "grad_norm": 0.0022735595703125,
       "learning_rate": 7.542668361381896e-05,
       "loss": 0.0001,
       "step": 2633
     },
     {
       "epoch": 5.795379537953796,
-      "grad_norm": 0.0201416015625,
+      "grad_norm": 0.0111083984375,
       "learning_rate": 7.535946516634627e-05,
       "loss": 0.0002,
       "step": 2634
     },
     {
       "epoch": 5.7975797579757975,
-      "grad_norm": 0.004364013671875,
+      "grad_norm": 0.01336669921875,
       "learning_rate": 7.52922585698315e-05,
       "loss": 0.0002,
       "step": 2635
     },
     {
       "epoch": 5.7997799779978,
-      "grad_norm": 0.004425048828125,
+      "grad_norm": 0.00341796875,
       "learning_rate": 7.522506385659782e-05,
       "loss": 0.0001,
       "step": 2636
     },
     {
       "epoch": 5.801980198019802,
-      "grad_norm": 0.022705078125,
+      "grad_norm": 0.007110595703125,
       "learning_rate": 7.515788105896285e-05,
       "loss": 0.0002,
       "step": 2637
     },
     {
       "epoch": 5.804180418041804,
-      "grad_norm": 0.08203125,
+      "grad_norm": 0.048095703125,
       "learning_rate": 7.509071020923838e-05,
-      "loss": 0.0004,
+      "loss": 0.0003,
       "step": 2638
     },
     {
       "epoch": 5.806380638063806,
-      "grad_norm": 0.0255126953125,
+      "grad_norm": 0.0020751953125,
       "learning_rate": 7.502355133973045e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2639
     },
     {
       "epoch": 5.808580858085809,
-      "grad_norm": 0.004302978515625,
+      "grad_norm": 0.002777099609375,
       "learning_rate": 7.495640448273946e-05,
       "loss": 0.0001,
       "step": 2640
     },
     {
       "epoch": 5.81078107810781,
-      "grad_norm": 0.0283203125,
+      "grad_norm": 0.0032196044921875,
       "learning_rate": 7.488926967055983e-05,
-      "loss": 0.0007,
+      "loss": 0.0001,
       "step": 2641
     },
     {
       "epoch": 5.812981298129813,
-      "grad_norm": 0.0111083984375,
+      "grad_norm": 0.00408935546875,
       "learning_rate": 7.482214693548042e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2642
     },
     {
       "epoch": 5.815181518151816,
-      "grad_norm": 0.01416015625,
+      "grad_norm": 0.0279541015625,
       "learning_rate": 7.475503630978406e-05,
       "loss": 0.0002,
       "step": 2643
     },
     {
       "epoch": 5.817381738173817,
-      "grad_norm": 0.0028533935546875,
+      "grad_norm": 0.005096435546875,
       "learning_rate": 7.468793782574787e-05,
       "loss": 0.0001,
       "step": 2644
     },
     {
       "epoch": 5.81958195819582,
-      "grad_norm": 0.00555419921875,
+      "grad_norm": 0.006195068359375,
       "learning_rate": 7.462085151564322e-05,
       "loss": 0.0002,
       "step": 2645
     },
     {
       "epoch": 5.821782178217822,
-      "grad_norm": 0.0034942626953125,
+      "grad_norm": 0.011962890625,
       "learning_rate": 7.455377741173539e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2646
     },
     {
       "epoch": 5.823982398239824,
-      "grad_norm": 0.0034027099609375,
+      "grad_norm": 0.0040283203125,
       "learning_rate": 7.448671554628401e-05,
       "loss": 0.0001,
       "step": 2647
     },
     {
       "epoch": 5.826182618261826,
-      "grad_norm": 0.00244140625,
+      "grad_norm": 0.0029754638671875,
       "learning_rate": 7.441966595154271e-05,
       "loss": 0.0001,
       "step": 2648
     },
     {
       "epoch": 5.8283828382838285,
-      "grad_norm": 0.017333984375,
+      "grad_norm": 0.00616455078125,
       "learning_rate": 7.43526286597592e-05,
       "loss": 0.0002,
       "step": 2649
     },
     {
       "epoch": 5.83058305830583,
-      "grad_norm": 0.005035400390625,
+      "grad_norm": 0.003265380859375,
       "learning_rate": 7.428560370317542e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2650
     },
     {
       "epoch": 5.832783278327833,
-      "grad_norm": 0.0189208984375,
+      "grad_norm": 0.00701904296875,
       "learning_rate": 7.421859111402715e-05,
       "loss": 0.0001,
       "step": 2651
     },
     {
       "epoch": 5.834983498349835,
-      "grad_norm": 0.0028076171875,
+      "grad_norm": 0.028076171875,
       "learning_rate": 7.415159092454449e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2652
     },
     {
       "epoch": 5.837183718371837,
-      "grad_norm": 0.006011962890625,
+      "grad_norm": 0.006134033203125,
       "learning_rate": 7.408460316695131e-05,
       "loss": 0.0001,
       "step": 2653
     },
     {
       "epoch": 5.83938393839384,
-      "grad_norm": 0.004638671875,
+      "grad_norm": 0.0027923583984375,
       "learning_rate": 7.401762787346564e-05,
       "loss": 0.0001,
       "step": 2654
     },
     {
       "epoch": 5.841584158415841,
-      "grad_norm": 0.0040283203125,
+      "grad_norm": 0.0283203125,
       "learning_rate": 7.39506650762996e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2655
     },
     {
       "epoch": 5.843784378437844,
-      "grad_norm": 0.010009765625,
+      "grad_norm": 0.00482177734375,
       "learning_rate": 7.388371480765905e-05,
       "loss": 0.0002,
       "step": 2656
     },
     {
       "epoch": 5.845984598459846,
-      "grad_norm": 0.0025177001953125,
+      "grad_norm": 0.00250244140625,
       "learning_rate": 7.381677709974412e-05,
       "loss": 0.0001,
       "step": 2657
     },
     {
       "epoch": 5.848184818481848,
-      "grad_norm": 0.0123291015625,
+      "grad_norm": 0.03857421875,
       "learning_rate": 7.374985198474864e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2658
     },
     {
       "epoch": 5.85038503850385,
-      "grad_norm": 0.0037841796875,
+      "grad_norm": 0.0047607421875,
       "learning_rate": 7.368293949486053e-05,
       "loss": 0.0001,
       "step": 2659
     },
     {
       "epoch": 5.852585258525853,
-      "grad_norm": 0.00848388671875,
+      "grad_norm": 0.01214599609375,
       "learning_rate": 7.361603966226164e-05,
       "loss": 0.0002,
       "step": 2660
     },
     {
       "epoch": 5.854785478547855,
-      "grad_norm": 0.007080078125,
+      "grad_norm": 0.0050048828125,
       "learning_rate": 7.354915251912765e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2661
     },
     {
       "epoch": 5.856985698569857,
-      "grad_norm": 0.0035858154296875,
+      "grad_norm": 0.0091552734375,
       "learning_rate": 7.348227809762822e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2662
     },
     {
       "epoch": 5.8591859185918596,
-      "grad_norm": 0.00567626953125,
+      "grad_norm": 0.0034027099609375,
       "learning_rate": 7.341541642992681e-05,
       "loss": 0.0001,
       "step": 2663
     },
     {
       "epoch": 5.861386138613861,
-      "grad_norm": 0.007720947265625,
+      "grad_norm": 0.00274658203125,
       "learning_rate": 7.334856754818085e-05,
       "loss": 0.0001,
       "step": 2664
     },
     {
       "epoch": 5.863586358635864,
-      "grad_norm": 0.0040283203125,
+      "grad_norm": 0.00555419921875,
       "learning_rate": 7.328173148454151e-05,
       "loss": 0.0001,
       "step": 2665
     },
     {
       "epoch": 5.865786578657866,
-      "grad_norm": 0.0030670166015625,
+      "grad_norm": 0.005096435546875,
       "learning_rate": 7.321490827115382e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2666
     },
     {
       "epoch": 5.867986798679868,
-      "grad_norm": 0.00396728515625,
+      "grad_norm": 0.005401611328125,
       "learning_rate": 7.314809794015677e-05,
       "loss": 0.0002,
       "step": 2667
     },
     {
       "epoch": 5.87018701870187,
-      "grad_norm": 0.003021240234375,
+      "grad_norm": 0.0032196044921875,
       "learning_rate": 7.308130052368291e-05,
       "loss": 0.0001,
       "step": 2668
     },
     {
       "epoch": 5.8723872387238725,
-      "grad_norm": 0.0079345703125,
+      "grad_norm": 0.005340576171875,
       "learning_rate": 7.301451605385881e-05,
       "loss": 0.0002,
       "step": 2669
     },
     {
       "epoch": 5.874587458745875,
-      "grad_norm": 0.0030670166015625,
+      "grad_norm": 0.02392578125,
       "learning_rate": 7.294774456280466e-05,
-      "loss": 0.0001,
+      "loss": 0.0004,
       "step": 2670
     },
     {
       "epoch": 5.876787678767877,
-      "grad_norm": 0.00946044921875,
+      "grad_norm": 0.00543212890625,
       "learning_rate": 7.288098608263443e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2671
     },
     {
       "epoch": 5.878987898789879,
-      "grad_norm": 0.1015625,
+      "grad_norm": 0.002838134765625,
       "learning_rate": 7.281424064545595e-05,
-      "loss": 0.0006,
+      "loss": 0.0001,
       "step": 2672
     },
     {
       "epoch": 5.881188118811881,
-      "grad_norm": 0.01043701171875,
+      "grad_norm": 0.01165771484375,
       "learning_rate": 7.27475082833706e-05,
       "loss": 0.0002,
       "step": 2673
     },
     {
       "epoch": 5.883388338833884,
-      "grad_norm": 0.01318359375,
+      "grad_norm": 0.0113525390625,
       "learning_rate": 7.268078902847364e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2674
     },
     {
       "epoch": 5.885588558855885,
-      "grad_norm": 0.0025177001953125,
+      "grad_norm": 0.006744384765625,
       "learning_rate": 7.261408291285389e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2675
     },
     {
       "epoch": 5.887788778877888,
-      "grad_norm": 0.002838134765625,
+      "grad_norm": 0.0026092529296875,
       "learning_rate": 7.25473899685939e-05,
       "loss": 0.0001,
       "step": 2676
     },
     {
       "epoch": 5.88998899889989,
-      "grad_norm": 0.01275634765625,
+      "grad_norm": 0.0025634765625,
       "learning_rate": 7.248071022776994e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2677
     },
     {
       "epoch": 5.892189218921892,
-      "grad_norm": 0.03173828125,
+      "grad_norm": 0.00439453125,
       "learning_rate": 7.241404372245184e-05,
-      "loss": 0.0003,
+      "loss": 0.0001,
       "step": 2678
     },
     {
       "epoch": 5.894389438943895,
-      "grad_norm": 0.010498046875,
+      "grad_norm": 0.00390625,
       "learning_rate": 7.234739048470316e-05,
       "loss": 0.0001,
       "step": 2679
     },
     {
       "epoch": 5.896589658965897,
-      "grad_norm": 0.0033416748046875,
+      "grad_norm": 0.00933837890625,
       "learning_rate": 7.228075054658096e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2680
     },
     {
       "epoch": 5.898789878987898,
-      "grad_norm": 0.004638671875,
+      "grad_norm": 0.01251220703125,
       "learning_rate": 7.221412394013602e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2681
     },
     {
       "epoch": 5.900990099009901,
-      "grad_norm": 0.007049560546875,
+      "grad_norm": 0.0145263671875,
       "learning_rate": 7.21475106974126e-05,
       "loss": 0.0002,
       "step": 2682
     },
     {
       "epoch": 5.9031903190319035,
-      "grad_norm": 0.11181640625,
+      "grad_norm": 0.00634765625,
       "learning_rate": 7.208091085044861e-05,
-      "loss": 0.0008,
+      "loss": 0.0002,
       "step": 2683
     },
     {
       "epoch": 5.905390539053905,
-      "grad_norm": 0.00640869140625,
+      "grad_norm": 0.01324462890625,
       "learning_rate": 7.201432443127556e-05,
       "loss": 0.0002,
       "step": 2684
     },
     {
       "epoch": 5.907590759075908,
-      "grad_norm": 0.003265380859375,
+      "grad_norm": 0.0030059814453125,
       "learning_rate": 7.194775147191839e-05,
       "loss": 0.0001,
       "step": 2685
     },
     {
       "epoch": 5.9097909790979095,
-      "grad_norm": 0.003753662109375,
+      "grad_norm": 0.0126953125,
       "learning_rate": 7.188119200439559e-05,
       "loss": 0.0001,
       "step": 2686
     },
     {
       "epoch": 5.911991199119912,
-      "grad_norm": 0.0064697265625,
+      "grad_norm": 0.003936767578125,
       "learning_rate": 7.181464606071923e-05,
       "loss": 0.0001,
       "step": 2687
     },
     {
       "epoch": 5.914191419141914,
-      "grad_norm": 0.003448486328125,
+      "grad_norm": 0.0084228515625,
       "learning_rate": 7.174811367289478e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2688
     },
     {
       "epoch": 5.916391639163916,
-      "grad_norm": 0.004425048828125,
+      "grad_norm": 0.0089111328125,
       "learning_rate": 7.168159487292132e-05,
       "loss": 0.0002,
       "step": 2689
     },
     {
       "epoch": 5.918591859185918,
-      "grad_norm": 0.003143310546875,
+      "grad_norm": 0.0032501220703125,
       "learning_rate": 7.161508969279122e-05,
       "loss": 0.0001,
       "step": 2690
     },
     {
       "epoch": 5.920792079207921,
-      "grad_norm": 0.0037994384765625,
+      "grad_norm": 0.003936767578125,
       "learning_rate": 7.15485981644905e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2691
     },
     {
       "epoch": 5.922992299229923,
-      "grad_norm": 0.00506591796875,
+      "grad_norm": 0.0057373046875,
       "learning_rate": 7.148212031999844e-05,
       "loss": 0.0002,
       "step": 2692
     },
     {
       "epoch": 5.925192519251925,
-      "grad_norm": 0.0031585693359375,
+      "grad_norm": 0.01409912109375,
       "learning_rate": 7.141565619128781e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2693
     },
     {
       "epoch": 5.927392739273928,
-      "grad_norm": 0.00775146484375,
+      "grad_norm": 0.004425048828125,
       "learning_rate": 7.134920581032482e-05,
       "loss": 0.0002,
       "step": 2694
     },
     {
       "epoch": 5.929592959295929,
-      "grad_norm": 0.00372314453125,
+      "grad_norm": 0.003814697265625,
       "learning_rate": 7.1282769209069e-05,
       "loss": 0.0001,
       "step": 2695
     },
     {
       "epoch": 5.931793179317932,
-      "grad_norm": 0.008544921875,
+      "grad_norm": 0.06787109375,
       "learning_rate": 7.121634641947332e-05,
-      "loss": 0.0002,
+      "loss": 0.0004,
       "step": 2696
     },
     {
       "epoch": 5.933993399339934,
-      "grad_norm": 0.0037841796875,
+      "grad_norm": 0.0086669921875,
       "learning_rate": 7.114993747348402e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2697
     },
     {
       "epoch": 5.936193619361936,
-      "grad_norm": 0.009033203125,
+      "grad_norm": 0.00677490234375,
       "learning_rate": 7.108354240304073e-05,
       "loss": 0.0002,
       "step": 2698
     },
     {
       "epoch": 5.938393839383938,
-      "grad_norm": 0.0028228759765625,
+      "grad_norm": 0.003021240234375,
       "learning_rate": 7.101716124007651e-05,
       "loss": 0.0001,
       "step": 2699
     },
     {
       "epoch": 5.9405940594059405,
-      "grad_norm": 0.007537841796875,
+      "grad_norm": 0.0130615234375,
       "learning_rate": 7.095079401651748e-05,
       "loss": 0.0002,
       "step": 2700
     },
     {
       "epoch": 5.942794279427943,
-      "grad_norm": 0.00360107421875,
+      "grad_norm": 0.00775146484375,
       "learning_rate": 7.088444076428334e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2701
     },
     {
       "epoch": 5.944994499449945,
-      "grad_norm": 0.003448486328125,
+      "grad_norm": 0.003631591796875,
       "learning_rate": 7.081810151528684e-05,
       "loss": 0.0001,
       "step": 2702
     },
     {
       "epoch": 5.947194719471947,
-      "grad_norm": 0.004913330078125,
+      "grad_norm": 0.0028533935546875,
       "learning_rate": 7.075177630143417e-05,
       "loss": 0.0001,
       "step": 2703
     },
     {
       "epoch": 5.949394939493949,
-      "grad_norm": 0.0133056640625,
+      "grad_norm": 0.005401611328125,
       "learning_rate": 7.068546515462461e-05,
       "loss": 0.0002,
       "step": 2704
     },
     {
       "epoch": 5.951595159515952,
-      "grad_norm": 0.0035247802734375,
+      "grad_norm": 0.003387451171875,
       "learning_rate": 7.061916810675078e-05,
       "loss": 0.0001,
       "step": 2705
     },
     {
       "epoch": 5.9537953795379535,
-      "grad_norm": 0.00439453125,
+      "grad_norm": 0.00421142578125,
       "learning_rate": 7.055288518969856e-05,
       "loss": 0.0001,
       "step": 2706
     },
     {
       "epoch": 5.955995599559956,
-      "grad_norm": 0.007781982421875,
+      "grad_norm": 0.005645751953125,
       "learning_rate": 7.04866164353469e-05,
       "loss": 0.0001,
       "step": 2707
     },
     {
       "epoch": 5.958195819581958,
-      "grad_norm": 0.1826171875,
+      "grad_norm": 0.00347900390625,
       "learning_rate": 7.042036187556804e-05,
-      "loss": 0.0006,
+      "loss": 0.0002,
       "step": 2708
     },
     {
       "epoch": 5.96039603960396,
-      "grad_norm": 0.006683349609375,
+      "grad_norm": 0.0038299560546875,
       "learning_rate": 7.035412154222729e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2709
     },
     {
       "epoch": 5.962596259625963,
-      "grad_norm": 0.0038299560546875,
+      "grad_norm": 0.0034942626953125,
       "learning_rate": 7.028789546718326e-05,
       "loss": 0.0001,
       "step": 2710
     },
     {
       "epoch": 5.964796479647965,
-      "grad_norm": 0.0027923583984375,
+      "grad_norm": 0.0026397705078125,
       "learning_rate": 7.022168368228764e-05,
       "loss": 0.0001,
       "step": 2711
     },
     {
       "epoch": 5.966996699669967,
-      "grad_norm": 0.005340576171875,
+      "grad_norm": 0.004791259765625,
       "learning_rate": 7.01554862193852e-05,
       "loss": 0.0001,
       "step": 2712
     },
     {
       "epoch": 5.969196919691969,
-      "grad_norm": 0.0019683837890625,
+      "grad_norm": 0.0028076171875,
       "learning_rate": 7.008930311031385e-05,
       "loss": 0.0001,
       "step": 2713
     },
     {
       "epoch": 5.971397139713972,
-      "grad_norm": 0.0111083984375,
+      "grad_norm": 0.00299072265625,
       "learning_rate": 7.002313438690462e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2714
     },
     {
       "epoch": 5.973597359735973,
-      "grad_norm": 0.010986328125,
+      "grad_norm": 0.004364013671875,
       "learning_rate": 6.99569800809816e-05,
       "loss": 0.0002,
       "step": 2715
     },
     {
       "epoch": 5.975797579757976,
-      "grad_norm": 0.007781982421875,
+      "grad_norm": 0.01519775390625,
       "learning_rate": 6.989084022436201e-05,
       "loss": 0.0002,
       "step": 2716
     },
     {
       "epoch": 5.977997799779978,
-      "grad_norm": 0.002716064453125,
+      "grad_norm": 0.0038299560546875,
       "learning_rate": 6.982471484885596e-05,
       "loss": 0.0001,
       "step": 2717
     },
     {
       "epoch": 5.98019801980198,
-      "grad_norm": 0.0036468505859375,
+      "grad_norm": 0.004364013671875,
       "learning_rate": 6.97586039862668e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2718
     },
     {
       "epoch": 5.982398239823983,
-      "grad_norm": 0.002532958984375,
+      "grad_norm": 0.0030517578125,
       "learning_rate": 6.96925076683907e-05,
       "loss": 0.0001,
       "step": 2719
     },
     {
       "epoch": 5.9845984598459845,
-      "grad_norm": 0.002532958984375,
+      "grad_norm": 0.004730224609375,
       "learning_rate": 6.962642592701703e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2720
     },
     {
       "epoch": 5.986798679867987,
-      "grad_norm": 0.0036163330078125,
+      "grad_norm": 0.0478515625,
       "learning_rate": 6.956035879392795e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2721
     },
     {
       "epoch": 5.988998899889989,
-      "grad_norm": 0.0027008056640625,
+      "grad_norm": 0.0164794921875,
       "learning_rate": 6.949430630089874e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2722
     },
     {
       "epoch": 5.991199119911991,
-      "grad_norm": 0.00579833984375,
+      "grad_norm": 0.0029754638671875,
       "learning_rate": 6.942826847969766e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2723
     },
     {
       "epoch": 5.993399339933993,
-      "grad_norm": 0.00262451171875,
+      "grad_norm": 0.00335693359375,
       "learning_rate": 6.936224536208577e-05,
       "loss": 0.0001,
       "step": 2724
     },
     {
       "epoch": 5.995599559955996,
-      "grad_norm": 0.0028076171875,
+      "grad_norm": 0.0035247802734375,
       "learning_rate": 6.929623697981718e-05,
       "loss": 0.0001,
       "step": 2725
     },
     {
       "epoch": 5.997799779977997,
-      "grad_norm": 0.00421142578125,
+      "grad_norm": 0.00909423828125,
       "learning_rate": 6.923024336463881e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2726
     },
     {
       "epoch": 6.0,
-      "grad_norm": 0.004058837890625,
+      "grad_norm": 0.0264892578125,
       "learning_rate": 6.916426454829057e-05,
-      "loss": 0.0001,
+      "loss": 0.0003,
       "step": 2727
     },
     {
       "epoch": 6.002200220022003,
-      "grad_norm": 0.004058837890625,
+      "grad_norm": 0.00836181640625,
       "learning_rate": 6.909830056250527e-05,
       "loss": 0.0002,
       "step": 2728
     },
     {
       "epoch": 6.004400440044004,
-      "grad_norm": 0.00555419921875,
+      "grad_norm": 0.00958251953125,
       "learning_rate": 6.903235143900847e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2729
     },
     {
       "epoch": 6.006600660066007,
-      "grad_norm": 0.0035400390625,
+      "grad_norm": 0.00194549560546875,
       "learning_rate": 6.896641720951869e-05,
       "loss": 0.0001,
       "step": 2730
     },
     {
       "epoch": 6.008800880088009,
-      "grad_norm": 0.005584716796875,
+      "grad_norm": 0.004791259765625,
       "learning_rate": 6.89004979057472e-05,
       "loss": 0.0002,
       "step": 2731
     },
     {
       "epoch": 6.011001100110011,
-      "grad_norm": 0.002288818359375,
+      "grad_norm": 0.005279541015625,
       "learning_rate": 6.883459355939814e-05,
       "loss": 0.0001,
       "step": 2732
     },
     {
       "epoch": 6.013201320132013,
-      "grad_norm": 0.003997802734375,
+      "grad_norm": 0.00469970703125,
       "learning_rate": 6.876870420216855e-05,
       "loss": 0.0001,
       "step": 2733
     },
     {
       "epoch": 6.0154015401540155,
-      "grad_norm": 0.0052490234375,
+      "grad_norm": 0.00787353515625,
       "learning_rate": 6.870282986574803e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2734
     },
     {
       "epoch": 6.017601760176017,
-      "grad_norm": 0.003692626953125,
+      "grad_norm": 0.01055908203125,
       "learning_rate": 6.86369705818192e-05,
       "loss": 0.0002,
       "step": 2735
     },
     {
       "epoch": 6.01980198019802,
-      "grad_norm": 0.002838134765625,
+      "grad_norm": 0.00421142578125,
       "learning_rate": 6.857112638205724e-05,
       "loss": 0.0001,
       "step": 2736
     },
     {
       "epoch": 6.01980198019802,
-      "eval_loss": 0.0002610852534417063,
-      "eval_runtime": 10.1484,
-      "eval_samples_per_second": 37.74,
-      "eval_steps_per_second": 4.73,
+      "eval_loss": 0.00038225698517635465,
+      "eval_runtime": 10.1135,
+      "eval_samples_per_second": 37.87,
+      "eval_steps_per_second": 4.746,
       "step": 2736
     },
     {
       "epoch": 6.022002200220022,
-      "grad_norm": 0.0037078857421875,
+      "grad_norm": 0.0052490234375,
       "learning_rate": 6.85052972981302e-05,
       "loss": 0.0002,
       "step": 2737
     },
     {
       "epoch": 6.024202420242024,
-      "grad_norm": 0.004638671875,
+      "grad_norm": 0.00311279296875,
       "learning_rate": 6.843948336169888e-05,
       "loss": 0.0001,
       "step": 2738
     },
     {
       "epoch": 6.026402640264027,
-      "grad_norm": 0.0032501220703125,
+      "grad_norm": 0.00567626953125,
       "learning_rate": 6.837368460441667e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2739
     },
     {
       "epoch": 6.028602860286028,
-      "grad_norm": 0.00421142578125,
+      "grad_norm": 0.005340576171875,
       "learning_rate": 6.830790105792973e-05,
       "loss": 0.0002,
       "step": 2740
     },
     {
       "epoch": 6.030803080308031,
-      "grad_norm": 0.0025482177734375,
+      "grad_norm": 0.00238037109375,
       "learning_rate": 6.824213275387696e-05,
       "loss": 0.0001,
       "step": 2741
     },
     {
       "epoch": 6.033003300330033,
-      "grad_norm": 0.0027313232421875,
+      "grad_norm": 0.027587890625,
       "learning_rate": 6.817637972388984e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2742
     },
     {
       "epoch": 6.035203520352035,
-      "grad_norm": 0.0030364990234375,
+      "grad_norm": 0.00311279296875,
       "learning_rate": 6.81106419995925e-05,
       "loss": 0.0001,
       "step": 2743
     },
     {
       "epoch": 6.037403740374037,
-      "grad_norm": 0.006072998046875,
+      "grad_norm": 0.0036163330078125,
       "learning_rate": 6.80449196126018e-05,
       "loss": 0.0001,
       "step": 2744
     },
     {
       "epoch": 6.03960396039604,
-      "grad_norm": 0.007476806640625,
+      "grad_norm": 0.0034332275390625,
       "learning_rate": 6.797921259452715e-05,
       "loss": 0.0001,
       "step": 2745
     },
     {
       "epoch": 6.041804180418042,
-      "grad_norm": 0.0020904541015625,
+      "grad_norm": 0.004364013671875,
       "learning_rate": 6.791352097697059e-05,
       "loss": 0.0001,
       "step": 2746
     },
     {
       "epoch": 6.044004400440044,
-      "grad_norm": 0.00775146484375,
+      "grad_norm": 0.004486083984375,
       "learning_rate": 6.784784479152677e-05,
       "loss": 0.0002,
       "step": 2747
     },
     {
       "epoch": 6.0462046204620465,
-      "grad_norm": 0.002777099609375,
+      "grad_norm": 0.002166748046875,
       "learning_rate": 6.778218406978282e-05,
       "loss": 0.0001,
       "step": 2748
     },
     {
       "epoch": 6.048404840484048,
-      "grad_norm": 0.0035858154296875,
+      "grad_norm": 0.0023651123046875,
       "learning_rate": 6.771653884331858e-05,
       "loss": 0.0001,
       "step": 2749
     },
     {
       "epoch": 6.050605060506051,
-      "grad_norm": 0.0035400390625,
+      "grad_norm": 0.00299072265625,
       "learning_rate": 6.765090914370642e-05,
       "loss": 0.0001,
       "step": 2750
     },
     {
       "epoch": 6.052805280528053,
-      "grad_norm": 0.004150390625,
+      "grad_norm": 0.0026092529296875,
       "learning_rate": 6.758529500251108e-05,
       "loss": 0.0001,
       "step": 2751
     },
     {
       "epoch": 6.055005500550055,
-      "grad_norm": 0.00579833984375,
+      "grad_norm": 0.0074462890625,
       "learning_rate": 6.751969645129002e-05,
       "loss": 0.0002,
       "step": 2752
     },
     {
       "epoch": 6.057205720572057,
-      "grad_norm": 0.00567626953125,
+      "grad_norm": 0.0032196044921875,
       "learning_rate": 6.745411352159299e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2753
     },
     {
       "epoch": 6.0594059405940595,
-      "grad_norm": 0.00323486328125,
+      "grad_norm": 0.004119873046875,
       "learning_rate": 6.738854624496244e-05,
       "loss": 0.0002,
       "step": 2754
     },
     {
       "epoch": 6.061606160616061,
-      "grad_norm": 0.002716064453125,
+      "grad_norm": 0.0036773681640625,
       "learning_rate": 6.732299465293322e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2755
     },
     {
       "epoch": 6.063806380638064,
-      "grad_norm": 0.00225830078125,
+      "grad_norm": 0.0030975341796875,
       "learning_rate": 6.725745877703255e-05,
       "loss": 0.0001,
       "step": 2756
     },
     {
       "epoch": 6.066006600660066,
-      "grad_norm": 0.00927734375,
+      "grad_norm": 0.0048828125,
       "learning_rate": 6.719193864878022e-05,
       "loss": 0.0002,
       "step": 2757
     },
     {
       "epoch": 6.068206820682068,
-      "grad_norm": 0.005035400390625,
+      "grad_norm": 0.017333984375,
       "learning_rate": 6.712643429968827e-05,
       "loss": 0.0002,
       "step": 2758
     },
     {
       "epoch": 6.070407040704071,
-      "grad_norm": 0.0024566650390625,
+      "grad_norm": 0.0022125244140625,
       "learning_rate": 6.706094576126139e-05,
       "loss": 0.0001,
       "step": 2759
     },
     {
       "epoch": 6.072607260726072,
-      "grad_norm": 0.00179290771484375,
+      "grad_norm": 0.00848388671875,
       "learning_rate": 6.699547306499646e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2760
     },
     {
       "epoch": 6.074807480748075,
-      "grad_norm": 0.0027618408203125,
+      "grad_norm": 0.00335693359375,
       "learning_rate": 6.693001624238284e-05,
       "loss": 0.0001,
       "step": 2761
     },
     {
       "epoch": 6.077007700770077,
-      "grad_norm": 0.002777099609375,
+      "grad_norm": 0.00274658203125,
       "learning_rate": 6.686457532490226e-05,
       "loss": 0.0001,
       "step": 2762
     },
     {
       "epoch": 6.079207920792079,
-      "grad_norm": 0.0034942626953125,
+      "grad_norm": 0.004302978515625,
       "learning_rate": 6.679915034402873e-05,
       "loss": 0.0001,
       "step": 2763
     },
     {
       "epoch": 6.081408140814081,
-      "grad_norm": 0.00238037109375,
+      "grad_norm": 0.005157470703125,
       "learning_rate": 6.673374133122869e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2764
     },
     {
       "epoch": 6.083608360836084,
-      "grad_norm": 0.0022735595703125,
+      "grad_norm": 0.0036773681640625,
       "learning_rate": 6.66683483179608e-05,
       "loss": 0.0001,
       "step": 2765
     },
     {
       "epoch": 6.085808580858086,
-      "grad_norm": 0.005096435546875,
+      "grad_norm": 0.0067138671875,
       "learning_rate": 6.66029713356761e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2766
     },
     {
       "epoch": 6.088008800880088,
-      "grad_norm": 0.0027008056640625,
+      "grad_norm": 0.004058837890625,
       "learning_rate": 6.653761041581791e-05,
       "loss": 0.0001,
       "step": 2767
     },
     {
       "epoch": 6.0902090209020905,
-      "grad_norm": 0.0034332275390625,
+      "grad_norm": 0.00433349609375,
       "learning_rate": 6.647226558982179e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2768
     },
     {
       "epoch": 6.092409240924092,
-      "grad_norm": 0.0035552978515625,
+      "grad_norm": 0.0035858154296875,
       "learning_rate": 6.64069368891156e-05,
       "loss": 0.0001,
       "step": 2769
     },
     {
       "epoch": 6.094609460946095,
-      "grad_norm": 0.0045166015625,
+      "grad_norm": 0.004486083984375,
       "learning_rate": 6.63416243451194e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2770
     },
     {
       "epoch": 6.0968096809680965,
-      "grad_norm": 0.002899169921875,
+      "grad_norm": 0.003814697265625,
       "learning_rate": 6.62763279892455e-05,
       "loss": 0.0001,
       "step": 2771
     },
     {
       "epoch": 6.099009900990099,
-      "grad_norm": 0.0025634765625,
+      "grad_norm": 0.002960205078125,
       "learning_rate": 6.621104785289845e-05,
       "loss": 0.0001,
       "step": 2772
     },
     {
       "epoch": 6.101210121012101,
-      "grad_norm": 0.0027618408203125,
+      "grad_norm": 0.004913330078125,
       "learning_rate": 6.614578396747497e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2773
     },
     {
       "epoch": 6.103410341034103,
-      "grad_norm": 0.002471923828125,
+      "grad_norm": 0.005859375,
       "learning_rate": 6.608053636436399e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2774
     },
     {
       "epoch": 6.105610561056106,
-      "grad_norm": 0.004913330078125,
+      "grad_norm": 0.006439208984375,
       "learning_rate": 6.601530507494653e-05,
       "loss": 0.0002,
       "step": 2775
     },
     {
       "epoch": 6.107810781078108,
-      "grad_norm": 0.003631591796875,
+      "grad_norm": 0.00494384765625,
       "learning_rate": 6.595009013059585e-05,
       "loss": 0.0001,
       "step": 2776
     },
     {
       "epoch": 6.11001100110011,
-      "grad_norm": 0.0024261474609375,
+      "grad_norm": 0.00390625,
       "learning_rate": 6.58848915626774e-05,
       "loss": 0.0001,
       "step": 2777
     },
     {
       "epoch": 6.112211221122112,
-      "grad_norm": 0.00323486328125,
+      "grad_norm": 0.00482177734375,
       "learning_rate": 6.581970940254857e-05,
       "loss": 0.0001,
       "step": 2778
     },
     {
       "epoch": 6.114411441144115,
-      "grad_norm": 0.0024566650390625,
+      "grad_norm": 0.0042724609375,
       "learning_rate": 6.575454368155905e-05,
       "loss": 0.0001,
       "step": 2779
     },
     {
       "epoch": 6.116611661166116,
-      "grad_norm": 0.0047607421875,
+      "grad_norm": 0.0030059814453125,
       "learning_rate": 6.568939443105044e-05,
       "loss": 0.0001,
       "step": 2780
     },
     {
       "epoch": 6.118811881188119,
-      "grad_norm": 0.0034027099609375,
+      "grad_norm": 0.008544921875,
       "learning_rate": 6.562426168235663e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2781
     },
     {
       "epoch": 6.121012101210121,
-      "grad_norm": 0.006195068359375,
+      "grad_norm": 0.0045166015625,
       "learning_rate": 6.555914546680334e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2782
     },
     {
       "epoch": 6.123212321232123,
-      "grad_norm": 0.0033416748046875,
+      "grad_norm": 0.003753662109375,
       "learning_rate": 6.549404581570855e-05,
       "loss": 0.0001,
       "step": 2783
     },
     {
       "epoch": 6.125412541254126,
-      "grad_norm": 0.006072998046875,
+      "grad_norm": 0.010009765625,
       "learning_rate": 6.542896276038217e-05,
       "loss": 0.0002,
       "step": 2784
     },
     {
       "epoch": 6.1276127612761275,
-      "grad_norm": 0.0033416748046875,
+      "grad_norm": 0.010009765625,
       "learning_rate": 6.536389633212609e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2785
     },
     {
       "epoch": 6.12981298129813,
-      "grad_norm": 0.00238037109375,
+      "grad_norm": 0.002105712890625,
       "learning_rate": 6.529884656223433e-05,
       "loss": 0.0001,
       "step": 2786
     },
     {
       "epoch": 6.132013201320132,
-      "grad_norm": 0.00439453125,
+      "grad_norm": 0.0029296875,
       "learning_rate": 6.523381348199274e-05,
       "loss": 0.0001,
       "step": 2787
     },
     {
       "epoch": 6.134213421342134,
-      "grad_norm": 0.00299072265625,
+      "grad_norm": 0.0030059814453125,
       "learning_rate": 6.516879712267928e-05,
       "loss": 0.0001,
       "step": 2788
     },
     {
       "epoch": 6.136413641364136,
-      "grad_norm": 0.005584716796875,
+      "grad_norm": 0.0040283203125,
       "learning_rate": 6.510379751556382e-05,
       "loss": 0.0001,
       "step": 2789
     },
     {
       "epoch": 6.138613861386139,
-      "grad_norm": 0.005279541015625,
+      "grad_norm": 0.00518798828125,
       "learning_rate": 6.503881469190809e-05,
       "loss": 0.0001,
       "step": 2790
     },
     {
       "epoch": 6.1408140814081404,
-      "grad_norm": 0.006866455078125,
+      "grad_norm": 0.00408935546875,
       "learning_rate": 6.497384868296592e-05,
       "loss": 0.0002,
       "step": 2791
     },
     {
       "epoch": 6.143014301430143,
-      "grad_norm": 0.00238037109375,
+      "grad_norm": 0.002716064453125,
       "learning_rate": 6.490889951998287e-05,
       "loss": 0.0001,
       "step": 2792
     },
     {
       "epoch": 6.145214521452146,
-      "grad_norm": 0.01348876953125,
+      "grad_norm": 0.004425048828125,
       "learning_rate": 6.484396723419653e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2793
     },
     {
       "epoch": 6.147414741474147,
-      "grad_norm": 0.00408935546875,
+      "grad_norm": 0.004913330078125,
       "learning_rate": 6.477905185683635e-05,
       "loss": 0.0002,
       "step": 2794
     },
     {
       "epoch": 6.14961496149615,
-      "grad_norm": 0.0023193359375,
+      "grad_norm": 0.0029144287109375,
       "learning_rate": 6.471415341912357e-05,
       "loss": 0.0001,
       "step": 2795
     },
     {
       "epoch": 6.151815181518152,
-      "grad_norm": 0.0111083984375,
+      "grad_norm": 0.00390625,
       "learning_rate": 6.46492719522714e-05,
       "loss": 0.0002,
       "step": 2796
     },
     {
       "epoch": 6.154015401540154,
-      "grad_norm": 0.0035858154296875,
+      "grad_norm": 0.0057373046875,
       "learning_rate": 6.458440748748474e-05,
       "loss": 0.0002,
       "step": 2797
     },
     {
       "epoch": 6.156215621562156,
-      "grad_norm": 0.0031890869140625,
+      "grad_norm": 0.00762939453125,
       "learning_rate": 6.45195600559605e-05,
       "loss": 0.0001,
       "step": 2798
     },
     {
       "epoch": 6.158415841584159,
-      "grad_norm": 0.006591796875,
+      "grad_norm": 0.004302978515625,
       "learning_rate": 6.44547296888872e-05,
       "loss": 0.0002,
       "step": 2799
     },
     {
       "epoch": 6.16061606160616,
-      "grad_norm": 0.005218505859375,
+      "grad_norm": 0.00421142578125,
       "learning_rate": 6.43899164174453e-05,
       "loss": 0.0001,
       "step": 2800
     },
     {
       "epoch": 6.162816281628163,
-      "grad_norm": 0.002532958984375,
+      "grad_norm": 0.0024261474609375,
       "learning_rate": 6.432512027280702e-05,
       "loss": 0.0001,
       "step": 2801
     },
     {
       "epoch": 6.165016501650165,
-      "grad_norm": 0.0035552978515625,
+      "grad_norm": 0.00677490234375,
       "learning_rate": 6.42603412861362e-05,
       "loss": 0.0001,
       "step": 2802
     },
     {
       "epoch": 6.167216721672167,
-      "grad_norm": 0.01068115234375,
+      "grad_norm": 0.005279541015625,
       "learning_rate": 6.419557948858867e-05,
       "loss": 0.0002,
       "step": 2803
     },
     {
       "epoch": 6.16941694169417,
-      "grad_norm": 0.004608154296875,
+      "grad_norm": 0.0028839111328125,
       "learning_rate": 6.413083491131176e-05,
       "loss": 0.0001,
       "step": 2804
     },
     {
       "epoch": 6.1716171617161715,
-      "grad_norm": 0.0023345947265625,
+      "grad_norm": 0.0032958984375,
       "learning_rate": 6.406610758544464e-05,
       "loss": 0.0001,
       "step": 2805
     },
     {
       "epoch": 6.173817381738174,
-      "grad_norm": 0.00335693359375,
+      "grad_norm": 0.003509521484375,
       "learning_rate": 6.400139754211823e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2806
     },
     {
       "epoch": 6.176017601760176,
-      "grad_norm": 0.003387451171875,
+      "grad_norm": 0.00274658203125,
       "learning_rate": 6.393670481245495e-05,
       "loss": 0.0001,
       "step": 2807
     },
     {
       "epoch": 6.178217821782178,
-      "grad_norm": 0.0033721923828125,
+      "grad_norm": 0.01171875,
       "learning_rate": 6.387202942756912e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2808
     },
     {
       "epoch": 6.18041804180418,
-      "grad_norm": 0.00360107421875,
+      "grad_norm": 0.0045166015625,
       "learning_rate": 6.380737141856652e-05,
       "loss": 0.0002,
       "step": 2809
     },
     {
       "epoch": 6.182618261826183,
-      "grad_norm": 0.005615234375,
+      "grad_norm": 0.003997802734375,
       "learning_rate": 6.374273081654473e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2810
     },
     {
       "epoch": 6.184818481848184,
-      "grad_norm": 0.0027923583984375,
+      "grad_norm": 0.002838134765625,
       "learning_rate": 6.36781076525929e-05,
       "loss": 0.0001,
       "step": 2811
     },
     {
       "epoch": 6.187018701870187,
-      "grad_norm": 0.00421142578125,
+      "grad_norm": 0.00897216796875,
       "learning_rate": 6.36135019577917e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2812
     },
     {
       "epoch": 6.18921892189219,
-      "grad_norm": 0.0108642578125,
+      "grad_norm": 0.003997802734375,
       "learning_rate": 6.354891376321357e-05,
       "loss": 0.0002,
       "step": 2813
     },
     {
       "epoch": 6.191419141914191,
-      "grad_norm": 0.003448486328125,
+      "grad_norm": 0.0037384033203125,
       "learning_rate": 6.348434309992238e-05,
       "loss": 0.0001,
       "step": 2814
     },
     {
       "epoch": 6.193619361936194,
-      "grad_norm": 0.00299072265625,
+      "grad_norm": 0.0027923583984375,
       "learning_rate": 6.341978999897365e-05,
       "loss": 0.0001,
       "step": 2815
     },
     {
       "epoch": 6.195819581958196,
-      "grad_norm": 0.0024566650390625,
+      "grad_norm": 0.003326416015625,
       "learning_rate": 6.335525449141449e-05,
       "loss": 0.0001,
       "step": 2816
     },
     {
       "epoch": 6.198019801980198,
-      "grad_norm": 0.004669189453125,
+      "grad_norm": 0.0037689208984375,
       "learning_rate": 6.329073660828341e-05,
       "loss": 0.0002,
       "step": 2817
     },
     {
       "epoch": 6.2002200220022,
-      "grad_norm": 0.003997802734375,
+      "grad_norm": 0.0076904296875,
       "learning_rate": 6.322623638061061e-05,
       "loss": 0.0001,
       "step": 2818
     },
     {
       "epoch": 6.2024202420242025,
-      "grad_norm": 0.0027008056640625,
+      "grad_norm": 0.003875732421875,
       "learning_rate": 6.316175383941763e-05,
       "loss": 0.0001,
       "step": 2819
     },
     {
       "epoch": 6.204620462046204,
-      "grad_norm": 0.00457763671875,
+      "grad_norm": 0.0027618408203125,
       "learning_rate": 6.30972890157177e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2820
     },
     {
       "epoch": 6.206820682068207,
-      "grad_norm": 0.002197265625,
+      "grad_norm": 0.0030975341796875,
       "learning_rate": 6.30328419405153e-05,
       "loss": 0.0001,
       "step": 2821
     },
     {
       "epoch": 6.209020902090209,
-      "grad_norm": 0.0036468505859375,
+      "grad_norm": 0.00811767578125,
       "learning_rate": 6.296841264480657e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2822
     },
     {
       "epoch": 6.211221122112211,
-      "grad_norm": 0.004119873046875,
+      "grad_norm": 0.016845703125,
       "learning_rate": 6.290400115957903e-05,
-      "loss": 0.0001,
+      "loss": 0.0003,
       "step": 2823
     },
     {
       "epoch": 6.213421342134214,
-      "grad_norm": 0.001983642578125,
+      "grad_norm": 0.0028076171875,
       "learning_rate": 6.283960751581155e-05,
       "loss": 0.0001,
       "step": 2824
     },
     {
       "epoch": 6.215621562156215,
-      "grad_norm": 0.0031890869140625,
+      "grad_norm": 0.007476806640625,
       "learning_rate": 6.27752317444746e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2825
     },
     {
       "epoch": 6.217821782178218,
-      "grad_norm": 0.0205078125,
+      "grad_norm": 0.01025390625,
       "learning_rate": 6.271087387652983e-05,
-      "loss": 0.0005,
+      "loss": 0.0002,
       "step": 2826
     },
     {
       "epoch": 6.22002200220022,
-      "grad_norm": 0.00628662109375,
+      "grad_norm": 0.0045166015625,
       "learning_rate": 6.26465339429305e-05,
       "loss": 0.0002,
       "step": 2827
     },
     {
       "epoch": 6.222222222222222,
-      "grad_norm": 0.00634765625,
+      "grad_norm": 0.00262451171875,
       "learning_rate": 6.25822119746211e-05,
       "loss": 0.0001,
       "step": 2828
     },
     {
       "epoch": 6.224422442244224,
-      "grad_norm": 0.003936767578125,
+      "grad_norm": 0.006011962890625,
       "learning_rate": 6.251790800253752e-05,
       "loss": 0.0002,
       "step": 2829
     },
     {
       "epoch": 6.226622662266227,
-      "grad_norm": 0.003631591796875,
+      "grad_norm": 0.0025482177734375,
       "learning_rate": 6.245362205760704e-05,
       "loss": 0.0001,
       "step": 2830
     },
     {
       "epoch": 6.228822882288229,
-      "grad_norm": 0.002197265625,
+      "grad_norm": 0.004638671875,
       "learning_rate": 6.238935417074812e-05,
       "loss": 0.0001,
       "step": 2831
     },
     {
       "epoch": 6.231023102310231,
-      "grad_norm": 0.004058837890625,
+      "grad_norm": 0.0040283203125,
       "learning_rate": 6.232510437287076e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2832
     },
     {
       "epoch": 6.2332233223322335,
-      "grad_norm": 0.003387451171875,
+      "grad_norm": 0.00799560546875,
       "learning_rate": 6.22608726948761e-05,
       "loss": 0.0002,
       "step": 2833
     },
     {
       "epoch": 6.235423542354235,
-      "grad_norm": 0.0213623046875,
+      "grad_norm": 0.005645751953125,
       "learning_rate": 6.219665916765657e-05,
       "loss": 0.0001,
       "step": 2834
     },
     {
       "epoch": 6.237623762376238,
-      "grad_norm": 0.003204345703125,
+      "grad_norm": 0.0031890869140625,
       "learning_rate": 6.213246382209599e-05,
       "loss": 0.0001,
       "step": 2835
     },
     {
       "epoch": 6.2398239823982395,
-      "grad_norm": 0.00347900390625,
+      "grad_norm": 0.006103515625,
       "learning_rate": 6.206828668906926e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2836
     },
     {
       "epoch": 6.242024202420242,
-      "grad_norm": 0.00537109375,
+      "grad_norm": 0.00628662109375,
       "learning_rate": 6.200412779944267e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2837
     },
     {
       "epoch": 6.244224422442244,
-      "grad_norm": 0.0025177001953125,
+      "grad_norm": 0.00457763671875,
       "learning_rate": 6.193998718407365e-05,
       "loss": 0.0001,
       "step": 2838
     },
     {
       "epoch": 6.2464246424642464,
-      "grad_norm": 0.006622314453125,
+      "grad_norm": 0.0135498046875,
       "learning_rate": 6.187586487381089e-05,
       "loss": 0.0002,
       "step": 2839
     },
     {
       "epoch": 6.248624862486249,
-      "grad_norm": 0.0024566650390625,
+      "grad_norm": 0.004547119140625,
       "learning_rate": 6.181176089949427e-05,
       "loss": 0.0001,
       "step": 2840
     },
     {
       "epoch": 6.250825082508251,
-      "grad_norm": 0.004730224609375,
+      "grad_norm": 0.003082275390625,
       "learning_rate": 6.174767529195478e-05,
       "loss": 0.0001,
       "step": 2841
     },
     {
       "epoch": 6.253025302530253,
-      "grad_norm": 0.002166748046875,
+      "grad_norm": 0.004852294921875,
       "learning_rate": 6.168360808201473e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2842
     },
     {
       "epoch": 6.255225522552255,
-      "grad_norm": 0.0033111572265625,
+      "grad_norm": 0.002960205078125,
       "learning_rate": 6.161955930048739e-05,
       "loss": 0.0001,
       "step": 2843
     },
     {
       "epoch": 6.257425742574258,
-      "grad_norm": 0.0037689208984375,
+      "grad_norm": 0.0030059814453125,
       "learning_rate": 6.155552897817726e-05,
       "loss": 0.0001,
       "step": 2844
     },
     {
       "epoch": 6.259625962596259,
-      "grad_norm": 0.007080078125,
+      "grad_norm": 0.003082275390625,
       "learning_rate": 6.149151714588009e-05,
       "loss": 0.0001,
       "step": 2845
     },
     {
       "epoch": 6.261826182618262,
-      "grad_norm": 0.0020751953125,
+      "grad_norm": 0.00457763671875,
       "learning_rate": 6.142752383438247e-05,
       "loss": 0.0001,
       "step": 2846
     },
     {
       "epoch": 6.264026402640264,
-      "grad_norm": 0.007293701171875,
+      "grad_norm": 0.0032958984375,
       "learning_rate": 6.136354907446229e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2847
     },
     {
       "epoch": 6.266226622662266,
-      "grad_norm": 0.0028533935546875,
+      "grad_norm": 0.007110595703125,
       "learning_rate": 6.129959289688844e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2848
     },
     {
       "epoch": 6.268426842684269,
-      "grad_norm": 0.004547119140625,
+      "grad_norm": 0.003326416015625,
       "learning_rate": 6.123565533242088e-05,
       "loss": 0.0001,
       "step": 2849
     },
     {
       "epoch": 6.270627062706271,
-      "grad_norm": 0.0028228759765625,
+      "grad_norm": 0.0032501220703125,
       "learning_rate": 6.117173641181064e-05,
       "loss": 0.0001,
       "step": 2850
     },
     {
       "epoch": 6.270627062706271,
-      "eval_loss": 0.00026186328614130616,
-      "eval_runtime": 10.1553,
-      "eval_samples_per_second": 37.714,
-      "eval_steps_per_second": 4.727,
+      "eval_loss": 0.00037015212001278996,
+      "eval_runtime": 10.259,
+      "eval_samples_per_second": 37.333,
+      "eval_steps_per_second": 4.679,
       "step": 2850
     },
     {
       "epoch": 6.272827282728273,
-      "grad_norm": 0.0026092529296875,
+      "grad_norm": 0.003448486328125,
       "learning_rate": 6.110783616579974e-05,
       "loss": 0.0001,
       "step": 2851
     },
     {
       "epoch": 6.275027502750275,
-      "grad_norm": 0.0036773681640625,
+      "grad_norm": 0.00811767578125,
       "learning_rate": 6.104395462512127e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2852
     },
     {
       "epoch": 6.2772277227722775,
-      "grad_norm": 0.005645751953125,
+      "grad_norm": 0.003570556640625,
       "learning_rate": 6.098009182049925e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2853
     },
     {
       "epoch": 6.279427942794279,
-      "grad_norm": 0.0037384033203125,
+      "grad_norm": 0.0032501220703125,
       "learning_rate": 6.0916247782648794e-05,
       "loss": 0.0001,
       "step": 2854
     },
     {
       "epoch": 6.281628162816282,
-      "grad_norm": 0.003753662109375,
+      "grad_norm": 0.004913330078125,
       "learning_rate": 6.085242254227592e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2855
     },
     {
       "epoch": 6.2838283828382835,
-      "grad_norm": 0.0030517578125,
+      "grad_norm": 0.00372314453125,
       "learning_rate": 6.078861613007757e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2856
     },
     {
       "epoch": 6.286028602860286,
-      "grad_norm": 0.003936767578125,
+      "grad_norm": 0.01214599609375,
       "learning_rate": 6.0724828576741754e-05,
       "loss": 0.0002,
       "step": 2857
     },
     {
       "epoch": 6.288228822882289,
-      "grad_norm": 0.004058837890625,
+      "grad_norm": 0.0036163330078125,
       "learning_rate": 6.066105991294724e-05,
       "loss": 0.0001,
       "step": 2858
     },
     {
       "epoch": 6.29042904290429,
-      "grad_norm": 0.003875732421875,
+      "grad_norm": 0.01470947265625,
       "learning_rate": 6.059731016936391e-05,
-      "loss": 0.0001,
+      "loss": 0.0003,
       "step": 2859
     },
     {
       "epoch": 6.292629262926293,
-      "grad_norm": 0.0030364990234375,
+      "grad_norm": 0.00262451171875,
       "learning_rate": 6.053357937665237e-05,
       "loss": 0.0001,
       "step": 2860
     },
     {
       "epoch": 6.294829482948295,
-      "grad_norm": 0.0032806396484375,
+      "grad_norm": 0.003875732421875,
       "learning_rate": 6.046986756546419e-05,
       "loss": 0.0001,
       "step": 2861
     },
     {
       "epoch": 6.297029702970297,
-      "grad_norm": 0.003509521484375,
+      "grad_norm": 0.0029144287109375,
       "learning_rate": 6.040617476644186e-05,
       "loss": 0.0001,
       "step": 2862
     },
     {
       "epoch": 6.299229922992299,
-      "grad_norm": 0.00193023681640625,
+      "grad_norm": 0.0019073486328125,
       "learning_rate": 6.034250101021859e-05,
       "loss": 0.0001,
       "step": 2863
     },
     {
       "epoch": 6.301430143014302,
-      "grad_norm": 0.0021820068359375,
+      "grad_norm": 0.0059814453125,
       "learning_rate": 6.02788463274186e-05,
       "loss": 0.0001,
       "step": 2864
     },
     {
       "epoch": 6.303630363036303,
-      "grad_norm": 0.0022125244140625,
+      "grad_norm": 0.0025482177734375,
       "learning_rate": 6.021521074865678e-05,
       "loss": 0.0001,
       "step": 2865
     },
     {
       "epoch": 6.305830583058306,
-      "grad_norm": 0.0235595703125,
+      "grad_norm": 0.0038604736328125,
       "learning_rate": 6.015159430453892e-05,
-      "loss": 0.0006,
+      "loss": 0.0001,
       "step": 2866
     },
     {
       "epoch": 6.3080308030803085,
-      "grad_norm": 0.003448486328125,
+      "grad_norm": 0.00347900390625,
       "learning_rate": 6.008799702566161e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2867
     },
     {
       "epoch": 6.31023102310231,
-      "grad_norm": 0.00201416015625,
+      "grad_norm": 0.0025177001953125,
       "learning_rate": 6.002441894261216e-05,
       "loss": 0.0001,
       "step": 2868
     },
     {
       "epoch": 6.312431243124313,
-      "grad_norm": 0.003265380859375,
+      "grad_norm": 0.003204345703125,
       "learning_rate": 5.996086008596875e-05,
       "loss": 0.0001,
       "step": 2869
     },
     {
       "epoch": 6.3146314631463145,
-      "grad_norm": 0.0033111572265625,
+      "grad_norm": 0.004241943359375,
       "learning_rate": 5.989732048630019e-05,
       "loss": 0.0002,
       "step": 2870
     },
     {
       "epoch": 6.316831683168317,
-      "grad_norm": 0.004119873046875,
+      "grad_norm": 0.0042724609375,
       "learning_rate": 5.983380017416609e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2871
     },
     {
       "epoch": 6.319031903190319,
-      "grad_norm": 0.005615234375,
+      "grad_norm": 0.01129150390625,
       "learning_rate": 5.977029918011685e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2872
     },
     {
       "epoch": 6.321232123212321,
-      "grad_norm": 0.0030975341796875,
+      "grad_norm": 0.004791259765625,
       "learning_rate": 5.9706817534693426e-05,
       "loss": 0.0001,
       "step": 2873
     },
     {
       "epoch": 6.323432343234323,
-      "grad_norm": 0.0018157958984375,
+      "grad_norm": 0.0036163330078125,
       "learning_rate": 5.964335526842764e-05,
       "loss": 0.0001,
       "step": 2874
@@ -20343,3637 +20343,3637 @@
     },
     {
       "epoch": 6.327832783278327,
-      "grad_norm": 0.0029296875,
+      "grad_norm": 0.002960205078125,
       "learning_rate": 5.9516488995449124e-05,
       "loss": 0.0001,
       "step": 2876
     },
     {
       "epoch": 6.33003300330033,
-      "grad_norm": 0.003570556640625,
+      "grad_norm": 0.0025634765625,
       "learning_rate": 5.9453085049753246e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2877
     },
     {
       "epoch": 6.332233223322333,
-      "grad_norm": 0.00531005859375,
+      "grad_norm": 0.005859375,
       "learning_rate": 5.938970060524851e-05,
       "loss": 0.0001,
       "step": 2878
     },
     {
       "epoch": 6.334433443344334,
-      "grad_norm": 0.0035552978515625,
+      "grad_norm": 0.00469970703125,
       "learning_rate": 5.9326335692419995e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2879
     },
     {
       "epoch": 6.336633663366337,
-      "grad_norm": 0.004241943359375,
+      "grad_norm": 0.003265380859375,
       "learning_rate": 5.9262990341743206e-05,
       "loss": 0.0001,
       "step": 2880
     },
     {
       "epoch": 6.338833883388339,
-      "grad_norm": 0.01153564453125,
+      "grad_norm": 0.004730224609375,
       "learning_rate": 5.919966458368441e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2881
     },
     {
       "epoch": 6.341034103410341,
-      "grad_norm": 0.00640869140625,
+      "grad_norm": 0.007781982421875,
       "learning_rate": 5.9136358448700305e-05,
       "loss": 0.0002,
       "step": 2882
     },
     {
       "epoch": 6.343234323432343,
-      "grad_norm": 0.0030059814453125,
+      "grad_norm": 0.003936767578125,
       "learning_rate": 5.9073071967238214e-05,
       "loss": 0.0001,
       "step": 2883
     },
     {
       "epoch": 6.3454345434543455,
-      "grad_norm": 0.00335693359375,
+      "grad_norm": 0.008544921875,
       "learning_rate": 5.9009805169736064e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2884
     },
     {
       "epoch": 6.347634763476347,
-      "grad_norm": 0.002288818359375,
+      "grad_norm": 0.007568359375,
       "learning_rate": 5.894655808662219e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2885
     },
     {
       "epoch": 6.34983498349835,
-      "grad_norm": 0.00262451171875,
+      "grad_norm": 0.005767822265625,
       "learning_rate": 5.88833307483156e-05,
       "loss": 0.0001,
       "step": 2886
     },
     {
       "epoch": 6.3520352035203524,
-      "grad_norm": 0.00543212890625,
+      "grad_norm": 0.0037841796875,
       "learning_rate": 5.882012318522563e-05,
       "loss": 0.0002,
       "step": 2887
     },
     {
       "epoch": 6.354235423542354,
-      "grad_norm": 0.004425048828125,
+      "grad_norm": 0.0079345703125,
       "learning_rate": 5.8756935427752223e-05,
       "loss": 0.0002,
       "step": 2888
     },
     {
       "epoch": 6.356435643564357,
-      "grad_norm": 0.0038909912109375,
+      "grad_norm": 0.00567626953125,
       "learning_rate": 5.869376750628582e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2889
     },
     {
       "epoch": 6.3586358635863585,
-      "grad_norm": 0.003204345703125,
+      "grad_norm": 0.00311279296875,
       "learning_rate": 5.863061945120719e-05,
       "loss": 0.0001,
       "step": 2890
     },
     {
       "epoch": 6.360836083608361,
-      "grad_norm": 0.003936767578125,
+      "grad_norm": 0.007080078125,
       "learning_rate": 5.856749129288771e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2891
     },
     {
       "epoch": 6.363036303630363,
-      "grad_norm": 0.0029449462890625,
+      "grad_norm": 0.002838134765625,
       "learning_rate": 5.850438306168904e-05,
       "loss": 0.0001,
       "step": 2892
     },
     {
       "epoch": 6.365236523652365,
-      "grad_norm": 0.002532958984375,
+      "grad_norm": 0.0030975341796875,
       "learning_rate": 5.844129478796332e-05,
       "loss": 0.0001,
       "step": 2893
     },
     {
       "epoch": 6.367436743674367,
-      "grad_norm": 0.0028228759765625,
+      "grad_norm": 0.0036468505859375,
       "learning_rate": 5.837822650205313e-05,
       "loss": 0.0001,
       "step": 2894
     },
     {
       "epoch": 6.36963696369637,
-      "grad_norm": 0.003631591796875,
+      "grad_norm": 0.006927490234375,
       "learning_rate": 5.83151782342914e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2895
     },
     {
       "epoch": 6.371837183718371,
-      "grad_norm": 0.0025482177734375,
+      "grad_norm": 0.00341796875,
       "learning_rate": 5.825215001500146e-05,
       "loss": 0.0001,
       "step": 2896
     },
     {
       "epoch": 6.374037403740374,
-      "grad_norm": 0.00347900390625,
+      "grad_norm": 0.0031890869140625,
       "learning_rate": 5.818914187449683e-05,
       "loss": 0.0001,
       "step": 2897
     },
     {
       "epoch": 6.376237623762377,
-      "grad_norm": 0.01373291015625,
+      "grad_norm": 0.005767822265625,
       "learning_rate": 5.81261538430817e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2898
     },
     {
       "epoch": 6.378437843784378,
-      "grad_norm": 0.003021240234375,
+      "grad_norm": 0.006988525390625,
       "learning_rate": 5.806318595105022e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2899
     },
     {
       "epoch": 6.380638063806381,
-      "grad_norm": 0.00732421875,
+      "grad_norm": 0.0177001953125,
       "learning_rate": 5.8000238228687164e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2900
     },
     {
       "epoch": 6.382838283828383,
-      "grad_norm": 0.003692626953125,
+      "grad_norm": 0.003814697265625,
       "learning_rate": 5.793731070626749e-05,
       "loss": 0.0001,
       "step": 2901
     },
     {
       "epoch": 6.385038503850385,
-      "grad_norm": 0.1337890625,
+      "grad_norm": 0.004608154296875,
       "learning_rate": 5.787440341405628e-05,
-      "loss": 0.0003,
+      "loss": 0.0001,
       "step": 2902
     },
     {
       "epoch": 6.387238723872387,
-      "grad_norm": 0.0054931640625,
+      "grad_norm": 0.005096435546875,
       "learning_rate": 5.781151638230919e-05,
       "loss": 0.0002,
       "step": 2903
     },
     {
       "epoch": 6.3894389438943895,
-      "grad_norm": 0.0028228759765625,
+      "grad_norm": 0.00347900390625,
       "learning_rate": 5.7748649641271845e-05,
       "loss": 0.0001,
       "step": 2904
     },
     {
       "epoch": 6.391639163916391,
-      "grad_norm": 0.0037994384765625,
+      "grad_norm": 0.0026092529296875,
       "learning_rate": 5.768580322118034e-05,
       "loss": 0.0001,
       "step": 2905
     },
     {
       "epoch": 6.393839383938394,
-      "grad_norm": 0.01544189453125,
+      "grad_norm": 0.00396728515625,
       "learning_rate": 5.762297715226089e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 2906
     },
     {
       "epoch": 6.396039603960396,
-      "grad_norm": 0.004150390625,
+      "grad_norm": 0.00433349609375,
       "learning_rate": 5.756017146472983e-05,
       "loss": 0.0001,
       "step": 2907
     },
     {
       "epoch": 6.398239823982398,
-      "grad_norm": 0.0024566650390625,
+      "grad_norm": 0.00396728515625,
       "learning_rate": 5.7497386188793945e-05,
       "loss": 0.0001,
       "step": 2908
     },
     {
       "epoch": 6.400440044004401,
-      "grad_norm": 0.00213623046875,
+      "grad_norm": 0.00494384765625,
       "learning_rate": 5.7434621354649874e-05,
       "loss": 0.0001,
       "step": 2909
     },
     {
       "epoch": 6.402640264026402,
-      "grad_norm": 0.0025787353515625,
+      "grad_norm": 0.0024566650390625,
       "learning_rate": 5.737187699248474e-05,
       "loss": 0.0001,
       "step": 2910
     },
     {
       "epoch": 6.404840484048405,
-      "grad_norm": 0.00457763671875,
+      "grad_norm": 0.0037689208984375,
       "learning_rate": 5.730915313247567e-05,
       "loss": 0.0001,
       "step": 2911
     },
     {
       "epoch": 6.407040704070407,
-      "grad_norm": 0.0023956298828125,
+      "grad_norm": 0.005401611328125,
       "learning_rate": 5.724644980478982e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2912
     },
     {
       "epoch": 6.409240924092409,
-      "grad_norm": 0.0064697265625,
+      "grad_norm": 0.004669189453125,
       "learning_rate": 5.718376703958476e-05,
       "loss": 0.0001,
       "step": 2913
     },
     {
       "epoch": 6.411441144114411,
-      "grad_norm": 0.003997802734375,
+      "grad_norm": 0.00390625,
       "learning_rate": 5.712110486700782e-05,
       "loss": 0.0001,
       "step": 2914
     },
     {
       "epoch": 6.413641364136414,
-      "grad_norm": 0.0026092529296875,
+      "grad_norm": 0.0130615234375,
       "learning_rate": 5.705846331719675e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2915
     },
     {
       "epoch": 6.415841584158416,
-      "grad_norm": 0.003814697265625,
+      "grad_norm": 0.0030670166015625,
       "learning_rate": 5.69958424202792e-05,
       "loss": 0.0001,
       "step": 2916
     },
     {
       "epoch": 6.418041804180418,
-      "grad_norm": 0.007080078125,
+      "grad_norm": 0.0120849609375,
       "learning_rate": 5.6933242206372904e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2917
     },
     {
       "epoch": 6.4202420242024205,
-      "grad_norm": 0.008056640625,
+      "grad_norm": 0.01177978515625,
       "learning_rate": 5.687066270558572e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2918
     },
     {
       "epoch": 6.422442244224422,
-      "grad_norm": 0.002471923828125,
+      "grad_norm": 0.00311279296875,
       "learning_rate": 5.680810394801539e-05,
       "loss": 0.0001,
       "step": 2919
     },
     {
       "epoch": 6.424642464246425,
-      "grad_norm": 0.004608154296875,
+      "grad_norm": 0.0036163330078125,
       "learning_rate": 5.6745565963749925e-05,
       "loss": 0.0001,
       "step": 2920
     },
     {
       "epoch": 6.4268426842684265,
-      "grad_norm": 0.001922607421875,
+      "grad_norm": 0.007232666015625,
       "learning_rate": 5.6683048782867054e-05,
       "loss": 0.0001,
       "step": 2921
     },
     {
       "epoch": 6.429042904290429,
-      "grad_norm": 0.002166748046875,
+      "grad_norm": 0.002410888671875,
       "learning_rate": 5.662055243543476e-05,
       "loss": 0.0001,
       "step": 2922
     },
     {
       "epoch": 6.431243124312431,
-      "grad_norm": 0.002716064453125,
+      "grad_norm": 0.0032501220703125,
       "learning_rate": 5.6558076951510894e-05,
       "loss": 0.0001,
       "step": 2923
     },
     {
       "epoch": 6.433443344334433,
-      "grad_norm": 0.0036468505859375,
+      "grad_norm": 0.00201416015625,
       "learning_rate": 5.649562236114317e-05,
       "loss": 0.0001,
       "step": 2924
     },
     {
       "epoch": 6.435643564356436,
-      "grad_norm": 0.003204345703125,
+      "grad_norm": 0.0023040771484375,
       "learning_rate": 5.643318869436951e-05,
       "loss": 0.0001,
       "step": 2925
     },
     {
       "epoch": 6.437843784378438,
-      "grad_norm": 0.005401611328125,
+      "grad_norm": 0.0038604736328125,
       "learning_rate": 5.637077598121747e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2926
     },
     {
       "epoch": 6.44004400440044,
-      "grad_norm": 0.0026397705078125,
+      "grad_norm": 0.0035858154296875,
       "learning_rate": 5.630838425170479e-05,
       "loss": 0.0001,
       "step": 2927
     },
     {
       "epoch": 6.442244224422442,
-      "grad_norm": 0.004791259765625,
+      "grad_norm": 0.005218505859375,
       "learning_rate": 5.6246013535839004e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2928
     },
     {
       "epoch": 6.444444444444445,
-      "grad_norm": 0.004852294921875,
+      "grad_norm": 0.00714111328125,
       "learning_rate": 5.618366386361742e-05,
       "loss": 0.0002,
       "step": 2929
     },
     {
       "epoch": 6.446644664466446,
-      "grad_norm": 0.00225830078125,
+      "grad_norm": 0.0030975341796875,
       "learning_rate": 5.612133526502752e-05,
       "loss": 0.0001,
       "step": 2930
     },
     {
       "epoch": 6.448844884488449,
-      "grad_norm": 0.002288818359375,
+      "grad_norm": 0.0034942626953125,
       "learning_rate": 5.60590277700463e-05,
       "loss": 0.0001,
       "step": 2931
     },
     {
       "epoch": 6.451045104510451,
-      "grad_norm": 0.00396728515625,
+      "grad_norm": 0.0047607421875,
       "learning_rate": 5.5996741408640926e-05,
       "loss": 0.0001,
       "step": 2932
     },
     {
       "epoch": 6.453245324532453,
-      "grad_norm": 0.002593994140625,
+      "grad_norm": 0.00421142578125,
       "learning_rate": 5.593447621076824e-05,
       "loss": 0.0001,
       "step": 2933
     },
     {
       "epoch": 6.455445544554456,
-      "grad_norm": 0.0026702880859375,
+      "grad_norm": 0.009521484375,
       "learning_rate": 5.587223220637481e-05,
       "loss": 0.0001,
       "step": 2934
     },
     {
       "epoch": 6.457645764576458,
-      "grad_norm": 0.0038299560546875,
+      "grad_norm": 0.0031585693359375,
       "learning_rate": 5.5810009425397294e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2935
     },
     {
       "epoch": 6.45984598459846,
-      "grad_norm": 0.005767822265625,
+      "grad_norm": 0.004486083984375,
       "learning_rate": 5.5747807897761785e-05,
       "loss": 0.0001,
       "step": 2936
     },
     {
       "epoch": 6.462046204620462,
-      "grad_norm": 0.00390625,
+      "grad_norm": 0.0068359375,
       "learning_rate": 5.568562765338452e-05,
       "loss": 0.0002,
       "step": 2937
     },
     {
       "epoch": 6.4642464246424645,
-      "grad_norm": 0.002777099609375,
+      "grad_norm": 0.003204345703125,
       "learning_rate": 5.562346872217121e-05,
       "loss": 0.0001,
       "step": 2938
     },
     {
       "epoch": 6.466446644664466,
-      "grad_norm": 0.006134033203125,
+      "grad_norm": 0.0037689208984375,
       "learning_rate": 5.5561331134017415e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2939
     },
     {
       "epoch": 6.468646864686469,
-      "grad_norm": 0.003143310546875,
+      "grad_norm": 0.003448486328125,
       "learning_rate": 5.5499214918808565e-05,
       "loss": 0.0001,
       "step": 2940
     },
     {
       "epoch": 6.4708470847084705,
-      "grad_norm": 0.0032958984375,
+      "grad_norm": 0.0034637451171875,
       "learning_rate": 5.543712010641952e-05,
       "loss": 0.0001,
       "step": 2941
     },
     {
       "epoch": 6.473047304730473,
-      "grad_norm": 0.001983642578125,
+      "grad_norm": 0.003875732421875,
       "learning_rate": 5.537504672671519e-05,
       "loss": 0.0001,
       "step": 2942
     },
     {
       "epoch": 6.475247524752476,
-      "grad_norm": 0.007354736328125,
+      "grad_norm": 0.005523681640625,
       "learning_rate": 5.531299480954981e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2943
     },
     {
       "epoch": 6.477447744774477,
-      "grad_norm": 0.0027923583984375,
+      "grad_norm": 0.003326416015625,
       "learning_rate": 5.525096438476762e-05,
       "loss": 0.0001,
       "step": 2944
     },
     {
       "epoch": 6.47964796479648,
-      "grad_norm": 0.0054931640625,
+      "grad_norm": 0.0027923583984375,
       "learning_rate": 5.51889554822024e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2945
     },
     {
       "epoch": 6.481848184818482,
-      "grad_norm": 0.0035858154296875,
+      "grad_norm": 0.00311279296875,
       "learning_rate": 5.512696813167744e-05,
       "loss": 0.0001,
       "step": 2946
     },
     {
       "epoch": 6.484048404840484,
-      "grad_norm": 0.00665283203125,
+      "grad_norm": 0.004058837890625,
       "learning_rate": 5.506500236300594e-05,
       "loss": 0.0002,
       "step": 2947
     },
     {
       "epoch": 6.486248624862486,
-      "grad_norm": 0.00408935546875,
+      "grad_norm": 0.0037841796875,
       "learning_rate": 5.50030582059904e-05,
       "loss": 0.0001,
       "step": 2948
     },
     {
       "epoch": 6.488448844884489,
-      "grad_norm": 0.00457763671875,
+      "grad_norm": 0.0032806396484375,
       "learning_rate": 5.494113569042323e-05,
       "loss": 0.0001,
       "step": 2949
     },
     {
       "epoch": 6.49064906490649,
-      "grad_norm": 0.0050048828125,
+      "grad_norm": 0.0025177001953125,
       "learning_rate": 5.487923484608629e-05,
       "loss": 0.0001,
       "step": 2950
     },
     {
       "epoch": 6.492849284928493,
-      "grad_norm": 0.002197265625,
+      "grad_norm": 0.002593994140625,
       "learning_rate": 5.481735570275092e-05,
       "loss": 0.0001,
       "step": 2951
     },
     {
       "epoch": 6.4950495049504955,
-      "grad_norm": 0.0028533935546875,
+      "grad_norm": 0.00933837890625,
       "learning_rate": 5.475549829017825e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2952
     },
     {
       "epoch": 6.497249724972497,
-      "grad_norm": 0.0029144287109375,
+      "grad_norm": 0.004241943359375,
       "learning_rate": 5.4693662638118726e-05,
       "loss": 0.0001,
       "step": 2953
     },
     {
       "epoch": 6.4994499449945,
-      "grad_norm": 0.00244140625,
+      "grad_norm": 0.0026397705078125,
       "learning_rate": 5.463184877631251e-05,
       "loss": 0.0001,
       "step": 2954
     },
     {
       "epoch": 6.5016501650165015,
-      "grad_norm": 0.0028228759765625,
+      "grad_norm": 0.0035858154296875,
       "learning_rate": 5.457005673448925e-05,
       "loss": 0.0001,
       "step": 2955
     },
     {
       "epoch": 6.503850385038504,
-      "grad_norm": 0.00628662109375,
+      "grad_norm": 0.00299072265625,
       "learning_rate": 5.4508286542367905e-05,
       "loss": 0.0002,
       "step": 2956
     },
     {
       "epoch": 6.506050605060506,
-      "grad_norm": 0.0037384033203125,
+      "grad_norm": 0.00408935546875,
       "learning_rate": 5.4446538229657275e-05,
       "loss": 0.0001,
       "step": 2957
     },
     {
       "epoch": 6.508250825082508,
-      "grad_norm": 0.00323486328125,
+      "grad_norm": 0.0028076171875,
       "learning_rate": 5.438481182605526e-05,
       "loss": 0.0001,
       "step": 2958
     },
     {
       "epoch": 6.51045104510451,
-      "grad_norm": 0.004547119140625,
+      "grad_norm": 0.006072998046875,
       "learning_rate": 5.432310736124957e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2959
     },
     {
       "epoch": 6.512651265126513,
-      "grad_norm": 0.003387451171875,
+      "grad_norm": 0.00347900390625,
       "learning_rate": 5.426142486491707e-05,
       "loss": 0.0001,
       "step": 2960
     },
     {
       "epoch": 6.514851485148515,
-      "grad_norm": 0.0036773681640625,
+      "grad_norm": 0.00543212890625,
       "learning_rate": 5.4199764366724226e-05,
       "loss": 0.0001,
       "step": 2961
     },
     {
       "epoch": 6.517051705170517,
-      "grad_norm": 0.0115966796875,
+      "grad_norm": 0.0040283203125,
       "learning_rate": 5.413812589632696e-05,
       "loss": 0.0002,
       "step": 2962
     },
     {
       "epoch": 6.51925192519252,
-      "grad_norm": 0.003936767578125,
+      "grad_norm": 0.002655029296875,
       "learning_rate": 5.40765094833704e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2963
     },
     {
       "epoch": 6.521452145214521,
-      "grad_norm": 0.0031890869140625,
+      "grad_norm": 0.0033416748046875,
       "learning_rate": 5.4014915157489335e-05,
       "loss": 0.0001,
       "step": 2964
     },
     {
       "epoch": 6.521452145214521,
-      "eval_loss": 0.00025018342421390116,
-      "eval_runtime": 10.9703,
-      "eval_samples_per_second": 34.913,
-      "eval_steps_per_second": 4.375,
+      "eval_loss": 0.0003639912174548954,
+      "eval_runtime": 10.1855,
+      "eval_samples_per_second": 37.602,
+      "eval_steps_per_second": 4.713,
       "step": 2964
     },
     {
       "epoch": 6.523652365236524,
-      "grad_norm": 0.003936767578125,
+      "grad_norm": 0.0042724609375,
       "learning_rate": 5.395334294830765e-05,
       "loss": 0.0001,
       "step": 2965
     },
     {
       "epoch": 6.525852585258526,
-      "grad_norm": 0.00262451171875,
+      "grad_norm": 0.002532958984375,
       "learning_rate": 5.389179288543877e-05,
       "loss": 0.0001,
       "step": 2966
     },
     {
       "epoch": 6.528052805280528,
-      "grad_norm": 0.0103759765625,
+      "grad_norm": 0.003509521484375,
       "learning_rate": 5.3830264998485535e-05,
       "loss": 0.0001,
       "step": 2967
     },
     {
       "epoch": 6.53025302530253,
-      "grad_norm": 0.005035400390625,
+      "grad_norm": 0.0057373046875,
       "learning_rate": 5.376875931703984e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2968
     },
     {
       "epoch": 6.5324532453245325,
-      "grad_norm": 0.005279541015625,
+      "grad_norm": 0.004669189453125,
       "learning_rate": 5.370727587068325e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2969
     },
     {
       "epoch": 6.534653465346535,
-      "grad_norm": 0.004058837890625,
+      "grad_norm": 0.00799560546875,
       "learning_rate": 5.364581468898628e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2970
     },
     {
       "epoch": 6.536853685368537,
-      "grad_norm": 0.00179290771484375,
+      "grad_norm": 0.003448486328125,
       "learning_rate": 5.3584375801509066e-05,
       "loss": 0.0001,
       "step": 2971
     },
     {
       "epoch": 6.539053905390539,
-      "grad_norm": 0.0084228515625,
+      "grad_norm": 0.0023040771484375,
       "learning_rate": 5.352295923780085e-05,
       "loss": 0.0001,
       "step": 2972
     },
     {
       "epoch": 6.541254125412541,
-      "grad_norm": 0.0201416015625,
+      "grad_norm": 0.01397705078125,
       "learning_rate": 5.346156502740004e-05,
-      "loss": 0.0005,
+      "loss": 0.0002,
       "step": 2973
     },
     {
       "epoch": 6.543454345434544,
-      "grad_norm": 0.00726318359375,
+      "grad_norm": 0.0024566650390625,
       "learning_rate": 5.340019319983458e-05,
       "loss": 0.0001,
       "step": 2974
     },
     {
       "epoch": 6.5456545654565454,
-      "grad_norm": 0.003448486328125,
+      "grad_norm": 0.003997802734375,
       "learning_rate": 5.3338843784621306e-05,
       "loss": 0.0001,
       "step": 2975
     },
     {
       "epoch": 6.547854785478548,
-      "grad_norm": 0.0042724609375,
+      "grad_norm": 0.0028076171875,
       "learning_rate": 5.327751681126664e-05,
       "loss": 0.0001,
       "step": 2976
     },
     {
       "epoch": 6.55005500550055,
-      "grad_norm": 0.00445556640625,
+      "grad_norm": 0.00421142578125,
       "learning_rate": 5.3216212309265864e-05,
       "loss": 0.0001,
       "step": 2977
     },
     {
       "epoch": 6.552255225522552,
-      "grad_norm": 0.0037078857421875,
+      "grad_norm": 0.0024566650390625,
       "learning_rate": 5.315493030810364e-05,
       "loss": 0.0001,
       "step": 2978
     },
     {
       "epoch": 6.554455445544555,
-      "grad_norm": 0.0025787353515625,
+      "grad_norm": 0.0022735595703125,
       "learning_rate": 5.309367083725387e-05,
       "loss": 0.0001,
       "step": 2979
     },
     {
       "epoch": 6.556655665566557,
-      "grad_norm": 0.00311279296875,
+      "grad_norm": 0.0052490234375,
       "learning_rate": 5.3032433926179395e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2980
     },
     {
       "epoch": 6.558855885588558,
-      "grad_norm": 0.002288818359375,
+      "grad_norm": 0.0032806396484375,
       "learning_rate": 5.2971219604332465e-05,
       "loss": 0.0001,
       "step": 2981
     },
     {
       "epoch": 6.561056105610561,
-      "grad_norm": 0.007232666015625,
+      "grad_norm": 0.0031280517578125,
       "learning_rate": 5.291002790115425e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2982
     },
     {
       "epoch": 6.563256325632564,
-      "grad_norm": 0.0038299560546875,
+      "grad_norm": 0.0031280517578125,
       "learning_rate": 5.284885884607511e-05,
       "loss": 0.0001,
       "step": 2983
     },
     {
       "epoch": 6.565456545654565,
-      "grad_norm": 0.003509521484375,
+      "grad_norm": 0.01373291015625,
       "learning_rate": 5.278771246851466e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2984
     },
     {
       "epoch": 6.567656765676568,
-      "grad_norm": 0.00408935546875,
+      "grad_norm": 0.0033111572265625,
       "learning_rate": 5.2726588797881324e-05,
       "loss": 0.0001,
       "step": 2985
     },
     {
       "epoch": 6.56985698569857,
-      "grad_norm": 0.007720947265625,
+      "grad_norm": 0.002685546875,
       "learning_rate": 5.26654878635729e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 2986
     },
     {
       "epoch": 6.572057205720572,
-      "grad_norm": 0.003021240234375,
+      "grad_norm": 0.00579833984375,
       "learning_rate": 5.260440969497602e-05,
       "loss": 0.0001,
       "step": 2987
     },
     {
       "epoch": 6.574257425742574,
-      "grad_norm": 0.00811767578125,
+      "grad_norm": 0.005615234375,
       "learning_rate": 5.254335432146644e-05,
       "loss": 0.0001,
       "step": 2988
     },
     {
       "epoch": 6.5764576457645765,
-      "grad_norm": 0.005828857421875,
+      "grad_norm": 0.002197265625,
       "learning_rate": 5.24823217724091e-05,
       "loss": 0.0001,
       "step": 2989
     },
     {
       "epoch": 6.578657865786578,
-      "grad_norm": 0.00433349609375,
+      "grad_norm": 0.0025787353515625,
       "learning_rate": 5.242131207715768e-05,
       "loss": 0.0001,
       "step": 2990
     },
     {
       "epoch": 6.580858085808581,
-      "grad_norm": 0.0032501220703125,
+      "grad_norm": 0.0029296875,
       "learning_rate": 5.2360325265055165e-05,
       "loss": 0.0001,
       "step": 2991
     },
     {
       "epoch": 6.583058305830583,
-      "grad_norm": 0.0030517578125,
+      "grad_norm": 0.00537109375,
       "learning_rate": 5.229936136543327e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2992
     },
     {
       "epoch": 6.585258525852585,
-      "grad_norm": 0.0032806396484375,
+      "grad_norm": 0.0035247802734375,
       "learning_rate": 5.223842040761284e-05,
       "loss": 0.0001,
       "step": 2993
     },
     {
       "epoch": 6.587458745874588,
-      "grad_norm": 0.00909423828125,
+      "grad_norm": 0.0034942626953125,
       "learning_rate": 5.2177502420903755e-05,
       "loss": 0.0001,
       "step": 2994
     },
     {
       "epoch": 6.589658965896589,
-      "grad_norm": 0.0220947265625,
+      "grad_norm": 0.00714111328125,
       "learning_rate": 5.211660743460458e-05,
-      "loss": 0.0005,
+      "loss": 0.0002,
       "step": 2995
     },
     {
       "epoch": 6.591859185918592,
-      "grad_norm": 0.001708984375,
+      "grad_norm": 0.00701904296875,
       "learning_rate": 5.2055735478003175e-05,
       "loss": 0.0001,
       "step": 2996
     },
     {
       "epoch": 6.594059405940594,
-      "grad_norm": 0.002838134765625,
+      "grad_norm": 0.00421142578125,
       "learning_rate": 5.1994886580375965e-05,
       "loss": 0.0001,
       "step": 2997
     },
     {
       "epoch": 6.596259625962596,
-      "grad_norm": 0.0028228759765625,
+      "grad_norm": 0.005126953125,
       "learning_rate": 5.193406077098861e-05,
       "loss": 0.0001,
       "step": 2998
     },
     {
       "epoch": 6.598459845984598,
-      "grad_norm": 0.003509521484375,
+      "grad_norm": 0.0054931640625,
       "learning_rate": 5.1873258079095397e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 2999
     },
     {
       "epoch": 6.600660066006601,
-      "grad_norm": 0.002960205078125,
+      "grad_norm": 0.004241943359375,
       "learning_rate": 5.181247853393961e-05,
       "loss": 0.0001,
       "step": 3000
     },
     {
       "epoch": 6.602860286028603,
-      "grad_norm": 0.0037384033203125,
+      "grad_norm": 0.00653076171875,
       "learning_rate": 5.1751722164753514e-05,
       "loss": 0.0001,
       "step": 3001
     },
     {
       "epoch": 6.605060506050605,
-      "grad_norm": 0.00360107421875,
+      "grad_norm": 0.00665283203125,
       "learning_rate": 5.169098900075796e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3002
     },
     {
       "epoch": 6.6072607260726075,
-      "grad_norm": 0.004058837890625,
+      "grad_norm": 0.00457763671875,
       "learning_rate": 5.163027907116296e-05,
       "loss": 0.0001,
       "step": 3003
     },
     {
       "epoch": 6.609460946094609,
-      "grad_norm": 0.0024261474609375,
+      "grad_norm": 0.0052490234375,
       "learning_rate": 5.1569592405167036e-05,
       "loss": 0.0001,
       "step": 3004
     },
     {
       "epoch": 6.611661166116612,
-      "grad_norm": 0.002655029296875,
+      "grad_norm": 0.005157470703125,
       "learning_rate": 5.150892903195769e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3005
     },
     {
       "epoch": 6.6138613861386135,
-      "grad_norm": 0.0036468505859375,
+      "grad_norm": 0.005584716796875,
       "learning_rate": 5.14482889807113e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3006
     },
     {
       "epoch": 6.616061606160616,
-      "grad_norm": 0.00732421875,
+      "grad_norm": 0.003326416015625,
       "learning_rate": 5.13876722805928e-05,
       "loss": 0.0001,
       "step": 3007
     },
     {
       "epoch": 6.618261826182618,
-      "grad_norm": 0.00433349609375,
+      "grad_norm": 0.003662109375,
       "learning_rate": 5.1327078960756145e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3008
     },
     {
       "epoch": 6.62046204620462,
-      "grad_norm": 0.0045166015625,
+      "grad_norm": 0.0023345947265625,
       "learning_rate": 5.126650905034382e-05,
       "loss": 0.0001,
       "step": 3009
     },
     {
       "epoch": 6.622662266226623,
-      "grad_norm": 0.0025482177734375,
+      "grad_norm": 0.00408935546875,
       "learning_rate": 5.1205962578487155e-05,
       "loss": 0.0001,
       "step": 3010
     },
     {
       "epoch": 6.624862486248625,
-      "grad_norm": 0.00323486328125,
+      "grad_norm": 0.00457763671875,
       "learning_rate": 5.114543957430631e-05,
       "loss": 0.0001,
       "step": 3011
     },
     {
       "epoch": 6.627062706270627,
-      "grad_norm": 0.001983642578125,
+      "grad_norm": 0.00567626953125,
       "learning_rate": 5.108494006690989e-05,
       "loss": 0.0001,
       "step": 3012
     },
     {
       "epoch": 6.629262926292629,
-      "grad_norm": 0.006011962890625,
+      "grad_norm": 0.0030670166015625,
       "learning_rate": 5.102446408539554e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3013
     },
     {
       "epoch": 6.631463146314632,
-      "grad_norm": 0.004669189453125,
+      "grad_norm": 0.00408935546875,
       "learning_rate": 5.096401165884926e-05,
       "loss": 0.0001,
       "step": 3014
     },
     {
       "epoch": 6.633663366336633,
-      "grad_norm": 0.00213623046875,
+      "grad_norm": 0.0040283203125,
       "learning_rate": 5.090358281634594e-05,
       "loss": 0.0001,
       "step": 3015
     },
     {
       "epoch": 6.635863586358636,
-      "grad_norm": 0.0030364990234375,
+      "grad_norm": 0.0045166015625,
       "learning_rate": 5.084317758694904e-05,
       "loss": 0.0001,
       "step": 3016
     },
     {
       "epoch": 6.638063806380638,
-      "grad_norm": 0.00384521484375,
+      "grad_norm": 0.0050048828125,
       "learning_rate": 5.0782795999710654e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3017
     },
     {
       "epoch": 6.64026402640264,
-      "grad_norm": 0.00299072265625,
+      "grad_norm": 0.006195068359375,
       "learning_rate": 5.0722438083671654e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3018
     },
     {
       "epoch": 6.642464246424643,
-      "grad_norm": 0.0032806396484375,
+      "grad_norm": 0.004180908203125,
       "learning_rate": 5.0662103867861276e-05,
       "loss": 0.0001,
       "step": 3019
     },
     {
       "epoch": 6.6446644664466445,
-      "grad_norm": 0.0026092529296875,
+      "grad_norm": 0.004486083984375,
       "learning_rate": 5.0601793381297534e-05,
       "loss": 0.0001,
       "step": 3020
     },
     {
       "epoch": 6.646864686468647,
-      "grad_norm": 0.00189208984375,
+      "grad_norm": 0.0032806396484375,
       "learning_rate": 5.054150665298698e-05,
       "loss": 0.0001,
       "step": 3021
     },
     {
       "epoch": 6.649064906490649,
-      "grad_norm": 0.0029296875,
+      "grad_norm": 0.00347900390625,
       "learning_rate": 5.0481243711924705e-05,
       "loss": 0.0001,
       "step": 3022
     },
     {
       "epoch": 6.6512651265126514,
-      "grad_norm": 0.0030364990234375,
+      "grad_norm": 0.00372314453125,
       "learning_rate": 5.042100458709452e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3023
     },
     {
       "epoch": 6.653465346534653,
-      "grad_norm": 0.00286865234375,
+      "grad_norm": 0.002288818359375,
       "learning_rate": 5.0360789307468505e-05,
       "loss": 0.0001,
       "step": 3024
     },
     {
       "epoch": 6.655665566556656,
-      "grad_norm": 0.0023956298828125,
+      "grad_norm": 0.0028533935546875,
       "learning_rate": 5.030059790200756e-05,
       "loss": 0.0001,
       "step": 3025
     },
     {
       "epoch": 6.6578657865786575,
-      "grad_norm": 0.0035858154296875,
+      "grad_norm": 0.005096435546875,
       "learning_rate": 5.0240430399660885e-05,
       "loss": 0.0002,
       "step": 3026
     },
     {
       "epoch": 6.66006600660066,
-      "grad_norm": 0.0027618408203125,
+      "grad_norm": 0.0038604736328125,
       "learning_rate": 5.018028682936624e-05,
       "loss": 0.0001,
       "step": 3027
     },
     {
       "epoch": 6.662266226622663,
-      "grad_norm": 0.002197265625,
+      "grad_norm": 0.003936767578125,
       "learning_rate": 5.0120167220050017e-05,
       "loss": 0.0001,
       "step": 3028
     },
     {
       "epoch": 6.664466446644664,
-      "grad_norm": 0.004150390625,
+      "grad_norm": 0.0126953125,
       "learning_rate": 5.006007160062682e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3029
     },
     {
       "epoch": 6.666666666666667,
-      "grad_norm": 0.003997802734375,
+      "grad_norm": 0.00433349609375,
       "learning_rate": 5.000000000000002e-05,
       "loss": 0.0001,
       "step": 3030
     },
     {
       "epoch": 6.668866886688669,
-      "grad_norm": 0.004150390625,
+      "grad_norm": 0.0036773681640625,
       "learning_rate": 4.9939952447061166e-05,
       "loss": 0.0001,
       "step": 3031
     },
     {
       "epoch": 6.671067106710671,
-      "grad_norm": 0.00347900390625,
+      "grad_norm": 0.0032501220703125,
       "learning_rate": 4.9879928970690345e-05,
       "loss": 0.0001,
       "step": 3032
     },
     {
       "epoch": 6.673267326732673,
-      "grad_norm": 0.00396728515625,
+      "grad_norm": 0.0020751953125,
       "learning_rate": 4.98199295997562e-05,
       "loss": 0.0001,
       "step": 3033
     },
     {
       "epoch": 6.675467546754676,
-      "grad_norm": 0.029296875,
+      "grad_norm": 0.0023193359375,
       "learning_rate": 4.975995436311549e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3034
     },
     {
       "epoch": 6.677667766776677,
-      "grad_norm": 0.003082275390625,
+      "grad_norm": 0.0042724609375,
       "learning_rate": 4.9700003289613694e-05,
       "loss": 0.0001,
       "step": 3035
     },
     {
       "epoch": 6.67986798679868,
-      "grad_norm": 0.0027923583984375,
+      "grad_norm": 0.003814697265625,
       "learning_rate": 4.964007640808438e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3036
     },
     {
       "epoch": 6.6820682068206825,
-      "grad_norm": 0.0035552978515625,
+      "grad_norm": 0.0034332275390625,
       "learning_rate": 4.9580173747349654e-05,
       "loss": 0.0001,
       "step": 3037
     },
     {
       "epoch": 6.684268426842684,
-      "grad_norm": 0.003082275390625,
+      "grad_norm": 0.00311279296875,
       "learning_rate": 4.9520295336219924e-05,
       "loss": 0.0001,
       "step": 3038
     },
     {
       "epoch": 6.686468646864687,
-      "grad_norm": 0.015625,
+      "grad_norm": 0.0023040771484375,
       "learning_rate": 4.94604412034939e-05,
       "loss": 0.0001,
       "step": 3039
     },
     {
       "epoch": 6.6886688668866885,
-      "grad_norm": 0.0042724609375,
+      "grad_norm": 0.0038604736328125,
       "learning_rate": 4.940061137795876e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3040
     },
     {
       "epoch": 6.690869086908691,
-      "grad_norm": 0.0107421875,
+      "grad_norm": 0.0140380859375,
       "learning_rate": 4.934080588838977e-05,
       "loss": 0.0003,
       "step": 3041
     },
     {
       "epoch": 6.693069306930693,
-      "grad_norm": 0.0018768310546875,
+      "grad_norm": 0.004302978515625,
       "learning_rate": 4.928102476355065e-05,
       "loss": 0.0001,
       "step": 3042
     },
     {
       "epoch": 6.695269526952695,
-      "grad_norm": 0.00439453125,
+      "grad_norm": 0.003387451171875,
       "learning_rate": 4.922126803219336e-05,
       "loss": 0.0001,
       "step": 3043
     },
     {
       "epoch": 6.697469746974697,
-      "grad_norm": 0.007537841796875,
+      "grad_norm": 0.00433349609375,
       "learning_rate": 4.916153572305807e-05,
       "loss": 0.0001,
       "step": 3044
     },
     {
       "epoch": 6.6996699669967,
-      "grad_norm": 0.0032501220703125,
+      "grad_norm": 0.004180908203125,
       "learning_rate": 4.9101827864873384e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3045
     },
     {
       "epoch": 6.701870187018702,
-      "grad_norm": 0.0023345947265625,
+      "grad_norm": 0.0045166015625,
       "learning_rate": 4.904214448635591e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3046
     },
     {
       "epoch": 6.704070407040704,
-      "grad_norm": 0.003021240234375,
+      "grad_norm": 0.006378173828125,
       "learning_rate": 4.898248561621063e-05,
       "loss": 0.0001,
       "step": 3047
     },
     {
       "epoch": 6.706270627062707,
-      "grad_norm": 0.007110595703125,
+      "grad_norm": 0.0042724609375,
       "learning_rate": 4.892285128313069e-05,
       "loss": 0.0002,
       "step": 3048
     },
     {
       "epoch": 6.708470847084708,
-      "grad_norm": 0.003448486328125,
+      "grad_norm": 0.01055908203125,
       "learning_rate": 4.886324151579742e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3049
     },
     {
       "epoch": 6.710671067106711,
-      "grad_norm": 0.0028228759765625,
+      "grad_norm": 0.003326416015625,
       "learning_rate": 4.8803656342880477e-05,
       "loss": 0.0001,
       "step": 3050
     },
     {
       "epoch": 6.712871287128713,
-      "grad_norm": 0.003143310546875,
+      "grad_norm": 0.00921630859375,
       "learning_rate": 4.8744095793037415e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3051
     },
     {
       "epoch": 6.715071507150715,
-      "grad_norm": 0.00592041015625,
+      "grad_norm": 0.004486083984375,
       "learning_rate": 4.868455989491426e-05,
       "loss": 0.0001,
       "step": 3052
     },
     {
       "epoch": 6.717271727172717,
-      "grad_norm": 0.003173828125,
+      "grad_norm": 0.0023193359375,
       "learning_rate": 4.862504867714491e-05,
       "loss": 0.0001,
       "step": 3053
     },
     {
       "epoch": 6.7194719471947195,
-      "grad_norm": 0.006561279296875,
+      "grad_norm": 0.0050048828125,
       "learning_rate": 4.856556216835155e-05,
       "loss": 0.0002,
       "step": 3054
     },
     {
       "epoch": 6.721672167216722,
-      "grad_norm": 0.0020294189453125,
+      "grad_norm": 0.005828857421875,
       "learning_rate": 4.850610039714444e-05,
       "loss": 0.0001,
       "step": 3055
     },
     {
       "epoch": 6.723872387238724,
-      "grad_norm": 0.003448486328125,
+      "grad_norm": 0.00421142578125,
       "learning_rate": 4.844666339212189e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3056
     },
     {
       "epoch": 6.726072607260726,
-      "grad_norm": 0.002716064453125,
+      "grad_norm": 0.0042724609375,
       "learning_rate": 4.83872511818705e-05,
       "loss": 0.0001,
       "step": 3057
     },
     {
       "epoch": 6.728272827282728,
-      "grad_norm": 0.0024871826171875,
+      "grad_norm": 0.01007080078125,
       "learning_rate": 4.832786379496465e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3058
     },
     {
       "epoch": 6.730473047304731,
-      "grad_norm": 0.003021240234375,
+      "grad_norm": 0.0021209716796875,
       "learning_rate": 4.826850125996697e-05,
       "loss": 0.0001,
       "step": 3059
     },
     {
       "epoch": 6.732673267326732,
-      "grad_norm": 0.0038299560546875,
+      "grad_norm": 0.004364013671875,
       "learning_rate": 4.82091636054281e-05,
       "loss": 0.0001,
       "step": 3060
     },
     {
       "epoch": 6.734873487348735,
-      "grad_norm": 0.0027008056640625,
+      "grad_norm": 0.00494384765625,
       "learning_rate": 4.8149850859886667e-05,
       "loss": 0.0001,
       "step": 3061
     },
     {
       "epoch": 6.737073707370737,
-      "grad_norm": 0.002899169921875,
+      "grad_norm": 0.003753662109375,
       "learning_rate": 4.809056305186949e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3062
     },
     {
       "epoch": 6.739273927392739,
-      "grad_norm": 0.00750732421875,
+      "grad_norm": 0.01904296875,
       "learning_rate": 4.8031300209891114e-05,
       "loss": 0.0002,
       "step": 3063
     },
     {
       "epoch": 6.741474147414742,
-      "grad_norm": 0.0034027099609375,
+      "grad_norm": 0.02490234375,
       "learning_rate": 4.797206236245429e-05,
-      "loss": 0.0001,
+      "loss": 0.0004,
       "step": 3064
     },
     {
       "epoch": 6.743674367436744,
-      "grad_norm": 0.0032501220703125,
+      "grad_norm": 0.004150390625,
       "learning_rate": 4.791284953804969e-05,
       "loss": 0.0001,
       "step": 3065
     },
     {
       "epoch": 6.745874587458746,
-      "grad_norm": 0.002227783203125,
+      "grad_norm": 0.0031280517578125,
       "learning_rate": 4.785366176515589e-05,
       "loss": 0.0001,
       "step": 3066
     },
     {
       "epoch": 6.748074807480748,
-      "grad_norm": 0.072265625,
+      "grad_norm": 0.005645751953125,
       "learning_rate": 4.7794499072239586e-05,
-      "loss": 0.0004,
+      "loss": 0.0002,
       "step": 3067
     },
     {
       "epoch": 6.7502750275027505,
-      "grad_norm": 0.00421142578125,
+      "grad_norm": 0.0026397705078125,
       "learning_rate": 4.773536148775518e-05,
       "loss": 0.0001,
       "step": 3068
     },
     {
       "epoch": 6.752475247524752,
-      "grad_norm": 0.0019683837890625,
+      "grad_norm": 0.0020751953125,
       "learning_rate": 4.767624904014517e-05,
       "loss": 0.0001,
       "step": 3069
     },
     {
       "epoch": 6.754675467546755,
-      "grad_norm": 0.003143310546875,
+      "grad_norm": 0.004638671875,
       "learning_rate": 4.761716175783989e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3070
     },
     {
       "epoch": 6.756875687568757,
-      "grad_norm": 0.0045166015625,
+      "grad_norm": 0.0027008056640625,
       "learning_rate": 4.755809966925757e-05,
       "loss": 0.0001,
       "step": 3071
     },
     {
       "epoch": 6.759075907590759,
-      "grad_norm": 0.00177001953125,
+      "grad_norm": 0.0020294189453125,
       "learning_rate": 4.749906280280446e-05,
       "loss": 0.0001,
       "step": 3072
     },
     {
       "epoch": 6.761276127612762,
-      "grad_norm": 0.002838134765625,
+      "grad_norm": 0.005096435546875,
       "learning_rate": 4.7440051186874434e-05,
       "loss": 0.0001,
       "step": 3073
     },
     {
       "epoch": 6.7634763476347635,
-      "grad_norm": 0.0025787353515625,
+      "grad_norm": 0.0108642578125,
       "learning_rate": 4.738106484984939e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3074
     },
     {
       "epoch": 6.765676567656766,
-      "grad_norm": 0.002227783203125,
+      "grad_norm": 0.0030059814453125,
       "learning_rate": 4.732210382009904e-05,
       "loss": 0.0001,
       "step": 3075
     },
     {
       "epoch": 6.767876787678768,
-      "grad_norm": 0.0027008056640625,
+      "grad_norm": 0.004669189453125,
       "learning_rate": 4.7263168125980895e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3076
     },
     {
       "epoch": 6.77007700770077,
-      "grad_norm": 0.00665283203125,
+      "grad_norm": 0.0057373046875,
       "learning_rate": 4.720425779584032e-05,
       "loss": 0.0002,
       "step": 3077
     },
     {
       "epoch": 6.772277227722772,
-      "grad_norm": 0.00408935546875,
+      "grad_norm": 0.00445556640625,
       "learning_rate": 4.714537285801041e-05,
       "loss": 0.0001,
       "step": 3078
     },
     {
       "epoch": 6.772277227722772,
-      "eval_loss": 0.0002463853161316365,
-      "eval_runtime": 10.2165,
-      "eval_samples_per_second": 37.488,
-      "eval_steps_per_second": 4.698,
+      "eval_loss": 0.00036034645745530725,
+      "eval_runtime": 10.0597,
+      "eval_samples_per_second": 38.073,
+      "eval_steps_per_second": 4.772,
       "step": 3078
     },
     {
       "epoch": 6.774477447744775,
-      "grad_norm": 0.003448486328125,
+      "grad_norm": 0.005523681640625,
       "learning_rate": 4.708651334081221e-05,
       "loss": 0.0001,
       "step": 3079
     },
     {
       "epoch": 6.776677667766776,
-      "grad_norm": 0.005035400390625,
+      "grad_norm": 0.01007080078125,
       "learning_rate": 4.702767927255432e-05,
       "loss": 0.0002,
       "step": 3080
     },
     {
       "epoch": 6.778877887788779,
-      "grad_norm": 0.0040283203125,
+      "grad_norm": 0.00433349609375,
       "learning_rate": 4.696887068153323e-05,
       "loss": 0.0002,
       "step": 3081
     },
     {
       "epoch": 6.781078107810782,
-      "grad_norm": 0.0034027099609375,
+      "grad_norm": 0.0089111328125,
       "learning_rate": 4.691008759603316e-05,
       "loss": 0.0001,
       "step": 3082
     },
     {
       "epoch": 6.783278327832783,
-      "grad_norm": 0.005950927734375,
+      "grad_norm": 0.006317138671875,
       "learning_rate": 4.6851330044326025e-05,
       "loss": 0.0002,
       "step": 3083
     },
     {
       "epoch": 6.785478547854785,
-      "grad_norm": 0.0022430419921875,
+      "grad_norm": 0.0050048828125,
       "learning_rate": 4.67925980546716e-05,
       "loss": 0.0001,
       "step": 3084
     },
     {
       "epoch": 6.787678767876788,
-      "grad_norm": 0.002685546875,
+      "grad_norm": 0.003997802734375,
       "learning_rate": 4.673389165531714e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3085
     },
     {
       "epoch": 6.78987898789879,
-      "grad_norm": 0.0048828125,
+      "grad_norm": 0.004669189453125,
       "learning_rate": 4.667521087449776e-05,
       "loss": 0.0002,
       "step": 3086
     },
     {
       "epoch": 6.792079207920792,
-      "grad_norm": 0.002655029296875,
+      "grad_norm": 0.003875732421875,
       "learning_rate": 4.6616555740436176e-05,
       "loss": 0.0001,
       "step": 3087
     },
     {
       "epoch": 6.7942794279427945,
-      "grad_norm": 0.003997802734375,
+      "grad_norm": 0.01434326171875,
       "learning_rate": 4.655792628134277e-05,
-      "loss": 0.0001,
+      "loss": 0.0003,
       "step": 3088
     },
     {
       "epoch": 6.796479647964796,
-      "grad_norm": 0.003509521484375,
+      "grad_norm": 0.00238037109375,
       "learning_rate": 4.649932252541572e-05,
       "loss": 0.0001,
       "step": 3089
     },
     {
       "epoch": 6.798679867986799,
-      "grad_norm": 0.00191497802734375,
+      "grad_norm": 0.0027008056640625,
       "learning_rate": 4.6440744500840605e-05,
       "loss": 0.0001,
       "step": 3090
     },
     {
       "epoch": 6.8008800880088005,
-      "grad_norm": 0.00439453125,
+      "grad_norm": 0.007293701171875,
       "learning_rate": 4.638219223579079e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3091
     },
     {
       "epoch": 6.803080308030803,
-      "grad_norm": 0.0022125244140625,
+      "grad_norm": 0.00225830078125,
       "learning_rate": 4.632366575842719e-05,
       "loss": 0.0001,
       "step": 3092
     },
     {
       "epoch": 6.805280528052805,
-      "grad_norm": 0.004241943359375,
+      "grad_norm": 0.005523681640625,
       "learning_rate": 4.6265165096898356e-05,
       "loss": 0.0001,
       "step": 3093
     },
     {
       "epoch": 6.807480748074807,
-      "grad_norm": 0.003631591796875,
+      "grad_norm": 0.003875732421875,
       "learning_rate": 4.620669027934038e-05,
       "loss": 0.0002,
       "step": 3094
     },
     {
       "epoch": 6.80968096809681,
-      "grad_norm": 0.0036163330078125,
+      "grad_norm": 0.00738525390625,
       "learning_rate": 4.6148241333876974e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3095
     },
     {
       "epoch": 6.811881188118812,
-      "grad_norm": 0.00213623046875,
+      "grad_norm": 0.0028228759765625,
       "learning_rate": 4.6089818288619344e-05,
       "loss": 0.0001,
       "step": 3096
     },
     {
       "epoch": 6.814081408140814,
-      "grad_norm": 0.00384521484375,
+      "grad_norm": 0.003082275390625,
       "learning_rate": 4.603142117166629e-05,
       "loss": 0.0001,
       "step": 3097
     },
     {
       "epoch": 6.816281628162816,
-      "grad_norm": 0.0029296875,
+      "grad_norm": 0.005279541015625,
       "learning_rate": 4.5973050011104126e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3098
     },
     {
       "epoch": 6.818481848184819,
-      "grad_norm": 0.00823974609375,
+      "grad_norm": 0.0069580078125,
       "learning_rate": 4.5914704835006686e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3099
     },
     {
       "epoch": 6.82068206820682,
-      "grad_norm": 0.005035400390625,
+      "grad_norm": 0.003936767578125,
       "learning_rate": 4.585638567143529e-05,
       "loss": 0.0001,
       "step": 3100
     },
     {
       "epoch": 6.822882288228823,
-      "grad_norm": 0.0019378662109375,
+      "grad_norm": 0.0022125244140625,
       "learning_rate": 4.5798092548438766e-05,
       "loss": 0.0001,
       "step": 3101
     },
     {
       "epoch": 6.825082508250825,
-      "grad_norm": 0.0035552978515625,
+      "grad_norm": 0.0025482177734375,
       "learning_rate": 4.57398254940534e-05,
       "loss": 0.0001,
       "step": 3102
     },
     {
       "epoch": 6.827282728272827,
-      "grad_norm": 0.0025634765625,
+      "grad_norm": 0.00518798828125,
       "learning_rate": 4.5681584536302966e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3103
     },
     {
       "epoch": 6.82948294829483,
-      "grad_norm": 0.00433349609375,
+      "grad_norm": 0.007598876953125,
       "learning_rate": 4.562336970319867e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3104
     },
     {
       "epoch": 6.8316831683168315,
-      "grad_norm": 0.0025634765625,
+      "grad_norm": 0.003173828125,
       "learning_rate": 4.55651810227391e-05,
       "loss": 0.0001,
       "step": 3105
     },
     {
       "epoch": 6.833883388338834,
-      "grad_norm": 0.004974365234375,
+      "grad_norm": 0.006683349609375,
       "learning_rate": 4.550701852291046e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3106
     },
     {
       "epoch": 6.836083608360836,
-      "grad_norm": 0.007598876953125,
+      "grad_norm": 0.00396728515625,
       "learning_rate": 4.5448882231686085e-05,
       "loss": 0.0001,
       "step": 3107
     },
     {
       "epoch": 6.838283828382838,
-      "grad_norm": 0.0019683837890625,
+      "grad_norm": 0.004364013671875,
       "learning_rate": 4.5390772177026893e-05,
       "loss": 0.0001,
       "step": 3108
     },
     {
       "epoch": 6.84048404840484,
-      "grad_norm": 0.0032958984375,
+      "grad_norm": 0.0036773681640625,
       "learning_rate": 4.533268838688114e-05,
       "loss": 0.0001,
       "step": 3109
     },
     {
       "epoch": 6.842684268426843,
-      "grad_norm": 0.001953125,
+      "grad_norm": 0.005096435546875,
       "learning_rate": 4.527463088918439e-05,
       "loss": 0.0001,
       "step": 3110
     },
     {
       "epoch": 6.8448844884488445,
-      "grad_norm": 0.0196533203125,
+      "grad_norm": 0.0030364990234375,
       "learning_rate": 4.5216599711859744e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3111
     },
     {
       "epoch": 6.847084708470847,
-      "grad_norm": 0.0034027099609375,
+      "grad_norm": 0.00445556640625,
       "learning_rate": 4.515859488281739e-05,
       "loss": 0.0001,
       "step": 3112
     },
     {
       "epoch": 6.84928492849285,
-      "grad_norm": 0.0037689208984375,
+      "grad_norm": 0.0106201171875,
       "learning_rate": 4.510061642995499e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3113
     },
     {
       "epoch": 6.851485148514851,
-      "grad_norm": 0.004302978515625,
+      "grad_norm": 0.004791259765625,
       "learning_rate": 4.504266438115754e-05,
       "loss": 0.0002,
       "step": 3114
     },
     {
       "epoch": 6.853685368536854,
-      "grad_norm": 0.0033111572265625,
+      "grad_norm": 0.0034027099609375,
       "learning_rate": 4.498473876429726e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3115
     },
     {
       "epoch": 6.855885588558856,
-      "grad_norm": 0.00433349609375,
+      "grad_norm": 0.0054931640625,
       "learning_rate": 4.492683960723372e-05,
       "loss": 0.0001,
       "step": 3116
     },
     {
       "epoch": 6.858085808580858,
-      "grad_norm": 0.0027008056640625,
+      "grad_norm": 0.00250244140625,
       "learning_rate": 4.486896693781372e-05,
       "loss": 0.0001,
       "step": 3117
     },
     {
       "epoch": 6.86028602860286,
-      "grad_norm": 0.002777099609375,
+      "grad_norm": 0.005859375,
       "learning_rate": 4.4811120783871344e-05,
       "loss": 0.0001,
       "step": 3118
     },
     {
       "epoch": 6.862486248624863,
-      "grad_norm": 0.0022430419921875,
+      "grad_norm": 0.0031585693359375,
       "learning_rate": 4.47533011732279e-05,
       "loss": 0.0001,
       "step": 3119
     },
     {
       "epoch": 6.864686468646864,
-      "grad_norm": 0.0038604736328125,
+      "grad_norm": 0.00445556640625,
       "learning_rate": 4.4695508133691975e-05,
       "loss": 0.0002,
       "step": 3120
     },
     {
       "epoch": 6.866886688668867,
-      "grad_norm": 0.002899169921875,
+      "grad_norm": 0.004608154296875,
       "learning_rate": 4.4637741693059346e-05,
       "loss": 0.0001,
       "step": 3121
     },
     {
       "epoch": 6.8690869086908695,
-      "grad_norm": 0.0030670166015625,
+      "grad_norm": 0.00433349609375,
       "learning_rate": 4.458000187911298e-05,
       "loss": 0.0001,
       "step": 3122
     },
     {
       "epoch": 6.871287128712871,
-      "grad_norm": 0.006072998046875,
+      "grad_norm": 0.01220703125,
       "learning_rate": 4.4522288719623064e-05,
       "loss": 0.0002,
       "step": 3123
     },
     {
       "epoch": 6.873487348734874,
-      "grad_norm": 0.0033721923828125,
+      "grad_norm": 0.01190185546875,
       "learning_rate": 4.446460224234696e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3124
     },
     {
       "epoch": 6.8756875687568755,
-      "grad_norm": 0.00732421875,
+      "grad_norm": 0.006103515625,
       "learning_rate": 4.4406942475029214e-05,
       "loss": 0.0001,
       "step": 3125
     },
     {
       "epoch": 6.877887788778878,
-      "grad_norm": 0.0159912109375,
+      "grad_norm": 0.0054931640625,
       "learning_rate": 4.434930944540149e-05,
-      "loss": 0.0004,
+      "loss": 0.0002,
       "step": 3126
     },
     {
       "epoch": 6.88008800880088,
-      "grad_norm": 0.00469970703125,
+      "grad_norm": 0.007080078125,
       "learning_rate": 4.42917031811826e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3127
     },
     {
       "epoch": 6.882288228822882,
-      "grad_norm": 0.0033721923828125,
+      "grad_norm": 0.004730224609375,
       "learning_rate": 4.4234123710078514e-05,
       "loss": 0.0001,
       "step": 3128
     },
     {
       "epoch": 6.884488448844884,
-      "grad_norm": 0.0024871826171875,
+      "grad_norm": 0.003570556640625,
       "learning_rate": 4.417657105978229e-05,
       "loss": 0.0001,
       "step": 3129
     },
     {
       "epoch": 6.886688668866887,
-      "grad_norm": 0.0034942626953125,
+      "grad_norm": 0.00299072265625,
       "learning_rate": 4.411904525797408e-05,
       "loss": 0.0001,
       "step": 3130
     },
     {
       "epoch": 6.888888888888889,
-      "grad_norm": 0.00225830078125,
+      "grad_norm": 0.00445556640625,
       "learning_rate": 4.406154633232114e-05,
       "loss": 0.0001,
       "step": 3131
     },
     {
       "epoch": 6.891089108910891,
-      "grad_norm": 0.0029296875,
+      "grad_norm": 0.00726318359375,
       "learning_rate": 4.4004074310477816e-05,
       "loss": 0.0001,
       "step": 3132
     },
     {
       "epoch": 6.893289328932894,
-      "grad_norm": 0.0032501220703125,
+      "grad_norm": 0.0019683837890625,
       "learning_rate": 4.394662922008547e-05,
       "loss": 0.0001,
       "step": 3133
     },
     {
       "epoch": 6.895489548954895,
-      "grad_norm": 0.0093994140625,
+      "grad_norm": 0.00360107421875,
       "learning_rate": 4.388921108877253e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3134
     },
     {
       "epoch": 6.897689768976898,
-      "grad_norm": 0.0023956298828125,
+      "grad_norm": 0.00830078125,
       "learning_rate": 4.3831819944154485e-05,
       "loss": 0.0001,
       "step": 3135
     },
     {
       "epoch": 6.8998899889989,
-      "grad_norm": 0.006072998046875,
+      "grad_norm": 0.0068359375,
       "learning_rate": 4.377445581383379e-05,
       "loss": 0.0002,
       "step": 3136
     },
     {
       "epoch": 6.902090209020902,
-      "grad_norm": 0.00188446044921875,
+      "grad_norm": 0.0042724609375,
       "learning_rate": 4.3717118725399976e-05,
       "loss": 0.0001,
       "step": 3137
     },
     {
       "epoch": 6.904290429042904,
-      "grad_norm": 0.0030059814453125,
+      "grad_norm": 0.004425048828125,
       "learning_rate": 4.36598087064295e-05,
       "loss": 0.0001,
       "step": 3138
     },
     {
       "epoch": 6.9064906490649065,
-      "grad_norm": 0.005279541015625,
+      "grad_norm": 0.004852294921875,
       "learning_rate": 4.360252578448585e-05,
       "loss": 0.0001,
       "step": 3139
     },
     {
       "epoch": 6.908690869086909,
-      "grad_norm": 0.003265380859375,
+      "grad_norm": 0.006103515625,
       "learning_rate": 4.354526998711945e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3140
     },
     {
       "epoch": 6.910891089108911,
-      "grad_norm": 0.0034942626953125,
+      "grad_norm": 0.003326416015625,
       "learning_rate": 4.34880413418677e-05,
       "loss": 0.0001,
       "step": 3141
     },
     {
       "epoch": 6.913091309130913,
-      "grad_norm": 0.0079345703125,
+      "grad_norm": 0.0031890869140625,
       "learning_rate": 4.343083987625494e-05,
       "loss": 0.0001,
       "step": 3142
     },
     {
       "epoch": 6.915291529152915,
-      "grad_norm": 0.003875732421875,
+      "grad_norm": 0.005126953125,
       "learning_rate": 4.33736656177924e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3143
     },
     {
       "epoch": 6.917491749174918,
-      "grad_norm": 0.00421142578125,
+      "grad_norm": 0.002593994140625,
       "learning_rate": 4.33165185939783e-05,
       "loss": 0.0001,
       "step": 3144
     },
     {
       "epoch": 6.919691969196919,
-      "grad_norm": 0.0034637451171875,
+      "grad_norm": 0.00396728515625,
       "learning_rate": 4.325939883229766e-05,
       "loss": 0.0001,
       "step": 3145
     },
     {
       "epoch": 6.921892189218922,
-      "grad_norm": 0.003936767578125,
+      "grad_norm": 0.005828857421875,
       "learning_rate": 4.320230636022249e-05,
       "loss": 0.0001,
       "step": 3146
     },
     {
       "epoch": 6.924092409240924,
-      "grad_norm": 0.005828857421875,
+      "grad_norm": 0.003082275390625,
       "learning_rate": 4.3145241205211614e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3147
     },
     {
       "epoch": 6.926292629262926,
-      "grad_norm": 0.0030975341796875,
+      "grad_norm": 0.002899169921875,
       "learning_rate": 4.3088203394710726e-05,
       "loss": 0.0001,
       "step": 3148
     },
     {
       "epoch": 6.928492849284929,
-      "grad_norm": 0.002166748046875,
+      "grad_norm": 0.005401611328125,
       "learning_rate": 4.303119295615239e-05,
       "loss": 0.0001,
       "step": 3149
     },
     {
       "epoch": 6.930693069306931,
-      "grad_norm": 0.0027313232421875,
+      "grad_norm": 0.0059814453125,
       "learning_rate": 4.2974209916955975e-05,
       "loss": 0.0001,
       "step": 3150
     },
     {
       "epoch": 6.932893289328933,
-      "grad_norm": 0.0025482177734375,
+      "grad_norm": 0.0045166015625,
       "learning_rate": 4.291725430452769e-05,
       "loss": 0.0001,
       "step": 3151
     },
     {
       "epoch": 6.935093509350935,
-      "grad_norm": 0.002105712890625,
+      "grad_norm": 0.0025634765625,
       "learning_rate": 4.2860326146260565e-05,
       "loss": 0.0001,
       "step": 3152
     },
     {
       "epoch": 6.9372937293729375,
-      "grad_norm": 0.00439453125,
+      "grad_norm": 0.0025787353515625,
       "learning_rate": 4.28034254695344e-05,
       "loss": 0.0001,
       "step": 3153
     },
     {
       "epoch": 6.939493949394939,
-      "grad_norm": 0.0020904541015625,
+      "grad_norm": 0.003753662109375,
       "learning_rate": 4.274655230171579e-05,
       "loss": 0.0001,
       "step": 3154
     },
     {
       "epoch": 6.941694169416942,
-      "grad_norm": 0.003936767578125,
+      "grad_norm": 0.007354736328125,
       "learning_rate": 4.2689706670158104e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3155
     },
     {
       "epoch": 6.9438943894389435,
-      "grad_norm": 0.005584716796875,
+      "grad_norm": 0.0038604736328125,
       "learning_rate": 4.263288860220146e-05,
       "loss": 0.0001,
       "step": 3156
     },
     {
       "epoch": 6.946094609460946,
-      "grad_norm": 0.01165771484375,
+      "grad_norm": 0.0032196044921875,
       "learning_rate": 4.2576098125172715e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3157
     },
     {
       "epoch": 6.948294829482949,
-      "grad_norm": 0.003753662109375,
+      "grad_norm": 0.004730224609375,
       "learning_rate": 4.251933526638547e-05,
       "loss": 0.0002,
       "step": 3158
     },
     {
       "epoch": 6.9504950495049505,
-      "grad_norm": 0.0032196044921875,
+      "grad_norm": 0.009521484375,
       "learning_rate": 4.246260005314003e-05,
       "loss": 0.0001,
       "step": 3159
     },
     {
       "epoch": 6.952695269526953,
-      "grad_norm": 0.006103515625,
+      "grad_norm": 0.00408935546875,
       "learning_rate": 4.240589251272342e-05,
       "loss": 0.0001,
       "step": 3160
     },
     {
       "epoch": 6.954895489548955,
-      "grad_norm": 0.00244140625,
+      "grad_norm": 0.0028076171875,
       "learning_rate": 4.2349212672409343e-05,
       "loss": 0.0001,
       "step": 3161
     },
     {
       "epoch": 6.957095709570957,
-      "grad_norm": 0.002593994140625,
+      "grad_norm": 0.005889892578125,
       "learning_rate": 4.229256055945816e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3162
     },
     {
       "epoch": 6.959295929592959,
-      "grad_norm": 0.0035552978515625,
+      "grad_norm": 0.003753662109375,
       "learning_rate": 4.223593620111695e-05,
       "loss": 0.0001,
       "step": 3163
     },
     {
       "epoch": 6.961496149614962,
-      "grad_norm": 0.0030670166015625,
+      "grad_norm": 0.007110595703125,
       "learning_rate": 4.217933962461937e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3164
     },
     {
       "epoch": 6.963696369636963,
-      "grad_norm": 0.0098876953125,
+      "grad_norm": 0.00286865234375,
       "learning_rate": 4.21227708571858e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3165
     },
     {
       "epoch": 6.965896589658966,
-      "grad_norm": 0.0023651123046875,
+      "grad_norm": 0.0025482177734375,
       "learning_rate": 4.206622992602318e-05,
       "loss": 0.0001,
       "step": 3166
     },
     {
       "epoch": 6.968096809680969,
-      "grad_norm": 0.0037384033203125,
+      "grad_norm": 0.0032958984375,
       "learning_rate": 4.200971685832508e-05,
       "loss": 0.0002,
       "step": 3167
     },
     {
       "epoch": 6.97029702970297,
-      "grad_norm": 0.0032196044921875,
+      "grad_norm": 0.007171630859375,
       "learning_rate": 4.195323168127168e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3168
     },
     {
       "epoch": 6.972497249724973,
-      "grad_norm": 0.0036163330078125,
+      "grad_norm": 0.003509521484375,
       "learning_rate": 4.189677442202974e-05,
       "loss": 0.0001,
       "step": 3169
     },
     {
       "epoch": 6.974697469746975,
-      "grad_norm": 0.0029144287109375,
+      "grad_norm": 0.0026397705078125,
       "learning_rate": 4.1840345107752575e-05,
       "loss": 0.0001,
       "step": 3170
     },
     {
       "epoch": 6.976897689768977,
-      "grad_norm": 0.0028076171875,
+      "grad_norm": 0.0031890869140625,
       "learning_rate": 4.17839437655801e-05,
       "loss": 0.0001,
       "step": 3171
     },
     {
       "epoch": 6.979097909790979,
-      "grad_norm": 0.00323486328125,
+      "grad_norm": 0.003204345703125,
       "learning_rate": 4.172757042263874e-05,
       "loss": 0.0001,
       "step": 3172
     },
     {
       "epoch": 6.9812981298129815,
-      "grad_norm": 0.006378173828125,
+      "grad_norm": 0.00787353515625,
       "learning_rate": 4.1671225106041454e-05,
       "loss": 0.0002,
       "step": 3173
     },
     {
       "epoch": 6.983498349834983,
-      "grad_norm": 0.003662109375,
+      "grad_norm": 0.006378173828125,
       "learning_rate": 4.1614907842887764e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3174
     },
     {
       "epoch": 6.985698569856986,
-      "grad_norm": 0.0032501220703125,
+      "grad_norm": 0.003173828125,
       "learning_rate": 4.155861866026364e-05,
       "loss": 0.0001,
       "step": 3175
     },
     {
       "epoch": 6.987898789878988,
-      "grad_norm": 0.0025634765625,
+      "grad_norm": 0.0026702880859375,
       "learning_rate": 4.150235758524159e-05,
       "loss": 0.0001,
       "step": 3176
     },
     {
       "epoch": 6.99009900990099,
-      "grad_norm": 0.00811767578125,
+      "grad_norm": 0.006134033203125,
       "learning_rate": 4.144612464488059e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3177
     },
     {
       "epoch": 6.992299229922993,
-      "grad_norm": 0.00299072265625,
+      "grad_norm": 0.004425048828125,
       "learning_rate": 4.138991986622608e-05,
       "loss": 0.0001,
       "step": 3178
     },
     {
       "epoch": 6.994499449944994,
-      "grad_norm": 0.00390625,
+      "grad_norm": 0.004241943359375,
       "learning_rate": 4.1333743276309965e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3179
     },
     {
       "epoch": 6.996699669966997,
-      "grad_norm": 0.004425048828125,
+      "grad_norm": 0.0042724609375,
       "learning_rate": 4.127759490215058e-05,
       "loss": 0.0001,
       "step": 3180
     },
     {
       "epoch": 6.998899889988999,
-      "grad_norm": 0.002899169921875,
+      "grad_norm": 0.00396728515625,
       "learning_rate": 4.12214747707527e-05,
       "loss": 0.0001,
       "step": 3181
     },
     {
       "epoch": 7.001100110011001,
-      "grad_norm": 0.00177001953125,
+      "grad_norm": 0.00433349609375,
       "learning_rate": 4.116538290910751e-05,
       "loss": 0.0001,
       "step": 3182
     },
     {
       "epoch": 7.003300330033003,
-      "grad_norm": 0.0029754638671875,
+      "grad_norm": 0.0030059814453125,
       "learning_rate": 4.110931934419261e-05,
       "loss": 0.0001,
       "step": 3183
     },
     {
       "epoch": 7.005500550055006,
-      "grad_norm": 0.0032196044921875,
+      "grad_norm": 0.002838134765625,
       "learning_rate": 4.1053284102971964e-05,
       "loss": 0.0001,
       "step": 3184
     },
     {
       "epoch": 7.007700770077007,
-      "grad_norm": 0.0040283203125,
+      "grad_norm": 0.004180908203125,
       "learning_rate": 4.099727721239597e-05,
       "loss": 0.0001,
       "step": 3185
     },
     {
       "epoch": 7.00990099009901,
-      "grad_norm": 0.004974365234375,
+      "grad_norm": 0.01104736328125,
       "learning_rate": 4.0941298699401307e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3186
     },
     {
       "epoch": 7.0121012101210125,
-      "grad_norm": 0.005767822265625,
+      "grad_norm": 0.0036773681640625,
       "learning_rate": 4.088534859091108e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3187
     },
     {
       "epoch": 7.014301430143014,
-      "grad_norm": 0.0021514892578125,
+      "grad_norm": 0.0029296875,
       "learning_rate": 4.08294269138347e-05,
       "loss": 0.0001,
       "step": 3188
     },
     {
       "epoch": 7.016501650165017,
-      "grad_norm": 0.0035858154296875,
+      "grad_norm": 0.003265380859375,
       "learning_rate": 4.07735336950679e-05,
       "loss": 0.0001,
       "step": 3189
     },
     {
       "epoch": 7.0187018701870185,
-      "grad_norm": 0.003814697265625,
+      "grad_norm": 0.0031585693359375,
       "learning_rate": 4.071766896149273e-05,
       "loss": 0.0001,
       "step": 3190
     },
     {
       "epoch": 7.020902090209021,
-      "grad_norm": 0.006134033203125,
+      "grad_norm": 0.004638671875,
       "learning_rate": 4.0661832739977556e-05,
       "loss": 0.0002,
       "step": 3191
     },
     {
       "epoch": 7.023102310231023,
-      "grad_norm": 0.003021240234375,
+      "grad_norm": 0.004852294921875,
       "learning_rate": 4.0606025057377015e-05,
       "loss": 0.0001,
       "step": 3192
     },
     {
       "epoch": 7.023102310231023,
-      "eval_loss": 0.00024721783120185137,
-      "eval_runtime": 11.1776,
-      "eval_samples_per_second": 34.265,
-      "eval_steps_per_second": 4.294,
+      "eval_loss": 0.0003570860717445612,
+      "eval_runtime": 10.1256,
+      "eval_samples_per_second": 37.825,
+      "eval_steps_per_second": 4.74,
       "step": 3192
     },
     {
       "epoch": 7.025302530253025,
-      "grad_norm": 0.002960205078125,
+      "grad_norm": 0.0057373046875,
       "learning_rate": 4.055024594053203e-05,
       "loss": 0.0001,
       "step": 3193
     },
     {
       "epoch": 7.027502750275027,
-      "grad_norm": 0.0027923583984375,
+      "grad_norm": 0.0023193359375,
       "learning_rate": 4.049449541626975e-05,
       "loss": 0.0001,
       "step": 3194
     },
     {
       "epoch": 7.02970297029703,
-      "grad_norm": 0.003997802734375,
+      "grad_norm": 0.003631591796875,
       "learning_rate": 4.0438773511403626e-05,
       "loss": 0.0001,
       "step": 3195
     },
     {
       "epoch": 7.031903190319032,
-      "grad_norm": 0.0021820068359375,
+      "grad_norm": 0.00311279296875,
       "learning_rate": 4.038308025273332e-05,
       "loss": 0.0001,
       "step": 3196
     },
     {
       "epoch": 7.034103410341034,
-      "grad_norm": 0.00191497802734375,
+      "grad_norm": 0.002410888671875,
       "learning_rate": 4.032741566704469e-05,
       "loss": 0.0001,
       "step": 3197
     },
     {
       "epoch": 7.036303630363037,
-      "grad_norm": 0.00173187255859375,
+      "grad_norm": 0.0030517578125,
       "learning_rate": 4.027177978110983e-05,
       "loss": 0.0001,
       "step": 3198
     },
     {
       "epoch": 7.038503850385038,
-      "grad_norm": 0.004150390625,
+      "grad_norm": 0.004364013671875,
       "learning_rate": 4.0216172621687055e-05,
       "loss": 0.0001,
       "step": 3199
     },
     {
       "epoch": 7.040704070407041,
-      "grad_norm": 0.0033721923828125,
+      "grad_norm": 0.0050048828125,
       "learning_rate": 4.016059421552082e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3200
     },
     {
       "epoch": 7.042904290429043,
-      "grad_norm": 0.003265380859375,
+      "grad_norm": 0.0027618408203125,
       "learning_rate": 4.0105044589341764e-05,
       "loss": 0.0001,
       "step": 3201
     },
     {
       "epoch": 7.045104510451045,
-      "grad_norm": 0.003173828125,
+      "grad_norm": 0.00244140625,
       "learning_rate": 4.0049523769866694e-05,
       "loss": 0.0001,
       "step": 3202
     },
     {
       "epoch": 7.047304730473047,
-      "grad_norm": 0.00225830078125,
+      "grad_norm": 0.00506591796875,
       "learning_rate": 3.999403178379856e-05,
       "loss": 0.0001,
       "step": 3203
     },
     {
       "epoch": 7.0495049504950495,
-      "grad_norm": 0.0054931640625,
+      "grad_norm": 0.00445556640625,
       "learning_rate": 3.993856865782644e-05,
       "loss": 0.0001,
       "step": 3204
     },
     {
       "epoch": 7.051705170517051,
-      "grad_norm": 0.00445556640625,
+      "grad_norm": 0.005645751953125,
       "learning_rate": 3.988313441862553e-05,
       "loss": 0.0002,
       "step": 3205
     },
     {
       "epoch": 7.053905390539054,
-      "grad_norm": 0.0024871826171875,
+      "grad_norm": 0.00191497802734375,
       "learning_rate": 3.9827729092857136e-05,
       "loss": 0.0001,
       "step": 3206
     },
     {
       "epoch": 7.0561056105610565,
-      "grad_norm": 0.00396728515625,
+      "grad_norm": 0.005584716796875,
       "learning_rate": 3.977235270716866e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3207
     },
     {
       "epoch": 7.058305830583058,
-      "grad_norm": 0.0030059814453125,
+      "grad_norm": 0.0020599365234375,
       "learning_rate": 3.971700528819359e-05,
       "loss": 0.0001,
       "step": 3208
     },
     {
       "epoch": 7.060506050605061,
-      "grad_norm": 0.00482177734375,
+      "grad_norm": 0.0028533935546875,
       "learning_rate": 3.966168686255145e-05,
       "loss": 0.0001,
       "step": 3209
     },
     {
       "epoch": 7.0627062706270625,
-      "grad_norm": 0.01251220703125,
+      "grad_norm": 0.00872802734375,
       "learning_rate": 3.960639745684788e-05,
       "loss": 0.0002,
       "step": 3210
     },
     {
       "epoch": 7.064906490649065,
-      "grad_norm": 0.0033721923828125,
+      "grad_norm": 0.01043701171875,
       "learning_rate": 3.9551137097674496e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3211
     },
     {
       "epoch": 7.067106710671067,
-      "grad_norm": 0.0037841796875,
+      "grad_norm": 0.00299072265625,
       "learning_rate": 3.949590581160899e-05,
       "loss": 0.0001,
       "step": 3212
     },
     {
       "epoch": 7.069306930693069,
-      "grad_norm": 0.002105712890625,
+      "grad_norm": 0.004974365234375,
       "learning_rate": 3.944070362521506e-05,
       "loss": 0.0001,
       "step": 3213
     },
     {
       "epoch": 7.071507150715071,
-      "grad_norm": 0.0025177001953125,
+      "grad_norm": 0.0057373046875,
       "learning_rate": 3.93855305650424e-05,
       "loss": 0.0001,
       "step": 3214
     },
     {
       "epoch": 7.073707370737074,
-      "grad_norm": 0.00531005859375,
+      "grad_norm": 0.0032196044921875,
       "learning_rate": 3.93303866576267e-05,
       "loss": 0.0001,
       "step": 3215
     },
     {
       "epoch": 7.075907590759076,
-      "grad_norm": 0.0048828125,
+      "grad_norm": 0.0057373046875,
       "learning_rate": 3.927527192948962e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3216
     },
     {
       "epoch": 7.078107810781078,
-      "grad_norm": 0.0031280517578125,
+      "grad_norm": 0.003326416015625,
       "learning_rate": 3.922018640713882e-05,
       "loss": 0.0001,
       "step": 3217
     },
     {
       "epoch": 7.080308030803081,
-      "grad_norm": 0.003173828125,
+      "grad_norm": 0.003387451171875,
       "learning_rate": 3.916513011706787e-05,
       "loss": 0.0001,
       "step": 3218
     },
     {
       "epoch": 7.082508250825082,
-      "grad_norm": 0.00250244140625,
+      "grad_norm": 0.0025177001953125,
       "learning_rate": 3.911010308575631e-05,
       "loss": 0.0001,
       "step": 3219
     },
     {
       "epoch": 7.084708470847085,
-      "grad_norm": 0.001983642578125,
+      "grad_norm": 0.0048828125,
       "learning_rate": 3.9055105339669595e-05,
       "loss": 0.0001,
       "step": 3220
     },
     {
       "epoch": 7.086908690869087,
-      "grad_norm": 0.0079345703125,
+      "grad_norm": 0.004150390625,
       "learning_rate": 3.900013690525909e-05,
       "loss": 0.0002,
       "step": 3221
     },
     {
       "epoch": 7.089108910891089,
-      "grad_norm": 0.00482177734375,
+      "grad_norm": 0.0035858154296875,
       "learning_rate": 3.894519780896209e-05,
       "loss": 0.0001,
       "step": 3222
     },
     {
       "epoch": 7.091309130913091,
-      "grad_norm": 0.00164031982421875,
+      "grad_norm": 0.00537109375,
       "learning_rate": 3.889028807720173e-05,
       "loss": 0.0001,
       "step": 3223
     },
     {
       "epoch": 7.0935093509350935,
-      "grad_norm": 0.0027618408203125,
+      "grad_norm": 0.0032501220703125,
       "learning_rate": 3.883540773638706e-05,
       "loss": 0.0001,
       "step": 3224
     },
     {
       "epoch": 7.095709570957096,
-      "grad_norm": 0.004547119140625,
+      "grad_norm": 0.006439208984375,
       "learning_rate": 3.8780556812913016e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3225
     },
     {
       "epoch": 7.097909790979098,
-      "grad_norm": 0.0018463134765625,
+      "grad_norm": 0.005126953125,
       "learning_rate": 3.8725735333160315e-05,
       "loss": 0.0001,
       "step": 3226
     },
     {
       "epoch": 7.1001100110011,
-      "grad_norm": 0.00194549560546875,
+      "grad_norm": 0.003326416015625,
       "learning_rate": 3.867094332349559e-05,
       "loss": 0.0001,
       "step": 3227
     },
     {
       "epoch": 7.102310231023102,
-      "grad_norm": 0.0018463134765625,
+      "grad_norm": 0.00347900390625,
       "learning_rate": 3.861618081027124e-05,
       "loss": 0.0001,
       "step": 3228
     },
     {
       "epoch": 7.104510451045105,
-      "grad_norm": 0.003082275390625,
+      "grad_norm": 0.00347900390625,
       "learning_rate": 3.8561447819825505e-05,
       "loss": 0.0001,
       "step": 3229
     },
     {
       "epoch": 7.106710671067106,
-      "grad_norm": 0.0024871826171875,
+      "grad_norm": 0.003082275390625,
       "learning_rate": 3.850674437848243e-05,
       "loss": 0.0001,
       "step": 3230
     },
     {
       "epoch": 7.108910891089109,
-      "grad_norm": 0.0030975341796875,
+      "grad_norm": 0.00299072265625,
       "learning_rate": 3.845207051255183e-05,
       "loss": 0.0001,
       "step": 3231
     },
     {
       "epoch": 7.111111111111111,
-      "grad_norm": 0.00360107421875,
+      "grad_norm": 0.0037078857421875,
       "learning_rate": 3.8397426248329315e-05,
       "loss": 0.0001,
       "step": 3232
     },
     {
       "epoch": 7.113311331133113,
-      "grad_norm": 0.003082275390625,
+      "grad_norm": 0.0022430419921875,
       "learning_rate": 3.8342811612096255e-05,
       "loss": 0.0001,
       "step": 3233
     },
     {
       "epoch": 7.115511551155116,
-      "grad_norm": 0.002288818359375,
+      "grad_norm": 0.002960205078125,
       "learning_rate": 3.828822663011975e-05,
       "loss": 0.0001,
       "step": 3234
     },
     {
       "epoch": 7.117711771177118,
-      "grad_norm": 0.0029449462890625,
+      "grad_norm": 0.004638671875,
       "learning_rate": 3.823367132865265e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3235
     },
     {
       "epoch": 7.11991199119912,
-      "grad_norm": 0.0035552978515625,
+      "grad_norm": 0.0036468505859375,
       "learning_rate": 3.817914573393356e-05,
       "loss": 0.0001,
       "step": 3236
     },
     {
       "epoch": 7.122112211221122,
-      "grad_norm": 0.00457763671875,
+      "grad_norm": 0.004547119140625,
       "learning_rate": 3.812464987218675e-05,
       "loss": 0.0001,
       "step": 3237
     },
     {
       "epoch": 7.1243124312431245,
-      "grad_norm": 0.003082275390625,
+      "grad_norm": 0.003753662109375,
       "learning_rate": 3.807018376962219e-05,
       "loss": 0.0001,
       "step": 3238
     },
     {
       "epoch": 7.126512651265126,
-      "grad_norm": 0.00311279296875,
+      "grad_norm": 0.003448486328125,
       "learning_rate": 3.80157474524356e-05,
       "loss": 0.0001,
       "step": 3239
     },
     {
       "epoch": 7.128712871287129,
-      "grad_norm": 0.0036468505859375,
+      "grad_norm": 0.00439453125,
       "learning_rate": 3.79613409468083e-05,
       "loss": 0.0001,
       "step": 3240
     },
     {
       "epoch": 7.1309130913091305,
-      "grad_norm": 0.0024261474609375,
+      "grad_norm": 0.00390625,
       "learning_rate": 3.790696427890732e-05,
       "loss": 0.0001,
       "step": 3241
     },
     {
       "epoch": 7.133113311331133,
-      "grad_norm": 0.0034332275390625,
+      "grad_norm": 0.0050048828125,
       "learning_rate": 3.785261747488531e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3242
     },
     {
       "epoch": 7.135313531353136,
-      "grad_norm": 0.0025634765625,
+      "grad_norm": 0.006317138671875,
       "learning_rate": 3.779830056088057e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3243
     },
     {
       "epoch": 7.137513751375137,
-      "grad_norm": 0.0021820068359375,
+      "grad_norm": 0.003753662109375,
       "learning_rate": 3.774401356301703e-05,
       "loss": 0.0001,
       "step": 3244
     },
     {
       "epoch": 7.13971397139714,
-      "grad_norm": 0.0023651123046875,
+      "grad_norm": 0.0021514892578125,
       "learning_rate": 3.768975650740423e-05,
       "loss": 0.0001,
       "step": 3245
     },
     {
       "epoch": 7.141914191419142,
-      "grad_norm": 0.002105712890625,
+      "grad_norm": 0.003814697265625,
       "learning_rate": 3.763552942013731e-05,
       "loss": 0.0001,
       "step": 3246
     },
     {
       "epoch": 7.144114411441144,
-      "grad_norm": 0.0023345947265625,
+      "grad_norm": 0.003265380859375,
       "learning_rate": 3.758133232729696e-05,
       "loss": 0.0001,
       "step": 3247
     },
     {
       "epoch": 7.146314631463146,
-      "grad_norm": 0.002227783203125,
+      "grad_norm": 0.00159454345703125,
       "learning_rate": 3.7527165254949504e-05,
       "loss": 0.0001,
       "step": 3248
     },
     {
       "epoch": 7.148514851485149,
-      "grad_norm": 0.00188446044921875,
+      "grad_norm": 0.004058837890625,
       "learning_rate": 3.74730282291468e-05,
       "loss": 0.0001,
       "step": 3249
     },
     {
       "epoch": 7.15071507150715,
-      "grad_norm": 0.004791259765625,
+      "grad_norm": 0.0030364990234375,
       "learning_rate": 3.741892127592625e-05,
       "loss": 0.0001,
       "step": 3250
     },
     {
       "epoch": 7.152915291529153,
-      "grad_norm": 0.0028076171875,
+      "grad_norm": 0.0059814453125,
       "learning_rate": 3.736484442131077e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3251
     },
     {
       "epoch": 7.1551155115511555,
-      "grad_norm": 0.00445556640625,
+      "grad_norm": 0.002838134765625,
       "learning_rate": 3.731079769130886e-05,
       "loss": 0.0001,
       "step": 3252
     },
     {
       "epoch": 7.157315731573157,
-      "grad_norm": 0.00531005859375,
+      "grad_norm": 0.005615234375,
       "learning_rate": 3.72567811119145e-05,
       "loss": 0.0001,
       "step": 3253
     },
     {
       "epoch": 7.15951595159516,
-      "grad_norm": 0.004913330078125,
+      "grad_norm": 0.005126953125,
       "learning_rate": 3.720279470910717e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3254
     },
     {
       "epoch": 7.161716171617162,
-      "grad_norm": 0.0030517578125,
+      "grad_norm": 0.0032196044921875,
       "learning_rate": 3.714883850885176e-05,
       "loss": 0.0001,
       "step": 3255
     },
     {
       "epoch": 7.163916391639164,
-      "grad_norm": 0.005401611328125,
+      "grad_norm": 0.00604248046875,
       "learning_rate": 3.7094912537098803e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3256
     },
     {
       "epoch": 7.166116611661166,
-      "grad_norm": 0.004425048828125,
+      "grad_norm": 0.01190185546875,
       "learning_rate": 3.7041016819784145e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 3257
     },
     {
       "epoch": 7.1683168316831685,
-      "grad_norm": 0.002227783203125,
+      "grad_norm": 0.00299072265625,
       "learning_rate": 3.6987151382829154e-05,
       "loss": 0.0001,
       "step": 3258
     },
     {
       "epoch": 7.17051705170517,
-      "grad_norm": 0.00238037109375,
+      "grad_norm": 0.0032196044921875,
       "learning_rate": 3.693331625214061e-05,
       "loss": 0.0001,
       "step": 3259
     },
     {
       "epoch": 7.172717271727173,
-      "grad_norm": 0.0028533935546875,
+      "grad_norm": 0.0029144287109375,
       "learning_rate": 3.687951145361073e-05,
       "loss": 0.0001,
       "step": 3260
     },
     {
       "epoch": 7.174917491749175,
-      "grad_norm": 0.0032196044921875,
+      "grad_norm": 0.0048828125,
       "learning_rate": 3.6825737013117115e-05,
       "loss": 0.0001,
       "step": 3261
     },
     {
       "epoch": 7.177117711771177,
-      "grad_norm": 0.004913330078125,
+      "grad_norm": 0.003143310546875,
       "learning_rate": 3.6771992956522795e-05,
       "loss": 0.0001,
       "step": 3262
     },
     {
       "epoch": 7.17931793179318,
-      "grad_norm": 0.003997802734375,
+      "grad_norm": 0.0033721923828125,
       "learning_rate": 3.6718279309676174e-05,
       "loss": 0.0001,
       "step": 3263
     },
     {
       "epoch": 7.181518151815181,
-      "grad_norm": 0.0024871826171875,
+      "grad_norm": 0.0040283203125,
       "learning_rate": 3.666459609841103e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3264
     },
     {
       "epoch": 7.183718371837184,
-      "grad_norm": 0.00238037109375,
+      "grad_norm": 0.005859375,
       "learning_rate": 3.6610943348546526e-05,
       "loss": 0.0001,
       "step": 3265
     },
     {
       "epoch": 7.185918591859186,
-      "grad_norm": 0.003082275390625,
+      "grad_norm": 0.005615234375,
       "learning_rate": 3.655732108588712e-05,
       "loss": 0.0001,
       "step": 3266
     },
     {
       "epoch": 7.188118811881188,
-      "grad_norm": 0.0035858154296875,
+      "grad_norm": 0.00311279296875,
       "learning_rate": 3.650372933622266e-05,
       "loss": 0.0001,
       "step": 3267
     },
     {
       "epoch": 7.19031903190319,
-      "grad_norm": 0.003662109375,
+      "grad_norm": 0.0045166015625,
       "learning_rate": 3.64501681253283e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3268
     },
     {
       "epoch": 7.192519251925193,
-      "grad_norm": 0.0035552978515625,
+      "grad_norm": 0.00274658203125,
       "learning_rate": 3.63966374789645e-05,
       "loss": 0.0001,
       "step": 3269
     },
     {
       "epoch": 7.194719471947194,
-      "grad_norm": 0.00823974609375,
+      "grad_norm": 0.0089111328125,
       "learning_rate": 3.634313742287703e-05,
       "loss": 0.0002,
       "step": 3270
     },
     {
       "epoch": 7.196919691969197,
-      "grad_norm": 0.0037841796875,
+      "grad_norm": 0.00347900390625,
       "learning_rate": 3.628966798279697e-05,
       "loss": 0.0001,
       "step": 3271
     },
     {
       "epoch": 7.1991199119911995,
-      "grad_norm": 0.0031585693359375,
+      "grad_norm": 0.003692626953125,
       "learning_rate": 3.623622918444055e-05,
       "loss": 0.0001,
       "step": 3272
     },
     {
       "epoch": 7.201320132013201,
-      "grad_norm": 0.00396728515625,
+      "grad_norm": 0.0027618408203125,
       "learning_rate": 3.6182821053509476e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3273
     },
     {
       "epoch": 7.203520352035204,
-      "grad_norm": 0.006591796875,
+      "grad_norm": 0.0040283203125,
       "learning_rate": 3.612944361569054e-05,
       "loss": 0.0002,
       "step": 3274
     },
     {
       "epoch": 7.2057205720572055,
-      "grad_norm": 0.005615234375,
+      "grad_norm": 0.00982666015625,
       "learning_rate": 3.607609689665582e-05,
       "loss": 0.0001,
       "step": 3275
     },
     {
       "epoch": 7.207920792079208,
-      "grad_norm": 0.017578125,
+      "grad_norm": 0.003143310546875,
       "learning_rate": 3.602278092206266e-05,
-      "loss": 0.0004,
+      "loss": 0.0001,
       "step": 3276
     },
     {
       "epoch": 7.21012101210121,
-      "grad_norm": 0.002655029296875,
+      "grad_norm": 0.003143310546875,
       "learning_rate": 3.596949571755348e-05,
       "loss": 0.0001,
       "step": 3277
     },
     {
       "epoch": 7.212321232123212,
-      "grad_norm": 0.0038299560546875,
+      "grad_norm": 0.00885009765625,
       "learning_rate": 3.5916241308756105e-05,
       "loss": 0.0001,
       "step": 3278
     },
     {
       "epoch": 7.214521452145214,
-      "grad_norm": 0.0025634765625,
+      "grad_norm": 0.00176239013671875,
       "learning_rate": 3.586301772128339e-05,
       "loss": 0.0001,
       "step": 3279
     },
     {
       "epoch": 7.216721672167217,
-      "grad_norm": 0.0031890869140625,
+      "grad_norm": 0.00482177734375,
       "learning_rate": 3.580982498073344e-05,
       "loss": 0.0001,
       "step": 3280
     },
     {
       "epoch": 7.218921892189219,
-      "grad_norm": 0.00506591796875,
+      "grad_norm": 0.0033416748046875,
       "learning_rate": 3.575666311268953e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3281
     },
     {
       "epoch": 7.221122112211221,
-      "grad_norm": 0.006439208984375,
+      "grad_norm": 0.00634765625,
       "learning_rate": 3.570353214271995e-05,
       "loss": 0.0002,
       "step": 3282
     },
     {
       "epoch": 7.223322332233224,
-      "grad_norm": 0.00179290771484375,
+      "grad_norm": 0.00482177734375,
       "learning_rate": 3.565043209637835e-05,
       "loss": 0.0001,
       "step": 3283
     },
     {
       "epoch": 7.225522552255225,
-      "grad_norm": 0.003021240234375,
+      "grad_norm": 0.003814697265625,
       "learning_rate": 3.5597362999203373e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3284
     },
     {
       "epoch": 7.227722772277228,
-      "grad_norm": 0.0030670166015625,
+      "grad_norm": 0.0042724609375,
       "learning_rate": 3.554432487671878e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3285
     },
     {
       "epoch": 7.22992299229923,
-      "grad_norm": 0.0024261474609375,
+      "grad_norm": 0.004425048828125,
       "learning_rate": 3.549131775443347e-05,
       "loss": 0.0001,
       "step": 3286
     },
     {
       "epoch": 7.232123212321232,
-      "grad_norm": 0.0035552978515625,
+      "grad_norm": 0.0034027099609375,
       "learning_rate": 3.543834165784141e-05,
       "loss": 0.0001,
       "step": 3287
     },
     {
       "epoch": 7.234323432343234,
-      "grad_norm": 0.003143310546875,
+      "grad_norm": 0.005035400390625,
       "learning_rate": 3.5385396612421664e-05,
       "loss": 0.0002,
       "step": 3288
     },
     {
       "epoch": 7.2365236523652365,
-      "grad_norm": 0.004852294921875,
+      "grad_norm": 0.00531005859375,
       "learning_rate": 3.533248264363833e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3289
     },
     {
       "epoch": 7.238723872387239,
-      "grad_norm": 0.002777099609375,
+      "grad_norm": 0.0034637451171875,
       "learning_rate": 3.527959977694061e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3290
     },
     {
       "epoch": 7.240924092409241,
-      "grad_norm": 0.003509521484375,
+      "grad_norm": 0.0052490234375,
       "learning_rate": 3.522674803776268e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3291
     },
     {
       "epoch": 7.243124312431243,
-      "grad_norm": 0.0021514892578125,
+      "grad_norm": 0.0040283203125,
       "learning_rate": 3.5173927451523814e-05,
       "loss": 0.0001,
       "step": 3292
     },
     {
       "epoch": 7.245324532453245,
-      "grad_norm": 0.006591796875,
+      "grad_norm": 0.0026702880859375,
       "learning_rate": 3.51211380436283e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3293
     },
     {
       "epoch": 7.247524752475248,
-      "grad_norm": 0.0030670166015625,
+      "grad_norm": 0.00176239013671875,
       "learning_rate": 3.506837983946529e-05,
       "loss": 0.0001,
       "step": 3294
     },
     {
       "epoch": 7.2497249724972495,
-      "grad_norm": 0.0028076171875,
+      "grad_norm": 0.00421142578125,
       "learning_rate": 3.501565286440914e-05,
       "loss": 0.0001,
       "step": 3295
     },
     {
       "epoch": 7.251925192519252,
-      "grad_norm": 0.00274658203125,
+      "grad_norm": 0.002960205078125,
       "learning_rate": 3.496295714381907e-05,
       "loss": 0.0001,
       "step": 3296
     },
     {
       "epoch": 7.254125412541254,
-      "grad_norm": 0.002593994140625,
+      "grad_norm": 0.002960205078125,
       "learning_rate": 3.491029270303927e-05,
       "loss": 0.0001,
       "step": 3297
     },
     {
       "epoch": 7.256325632563256,
-      "grad_norm": 0.0028228759765625,
+      "grad_norm": 0.0019989013671875,
       "learning_rate": 3.485765956739895e-05,
       "loss": 0.0001,
       "step": 3298
     },
     {
       "epoch": 7.258525852585258,
-      "grad_norm": 0.00384521484375,
+      "grad_norm": 0.0034332275390625,
       "learning_rate": 3.480505776221209e-05,
       "loss": 0.0001,
       "step": 3299
     },
     {
       "epoch": 7.260726072607261,
-      "grad_norm": 0.004547119140625,
+      "grad_norm": 0.002593994140625,
       "learning_rate": 3.475248731277785e-05,
       "loss": 0.0001,
       "step": 3300
     },
     {
       "epoch": 7.262926292629263,
-      "grad_norm": 0.00836181640625,
+      "grad_norm": 0.0040283203125,
       "learning_rate": 3.469994824438014e-05,
       "loss": 0.0002,
       "step": 3301
     },
     {
       "epoch": 7.265126512651265,
-      "grad_norm": 0.007415771484375,
+      "grad_norm": 0.016357421875,
       "learning_rate": 3.4647440582287825e-05,
       "loss": 0.0002,
       "step": 3302
     },
     {
       "epoch": 7.267326732673268,
-      "grad_norm": 0.00262451171875,
+      "grad_norm": 0.005859375,
       "learning_rate": 3.459496435175469e-05,
       "loss": 0.0001,
       "step": 3303
     },
     {
       "epoch": 7.269526952695269,
-      "grad_norm": 0.00408935546875,
+      "grad_norm": 0.0064697265625,
       "learning_rate": 3.454251957801928e-05,
       "loss": 0.0002,
       "step": 3304
     },
     {
       "epoch": 7.271727172717272,
-      "grad_norm": 0.0025634765625,
+      "grad_norm": 0.00457763671875,
       "learning_rate": 3.4490106286305214e-05,
       "loss": 0.0001,
       "step": 3305
     },
     {
       "epoch": 7.273927392739274,
-      "grad_norm": 0.00445556640625,
+      "grad_norm": 0.0034942626953125,
       "learning_rate": 3.443772450182081e-05,
       "loss": 0.0001,
       "step": 3306
     },
     {
       "epoch": 7.273927392739274,
-      "eval_loss": 0.0002458308299537748,
-      "eval_runtime": 10.3235,
-      "eval_samples_per_second": 37.1,
-      "eval_steps_per_second": 4.65,
+      "eval_loss": 0.00035453730379231274,
+      "eval_runtime": 10.0914,
+      "eval_samples_per_second": 37.953,
+      "eval_steps_per_second": 4.757,
       "step": 3306
     },
     {
       "epoch": 7.276127612761276,
-      "grad_norm": 0.0027313232421875,
+      "grad_norm": 0.0037384033203125,
       "learning_rate": 3.4385374249759306e-05,
       "loss": 0.0001,
       "step": 3307
     },
     {
       "epoch": 7.278327832783278,
-      "grad_norm": 0.0034332275390625,
+      "grad_norm": 0.002532958984375,
       "learning_rate": 3.4333055555298776e-05,
       "loss": 0.0001,
       "step": 3308
     },
     {
       "epoch": 7.2805280528052805,
-      "grad_norm": 0.004730224609375,
+      "grad_norm": 0.003173828125,
       "learning_rate": 3.428076844360199e-05,
       "loss": 0.0001,
       "step": 3309
     },
     {
       "epoch": 7.282728272827283,
-      "grad_norm": 0.005615234375,
+      "grad_norm": 0.0040283203125,
       "learning_rate": 3.422851293981676e-05,
       "loss": 0.0001,
       "step": 3310
     },
     {
       "epoch": 7.284928492849285,
-      "grad_norm": 0.00775146484375,
+      "grad_norm": 0.007568359375,
       "learning_rate": 3.417628906907544e-05,
       "loss": 0.0002,
       "step": 3311
     },
     {
       "epoch": 7.287128712871287,
-      "grad_norm": 0.003936767578125,
+      "grad_norm": 0.00439453125,
       "learning_rate": 3.41240968564954e-05,
       "loss": 0.0001,
       "step": 3312
     },
     {
       "epoch": 7.289328932893289,
-      "grad_norm": 0.0042724609375,
+      "grad_norm": 0.004150390625,
       "learning_rate": 3.407193632717865e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3313
     },
     {
       "epoch": 7.291529152915292,
-      "grad_norm": 0.0032958984375,
+      "grad_norm": 0.00579833984375,
       "learning_rate": 3.401980750621199e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3314
     },
     {
       "epoch": 7.293729372937293,
-      "grad_norm": 0.002197265625,
+      "grad_norm": 0.004425048828125,
       "learning_rate": 3.3967710418666984e-05,
       "loss": 0.0001,
       "step": 3315
     },
     {
       "epoch": 7.295929592959296,
-      "grad_norm": 0.0030975341796875,
+      "grad_norm": 0.005889892578125,
       "learning_rate": 3.391564508959986e-05,
       "loss": 0.0001,
       "step": 3316
     },
     {
       "epoch": 7.298129812981298,
-      "grad_norm": 0.0028228759765625,
+      "grad_norm": 0.003021240234375,
       "learning_rate": 3.38636115440517e-05,
       "loss": 0.0001,
       "step": 3317
     },
     {
       "epoch": 7.3003300330033,
-      "grad_norm": 0.00396728515625,
+      "grad_norm": 0.0091552734375,
       "learning_rate": 3.381160980704822e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3318
     },
     {
       "epoch": 7.302530253025303,
-      "grad_norm": 0.005126953125,
+      "grad_norm": 0.007598876953125,
       "learning_rate": 3.375963990359985e-05,
       "loss": 0.0002,
       "step": 3319
     },
     {
       "epoch": 7.304730473047305,
-      "grad_norm": 0.00335693359375,
+      "grad_norm": 0.006256103515625,
       "learning_rate": 3.370770185870173e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3320
     },
     {
       "epoch": 7.306930693069307,
-      "grad_norm": 0.005340576171875,
+      "grad_norm": 0.0033111572265625,
       "learning_rate": 3.365579569733358e-05,
       "loss": 0.0001,
       "step": 3321
     },
     {
       "epoch": 7.309130913091309,
-      "grad_norm": 0.0027618408203125,
+      "grad_norm": 0.0032501220703125,
       "learning_rate": 3.360392144445994e-05,
       "loss": 0.0001,
       "step": 3322
     },
     {
       "epoch": 7.3113311331133115,
-      "grad_norm": 0.0038604736328125,
+      "grad_norm": 0.0125732421875,
       "learning_rate": 3.355207912502991e-05,
-      "loss": 0.0001,
+      "loss": 0.0003,
       "step": 3323
     },
     {
       "epoch": 7.313531353135313,
-      "grad_norm": 0.0023345947265625,
+      "grad_norm": 0.0024566650390625,
       "learning_rate": 3.350026876397725e-05,
       "loss": 0.0001,
       "step": 3324
     },
     {
       "epoch": 7.315731573157316,
-      "grad_norm": 0.0029754638671875,
+      "grad_norm": 0.01007080078125,
       "learning_rate": 3.3448490386220355e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3325
     },
     {
       "epoch": 7.3179317931793175,
-      "grad_norm": 0.0029754638671875,
+      "grad_norm": 0.00311279296875,
       "learning_rate": 3.339674401666216e-05,
       "loss": 0.0001,
       "step": 3326
     },
     {
       "epoch": 7.32013201320132,
-      "grad_norm": 0.002838134765625,
+      "grad_norm": 0.00286865234375,
       "learning_rate": 3.334502968019037e-05,
       "loss": 0.0001,
       "step": 3327
     },
     {
       "epoch": 7.322332233223323,
-      "grad_norm": 0.005340576171875,
+      "grad_norm": 0.00262451171875,
       "learning_rate": 3.3293347401677156e-05,
       "loss": 0.0001,
       "step": 3328
     },
     {
       "epoch": 7.324532453245324,
-      "grad_norm": 0.0025177001953125,
+      "grad_norm": 0.003692626953125,
       "learning_rate": 3.3241697205979286e-05,
       "loss": 0.0001,
       "step": 3329
     },
     {
       "epoch": 7.326732673267327,
-      "grad_norm": 0.00518798828125,
+      "grad_norm": 0.005340576171875,
       "learning_rate": 3.3190079117938166e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3330
     },
     {
       "epoch": 7.328932893289329,
-      "grad_norm": 0.004486083984375,
+      "grad_norm": 0.002349853515625,
       "learning_rate": 3.313849316237961e-05,
       "loss": 0.0001,
       "step": 3331
     },
     {
       "epoch": 7.331133113311331,
-      "grad_norm": 0.00238037109375,
+      "grad_norm": 0.00372314453125,
       "learning_rate": 3.308693936411421e-05,
       "loss": 0.0001,
       "step": 3332
     },
     {
       "epoch": 7.333333333333333,
-      "grad_norm": 0.003173828125,
+      "grad_norm": 0.0074462890625,
       "learning_rate": 3.303541774793681e-05,
       "loss": 0.0001,
       "step": 3333
     },
     {
       "epoch": 7.335533553355336,
-      "grad_norm": 0.004150390625,
+      "grad_norm": 0.00238037109375,
       "learning_rate": 3.298392833862706e-05,
       "loss": 0.0001,
       "step": 3334
     },
     {
       "epoch": 7.337733773377337,
-      "grad_norm": 0.004974365234375,
+      "grad_norm": 0.007659912109375,
       "learning_rate": 3.293247116094895e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3335
     },
     {
       "epoch": 7.33993399339934,
-      "grad_norm": 0.0033721923828125,
+      "grad_norm": 0.0030517578125,
       "learning_rate": 3.2881046239650915e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3336
     },
     {
       "epoch": 7.3421342134213425,
-      "grad_norm": 0.004730224609375,
+      "grad_norm": 0.0040283203125,
       "learning_rate": 3.28296535994661e-05,
       "loss": 0.0001,
       "step": 3337
     },
     {
       "epoch": 7.344334433443344,
-      "grad_norm": 0.0033721923828125,
+      "grad_norm": 0.0022125244140625,
       "learning_rate": 3.277829326511186e-05,
       "loss": 0.0001,
       "step": 3338
     },
     {
       "epoch": 7.346534653465347,
-      "grad_norm": 0.002685546875,
+      "grad_norm": 0.0034942626953125,
       "learning_rate": 3.272696526129022e-05,
       "loss": 0.0001,
       "step": 3339
     },
     {
       "epoch": 7.3487348734873486,
-      "grad_norm": 0.0022125244140625,
+      "grad_norm": 0.00311279296875,
       "learning_rate": 3.2675669612687565e-05,
       "loss": 0.0001,
       "step": 3340
     },
     {
       "epoch": 7.350935093509351,
-      "grad_norm": 0.0034942626953125,
+      "grad_norm": 0.00421142578125,
       "learning_rate": 3.2624406343974735e-05,
       "loss": 0.0002,
       "step": 3341
     },
     {
       "epoch": 7.353135313531353,
-      "grad_norm": 0.004638671875,
+      "grad_norm": 0.00408935546875,
       "learning_rate": 3.2573175479807e-05,
       "loss": 0.0001,
       "step": 3342
     },
     {
       "epoch": 7.3553355335533555,
-      "grad_norm": 0.003143310546875,
+      "grad_norm": 0.003326416015625,
       "learning_rate": 3.252197704482395e-05,
       "loss": 0.0001,
       "step": 3343
     },
     {
       "epoch": 7.357535753575357,
-      "grad_norm": 0.003082275390625,
+      "grad_norm": 0.002349853515625,
       "learning_rate": 3.247081106364977e-05,
       "loss": 0.0001,
       "step": 3344
     },
     {
       "epoch": 7.35973597359736,
-      "grad_norm": 0.0035247802734375,
+      "grad_norm": 0.0026397705078125,
       "learning_rate": 3.241967756089288e-05,
       "loss": 0.0001,
       "step": 3345
     },
     {
       "epoch": 7.361936193619362,
-      "grad_norm": 0.0029754638671875,
+      "grad_norm": 0.00738525390625,
       "learning_rate": 3.2368576561146156e-05,
       "loss": 0.0001,
       "step": 3346
     },
     {
       "epoch": 7.364136413641364,
-      "grad_norm": 0.0034637451171875,
+      "grad_norm": 0.0038604736328125,
       "learning_rate": 3.231750808898683e-05,
       "loss": 0.0001,
       "step": 3347
     },
     {
       "epoch": 7.366336633663367,
-      "grad_norm": 0.00323486328125,
+      "grad_norm": 0.003326416015625,
       "learning_rate": 3.226647216897638e-05,
       "loss": 0.0001,
       "step": 3348
     },
     {
       "epoch": 7.368536853685368,
-      "grad_norm": 0.0084228515625,
+      "grad_norm": 0.0118408203125,
       "learning_rate": 3.2215468825660865e-05,
       "loss": 0.0002,
       "step": 3349
     },
     {
       "epoch": 7.370737073707371,
-      "grad_norm": 0.00311279296875,
+      "grad_norm": 0.002716064453125,
       "learning_rate": 3.216449808357039e-05,
       "loss": 0.0001,
       "step": 3350
     },
     {
       "epoch": 7.372937293729373,
-      "grad_norm": 0.00244140625,
+      "grad_norm": 0.0034027099609375,
       "learning_rate": 3.211355996721963e-05,
       "loss": 0.0001,
       "step": 3351
     },
     {
       "epoch": 7.375137513751375,
-      "grad_norm": 0.0024261474609375,
+      "grad_norm": 0.005584716796875,
       "learning_rate": 3.206265450110746e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3352
     },
     {
       "epoch": 7.377337733773377,
-      "grad_norm": 0.003326416015625,
+      "grad_norm": 0.00872802734375,
       "learning_rate": 3.201178170971697e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3353
     },
     {
       "epoch": 7.37953795379538,
-      "grad_norm": 0.0107421875,
+      "grad_norm": 0.0033111572265625,
       "learning_rate": 3.196094161751573e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3354
     },
     {
       "epoch": 7.381738173817382,
-      "grad_norm": 0.002685546875,
+      "grad_norm": 0.002777099609375,
       "learning_rate": 3.191013424895536e-05,
       "loss": 0.0001,
       "step": 3355
     },
     {
       "epoch": 7.383938393839384,
-      "grad_norm": 0.002593994140625,
+      "grad_norm": 0.0048828125,
       "learning_rate": 3.185935962847194e-05,
       "loss": 0.0001,
       "step": 3356
     },
     {
       "epoch": 7.3861386138613865,
-      "grad_norm": 0.0027313232421875,
+      "grad_norm": 0.00579833984375,
       "learning_rate": 3.180861778048573e-05,
       "loss": 0.0001,
       "step": 3357
     },
     {
       "epoch": 7.388338833883388,
-      "grad_norm": 0.0036163330078125,
+      "grad_norm": 0.005828857421875,
       "learning_rate": 3.175790872940111e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3358
     },
     {
       "epoch": 7.390539053905391,
-      "grad_norm": 0.004302978515625,
+      "grad_norm": 0.00421142578125,
       "learning_rate": 3.17072324996069e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3359
     },
     {
       "epoch": 7.3927392739273925,
-      "grad_norm": 0.0021514892578125,
+      "grad_norm": 0.0184326171875,
       "learning_rate": 3.165658911547592e-05,
       "loss": 0.0001,
       "step": 3360
     },
     {
       "epoch": 7.394939493949395,
-      "grad_norm": 0.002716064453125,
+      "grad_norm": 0.0030517578125,
       "learning_rate": 3.160597860136538e-05,
       "loss": 0.0001,
       "step": 3361
     },
     {
       "epoch": 7.397139713971397,
-      "grad_norm": 0.0020599365234375,
+      "grad_norm": 0.00360107421875,
       "learning_rate": 3.155540098161659e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3362
     },
     {
       "epoch": 7.399339933993399,
-      "grad_norm": 0.004486083984375,
+      "grad_norm": 0.00445556640625,
       "learning_rate": 3.1504856280554975e-05,
       "loss": 0.0002,
       "step": 3363
     },
     {
       "epoch": 7.401540154015402,
-      "grad_norm": 0.00860595703125,
+      "grad_norm": 0.004302978515625,
       "learning_rate": 3.1454344522490297e-05,
       "loss": 0.0002,
       "step": 3364
     },
     {
       "epoch": 7.403740374037404,
-      "grad_norm": 0.002410888671875,
+      "grad_norm": 0.0067138671875,
       "learning_rate": 3.1403865731716266e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3365
     },
     {
       "epoch": 7.405940594059406,
-      "grad_norm": 0.005523681640625,
+      "grad_norm": 0.005279541015625,
       "learning_rate": 3.135341993251093e-05,
       "loss": 0.0002,
       "step": 3366
     },
     {
       "epoch": 7.408140814081408,
-      "grad_norm": 0.0026092529296875,
+      "grad_norm": 0.00482177734375,
       "learning_rate": 3.1303007149136374e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3367
     },
     {
       "epoch": 7.410341034103411,
-      "grad_norm": 0.00213623046875,
+      "grad_norm": 0.003387451171875,
       "learning_rate": 3.1252627405838794e-05,
       "loss": 0.0001,
       "step": 3368
     },
     {
       "epoch": 7.412541254125412,
-      "grad_norm": 0.002593994140625,
+      "grad_norm": 0.006011962890625,
       "learning_rate": 3.1202280726848564e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3369
     },
     {
       "epoch": 7.414741474147415,
-      "grad_norm": 0.004852294921875,
+      "grad_norm": 0.00537109375,
       "learning_rate": 3.115196713638e-05,
       "loss": 0.0002,
       "step": 3370
     },
     {
       "epoch": 7.416941694169417,
-      "grad_norm": 0.004119873046875,
+      "grad_norm": 0.005767822265625,
       "learning_rate": 3.1101686658631755e-05,
       "loss": 0.0002,
       "step": 3371
     },
     {
       "epoch": 7.419141914191419,
-      "grad_norm": 0.00830078125,
+      "grad_norm": 0.00634765625,
       "learning_rate": 3.105143931778628e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3372
     },
     {
       "epoch": 7.421342134213421,
-      "grad_norm": 0.003631591796875,
+      "grad_norm": 0.00323486328125,
       "learning_rate": 3.1001225138010316e-05,
       "loss": 0.0001,
       "step": 3373
     },
     {
       "epoch": 7.4235423542354235,
-      "grad_norm": 0.0032501220703125,
+      "grad_norm": 0.0019683837890625,
       "learning_rate": 3.095104414345455e-05,
       "loss": 0.0001,
       "step": 3374
     },
     {
       "epoch": 7.425742574257426,
-      "grad_norm": 0.0037841796875,
+      "grad_norm": 0.006988525390625,
       "learning_rate": 3.090089635825365e-05,
       "loss": 0.0001,
       "step": 3375
     },
     {
       "epoch": 7.427942794279428,
-      "grad_norm": 0.005828857421875,
+      "grad_norm": 0.0023651123046875,
       "learning_rate": 3.0850781806526484e-05,
       "loss": 0.0001,
       "step": 3376
     },
     {
       "epoch": 7.43014301430143,
-      "grad_norm": 0.0036163330078125,
+      "grad_norm": 0.002471923828125,
       "learning_rate": 3.080070051237572e-05,
       "loss": 0.0001,
       "step": 3377
     },
     {
       "epoch": 7.432343234323432,
-      "grad_norm": 0.010009765625,
+      "grad_norm": 0.010498046875,
       "learning_rate": 3.0750652499888256e-05,
       "loss": 0.0002,
       "step": 3378
     },
     {
       "epoch": 7.434543454345435,
-      "grad_norm": 0.0020599365234375,
+      "grad_norm": 0.0228271484375,
       "learning_rate": 3.0700637793134835e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3379
     },
     {
       "epoch": 7.436743674367436,
-      "grad_norm": 0.0030517578125,
+      "grad_norm": 0.0045166015625,
       "learning_rate": 3.065065641617015e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3380
     },
     {
       "epoch": 7.438943894389439,
-      "grad_norm": 0.00372314453125,
+      "grad_norm": 0.004486083984375,
       "learning_rate": 3.0600708393033054e-05,
       "loss": 0.0002,
       "step": 3381
     },
     {
       "epoch": 7.441144114411441,
-      "grad_norm": 0.00360107421875,
+      "grad_norm": 0.0045166015625,
       "learning_rate": 3.0550793747746096e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3382
     },
     {
       "epoch": 7.443344334433443,
-      "grad_norm": 0.0031890869140625,
+      "grad_norm": 0.0081787109375,
       "learning_rate": 3.0500912504316025e-05,
       "loss": 0.0001,
       "step": 3383
     },
     {
       "epoch": 7.445544554455446,
-      "grad_norm": 0.009033203125,
+      "grad_norm": 0.002593994140625,
       "learning_rate": 3.045106468673341e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3384
     },
     {
       "epoch": 7.447744774477448,
-      "grad_norm": 0.00494384765625,
+      "grad_norm": 0.014892578125,
       "learning_rate": 3.040125031897264e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 3385
     },
     {
       "epoch": 7.44994499449945,
-      "grad_norm": 0.0057373046875,
+      "grad_norm": 0.0072021484375,
       "learning_rate": 3.0351469424992263e-05,
       "loss": 0.0001,
       "step": 3386
     },
     {
       "epoch": 7.452145214521452,
-      "grad_norm": 0.0115966796875,
+      "grad_norm": 0.00286865234375,
       "learning_rate": 3.030172202873446e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3387
     },
     {
       "epoch": 7.4543454345434546,
-      "grad_norm": 0.0032196044921875,
+      "grad_norm": 0.002777099609375,
       "learning_rate": 3.0252008154125532e-05,
       "loss": 0.0001,
       "step": 3388
     },
     {
       "epoch": 7.456545654565456,
-      "grad_norm": 0.004058837890625,
+      "grad_norm": 0.00482177734375,
       "learning_rate": 3.0202327825075483e-05,
       "loss": 0.0002,
       "step": 3389
     },
     {
       "epoch": 7.458745874587459,
-      "grad_norm": 0.00238037109375,
+      "grad_norm": 0.002716064453125,
       "learning_rate": 3.015268106547825e-05,
       "loss": 0.0001,
       "step": 3390
@@ -23987,1220 +23987,1220 @@
     },
     {
       "epoch": 7.463146314631463,
-      "grad_norm": 0.0032196044921875,
+      "grad_norm": 0.0198974609375,
       "learning_rate": 3.0053488350137403e-05,
-      "loss": 0.0001,
+      "loss": 0.0003,
       "step": 3392
     },
     {
       "epoch": 7.465346534653466,
-      "grad_norm": 0.0029144287109375,
+      "grad_norm": 0.00762939453125,
       "learning_rate": 3.0003942442100896e-05,
       "loss": 0.0001,
       "step": 3393
     },
     {
       "epoch": 7.4675467546754675,
-      "grad_norm": 0.004364013671875,
+      "grad_norm": 0.00579833984375,
       "learning_rate": 2.995443019893137e-05,
       "loss": 0.0001,
       "step": 3394
     },
     {
       "epoch": 7.46974697469747,
-      "grad_norm": 0.0036773681640625,
+      "grad_norm": 0.0038604736328125,
       "learning_rate": 2.9904951644442014e-05,
       "loss": 0.0001,
       "step": 3395
     },
     {
       "epoch": 7.471947194719472,
-      "grad_norm": 0.0028228759765625,
+      "grad_norm": 0.003814697265625,
       "learning_rate": 2.9855506802429722e-05,
       "loss": 0.0001,
       "step": 3396
     },
     {
       "epoch": 7.474147414741474,
-      "grad_norm": 0.0030975341796875,
+      "grad_norm": 0.0034027099609375,
       "learning_rate": 2.9806095696675098e-05,
       "loss": 0.0001,
       "step": 3397
     },
     {
       "epoch": 7.476347634763476,
-      "grad_norm": 0.00335693359375,
+      "grad_norm": 0.0029449462890625,
       "learning_rate": 2.9756718350942703e-05,
       "loss": 0.0001,
       "step": 3398
     },
     {
       "epoch": 7.478547854785479,
-      "grad_norm": 0.0024566650390625,
+      "grad_norm": 0.007476806640625,
       "learning_rate": 2.9707374788980635e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3399
     },
     {
       "epoch": 7.48074807480748,
-      "grad_norm": 0.0157470703125,
+      "grad_norm": 0.0035552978515625,
       "learning_rate": 2.9658065034520978e-05,
-      "loss": 0.0004,
+      "loss": 0.0002,
       "step": 3400
     },
     {
       "epoch": 7.482948294829483,
-      "grad_norm": 0.0034637451171875,
+      "grad_norm": 0.00347900390625,
       "learning_rate": 2.9608789111279435e-05,
       "loss": 0.0001,
       "step": 3401
     },
     {
       "epoch": 7.485148514851485,
-      "grad_norm": 0.00372314453125,
+      "grad_norm": 0.00323486328125,
       "learning_rate": 2.9559547042955337e-05,
       "loss": 0.0001,
       "step": 3402
     },
     {
       "epoch": 7.487348734873487,
-      "grad_norm": 0.0038909912109375,
+      "grad_norm": 0.006561279296875,
       "learning_rate": 2.951033885323199e-05,
       "loss": 0.0001,
       "step": 3403
     },
     {
       "epoch": 7.48954895489549,
-      "grad_norm": 0.0021820068359375,
+      "grad_norm": 0.004119873046875,
       "learning_rate": 2.946116456577611e-05,
       "loss": 0.0001,
       "step": 3404
     },
     {
       "epoch": 7.491749174917492,
-      "grad_norm": 0.0018310546875,
+      "grad_norm": 0.00311279296875,
       "learning_rate": 2.9412024204238374e-05,
       "loss": 0.0001,
       "step": 3405
     },
     {
       "epoch": 7.493949394939494,
-      "grad_norm": 0.004150390625,
+      "grad_norm": 0.0037384033203125,
       "learning_rate": 2.936291779225301e-05,
       "loss": 0.0002,
       "step": 3406
     },
     {
       "epoch": 7.496149614961496,
-      "grad_norm": 0.00970458984375,
+      "grad_norm": 0.0030364990234375,
       "learning_rate": 2.931384535343783e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3407
     },
     {
       "epoch": 7.4983498349834985,
-      "grad_norm": 0.002838134765625,
+      "grad_norm": 0.0035247802734375,
       "learning_rate": 2.926480691139455e-05,
       "loss": 0.0001,
       "step": 3408
     },
     {
       "epoch": 7.5005500550055,
-      "grad_norm": 0.0024566650390625,
+      "grad_norm": 0.002410888671875,
       "learning_rate": 2.9215802489708268e-05,
       "loss": 0.0001,
       "step": 3409
     },
     {
       "epoch": 7.502750275027503,
-      "grad_norm": 0.00604248046875,
+      "grad_norm": 0.00482177734375,
       "learning_rate": 2.916683211194795e-05,
       "loss": 0.0002,
       "step": 3410
     },
     {
       "epoch": 7.5049504950495045,
-      "grad_norm": 0.0031890869140625,
+      "grad_norm": 0.0074462890625,
       "learning_rate": 2.9117895801665996e-05,
       "loss": 0.0001,
       "step": 3411
     },
     {
       "epoch": 7.507150715071507,
-      "grad_norm": 0.002532958984375,
+      "grad_norm": 0.00604248046875,
       "learning_rate": 2.906899358239853e-05,
       "loss": 0.0001,
       "step": 3412
     },
     {
       "epoch": 7.50935093509351,
-      "grad_norm": 0.0030975341796875,
+      "grad_norm": 0.0023956298828125,
       "learning_rate": 2.9020125477665317e-05,
       "loss": 0.0001,
       "step": 3413
     },
     {
       "epoch": 7.511551155115511,
-      "grad_norm": 0.0250244140625,
+      "grad_norm": 0.006500244140625,
       "learning_rate": 2.8971291510969566e-05,
-      "loss": 0.0006,
+      "loss": 0.0002,
       "step": 3414
     },
     {
       "epoch": 7.513751375137514,
-      "grad_norm": 0.0035552978515625,
+      "grad_norm": 0.004241943359375,
       "learning_rate": 2.892249170579826e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3415
     },
     {
       "epoch": 7.515951595159516,
-      "grad_norm": 0.004058837890625,
+      "grad_norm": 0.003387451171875,
       "learning_rate": 2.8873726085621755e-05,
       "loss": 0.0001,
       "step": 3416
     },
     {
       "epoch": 7.518151815181518,
-      "grad_norm": 0.0024566650390625,
+      "grad_norm": 0.0030364990234375,
       "learning_rate": 2.882499467389407e-05,
       "loss": 0.0001,
       "step": 3417
     },
     {
       "epoch": 7.52035203520352,
-      "grad_norm": 0.00225830078125,
+      "grad_norm": 0.0035247802734375,
       "learning_rate": 2.877629749405285e-05,
       "loss": 0.0001,
       "step": 3418
     },
     {
       "epoch": 7.522552255225523,
-      "grad_norm": 0.002685546875,
+      "grad_norm": 0.0040283203125,
       "learning_rate": 2.8727634569519057e-05,
       "loss": 0.0001,
       "step": 3419
     },
     {
       "epoch": 7.524752475247524,
-      "grad_norm": 0.002166748046875,
+      "grad_norm": 0.0032806396484375,
       "learning_rate": 2.8679005923697443e-05,
       "loss": 0.0001,
       "step": 3420
     },
     {
       "epoch": 7.524752475247524,
-      "eval_loss": 0.00024103005125653,
-      "eval_runtime": 10.1625,
-      "eval_samples_per_second": 37.688,
-      "eval_steps_per_second": 4.723,
+      "eval_loss": 0.0003504643391352147,
+      "eval_runtime": 10.1507,
+      "eval_samples_per_second": 37.731,
+      "eval_steps_per_second": 4.729,
       "step": 3420
     },
     {
       "epoch": 7.526952695269527,
-      "grad_norm": 0.0035858154296875,
+      "grad_norm": 0.00933837890625,
       "learning_rate": 2.8630411579976025e-05,
       "loss": 0.0002,
       "step": 3421
     },
     {
       "epoch": 7.5291529152915295,
-      "grad_norm": 0.00445556640625,
+      "grad_norm": 0.0034637451171875,
       "learning_rate": 2.8581851561726458e-05,
       "loss": 0.0002,
       "step": 3422
     },
     {
       "epoch": 7.531353135313531,
-      "grad_norm": 0.004669189453125,
+      "grad_norm": 0.0036468505859375,
       "learning_rate": 2.8533325892303953e-05,
       "loss": 0.0001,
       "step": 3423
     },
     {
       "epoch": 7.533553355335534,
-      "grad_norm": 0.00445556640625,
+      "grad_norm": 0.0034027099609375,
       "learning_rate": 2.8484834595046973e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3424
     },
     {
       "epoch": 7.5357535753575355,
-      "grad_norm": 0.0027618408203125,
+      "grad_norm": 0.00286865234375,
       "learning_rate": 2.8436377693277726e-05,
       "loss": 0.0001,
       "step": 3425
     },
     {
       "epoch": 7.537953795379538,
-      "grad_norm": 0.0027923583984375,
+      "grad_norm": 0.00592041015625,
       "learning_rate": 2.8387955210301632e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3426
     },
     {
       "epoch": 7.54015401540154,
-      "grad_norm": 0.0027313232421875,
+      "grad_norm": 0.002716064453125,
       "learning_rate": 2.8339567169407733e-05,
       "loss": 0.0001,
       "step": 3427
     },
     {
       "epoch": 7.542354235423542,
-      "grad_norm": 0.0035400390625,
+      "grad_norm": 0.003662109375,
       "learning_rate": 2.829121359386845e-05,
       "loss": 0.0001,
       "step": 3428
     },
     {
       "epoch": 7.544554455445544,
-      "grad_norm": 0.0029296875,
+      "grad_norm": 0.00469970703125,
       "learning_rate": 2.8242894506939523e-05,
       "loss": 0.0001,
       "step": 3429
     },
     {
       "epoch": 7.546754675467547,
-      "grad_norm": 0.004241943359375,
+      "grad_norm": 0.01513671875,
       "learning_rate": 2.8194609931860316e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 3430
     },
     {
       "epoch": 7.548954895489549,
-      "grad_norm": 0.01043701171875,
+      "grad_norm": 0.0030670166015625,
       "learning_rate": 2.814635989185337e-05,
       "loss": 0.0001,
       "step": 3431
     },
     {
       "epoch": 7.551155115511551,
-      "grad_norm": 0.0029754638671875,
+      "grad_norm": 0.0020904541015625,
       "learning_rate": 2.8098144410124837e-05,
       "loss": 0.0001,
       "step": 3432
     },
     {
       "epoch": 7.553355335533554,
-      "grad_norm": 0.0025177001953125,
+      "grad_norm": 0.003509521484375,
       "learning_rate": 2.8049963509864042e-05,
       "loss": 0.0001,
       "step": 3433
     },
     {
       "epoch": 7.555555555555555,
-      "grad_norm": 0.002105712890625,
+      "grad_norm": 0.0028839111328125,
       "learning_rate": 2.8001817214243774e-05,
       "loss": 0.0001,
       "step": 3434
     },
     {
       "epoch": 7.557755775577558,
-      "grad_norm": 0.004119873046875,
+      "grad_norm": 0.006805419921875,
       "learning_rate": 2.795370554642025e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3435
     },
     {
       "epoch": 7.55995599559956,
-      "grad_norm": 0.0027923583984375,
+      "grad_norm": 0.004669189453125,
       "learning_rate": 2.7905628529532858e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3436
     },
     {
       "epoch": 7.562156215621562,
-      "grad_norm": 0.0037078857421875,
+      "grad_norm": 0.005218505859375,
       "learning_rate": 2.785758618670453e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3437
     },
     {
       "epoch": 7.564356435643564,
-      "grad_norm": 0.002410888671875,
+      "grad_norm": 0.0034637451171875,
       "learning_rate": 2.7809578541041325e-05,
       "loss": 0.0001,
       "step": 3438
     },
     {
       "epoch": 7.566556655665567,
-      "grad_norm": 0.004669189453125,
+      "grad_norm": 0.00628662109375,
       "learning_rate": 2.776160561563269e-05,
       "loss": 0.0002,
       "step": 3439
     },
     {
       "epoch": 7.568756875687569,
-      "grad_norm": 0.004425048828125,
+      "grad_norm": 0.017822265625,
       "learning_rate": 2.7713667433551495e-05,
-      "loss": 0.0001,
+      "loss": 0.0003,
       "step": 3440
     },
     {
       "epoch": 7.570957095709571,
-      "grad_norm": 0.00579833984375,
+      "grad_norm": 0.002838134765625,
       "learning_rate": 2.7665764017853645e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3441
     },
     {
       "epoch": 7.5731573157315735,
-      "grad_norm": 0.00335693359375,
+      "grad_norm": 0.0098876953125,
       "learning_rate": 2.7617895391578607e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3442
     },
     {
       "epoch": 7.575357535753575,
-      "grad_norm": 0.001983642578125,
+      "grad_norm": 0.0027008056640625,
       "learning_rate": 2.7570061577748873e-05,
       "loss": 0.0001,
       "step": 3443
     },
     {
       "epoch": 7.577557755775578,
-      "grad_norm": 0.003936767578125,
+      "grad_norm": 0.0030975341796875,
       "learning_rate": 2.7522262599370307e-05,
       "loss": 0.0001,
       "step": 3444
     },
     {
       "epoch": 7.5797579757975795,
-      "grad_norm": 0.002899169921875,
+      "grad_norm": 0.0023651123046875,
       "learning_rate": 2.7474498479432087e-05,
       "loss": 0.0001,
       "step": 3445
     },
     {
       "epoch": 7.581958195819582,
-      "grad_norm": 0.007720947265625,
+      "grad_norm": 0.0037384033203125,
       "learning_rate": 2.7426769240906424e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3446
     },
     {
       "epoch": 7.584158415841584,
-      "grad_norm": 0.0023345947265625,
+      "grad_norm": 0.0034332275390625,
       "learning_rate": 2.7379074906749013e-05,
       "loss": 0.0001,
       "step": 3447
     },
     {
       "epoch": 7.586358635863586,
-      "grad_norm": 0.0042724609375,
+      "grad_norm": 0.00579833984375,
       "learning_rate": 2.7331415499898515e-05,
       "loss": 0.0002,
       "step": 3448
     },
     {
       "epoch": 7.588558855885589,
-      "grad_norm": 0.0062255859375,
+      "grad_norm": 0.0032806396484375,
       "learning_rate": 2.728379104327694e-05,
       "loss": 0.0001,
       "step": 3449
     },
     {
       "epoch": 7.590759075907591,
-      "grad_norm": 0.0037384033203125,
+      "grad_norm": 0.0047607421875,
       "learning_rate": 2.7236201559789453e-05,
       "loss": 0.0002,
       "step": 3450
     },
     {
       "epoch": 7.592959295929593,
-      "grad_norm": 0.004669189453125,
+      "grad_norm": 0.0084228515625,
       "learning_rate": 2.7188647072324346e-05,
       "loss": 0.0002,
       "step": 3451
     },
     {
       "epoch": 7.595159515951595,
-      "grad_norm": 0.0018310546875,
+      "grad_norm": 0.004638671875,
       "learning_rate": 2.7141127603753247e-05,
       "loss": 0.0001,
       "step": 3452
     },
     {
       "epoch": 7.597359735973598,
-      "grad_norm": 0.00518798828125,
+      "grad_norm": 0.003631591796875,
       "learning_rate": 2.7093643176930673e-05,
       "loss": 0.0001,
       "step": 3453
     },
     {
       "epoch": 7.599559955995599,
-      "grad_norm": 0.0024566650390625,
+      "grad_norm": 0.0032196044921875,
       "learning_rate": 2.7046193814694587e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3454
     },
     {
       "epoch": 7.601760176017602,
-      "grad_norm": 0.0035400390625,
+      "grad_norm": 0.00286865234375,
       "learning_rate": 2.6998779539865816e-05,
       "loss": 0.0001,
       "step": 3455
     },
     {
       "epoch": 7.603960396039604,
-      "grad_norm": 0.003082275390625,
+      "grad_norm": 0.0052490234375,
       "learning_rate": 2.6951400375248458e-05,
       "loss": 0.0001,
       "step": 3456
     },
     {
       "epoch": 7.606160616061606,
-      "grad_norm": 0.003143310546875,
+      "grad_norm": 0.004608154296875,
       "learning_rate": 2.690405634362979e-05,
       "loss": 0.0001,
       "step": 3457
     },
     {
       "epoch": 7.608360836083609,
-      "grad_norm": 0.00653076171875,
+      "grad_norm": 0.0032196044921875,
       "learning_rate": 2.6856747467779974e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3458
     },
     {
       "epoch": 7.6105610561056105,
-      "grad_norm": 0.00469970703125,
+      "grad_norm": 0.00494384765625,
       "learning_rate": 2.680947377045251e-05,
       "loss": 0.0002,
       "step": 3459
     },
     {
       "epoch": 7.612761276127613,
-      "grad_norm": 0.0024871826171875,
+      "grad_norm": 0.00347900390625,
       "learning_rate": 2.6762235274383772e-05,
       "loss": 0.0001,
       "step": 3460
     },
     {
       "epoch": 7.614961496149615,
-      "grad_norm": 0.0025177001953125,
+      "grad_norm": 0.0023345947265625,
       "learning_rate": 2.6715032002293306e-05,
       "loss": 0.0001,
       "step": 3461
     },
     {
       "epoch": 7.617161716171617,
-      "grad_norm": 0.00482177734375,
+      "grad_norm": 0.005706787109375,
       "learning_rate": 2.6667863976883767e-05,
       "loss": 0.0001,
       "step": 3462
     },
     {
       "epoch": 7.619361936193619,
-      "grad_norm": 0.0034942626953125,
+      "grad_norm": 0.01080322265625,
       "learning_rate": 2.662073122084069e-05,
       "loss": 0.0002,
       "step": 3463
     },
     {
       "epoch": 7.621562156215622,
-      "grad_norm": 0.004364013671875,
+      "grad_norm": 0.005035400390625,
       "learning_rate": 2.6573633756832872e-05,
       "loss": 0.0001,
       "step": 3464
     },
     {
       "epoch": 7.623762376237623,
-      "grad_norm": 0.00408935546875,
+      "grad_norm": 0.0084228515625,
       "learning_rate": 2.652657160751193e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3465
     },
     {
       "epoch": 7.625962596259626,
-      "grad_norm": 0.0062255859375,
+      "grad_norm": 0.0033416748046875,
       "learning_rate": 2.6479544795512568e-05,
       "loss": 0.0001,
       "step": 3466
     },
     {
       "epoch": 7.628162816281629,
-      "grad_norm": 0.0025482177734375,
+      "grad_norm": 0.0027618408203125,
       "learning_rate": 2.6432553343452605e-05,
       "loss": 0.0001,
       "step": 3467
     },
     {
       "epoch": 7.63036303630363,
-      "grad_norm": 0.007476806640625,
+      "grad_norm": 0.005096435546875,
       "learning_rate": 2.638559727393264e-05,
       "loss": 0.0001,
       "step": 3468
     },
     {
       "epoch": 7.632563256325633,
-      "grad_norm": 0.0028533935546875,
+      "grad_norm": 0.00439453125,
       "learning_rate": 2.6338676609536473e-05,
       "loss": 0.0001,
       "step": 3469
     },
     {
       "epoch": 7.634763476347635,
-      "grad_norm": 0.001617431640625,
+      "grad_norm": 0.0027008056640625,
       "learning_rate": 2.629179137283071e-05,
       "loss": 0.0001,
       "step": 3470
     },
     {
       "epoch": 7.636963696369637,
-      "grad_norm": 0.0025177001953125,
+      "grad_norm": 0.003814697265625,
       "learning_rate": 2.6244941586365e-05,
       "loss": 0.0001,
       "step": 3471
     },
     {
       "epoch": 7.639163916391639,
-      "grad_norm": 0.0093994140625,
+      "grad_norm": 0.00640869140625,
       "learning_rate": 2.6198127272671923e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3472
     },
     {
       "epoch": 7.6413641364136415,
-      "grad_norm": 0.0038604736328125,
+      "grad_norm": 0.005950927734375,
       "learning_rate": 2.6151348454266967e-05,
       "loss": 0.0002,
       "step": 3473
     },
     {
       "epoch": 7.643564356435643,
-      "grad_norm": 0.0030975341796875,
+      "grad_norm": 0.00384521484375,
       "learning_rate": 2.6104605153648674e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3474
     },
     {
       "epoch": 7.645764576457646,
-      "grad_norm": 0.0023040771484375,
+      "grad_norm": 0.0026092529296875,
       "learning_rate": 2.6057897393298324e-05,
       "loss": 0.0001,
       "step": 3475
     },
     {
       "epoch": 7.647964796479648,
-      "grad_norm": 0.002044677734375,
+      "grad_norm": 0.0030670166015625,
       "learning_rate": 2.601122519568022e-05,
       "loss": 0.0001,
       "step": 3476
     },
     {
       "epoch": 7.65016501650165,
-      "grad_norm": 0.00189971923828125,
+      "grad_norm": 0.0042724609375,
       "learning_rate": 2.5964588583241523e-05,
       "loss": 0.0001,
       "step": 3477
     },
     {
       "epoch": 7.652365236523653,
-      "grad_norm": 0.0026702880859375,
+      "grad_norm": 0.002838134765625,
       "learning_rate": 2.5917987578412263e-05,
       "loss": 0.0001,
       "step": 3478
     },
     {
       "epoch": 7.6545654565456545,
-      "grad_norm": 0.00390625,
+      "grad_norm": 0.0031585693359375,
       "learning_rate": 2.5871422203605444e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3479
     },
     {
       "epoch": 7.656765676567657,
-      "grad_norm": 0.00238037109375,
+      "grad_norm": 0.00811767578125,
       "learning_rate": 2.5824892481216767e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3480
     },
     {
       "epoch": 7.658965896589659,
-      "grad_norm": 0.0020904541015625,
+      "grad_norm": 0.003875732421875,
       "learning_rate": 2.5778398433624973e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3481
     },
     {
       "epoch": 7.661166116611661,
-      "grad_norm": 0.0023193359375,
+      "grad_norm": 0.00482177734375,
       "learning_rate": 2.5731940083191462e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3482
     },
     {
       "epoch": 7.663366336633663,
-      "grad_norm": 0.002349853515625,
+      "grad_norm": 0.00250244140625,
       "learning_rate": 2.5685517452260567e-05,
       "loss": 0.0001,
       "step": 3483
     },
     {
       "epoch": 7.665566556655666,
-      "grad_norm": 0.0035858154296875,
+      "grad_norm": 0.0030975341796875,
       "learning_rate": 2.5639130563159496e-05,
       "loss": 0.0001,
       "step": 3484
     },
     {
       "epoch": 7.667766776677668,
-      "grad_norm": 0.0037078857421875,
+      "grad_norm": 0.0047607421875,
       "learning_rate": 2.55927794381981e-05,
       "loss": 0.0001,
       "step": 3485
     },
     {
       "epoch": 7.66996699669967,
-      "grad_norm": 0.0022125244140625,
+      "grad_norm": 0.00738525390625,
       "learning_rate": 2.554646409966922e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3486
     },
     {
       "epoch": 7.672167216721672,
-      "grad_norm": 0.00262451171875,
+      "grad_norm": 0.002899169921875,
       "learning_rate": 2.550018456984833e-05,
       "loss": 0.0001,
       "step": 3487
     },
     {
       "epoch": 7.674367436743674,
-      "grad_norm": 0.008544921875,
+      "grad_norm": 0.005340576171875,
       "learning_rate": 2.5453940870993754e-05,
       "loss": 0.0002,
       "step": 3488
     },
     {
       "epoch": 7.676567656765677,
-      "grad_norm": 0.00439453125,
+      "grad_norm": 0.00579833984375,
       "learning_rate": 2.540773302534658e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3489
     },
     {
       "epoch": 7.678767876787679,
-      "grad_norm": 0.003570556640625,
+      "grad_norm": 0.004486083984375,
       "learning_rate": 2.536156105513062e-05,
       "loss": 0.0001,
       "step": 3490
     },
     {
       "epoch": 7.680968096809681,
-      "grad_norm": 0.0068359375,
+      "grad_norm": 0.0035858154296875,
       "learning_rate": 2.5315424982552537e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3491
     },
     {
       "epoch": 7.683168316831683,
-      "grad_norm": 0.0033111572265625,
+      "grad_norm": 0.00677490234375,
       "learning_rate": 2.5269324829801567e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3492
     },
     {
       "epoch": 7.6853685368536855,
-      "grad_norm": 0.00396728515625,
+      "grad_norm": 0.006591796875,
       "learning_rate": 2.522326061904978e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3493
     },
     {
       "epoch": 7.687568756875687,
-      "grad_norm": 0.00518798828125,
+      "grad_norm": 0.0048828125,
       "learning_rate": 2.517723237245193e-05,
       "loss": 0.0002,
       "step": 3494
     },
     {
       "epoch": 7.68976897689769,
-      "grad_norm": 0.006317138671875,
+      "grad_norm": 0.0028533935546875,
       "learning_rate": 2.5131240112145438e-05,
       "loss": 0.0001,
       "step": 3495
     },
     {
       "epoch": 7.6919691969196915,
-      "grad_norm": 0.0023345947265625,
+      "grad_norm": 0.00372314453125,
       "learning_rate": 2.5085283860250553e-05,
       "loss": 0.0001,
       "step": 3496
     },
     {
       "epoch": 7.694169416941694,
-      "grad_norm": 0.0026397705078125,
+      "grad_norm": 0.00567626953125,
       "learning_rate": 2.5039363638870018e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3497
     },
     {
       "epoch": 7.696369636963697,
-      "grad_norm": 0.00250244140625,
+      "grad_norm": 0.00482177734375,
       "learning_rate": 2.4993479470089364e-05,
       "loss": 0.0001,
       "step": 3498
     },
     {
       "epoch": 7.698569856985698,
-      "grad_norm": 0.004669189453125,
+      "grad_norm": 0.0101318359375,
       "learning_rate": 2.4947631375976742e-05,
       "loss": 0.0002,
       "step": 3499
     },
     {
       "epoch": 7.700770077007701,
-      "grad_norm": 0.00592041015625,
+      "grad_norm": 0.004302978515625,
       "learning_rate": 2.490181937858296e-05,
       "loss": 0.0001,
       "step": 3500
     },
     {
       "epoch": 7.702970297029703,
-      "grad_norm": 0.00927734375,
+      "grad_norm": 0.0038299560546875,
       "learning_rate": 2.485604349994155e-05,
       "loss": 0.0001,
       "step": 3501
     },
     {
       "epoch": 7.705170517051705,
-      "grad_norm": 0.0034942626953125,
+      "grad_norm": 0.00567626953125,
       "learning_rate": 2.481030376206849e-05,
       "loss": 0.0002,
       "step": 3502
     },
     {
       "epoch": 7.707370737073707,
-      "grad_norm": 0.003265380859375,
+      "grad_norm": 0.0026702880859375,
       "learning_rate": 2.4764600186962526e-05,
       "loss": 0.0001,
       "step": 3503
     },
     {
       "epoch": 7.70957095709571,
-      "grad_norm": 0.0050048828125,
+      "grad_norm": 0.00396728515625,
       "learning_rate": 2.4718932796604955e-05,
       "loss": 0.0001,
       "step": 3504
     },
     {
       "epoch": 7.711771177117711,
-      "grad_norm": 0.003021240234375,
+      "grad_norm": 0.003082275390625,
       "learning_rate": 2.4673301612959654e-05,
       "loss": 0.0001,
       "step": 3505
     },
     {
       "epoch": 7.713971397139714,
-      "grad_norm": 0.005889892578125,
+      "grad_norm": 0.00311279296875,
       "learning_rate": 2.4627706657973206e-05,
       "loss": 0.0002,
       "step": 3506
     },
     {
       "epoch": 7.7161716171617165,
-      "grad_norm": 0.0025787353515625,
+      "grad_norm": 0.0024566650390625,
       "learning_rate": 2.4582147953574542e-05,
       "loss": 0.0001,
       "step": 3507
     },
     {
       "epoch": 7.718371837183718,
-      "grad_norm": 0.005401611328125,
+      "grad_norm": 0.005828857421875,
       "learning_rate": 2.453662552167543e-05,
       "loss": 0.0002,
       "step": 3508
     },
     {
       "epoch": 7.720572057205721,
-      "grad_norm": 0.0034027099609375,
+      "grad_norm": 0.0054931640625,
       "learning_rate": 2.4491139384169947e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3509
     },
     {
       "epoch": 7.7227722772277225,
-      "grad_norm": 0.002288818359375,
+      "grad_norm": 0.002777099609375,
       "learning_rate": 2.444568956293486e-05,
       "loss": 0.0001,
       "step": 3510
     },
     {
       "epoch": 7.724972497249725,
-      "grad_norm": 0.00323486328125,
+      "grad_norm": 0.007568359375,
       "learning_rate": 2.4400276079829442e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3511
     },
     {
       "epoch": 7.727172717271727,
-      "grad_norm": 0.00347900390625,
+      "grad_norm": 0.00946044921875,
       "learning_rate": 2.4354898956695427e-05,
       "loss": 0.0001,
       "step": 3512
     },
     {
       "epoch": 7.729372937293729,
-      "grad_norm": 0.004486083984375,
+      "grad_norm": 0.0037689208984375,
       "learning_rate": 2.430955821535722e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3513
     },
     {
       "epoch": 7.731573157315731,
-      "grad_norm": 0.00262451171875,
+      "grad_norm": 0.0032501220703125,
       "learning_rate": 2.4264253877621535e-05,
       "loss": 0.0002,
       "step": 3514
     },
     {
       "epoch": 7.733773377337734,
-      "grad_norm": 0.0024871826171875,
+      "grad_norm": 0.0034942626953125,
       "learning_rate": 2.4218985965277675e-05,
       "loss": 0.0001,
       "step": 3515
     },
     {
       "epoch": 7.735973597359736,
-      "grad_norm": 0.00341796875,
+      "grad_norm": 0.00396728515625,
       "learning_rate": 2.4173754500097423e-05,
       "loss": 0.0001,
       "step": 3516
     },
     {
       "epoch": 7.738173817381738,
-      "grad_norm": 0.00299072265625,
+      "grad_norm": 0.0042724609375,
       "learning_rate": 2.412855950383499e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3517
     },
     {
       "epoch": 7.740374037403741,
-      "grad_norm": 0.01409912109375,
+      "grad_norm": 0.003387451171875,
       "learning_rate": 2.4083400998227167e-05,
-      "loss": 0.0003,
+      "loss": 0.0001,
       "step": 3518
     },
     {
       "epoch": 7.742574257425742,
-      "grad_norm": 0.0037689208984375,
+      "grad_norm": 0.0026702880859375,
       "learning_rate": 2.4038279004993025e-05,
       "loss": 0.0001,
       "step": 3519
     },
     {
       "epoch": 7.744774477447745,
-      "grad_norm": 0.00433349609375,
+      "grad_norm": 0.0033111572265625,
       "learning_rate": 2.399319354583418e-05,
       "loss": 0.0001,
       "step": 3520
     },
     {
       "epoch": 7.746974697469747,
-      "grad_norm": 0.0030364990234375,
+      "grad_norm": 0.00665283203125,
       "learning_rate": 2.3948144642434655e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3521
     },
     {
       "epoch": 7.749174917491749,
-      "grad_norm": 0.00225830078125,
+      "grad_norm": 0.003265380859375,
       "learning_rate": 2.3903132316460863e-05,
       "loss": 0.0001,
       "step": 3522
     },
     {
       "epoch": 7.751375137513751,
-      "grad_norm": 0.0026702880859375,
+      "grad_norm": 0.0032806396484375,
       "learning_rate": 2.3858156589561743e-05,
       "loss": 0.0001,
       "step": 3523
     },
     {
       "epoch": 7.7535753575357536,
-      "grad_norm": 0.0030059814453125,
+      "grad_norm": 0.003448486328125,
       "learning_rate": 2.3813217483368432e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3524
     },
     {
       "epoch": 7.755775577557756,
-      "grad_norm": 0.003387451171875,
+      "grad_norm": 0.003936767578125,
       "learning_rate": 2.3768315019494614e-05,
       "loss": 0.0001,
       "step": 3525
     },
     {
       "epoch": 7.757975797579758,
-      "grad_norm": 0.00186920166015625,
+      "grad_norm": 0.003936767578125,
       "learning_rate": 2.3723449219536274e-05,
       "loss": 0.0001,
       "step": 3526
     },
     {
       "epoch": 7.7601760176017605,
-      "grad_norm": 0.0022735595703125,
+      "grad_norm": 0.004638671875,
       "learning_rate": 2.3678620105071813e-05,
       "loss": 0.0001,
       "step": 3527
     },
     {
       "epoch": 7.762376237623762,
-      "grad_norm": 0.007354736328125,
+      "grad_norm": 0.004486083984375,
       "learning_rate": 2.3633827697661958e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3528
     },
     {
       "epoch": 7.764576457645765,
-      "grad_norm": 0.0027923583984375,
+      "grad_norm": 0.0020904541015625,
       "learning_rate": 2.3589072018849767e-05,
       "loss": 0.0001,
       "step": 3529
     },
     {
       "epoch": 7.7667766776677665,
-      "grad_norm": 0.00238037109375,
+      "grad_norm": 0.00299072265625,
       "learning_rate": 2.3544353090160666e-05,
       "loss": 0.0001,
       "step": 3530
     },
     {
       "epoch": 7.768976897689769,
-      "grad_norm": 0.0022125244140625,
+      "grad_norm": 0.0033416748046875,
       "learning_rate": 2.349967093310239e-05,
       "loss": 0.0001,
       "step": 3531
     },
     {
       "epoch": 7.771177117711771,
-      "grad_norm": 0.00396728515625,
+      "grad_norm": 0.0028228759765625,
       "learning_rate": 2.345502556916499e-05,
       "loss": 0.0001,
       "step": 3532
     },
     {
       "epoch": 7.773377337733773,
-      "grad_norm": 0.004150390625,
+      "grad_norm": 0.01287841796875,
       "learning_rate": 2.341041701982083e-05,
-      "loss": 0.0001,
+      "loss": 0.0003,
       "step": 3533
     },
     {
       "epoch": 7.775577557755776,
-      "grad_norm": 0.00274658203125,
+      "grad_norm": 0.0031280517578125,
       "learning_rate": 2.3365845306524516e-05,
       "loss": 0.0001,
       "step": 3534
     },
     {
       "epoch": 7.775577557755776,
-      "eval_loss": 0.0002438455558149144,
-      "eval_runtime": 10.1823,
-      "eval_samples_per_second": 37.614,
-      "eval_steps_per_second": 4.714,
+      "eval_loss": 0.0003531946858856827,
+      "eval_runtime": 10.0398,
+      "eval_samples_per_second": 38.148,
+      "eval_steps_per_second": 4.781,
       "step": 3534
     },
     {
       "epoch": 7.777777777777778,
-      "grad_norm": 0.0068359375,
+      "grad_norm": 0.005615234375,
       "learning_rate": 2.3321310450713062e-05,
       "loss": 0.0002,
       "step": 3535
     },
     {
       "epoch": 7.77997799779978,
-      "grad_norm": 0.0021820068359375,
+      "grad_norm": 0.0025482177734375,
       "learning_rate": 2.327681247380562e-05,
       "loss": 0.0001,
       "step": 3536
     },
     {
       "epoch": 7.782178217821782,
-      "grad_norm": 0.005462646484375,
+      "grad_norm": 0.0025634765625,
       "learning_rate": 2.3232351397203657e-05,
       "loss": 0.0001,
       "step": 3537
     },
     {
       "epoch": 7.784378437843785,
-      "grad_norm": 0.0024566650390625,
+      "grad_norm": 0.0032806396484375,
       "learning_rate": 2.3187927242290897e-05,
       "loss": 0.0001,
       "step": 3538
     },
     {
       "epoch": 7.786578657865786,
-      "grad_norm": 0.0026702880859375,
+      "grad_norm": 0.002593994140625,
       "learning_rate": 2.3143540030433296e-05,
       "loss": 0.0001,
       "step": 3539
     },
     {
       "epoch": 7.788778877887789,
-      "grad_norm": 0.00390625,
+      "grad_norm": 0.0062255859375,
       "learning_rate": 2.3099189782979124e-05,
       "loss": 0.0002,
       "step": 3540
     },
     {
       "epoch": 7.790979097909791,
-      "grad_norm": 0.0029144287109375,
+      "grad_norm": 0.00445556640625,
       "learning_rate": 2.305487652125872e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3541
     },
     {
       "epoch": 7.793179317931793,
-      "grad_norm": 0.0240478515625,
+      "grad_norm": 0.002777099609375,
       "learning_rate": 2.3010600266584724e-05,
-      "loss": 0.0006,
+      "loss": 0.0001,
       "step": 3542
     },
     {
       "epoch": 7.795379537953796,
-      "grad_norm": 0.0032958984375,
+      "grad_norm": 0.002960205078125,
       "learning_rate": 2.2966361040251982e-05,
       "loss": 0.0001,
       "step": 3543
     },
     {
       "epoch": 7.7975797579757975,
-      "grad_norm": 0.012939453125,
+      "grad_norm": 0.01507568359375,
       "learning_rate": 2.292215886353749e-05,
       "loss": 0.0002,
       "step": 3544
     },
     {
       "epoch": 7.7997799779978,
-      "grad_norm": 0.0023040771484375,
+      "grad_norm": 0.0034332275390625,
       "learning_rate": 2.2877993757700534e-05,
       "loss": 0.0001,
       "step": 3545
     },
     {
       "epoch": 7.801980198019802,
-      "grad_norm": 0.00482177734375,
+      "grad_norm": 0.005950927734375,
       "learning_rate": 2.2833865743982396e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3546
     },
     {
       "epoch": 7.804180418041804,
-      "grad_norm": 0.00726318359375,
+      "grad_norm": 0.0030670166015625,
       "learning_rate": 2.2789774843606637e-05,
       "loss": 0.0001,
       "step": 3547
     },
     {
       "epoch": 7.806380638063806,
-      "grad_norm": 0.002685546875,
+      "grad_norm": 0.002838134765625,
       "learning_rate": 2.274572107777896e-05,
       "loss": 0.0001,
       "step": 3548
     },
     {
       "epoch": 7.808580858085809,
-      "grad_norm": 0.002105712890625,
+      "grad_norm": 0.003936767578125,
       "learning_rate": 2.270170446768717e-05,
       "loss": 0.0001,
       "step": 3549
     },
     {
       "epoch": 7.81078107810781,
-      "grad_norm": 0.00811767578125,
+      "grad_norm": 0.00421142578125,
       "learning_rate": 2.265772503450122e-05,
       "loss": 0.0002,
       "step": 3550
     },
     {
       "epoch": 7.812981298129813,
-      "grad_norm": 0.00274658203125,
+      "grad_norm": 0.0023651123046875,
       "learning_rate": 2.2613782799373196e-05,
       "loss": 0.0001,
       "step": 3551
     },
     {
       "epoch": 7.815181518151816,
-      "grad_norm": 0.0050048828125,
+      "grad_norm": 0.0028228759765625,
       "learning_rate": 2.256987778343729e-05,
       "loss": 0.0001,
       "step": 3552
     },
     {
       "epoch": 7.817381738173817,
-      "grad_norm": 0.003631591796875,
+      "grad_norm": 0.0068359375,
       "learning_rate": 2.252601000780975e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3553
     },
     {
       "epoch": 7.81958195819582,
-      "grad_norm": 0.003143310546875,
+      "grad_norm": 0.010009765625,
       "learning_rate": 2.2482179493588984e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3554
     },
     {
       "epoch": 7.821782178217822,
-      "grad_norm": 0.003814697265625,
+      "grad_norm": 0.007568359375,
       "learning_rate": 2.2438386261855415e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3555
     },
     {
       "epoch": 7.823982398239824,
-      "grad_norm": 0.004486083984375,
+      "grad_norm": 0.006256103515625,
       "learning_rate": 2.2394630333671583e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3556
     },
     {
       "epoch": 7.826182618261826,
-      "grad_norm": 0.00390625,
+      "grad_norm": 0.0027313232421875,
       "learning_rate": 2.2350911730082058e-05,
       "loss": 0.0001,
       "step": 3557
     },
     {
       "epoch": 7.8283828382838285,
-      "grad_norm": 0.0029449462890625,
+      "grad_norm": 0.0034332275390625,
       "learning_rate": 2.2307230472113462e-05,
       "loss": 0.0001,
       "step": 3558
     },
     {
       "epoch": 7.83058305830583,
-      "grad_norm": 0.00543212890625,
+      "grad_norm": 0.0032196044921875,
       "learning_rate": 2.2263586580774466e-05,
       "loss": 0.0001,
       "step": 3559
     },
     {
       "epoch": 7.832783278327833,
-      "grad_norm": 0.004730224609375,
+      "grad_norm": 0.00506591796875,
       "learning_rate": 2.221998007705576e-05,
       "loss": 0.0002,
       "step": 3560
     },
     {
       "epoch": 7.834983498349835,
-      "grad_norm": 0.0020904541015625,
+      "grad_norm": 0.00421142578125,
       "learning_rate": 2.217641098193004e-05,
       "loss": 0.0001,
       "step": 3561
     },
     {
       "epoch": 7.837183718371837,
-      "grad_norm": 0.0031890869140625,
+      "grad_norm": 0.0036468505859375,
       "learning_rate": 2.2132879316352083e-05,
       "loss": 0.0001,
       "step": 3562
     },
     {
       "epoch": 7.83938393839384,
-      "grad_norm": 0.003936767578125,
+      "grad_norm": 0.0042724609375,
       "learning_rate": 2.208938510125855e-05,
       "loss": 0.0002,
       "step": 3563
@@ -25214,112 +25214,112 @@
     },
     {
       "epoch": 7.843784378437844,
-      "grad_norm": 0.003387451171875,
+      "grad_norm": 0.004119873046875,
       "learning_rate": 2.2002509106181624e-05,
       "loss": 0.0001,
       "step": 3565
     },
     {
       "epoch": 7.845984598459846,
-      "grad_norm": 0.0022735595703125,
+      "grad_norm": 0.0026702880859375,
       "learning_rate": 2.1959127367981556e-05,
       "loss": 0.0001,
       "step": 3566
     },
     {
       "epoch": 7.848184818481848,
-      "grad_norm": 0.005828857421875,
+      "grad_norm": 0.01165771484375,
       "learning_rate": 2.1915783163832592e-05,
       "loss": 0.0002,
       "step": 3567
     },
     {
       "epoch": 7.85038503850385,
-      "grad_norm": 0.0020751953125,
+      "grad_norm": 0.004852294921875,
       "learning_rate": 2.1872476514581276e-05,
       "loss": 0.0001,
       "step": 3568
     },
     {
       "epoch": 7.852585258525853,
-      "grad_norm": 0.00396728515625,
+      "grad_norm": 0.005889892578125,
       "learning_rate": 2.182920744105611e-05,
       "loss": 0.0002,
       "step": 3569
     },
     {
       "epoch": 7.854785478547855,
-      "grad_norm": 0.0029296875,
+      "grad_norm": 0.002471923828125,
       "learning_rate": 2.1785975964067516e-05,
       "loss": 0.0001,
       "step": 3570
     },
     {
       "epoch": 7.856985698569857,
-      "grad_norm": 0.0052490234375,
+      "grad_norm": 0.00555419921875,
       "learning_rate": 2.1742782104407834e-05,
       "loss": 0.0001,
       "step": 3571
     },
     {
       "epoch": 7.8591859185918596,
-      "grad_norm": 0.00274658203125,
+      "grad_norm": 0.00286865234375,
       "learning_rate": 2.1699625882851314e-05,
       "loss": 0.0001,
       "step": 3572
     },
     {
       "epoch": 7.861386138613861,
-      "grad_norm": 0.0052490234375,
+      "grad_norm": 0.005035400390625,
       "learning_rate": 2.1656507320154108e-05,
       "loss": 0.0002,
       "step": 3573
     },
     {
       "epoch": 7.863586358635864,
-      "grad_norm": 0.003753662109375,
+      "grad_norm": 0.0028076171875,
       "learning_rate": 2.1613426437054252e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3574
     },
     {
       "epoch": 7.865786578657866,
-      "grad_norm": 0.0024566650390625,
+      "grad_norm": 0.0030670166015625,
       "learning_rate": 2.1570383254271643e-05,
       "loss": 0.0001,
       "step": 3575
     },
     {
       "epoch": 7.867986798679868,
-      "grad_norm": 0.0023956298828125,
+      "grad_norm": 0.00244140625,
       "learning_rate": 2.1527377792508085e-05,
       "loss": 0.0001,
       "step": 3576
     },
     {
       "epoch": 7.87018701870187,
-      "grad_norm": 0.004974365234375,
+      "grad_norm": 0.0025634765625,
       "learning_rate": 2.1484410072447193e-05,
       "loss": 0.0001,
       "step": 3577
     },
     {
       "epoch": 7.8723872387238725,
-      "grad_norm": 0.0022430419921875,
+      "grad_norm": 0.01068115234375,
       "learning_rate": 2.1441480114754465e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3578
     },
     {
       "epoch": 7.874587458745875,
-      "grad_norm": 0.0027923583984375,
+      "grad_norm": 0.00274658203125,
       "learning_rate": 2.139858794007723e-05,
       "loss": 0.0001,
       "step": 3579
     },
     {
       "epoch": 7.876787678767877,
-      "grad_norm": 0.00250244140625,
+      "grad_norm": 0.0032196044921875,
       "learning_rate": 2.1355733569044635e-05,
       "loss": 0.0001,
       "step": 3580
@@ -25333,2937 +25333,2937 @@
     },
     {
       "epoch": 7.881188118811881,
-      "grad_norm": 0.00439453125,
+      "grad_norm": 0.0026092529296875,
       "learning_rate": 2.127013832033905e-05,
       "loss": 0.0001,
       "step": 3582
     },
     {
       "epoch": 7.883388338833884,
-      "grad_norm": 0.0032958984375,
+      "grad_norm": 0.00604248046875,
       "learning_rate": 2.1227397483833434e-05,
       "loss": 0.0001,
       "step": 3583
     },
     {
       "epoch": 7.885588558855885,
-      "grad_norm": 0.0031585693359375,
+      "grad_norm": 0.0022735595703125,
       "learning_rate": 2.1184694533307137e-05,
       "loss": 0.0001,
       "step": 3584
     },
     {
       "epoch": 7.887788778877888,
-      "grad_norm": 0.003021240234375,
+      "grad_norm": 0.00433349609375,
       "learning_rate": 2.114202948929832e-05,
       "loss": 0.0001,
       "step": 3585
     },
     {
       "epoch": 7.88998899889989,
-      "grad_norm": 0.0029754638671875,
+      "grad_norm": 0.007537841796875,
       "learning_rate": 2.1099402372326904e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3586
     },
     {
       "epoch": 7.892189218921892,
-      "grad_norm": 0.0033111572265625,
+      "grad_norm": 0.0035247802734375,
       "learning_rate": 2.1056813202894542e-05,
       "loss": 0.0001,
       "step": 3587
     },
     {
       "epoch": 7.894389438943895,
-      "grad_norm": 0.0031890869140625,
+      "grad_norm": 0.0035247802734375,
       "learning_rate": 2.1014262001484675e-05,
       "loss": 0.0001,
       "step": 3588
     },
     {
       "epoch": 7.896589658965897,
-      "grad_norm": 0.00183868408203125,
+      "grad_norm": 0.0087890625,
       "learning_rate": 2.0971748788562452e-05,
       "loss": 0.0001,
       "step": 3589
     },
     {
       "epoch": 7.898789878987898,
-      "grad_norm": 0.003692626953125,
+      "grad_norm": 0.005859375,
       "learning_rate": 2.092927358457476e-05,
       "loss": 0.0002,
       "step": 3590
     },
     {
       "epoch": 7.900990099009901,
-      "grad_norm": 0.005615234375,
+      "grad_norm": 0.0035858154296875,
       "learning_rate": 2.0886836409950218e-05,
       "loss": 0.0001,
       "step": 3591
     },
     {
       "epoch": 7.9031903190319035,
-      "grad_norm": 0.002532958984375,
+      "grad_norm": 0.002288818359375,
       "learning_rate": 2.0844437285099138e-05,
       "loss": 0.0001,
       "step": 3592
     },
     {
       "epoch": 7.905390539053905,
-      "grad_norm": 0.00445556640625,
+      "grad_norm": 0.00567626953125,
       "learning_rate": 2.0802076230413537e-05,
       "loss": 0.0001,
       "step": 3593
     },
     {
       "epoch": 7.907590759075908,
-      "grad_norm": 0.007171630859375,
+      "grad_norm": 0.0048828125,
       "learning_rate": 2.0759753266267123e-05,
       "loss": 0.0002,
       "step": 3594
     },
     {
       "epoch": 7.9097909790979095,
-      "grad_norm": 0.0084228515625,
+      "grad_norm": 0.0036163330078125,
       "learning_rate": 2.0717468413015283e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3595
     },
     {
       "epoch": 7.911991199119912,
-      "grad_norm": 0.0023651123046875,
+      "grad_norm": 0.0042724609375,
       "learning_rate": 2.0675221690995095e-05,
       "loss": 0.0001,
       "step": 3596
     },
     {
       "epoch": 7.914191419141914,
-      "grad_norm": 0.003204345703125,
+      "grad_norm": 0.01239013671875,
       "learning_rate": 2.0633013120525256e-05,
-      "loss": 0.0001,
+      "loss": 0.0003,
       "step": 3597
     },
     {
       "epoch": 7.916391639163916,
-      "grad_norm": 0.0029449462890625,
+      "grad_norm": 0.0034027099609375,
       "learning_rate": 2.0590842721906145e-05,
       "loss": 0.0001,
       "step": 3598
     },
     {
       "epoch": 7.918591859185918,
-      "grad_norm": 0.0037689208984375,
+      "grad_norm": 0.002777099609375,
       "learning_rate": 2.0548710515419778e-05,
       "loss": 0.0001,
       "step": 3599
     },
     {
       "epoch": 7.920792079207921,
-      "grad_norm": 0.00250244140625,
+      "grad_norm": 0.00274658203125,
       "learning_rate": 2.0506616521329804e-05,
       "loss": 0.0001,
       "step": 3600
     },
     {
       "epoch": 7.922992299229923,
-      "grad_norm": 0.002288818359375,
+      "grad_norm": 0.0028228759765625,
       "learning_rate": 2.0464560759881457e-05,
       "loss": 0.0001,
       "step": 3601
     },
     {
       "epoch": 7.925192519251925,
-      "grad_norm": 0.00262451171875,
+      "grad_norm": 0.0035400390625,
       "learning_rate": 2.0422543251301652e-05,
       "loss": 0.0001,
       "step": 3602
     },
     {
       "epoch": 7.927392739273928,
-      "grad_norm": 0.0026702880859375,
+      "grad_norm": 0.0030364990234375,
       "learning_rate": 2.0380564015798842e-05,
       "loss": 0.0001,
       "step": 3603
     },
     {
       "epoch": 7.929592959295929,
-      "grad_norm": 0.0025177001953125,
+      "grad_norm": 0.005157470703125,
       "learning_rate": 2.0338623073563102e-05,
       "loss": 0.0001,
       "step": 3604
     },
     {
       "epoch": 7.931793179317932,
-      "grad_norm": 0.004638671875,
+      "grad_norm": 0.005126953125,
       "learning_rate": 2.0296720444766094e-05,
       "loss": 0.0001,
       "step": 3605
     },
     {
       "epoch": 7.933993399339934,
-      "grad_norm": 0.0021820068359375,
+      "grad_norm": 0.00384521484375,
       "learning_rate": 2.0254856149561042e-05,
       "loss": 0.0001,
       "step": 3606
     },
     {
       "epoch": 7.936193619361936,
-      "grad_norm": 0.00323486328125,
+      "grad_norm": 0.00439453125,
       "learning_rate": 2.021303020808274e-05,
       "loss": 0.0001,
       "step": 3607
     },
     {
       "epoch": 7.938393839383938,
-      "grad_norm": 0.002532958984375,
+      "grad_norm": 0.002593994140625,
       "learning_rate": 2.017124264044752e-05,
       "loss": 0.0001,
       "step": 3608
     },
     {
       "epoch": 7.9405940594059405,
-      "grad_norm": 0.0028228759765625,
+      "grad_norm": 0.002410888671875,
       "learning_rate": 2.012949346675328e-05,
       "loss": 0.0001,
       "step": 3609
     },
     {
       "epoch": 7.942794279427943,
-      "grad_norm": 0.0034027099609375,
+      "grad_norm": 0.003997802734375,
       "learning_rate": 2.008778270707944e-05,
       "loss": 0.0001,
       "step": 3610
     },
     {
       "epoch": 7.944994499449945,
-      "grad_norm": 0.002349853515625,
+      "grad_norm": 0.002532958984375,
       "learning_rate": 2.0046110381486948e-05,
       "loss": 0.0001,
       "step": 3611
     },
     {
       "epoch": 7.947194719471947,
-      "grad_norm": 0.00494384765625,
+      "grad_norm": 0.00360107421875,
       "learning_rate": 2.0004476510018255e-05,
       "loss": 0.0001,
       "step": 3612
     },
     {
       "epoch": 7.949394939493949,
-      "grad_norm": 0.0024566650390625,
+      "grad_norm": 0.0042724609375,
       "learning_rate": 1.996288111269734e-05,
       "loss": 0.0001,
       "step": 3613
     },
     {
       "epoch": 7.951595159515952,
-      "grad_norm": 0.002655029296875,
+      "grad_norm": 0.0028839111328125,
       "learning_rate": 1.9921324209529667e-05,
       "loss": 0.0001,
       "step": 3614
     },
     {
       "epoch": 7.9537953795379535,
-      "grad_norm": 0.00567626953125,
+      "grad_norm": 0.005645751953125,
       "learning_rate": 1.9879805820502174e-05,
       "loss": 0.0002,
       "step": 3615
     },
     {
       "epoch": 7.955995599559956,
-      "grad_norm": 0.004974365234375,
+      "grad_norm": 0.01495361328125,
       "learning_rate": 1.9838325965583293e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3616
     },
     {
       "epoch": 7.958195819581958,
-      "grad_norm": 0.0026397705078125,
+      "grad_norm": 0.0025787353515625,
       "learning_rate": 1.9796884664722915e-05,
       "loss": 0.0001,
       "step": 3617
     },
     {
       "epoch": 7.96039603960396,
-      "grad_norm": 0.0020751953125,
+      "grad_norm": 0.0029144287109375,
       "learning_rate": 1.9755481937852395e-05,
       "loss": 0.0001,
       "step": 3618
     },
     {
       "epoch": 7.962596259625963,
-      "grad_norm": 0.0027313232421875,
+      "grad_norm": 0.00689697265625,
       "learning_rate": 1.9714117804884515e-05,
       "loss": 0.0001,
       "step": 3619
     },
     {
       "epoch": 7.964796479647965,
-      "grad_norm": 0.0025787353515625,
+      "grad_norm": 0.003143310546875,
       "learning_rate": 1.9672792285713528e-05,
       "loss": 0.0001,
       "step": 3620
     },
     {
       "epoch": 7.966996699669967,
-      "grad_norm": 0.002166748046875,
+      "grad_norm": 0.00225830078125,
       "learning_rate": 1.963150540021509e-05,
       "loss": 0.0001,
       "step": 3621
     },
     {
       "epoch": 7.969196919691969,
-      "grad_norm": 0.0023956298828125,
+      "grad_norm": 0.0024871826171875,
       "learning_rate": 1.9590257168246273e-05,
       "loss": 0.0001,
       "step": 3622
     },
     {
       "epoch": 7.971397139713972,
-      "grad_norm": 0.002777099609375,
+      "grad_norm": 0.00225830078125,
       "learning_rate": 1.9549047609645598e-05,
       "loss": 0.0001,
       "step": 3623
     },
     {
       "epoch": 7.973597359735973,
-      "grad_norm": 0.005279541015625,
+      "grad_norm": 0.002593994140625,
       "learning_rate": 1.9507876744232923e-05,
       "loss": 0.0001,
       "step": 3624
     },
     {
       "epoch": 7.975797579757976,
-      "grad_norm": 0.0032501220703125,
+      "grad_norm": 0.00653076171875,
       "learning_rate": 1.946674459180955e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3625
     },
     {
       "epoch": 7.977997799779978,
-      "grad_norm": 0.002105712890625,
+      "grad_norm": 0.00457763671875,
       "learning_rate": 1.9425651172158144e-05,
       "loss": 0.0001,
       "step": 3626
     },
     {
       "epoch": 7.98019801980198,
-      "grad_norm": 0.0023040771484375,
+      "grad_norm": 0.0018768310546875,
       "learning_rate": 1.938459650504274e-05,
       "loss": 0.0001,
       "step": 3627
     },
     {
       "epoch": 7.982398239823983,
-      "grad_norm": 0.0028228759765625,
+      "grad_norm": 0.004974365234375,
       "learning_rate": 1.9343580610208713e-05,
       "loss": 0.0001,
       "step": 3628
     },
     {
       "epoch": 7.9845984598459845,
-      "grad_norm": 0.005126953125,
+      "grad_norm": 0.00970458984375,
       "learning_rate": 1.9302603507382844e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3629
     },
     {
       "epoch": 7.986798679867987,
-      "grad_norm": 0.0036163330078125,
+      "grad_norm": 0.003143310546875,
       "learning_rate": 1.9261665216273193e-05,
       "loss": 0.0001,
       "step": 3630
     },
     {
       "epoch": 7.988998899889989,
-      "grad_norm": 0.003326416015625,
+      "grad_norm": 0.0050048828125,
       "learning_rate": 1.9220765756569213e-05,
       "loss": 0.0002,
       "step": 3631
     },
     {
       "epoch": 7.991199119911991,
-      "grad_norm": 0.00250244140625,
+      "grad_norm": 0.002349853515625,
       "learning_rate": 1.9179905147941636e-05,
       "loss": 0.0001,
       "step": 3632
     },
     {
       "epoch": 7.993399339933993,
-      "grad_norm": 0.003204345703125,
+      "grad_norm": 0.0029754638671875,
       "learning_rate": 1.9139083410042535e-05,
       "loss": 0.0001,
       "step": 3633
     },
     {
       "epoch": 7.995599559955996,
-      "grad_norm": 0.002349853515625,
+      "grad_norm": 0.005645751953125,
       "learning_rate": 1.9098300562505266e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3634
     },
     {
       "epoch": 7.997799779977997,
-      "grad_norm": 0.002838134765625,
+      "grad_norm": 0.004058837890625,
       "learning_rate": 1.9057556624944505e-05,
       "loss": 0.0001,
       "step": 3635
     },
     {
       "epoch": 8.0,
-      "grad_norm": 0.0054931640625,
+      "grad_norm": 0.00347900390625,
       "learning_rate": 1.901685161695619e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3636
     },
     {
       "epoch": 8.002200220022003,
-      "grad_norm": 0.0032958984375,
+      "grad_norm": 0.006988525390625,
       "learning_rate": 1.8976185558117567e-05,
       "loss": 0.0001,
       "step": 3637
     },
     {
       "epoch": 8.004400440044005,
-      "grad_norm": 0.00164794921875,
+      "grad_norm": 0.0025634765625,
       "learning_rate": 1.8935558467987103e-05,
       "loss": 0.0001,
       "step": 3638
     },
     {
       "epoch": 8.006600660066006,
-      "grad_norm": 0.002410888671875,
+      "grad_norm": 0.0026702880859375,
       "learning_rate": 1.889497036610457e-05,
       "loss": 0.0001,
       "step": 3639
     },
     {
       "epoch": 8.008800880088009,
-      "grad_norm": 0.004119873046875,
+      "grad_norm": 0.0069580078125,
       "learning_rate": 1.8854421271990964e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3640
     },
     {
       "epoch": 8.011001100110011,
-      "grad_norm": 0.003021240234375,
+      "grad_norm": 0.0026702880859375,
       "learning_rate": 1.8813911205148527e-05,
       "loss": 0.0001,
       "step": 3641
     },
     {
       "epoch": 8.013201320132014,
-      "grad_norm": 0.003265380859375,
+      "grad_norm": 0.0040283203125,
       "learning_rate": 1.877344018506071e-05,
       "loss": 0.0001,
       "step": 3642
     },
     {
       "epoch": 8.015401540154015,
-      "grad_norm": 0.00170135498046875,
+      "grad_norm": 0.0027618408203125,
       "learning_rate": 1.873300823119223e-05,
       "loss": 0.0001,
       "step": 3643
     },
     {
       "epoch": 8.017601760176017,
-      "grad_norm": 0.0029754638671875,
+      "grad_norm": 0.002166748046875,
       "learning_rate": 1.8692615362988962e-05,
       "loss": 0.0001,
       "step": 3644
     },
     {
       "epoch": 8.01980198019802,
-      "grad_norm": 0.002044677734375,
+      "grad_norm": 0.002593994140625,
       "learning_rate": 1.8652261599878028e-05,
       "loss": 0.0001,
       "step": 3645
     },
     {
       "epoch": 8.022002200220022,
-      "grad_norm": 0.0023956298828125,
+      "grad_norm": 0.00250244140625,
       "learning_rate": 1.8611946961267713e-05,
       "loss": 0.0001,
       "step": 3646
     },
     {
       "epoch": 8.024202420242025,
-      "grad_norm": 0.0025634765625,
+      "grad_norm": 0.004974365234375,
       "learning_rate": 1.857167146654749e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3647
     },
     {
       "epoch": 8.026402640264026,
-      "grad_norm": 0.005157470703125,
+      "grad_norm": 0.003021240234375,
       "learning_rate": 1.8531435135088015e-05,
       "loss": 0.0002,
       "step": 3648
     },
     {
       "epoch": 8.026402640264026,
-      "eval_loss": 0.00023969805624801666,
-      "eval_runtime": 10.2487,
-      "eval_samples_per_second": 37.371,
-      "eval_steps_per_second": 4.684,
+      "eval_loss": 0.00035373761784285307,
+      "eval_runtime": 11.2282,
+      "eval_samples_per_second": 34.111,
+      "eval_steps_per_second": 4.275,
       "step": 3648
     },
     {
       "epoch": 8.028602860286028,
-      "grad_norm": 0.003082275390625,
+      "grad_norm": 0.0023651123046875,
       "learning_rate": 1.8491237986241106e-05,
       "loss": 0.0001,
       "step": 3649
     },
     {
       "epoch": 8.030803080308031,
-      "grad_norm": 0.00537109375,
+      "grad_norm": 0.0037994384765625,
       "learning_rate": 1.845108003933972e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3650
     },
     {
       "epoch": 8.033003300330034,
-      "grad_norm": 0.004608154296875,
+      "grad_norm": 0.0033721923828125,
       "learning_rate": 1.841096131369798e-05,
       "loss": 0.0001,
       "step": 3651
     },
     {
       "epoch": 8.035203520352034,
-      "grad_norm": 0.0024261474609375,
+      "grad_norm": 0.0029449462890625,
       "learning_rate": 1.837088182861115e-05,
       "loss": 0.0001,
       "step": 3652
     },
     {
       "epoch": 8.037403740374037,
-      "grad_norm": 0.0035552978515625,
+      "grad_norm": 0.0031280517578125,
       "learning_rate": 1.8330841603355588e-05,
       "loss": 0.0001,
       "step": 3653
     },
     {
       "epoch": 8.03960396039604,
-      "grad_norm": 0.003631591796875,
+      "grad_norm": 0.005584716796875,
       "learning_rate": 1.8290840657188802e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3654
     },
     {
       "epoch": 8.041804180418042,
-      "grad_norm": 0.00244140625,
+      "grad_norm": 0.006683349609375,
       "learning_rate": 1.8250879009349398e-05,
       "loss": 0.0001,
       "step": 3655
     },
     {
       "epoch": 8.044004400440045,
-      "grad_norm": 0.00439453125,
+      "grad_norm": 0.0057373046875,
       "learning_rate": 1.821095667905708e-05,
       "loss": 0.0001,
       "step": 3656
     },
     {
       "epoch": 8.046204620462046,
-      "grad_norm": 0.01171875,
+      "grad_norm": 0.02001953125,
       "learning_rate": 1.817107368551264e-05,
-      "loss": 0.0001,
+      "loss": 0.0003,
       "step": 3657
     },
     {
       "epoch": 8.048404840484048,
-      "grad_norm": 0.00494384765625,
+      "grad_norm": 0.002471923828125,
       "learning_rate": 1.8131230047897964e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3658
     },
     {
       "epoch": 8.05060506050605,
-      "grad_norm": 0.004150390625,
+      "grad_norm": 0.0025787353515625,
       "learning_rate": 1.8091425785375982e-05,
       "loss": 0.0001,
       "step": 3659
     },
     {
       "epoch": 8.052805280528053,
-      "grad_norm": 0.0024261474609375,
+      "grad_norm": 0.00384521484375,
       "learning_rate": 1.805166091709072e-05,
       "loss": 0.0001,
       "step": 3660
     },
     {
       "epoch": 8.055005500550054,
-      "grad_norm": 0.0018157958984375,
+      "grad_norm": 0.0023193359375,
       "learning_rate": 1.801193546216723e-05,
       "loss": 0.0001,
       "step": 3661
     },
     {
       "epoch": 8.057205720572057,
-      "grad_norm": 0.0032806396484375,
+      "grad_norm": 0.0030364990234375,
       "learning_rate": 1.797224943971162e-05,
       "loss": 0.0001,
       "step": 3662
     },
     {
       "epoch": 8.05940594059406,
-      "grad_norm": 0.0026397705078125,
+      "grad_norm": 0.0025787353515625,
       "learning_rate": 1.7932602868811032e-05,
       "loss": 0.0001,
       "step": 3663
     },
     {
       "epoch": 8.061606160616062,
-      "grad_norm": 0.004241943359375,
+      "grad_norm": 0.0037689208984375,
       "learning_rate": 1.789299576853364e-05,
       "loss": 0.0001,
       "step": 3664
     },
     {
       "epoch": 8.063806380638065,
-      "grad_norm": 0.0034942626953125,
+      "grad_norm": 0.0042724609375,
       "learning_rate": 1.785342815792862e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3665
     },
     {
       "epoch": 8.066006600660065,
-      "grad_norm": 0.00191497802734375,
+      "grad_norm": 0.00360107421875,
       "learning_rate": 1.7813900056026167e-05,
       "loss": 0.0001,
       "step": 3666
     },
     {
       "epoch": 8.068206820682068,
-      "grad_norm": 0.00173187255859375,
+      "grad_norm": 0.0047607421875,
       "learning_rate": 1.777441148183746e-05,
       "loss": 0.0001,
       "step": 3667
     },
     {
       "epoch": 8.07040704070407,
-      "grad_norm": 0.00225830078125,
+      "grad_norm": 0.003997802734375,
       "learning_rate": 1.77349624543547e-05,
       "loss": 0.0001,
       "step": 3668
     },
     {
       "epoch": 8.072607260726073,
-      "grad_norm": 0.004730224609375,
+      "grad_norm": 0.004302978515625,
       "learning_rate": 1.7695552992551035e-05,
       "loss": 0.0001,
       "step": 3669
     },
     {
       "epoch": 8.074807480748074,
-      "grad_norm": 0.0025634765625,
+      "grad_norm": 0.003326416015625,
       "learning_rate": 1.7656183115380577e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3670
     },
     {
       "epoch": 8.077007700770077,
-      "grad_norm": 0.0035400390625,
+      "grad_norm": 0.01458740234375,
       "learning_rate": 1.7616852841778442e-05,
       "loss": 0.0001,
       "step": 3671
     },
     {
       "epoch": 8.07920792079208,
-      "grad_norm": 0.003021240234375,
+      "grad_norm": 0.003570556640625,
       "learning_rate": 1.7577562190660657e-05,
       "loss": 0.0001,
       "step": 3672
     },
     {
       "epoch": 8.081408140814082,
-      "grad_norm": 0.00193023681640625,
+      "grad_norm": 0.0025634765625,
       "learning_rate": 1.7538311180924206e-05,
       "loss": 0.0001,
       "step": 3673
     },
     {
       "epoch": 8.083608360836084,
-      "grad_norm": 0.00250244140625,
+      "grad_norm": 0.00396728515625,
       "learning_rate": 1.749909983144703e-05,
       "loss": 0.0001,
       "step": 3674
     },
     {
       "epoch": 8.085808580858085,
-      "grad_norm": 0.004058837890625,
+      "grad_norm": 0.01129150390625,
       "learning_rate": 1.7459928161087945e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3675
     },
     {
       "epoch": 8.088008800880088,
-      "grad_norm": 0.0040283203125,
+      "grad_norm": 0.002349853515625,
       "learning_rate": 1.742079618868674e-05,
       "loss": 0.0001,
       "step": 3676
     },
     {
       "epoch": 8.09020902090209,
-      "grad_norm": 0.0027313232421875,
+      "grad_norm": 0.00457763671875,
       "learning_rate": 1.7381703933064063e-05,
       "loss": 0.0001,
       "step": 3677
     },
     {
       "epoch": 8.092409240924093,
-      "grad_norm": 0.00244140625,
+      "grad_norm": 0.004425048828125,
       "learning_rate": 1.7342651413021503e-05,
       "loss": 0.0001,
       "step": 3678
     },
     {
       "epoch": 8.094609460946094,
-      "grad_norm": 0.005401611328125,
+      "grad_norm": 0.0037994384765625,
       "learning_rate": 1.7303638647341503e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3679
     },
     {
       "epoch": 8.096809680968097,
-      "grad_norm": 0.003265380859375,
+      "grad_norm": 0.0037384033203125,
       "learning_rate": 1.7264665654787404e-05,
       "loss": 0.0001,
       "step": 3680
     },
     {
       "epoch": 8.099009900990099,
-      "grad_norm": 0.003448486328125,
+      "grad_norm": 0.00433349609375,
       "learning_rate": 1.7225732454103404e-05,
       "loss": 0.0001,
       "step": 3681
     },
     {
       "epoch": 8.101210121012102,
-      "grad_norm": 0.00543212890625,
+      "grad_norm": 0.002532958984375,
       "learning_rate": 1.7186839064014582e-05,
       "loss": 0.0001,
       "step": 3682
     },
     {
       "epoch": 8.103410341034103,
-      "grad_norm": 0.0022430419921875,
+      "grad_norm": 0.002349853515625,
       "learning_rate": 1.714798550322685e-05,
       "loss": 0.0001,
       "step": 3683
     },
     {
       "epoch": 8.105610561056105,
-      "grad_norm": 0.0024261474609375,
+      "grad_norm": 0.012939453125,
       "learning_rate": 1.7109171790426992e-05,
       "loss": 0.0001,
       "step": 3684
     },
     {
       "epoch": 8.107810781078108,
-      "grad_norm": 0.0025787353515625,
+      "grad_norm": 0.005218505859375,
       "learning_rate": 1.707039794428259e-05,
       "loss": 0.0001,
       "step": 3685
     },
     {
       "epoch": 8.11001100110011,
-      "grad_norm": 0.0029449462890625,
+      "grad_norm": 0.0022430419921875,
       "learning_rate": 1.703166398344208e-05,
       "loss": 0.0001,
       "step": 3686
     },
     {
       "epoch": 8.112211221122113,
-      "grad_norm": 0.0024566650390625,
+      "grad_norm": 0.003814697265625,
       "learning_rate": 1.6992969926534718e-05,
       "loss": 0.0001,
       "step": 3687
     },
     {
       "epoch": 8.114411441144114,
-      "grad_norm": 0.0048828125,
+      "grad_norm": 0.00531005859375,
       "learning_rate": 1.6954315792170527e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3688
     },
     {
       "epoch": 8.116611661166116,
-      "grad_norm": 0.0186767578125,
+      "grad_norm": 0.005615234375,
       "learning_rate": 1.6915701598940383e-05,
-      "loss": 0.0004,
+      "loss": 0.0002,
       "step": 3689
     },
     {
       "epoch": 8.118811881188119,
-      "grad_norm": 0.0032196044921875,
+      "grad_norm": 0.003021240234375,
       "learning_rate": 1.6877127365415924e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3690
     },
     {
       "epoch": 8.121012101210122,
-      "grad_norm": 0.00189208984375,
+      "grad_norm": 0.006805419921875,
       "learning_rate": 1.683859311014956e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3691
     },
     {
       "epoch": 8.123212321232122,
-      "grad_norm": 0.0030517578125,
+      "grad_norm": 0.0030670166015625,
       "learning_rate": 1.6800098851674495e-05,
       "loss": 0.0001,
       "step": 3692
     },
     {
       "epoch": 8.125412541254125,
-      "grad_norm": 0.003936767578125,
+      "grad_norm": 0.0033721923828125,
       "learning_rate": 1.676164460850469e-05,
       "loss": 0.0001,
       "step": 3693
     },
     {
       "epoch": 8.127612761276128,
-      "grad_norm": 0.003631591796875,
+      "grad_norm": 0.00811767578125,
       "learning_rate": 1.672323039913485e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3694
     },
     {
       "epoch": 8.12981298129813,
-      "grad_norm": 0.0037078857421875,
+      "grad_norm": 0.002838134765625,
       "learning_rate": 1.6684856242040437e-05,
       "loss": 0.0001,
       "step": 3695
     },
     {
       "epoch": 8.132013201320133,
-      "grad_norm": 0.002166748046875,
+      "grad_norm": 0.0032501220703125,
       "learning_rate": 1.6646522155677646e-05,
       "loss": 0.0001,
       "step": 3696
     },
     {
       "epoch": 8.134213421342134,
-      "grad_norm": 0.002197265625,
+      "grad_norm": 0.0033416748046875,
       "learning_rate": 1.6608228158483406e-05,
       "loss": 0.0001,
       "step": 3697
     },
     {
       "epoch": 8.136413641364136,
-      "grad_norm": 0.01153564453125,
+      "grad_norm": 0.004119873046875,
       "learning_rate": 1.6569974268875353e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3698
     },
     {
       "epoch": 8.138613861386139,
-      "grad_norm": 0.00311279296875,
+      "grad_norm": 0.007110595703125,
       "learning_rate": 1.6531760505251836e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3699
     },
     {
       "epoch": 8.140814081408141,
-      "grad_norm": 0.002838134765625,
+      "grad_norm": 0.0026702880859375,
       "learning_rate": 1.649358688599191e-05,
       "loss": 0.0001,
       "step": 3700
     },
     {
       "epoch": 8.143014301430142,
-      "grad_norm": 0.01348876953125,
+      "grad_norm": 0.004058837890625,
       "learning_rate": 1.6455453429455327e-05,
       "loss": 0.0002,
       "step": 3701
     },
     {
       "epoch": 8.145214521452145,
-      "grad_norm": 0.00286865234375,
+      "grad_norm": 0.00433349609375,
       "learning_rate": 1.641736015398251e-05,
       "loss": 0.0001,
       "step": 3702
     },
     {
       "epoch": 8.147414741474147,
-      "grad_norm": 0.004425048828125,
+      "grad_norm": 0.00421142578125,
       "learning_rate": 1.637930707789457e-05,
       "loss": 0.0001,
       "step": 3703
     },
     {
       "epoch": 8.14961496149615,
-      "grad_norm": 0.00244140625,
+      "grad_norm": 0.006439208984375,
       "learning_rate": 1.6341294219493263e-05,
       "loss": 0.0001,
       "step": 3704
     },
     {
       "epoch": 8.151815181518153,
-      "grad_norm": 0.004608154296875,
+      "grad_norm": 0.00787353515625,
       "learning_rate": 1.630332159706105e-05,
       "loss": 0.0002,
       "step": 3705
     },
     {
       "epoch": 8.154015401540153,
-      "grad_norm": 0.0020599365234375,
+      "grad_norm": 0.00323486328125,
       "learning_rate": 1.626538922886094e-05,
       "loss": 0.0001,
       "step": 3706
     },
     {
       "epoch": 8.156215621562156,
-      "grad_norm": 0.0029144287109375,
+      "grad_norm": 0.003082275390625,
       "learning_rate": 1.6227497133136705e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3707
     },
     {
       "epoch": 8.158415841584159,
-      "grad_norm": 0.0034942626953125,
+      "grad_norm": 0.005096435546875,
       "learning_rate": 1.618964532811268e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3708
     },
     {
       "epoch": 8.160616061606161,
-      "grad_norm": 0.0031280517578125,
+      "grad_norm": 0.0048828125,
       "learning_rate": 1.6151833831993834e-05,
       "loss": 0.0001,
       "step": 3709
     },
     {
       "epoch": 8.162816281628162,
-      "grad_norm": 0.0032501220703125,
+      "grad_norm": 0.01025390625,
       "learning_rate": 1.6114062662965757e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3710
     },
     {
       "epoch": 8.165016501650165,
-      "grad_norm": 0.002349853515625,
+      "grad_norm": 0.0035552978515625,
       "learning_rate": 1.607633183919458e-05,
       "loss": 0.0001,
       "step": 3711
     },
     {
       "epoch": 8.167216721672167,
-      "grad_norm": 0.0033111572265625,
+      "grad_norm": 0.00396728515625,
       "learning_rate": 1.6038641378827145e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3712
     },
     {
       "epoch": 8.16941694169417,
-      "grad_norm": 0.0019989013671875,
+      "grad_norm": 0.0030975341796875,
       "learning_rate": 1.600099129999081e-05,
       "loss": 0.0001,
       "step": 3713
     },
     {
       "epoch": 8.171617161716172,
-      "grad_norm": 0.0048828125,
+      "grad_norm": 0.006439208984375,
       "learning_rate": 1.5963381620793495e-05,
       "loss": 0.0001,
       "step": 3714
     },
     {
       "epoch": 8.173817381738173,
-      "grad_norm": 0.006622314453125,
+      "grad_norm": 0.0025787353515625,
       "learning_rate": 1.5925812359323745e-05,
       "loss": 0.0001,
       "step": 3715
     },
     {
       "epoch": 8.176017601760176,
-      "grad_norm": 0.007354736328125,
+      "grad_norm": 0.002777099609375,
       "learning_rate": 1.5888283533650627e-05,
       "loss": 0.0001,
       "step": 3716
     },
     {
       "epoch": 8.178217821782178,
-      "grad_norm": 0.004241943359375,
+      "grad_norm": 0.0052490234375,
       "learning_rate": 1.585079516182376e-05,
       "loss": 0.0002,
       "step": 3717
     },
     {
       "epoch": 8.180418041804181,
-      "grad_norm": 0.003265380859375,
+      "grad_norm": 0.0024871826171875,
       "learning_rate": 1.5813347261873346e-05,
       "loss": 0.0001,
       "step": 3718
     },
     {
       "epoch": 8.182618261826182,
-      "grad_norm": 0.039306640625,
+      "grad_norm": 0.0032501220703125,
       "learning_rate": 1.5775939851810063e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3719
     },
     {
       "epoch": 8.184818481848184,
       "grad_norm": 0.004364013671875,
       "learning_rate": 1.5738572949625163e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3720
     },
     {
       "epoch": 8.187018701870187,
-      "grad_norm": 0.0032501220703125,
+      "grad_norm": 0.0031890869140625,
       "learning_rate": 1.5701246573290385e-05,
       "loss": 0.0001,
       "step": 3721
     },
     {
       "epoch": 8.18921892189219,
-      "grad_norm": 0.0037384033203125,
+      "grad_norm": 0.00531005859375,
       "learning_rate": 1.5663960740758e-05,
       "loss": 0.0001,
       "step": 3722
     },
     {
       "epoch": 8.191419141914192,
-      "grad_norm": 0.003631591796875,
+      "grad_norm": 0.004364013671875,
       "learning_rate": 1.5626715469960772e-05,
       "loss": 0.0001,
       "step": 3723
     },
     {
       "epoch": 8.193619361936193,
-      "grad_norm": 0.007659912109375,
+      "grad_norm": 0.003692626953125,
       "learning_rate": 1.5589510778811946e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3724
     },
     {
       "epoch": 8.195819581958196,
-      "grad_norm": 0.00185394287109375,
+      "grad_norm": 0.002838134765625,
       "learning_rate": 1.5552346685205266e-05,
       "loss": 0.0001,
       "step": 3725
     },
     {
       "epoch": 8.198019801980198,
-      "grad_norm": 0.0034637451171875,
+      "grad_norm": 0.003082275390625,
       "learning_rate": 1.551522320701493e-05,
       "loss": 0.0001,
       "step": 3726
     },
     {
       "epoch": 8.2002200220022,
-      "grad_norm": 0.005157470703125,
+      "grad_norm": 0.007659912109375,
       "learning_rate": 1.547814036209565e-05,
       "loss": 0.0002,
       "step": 3727
     },
     {
       "epoch": 8.202420242024202,
-      "grad_norm": 0.002685546875,
+      "grad_norm": 0.006591796875,
       "learning_rate": 1.5441098168282487e-05,
       "loss": 0.0001,
       "step": 3728
     },
     {
       "epoch": 8.204620462046204,
-      "grad_norm": 0.00531005859375,
+      "grad_norm": 0.0096435546875,
       "learning_rate": 1.5404096643391085e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3729
     },
     {
       "epoch": 8.206820682068207,
-      "grad_norm": 0.00482177734375,
+      "grad_norm": 0.00360107421875,
       "learning_rate": 1.5367135805217458e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3730
     },
     {
       "epoch": 8.20902090209021,
-      "grad_norm": 0.002838134765625,
+      "grad_norm": 0.006591796875,
       "learning_rate": 1.533021567153804e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3731
     },
     {
       "epoch": 8.211221122112212,
-      "grad_norm": 0.0034637451171875,
+      "grad_norm": 0.00390625,
       "learning_rate": 1.5293336260109757e-05,
       "loss": 0.0001,
       "step": 3732
     },
     {
       "epoch": 8.213421342134213,
-      "grad_norm": 0.01531982421875,
+      "grad_norm": 0.005157470703125,
       "learning_rate": 1.5256497588669805e-05,
-      "loss": 0.0003,
+      "loss": 0.0002,
       "step": 3733
     },
     {
       "epoch": 8.215621562156215,
-      "grad_norm": 0.0028839111328125,
+      "grad_norm": 0.00830078125,
       "learning_rate": 1.5219699674935973e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3734
     },
     {
       "epoch": 8.217821782178218,
-      "grad_norm": 0.00185394287109375,
+      "grad_norm": 0.00244140625,
       "learning_rate": 1.5182942536606316e-05,
       "loss": 0.0001,
       "step": 3735
     },
     {
       "epoch": 8.22002200220022,
-      "grad_norm": 0.0030059814453125,
+      "grad_norm": 0.007537841796875,
       "learning_rate": 1.5146226191359326e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3736
     },
     {
       "epoch": 8.222222222222221,
-      "grad_norm": 0.006500244140625,
+      "grad_norm": 0.004852294921875,
       "learning_rate": 1.51095506568539e-05,
       "loss": 0.0002,
       "step": 3737
     },
     {
       "epoch": 8.224422442244224,
-      "grad_norm": 0.003387451171875,
+      "grad_norm": 0.003326416015625,
       "learning_rate": 1.5072915950729182e-05,
       "loss": 0.0001,
       "step": 3738
     },
     {
       "epoch": 8.226622662266227,
-      "grad_norm": 0.01190185546875,
+      "grad_norm": 0.00665283203125,
       "learning_rate": 1.5036322090604848e-05,
       "loss": 0.0002,
       "step": 3739
     },
     {
       "epoch": 8.22882288228823,
-      "grad_norm": 0.003082275390625,
+      "grad_norm": 0.015869140625,
       "learning_rate": 1.499976909408085e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3740
     },
     {
       "epoch": 8.231023102310232,
-      "grad_norm": 0.0029449462890625,
+      "grad_norm": 0.004150390625,
       "learning_rate": 1.496325697873746e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3741
     },
     {
       "epoch": 8.233223322332233,
-      "grad_norm": 0.0023345947265625,
+      "grad_norm": 0.0033721923828125,
       "learning_rate": 1.4926785762135343e-05,
       "loss": 0.0001,
       "step": 3742
     },
     {
       "epoch": 8.235423542354235,
-      "grad_norm": 0.0036163330078125,
+      "grad_norm": 0.0030670166015625,
       "learning_rate": 1.4890355461815452e-05,
       "loss": 0.0001,
       "step": 3743
     },
     {
       "epoch": 8.237623762376238,
-      "grad_norm": 0.00360107421875,
+      "grad_norm": 0.0033416748046875,
       "learning_rate": 1.4853966095299098e-05,
       "loss": 0.0001,
       "step": 3744
     },
     {
       "epoch": 8.23982398239824,
-      "grad_norm": 0.004119873046875,
+      "grad_norm": 0.0052490234375,
       "learning_rate": 1.4817617680087825e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3745
     },
     {
       "epoch": 8.242024202420241,
-      "grad_norm": 0.0028228759765625,
+      "grad_norm": 0.00238037109375,
       "learning_rate": 1.4781310233663604e-05,
       "loss": 0.0001,
       "step": 3746
     },
     {
       "epoch": 8.244224422442244,
-      "grad_norm": 0.002227783203125,
+      "grad_norm": 0.002105712890625,
       "learning_rate": 1.474504377348861e-05,
       "loss": 0.0001,
       "step": 3747
     },
     {
       "epoch": 8.246424642464246,
-      "grad_norm": 0.0037384033203125,
+      "grad_norm": 0.005096435546875,
       "learning_rate": 1.4708818317005335e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3748
     },
     {
       "epoch": 8.248624862486249,
-      "grad_norm": 0.005859375,
+      "grad_norm": 0.007293701171875,
       "learning_rate": 1.4672633881636589e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3749
     },
     {
       "epoch": 8.250825082508252,
-      "grad_norm": 0.00433349609375,
+      "grad_norm": 0.006256103515625,
       "learning_rate": 1.4636490484785315e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3750
     },
     {
       "epoch": 8.253025302530252,
-      "grad_norm": 0.0059814453125,
+      "grad_norm": 0.007476806640625,
       "learning_rate": 1.460038814383491e-05,
       "loss": 0.0002,
       "step": 3751
     },
     {
       "epoch": 8.255225522552255,
-      "grad_norm": 0.0038604736328125,
+      "grad_norm": 0.005462646484375,
       "learning_rate": 1.4564326876148893e-05,
       "loss": 0.0001,
       "step": 3752
     },
     {
       "epoch": 8.257425742574258,
-      "grad_norm": 0.005340576171875,
+      "grad_norm": 0.007080078125,
       "learning_rate": 1.452830669907106e-05,
       "loss": 0.0002,
       "step": 3753
     },
     {
       "epoch": 8.25962596259626,
-      "grad_norm": 0.005096435546875,
+      "grad_norm": 0.018798828125,
       "learning_rate": 1.4492327629925495e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 3754
     },
     {
       "epoch": 8.261826182618261,
-      "grad_norm": 0.003082275390625,
+      "grad_norm": 0.00225830078125,
       "learning_rate": 1.4456389686016381e-05,
       "loss": 0.0001,
       "step": 3755
     },
     {
       "epoch": 8.264026402640264,
-      "grad_norm": 0.0024871826171875,
+      "grad_norm": 0.006072998046875,
       "learning_rate": 1.4420492884628278e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3756
     },
     {
       "epoch": 8.266226622662266,
-      "grad_norm": 0.00372314453125,
+      "grad_norm": 0.0032196044921875,
       "learning_rate": 1.4384637243025868e-05,
       "loss": 0.0001,
       "step": 3757
     },
     {
       "epoch": 8.268426842684269,
-      "grad_norm": 0.0028228759765625,
+      "grad_norm": 0.006072998046875,
       "learning_rate": 1.4348822778454062e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3758
     },
     {
       "epoch": 8.270627062706271,
-      "grad_norm": 0.0020904541015625,
+      "grad_norm": 0.002899169921875,
       "learning_rate": 1.4313049508137976e-05,
       "loss": 0.0001,
       "step": 3759
     },
     {
       "epoch": 8.272827282728272,
-      "grad_norm": 0.0023040771484375,
+      "grad_norm": 0.004425048828125,
       "learning_rate": 1.4277317449282834e-05,
       "loss": 0.0001,
       "step": 3760
     },
     {
       "epoch": 8.275027502750275,
-      "grad_norm": 0.0033111572265625,
+      "grad_norm": 0.0026397705078125,
       "learning_rate": 1.4241626619074167e-05,
       "loss": 0.0001,
       "step": 3761
     },
     {
       "epoch": 8.277227722772277,
-      "grad_norm": 0.004547119140625,
+      "grad_norm": 0.006744384765625,
       "learning_rate": 1.4205977034677598e-05,
       "loss": 0.0002,
       "step": 3762
     },
     {
       "epoch": 8.277227722772277,
-      "eval_loss": 0.00024230054987128824,
-      "eval_runtime": 10.3203,
-      "eval_samples_per_second": 37.111,
-      "eval_steps_per_second": 4.651,
+      "eval_loss": 0.0003495768178254366,
+      "eval_runtime": 10.1596,
+      "eval_samples_per_second": 37.698,
+      "eval_steps_per_second": 4.725,
       "step": 3762
     },
     {
       "epoch": 8.27942794279428,
-      "grad_norm": 0.0036163330078125,
+      "grad_norm": 0.004150390625,
       "learning_rate": 1.4170368713238924e-05,
       "loss": 0.0001,
       "step": 3763
     },
     {
       "epoch": 8.281628162816281,
-      "grad_norm": 0.0037384033203125,
+      "grad_norm": 0.0029296875,
       "learning_rate": 1.4134801671884123e-05,
       "loss": 0.0001,
       "step": 3764
     },
     {
       "epoch": 8.283828382838283,
-      "grad_norm": 0.002899169921875,
+      "grad_norm": 0.0028533935546875,
       "learning_rate": 1.4099275927719235e-05,
       "loss": 0.0001,
       "step": 3765
     },
     {
       "epoch": 8.286028602860286,
-      "grad_norm": 0.007415771484375,
+      "grad_norm": 0.006591796875,
       "learning_rate": 1.4063791497830592e-05,
       "loss": 0.0002,
       "step": 3766
     },
     {
       "epoch": 8.288228822882289,
-      "grad_norm": 0.0028839111328125,
+      "grad_norm": 0.0033721923828125,
       "learning_rate": 1.4028348399284464e-05,
       "loss": 0.0001,
       "step": 3767
     },
     {
       "epoch": 8.290429042904291,
-      "grad_norm": 0.0026092529296875,
+      "grad_norm": 0.006927490234375,
       "learning_rate": 1.3992946649127415e-05,
       "loss": 0.0001,
       "step": 3768
     },
     {
       "epoch": 8.292629262926292,
-      "grad_norm": 0.005035400390625,
+      "grad_norm": 0.0031280517578125,
       "learning_rate": 1.3957586264386035e-05,
       "loss": 0.0001,
       "step": 3769
     },
     {
       "epoch": 8.294829482948295,
-      "grad_norm": 0.0025634765625,
+      "grad_norm": 0.0034332275390625,
       "learning_rate": 1.3922267262067024e-05,
       "loss": 0.0001,
       "step": 3770
     },
     {
       "epoch": 8.297029702970297,
-      "grad_norm": 0.0027313232421875,
+      "grad_norm": 0.00372314453125,
       "learning_rate": 1.3886989659157212e-05,
       "loss": 0.0001,
       "step": 3771
     },
     {
       "epoch": 8.2992299229923,
-      "grad_norm": 0.003387451171875,
+      "grad_norm": 0.002471923828125,
       "learning_rate": 1.385175347262343e-05,
       "loss": 0.0001,
       "step": 3772
     },
     {
       "epoch": 8.3014301430143,
-      "grad_norm": 0.0022430419921875,
+      "grad_norm": 0.00408935546875,
       "learning_rate": 1.381655871941272e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3773
     },
     {
       "epoch": 8.303630363036303,
-      "grad_norm": 0.002960205078125,
+      "grad_norm": 0.0028839111328125,
       "learning_rate": 1.3781405416452109e-05,
       "loss": 0.0001,
       "step": 3774
     },
     {
       "epoch": 8.305830583058306,
-      "grad_norm": 0.00433349609375,
+      "grad_norm": 0.00390625,
       "learning_rate": 1.3746293580648717e-05,
       "loss": 0.0001,
       "step": 3775
     },
     {
       "epoch": 8.308030803080309,
-      "grad_norm": 0.0025634765625,
+      "grad_norm": 0.00286865234375,
       "learning_rate": 1.3711223228889725e-05,
       "loss": 0.0001,
       "step": 3776
     },
     {
       "epoch": 8.310231023102311,
-      "grad_norm": 0.00604248046875,
+      "grad_norm": 0.0030059814453125,
       "learning_rate": 1.3676194378042285e-05,
       "loss": 0.0001,
       "step": 3777
     },
     {
       "epoch": 8.312431243124312,
-      "grad_norm": 0.00482177734375,
+      "grad_norm": 0.00286865234375,
       "learning_rate": 1.3641207044953729e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3778
     },
     {
       "epoch": 8.314631463146315,
-      "grad_norm": 0.0019989013671875,
+      "grad_norm": 0.0037078857421875,
       "learning_rate": 1.3606261246451324e-05,
       "loss": 0.0001,
       "step": 3779
     },
     {
       "epoch": 8.316831683168317,
-      "grad_norm": 0.003173828125,
+      "grad_norm": 0.0023193359375,
       "learning_rate": 1.3571356999342366e-05,
       "loss": 0.0001,
       "step": 3780
     },
     {
       "epoch": 8.31903190319032,
-      "grad_norm": 0.003143310546875,
+      "grad_norm": 0.0031890869140625,
       "learning_rate": 1.3536494320414228e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3781
     },
     {
       "epoch": 8.32123212321232,
-      "grad_norm": 0.003692626953125,
+      "grad_norm": 0.003204345703125,
       "learning_rate": 1.3501673226434175e-05,
       "loss": 0.0002,
       "step": 3782
     },
     {
       "epoch": 8.323432343234323,
-      "grad_norm": 0.0020294189453125,
+      "grad_norm": 0.0019989013671875,
       "learning_rate": 1.3466893734149621e-05,
       "loss": 0.0001,
       "step": 3783
     },
     {
       "epoch": 8.325632563256326,
-      "grad_norm": 0.003082275390625,
+      "grad_norm": 0.0033111572265625,
       "learning_rate": 1.343215586028782e-05,
       "loss": 0.0001,
       "step": 3784
     },
     {
       "epoch": 8.327832783278328,
-      "grad_norm": 0.0022735595703125,
+      "grad_norm": 0.0026092529296875,
       "learning_rate": 1.339745962155613e-05,
       "loss": 0.0001,
       "step": 3785
     },
     {
       "epoch": 8.33003300330033,
-      "grad_norm": 0.006317138671875,
+      "grad_norm": 0.009033203125,
       "learning_rate": 1.3362805034641856e-05,
       "loss": 0.0002,
       "step": 3786
     },
     {
       "epoch": 8.332233223322332,
-      "grad_norm": 0.0024261474609375,
+      "grad_norm": 0.004180908203125,
       "learning_rate": 1.3328192116212179e-05,
       "loss": 0.0001,
       "step": 3787
     },
     {
       "epoch": 8.334433443344334,
-      "grad_norm": 0.00244140625,
+      "grad_norm": 0.0029144287109375,
       "learning_rate": 1.3293620882914393e-05,
       "loss": 0.0001,
       "step": 3788
     },
     {
       "epoch": 8.336633663366337,
-      "grad_norm": 0.0079345703125,
+      "grad_norm": 0.00634765625,
       "learning_rate": 1.3259091351375586e-05,
       "loss": 0.0002,
       "step": 3789
     },
     {
       "epoch": 8.33883388338834,
-      "grad_norm": 0.0029144287109375,
+      "grad_norm": 0.002655029296875,
       "learning_rate": 1.3224603538202929e-05,
       "loss": 0.0001,
       "step": 3790
     },
     {
       "epoch": 8.34103410341034,
-      "grad_norm": 0.0030670166015625,
+      "grad_norm": 0.0037689208984375,
       "learning_rate": 1.3190157459983465e-05,
       "loss": 0.0001,
       "step": 3791
     },
     {
       "epoch": 8.343234323432343,
-      "grad_norm": 0.004119873046875,
+      "grad_norm": 0.00299072265625,
       "learning_rate": 1.3155753133284089e-05,
       "loss": 0.0001,
       "step": 3792
     },
     {
       "epoch": 8.345434543454346,
-      "grad_norm": 0.0218505859375,
+      "grad_norm": 0.002471923828125,
       "learning_rate": 1.3121390574651794e-05,
-      "loss": 0.0005,
+      "loss": 0.0001,
       "step": 3793
     },
     {
       "epoch": 8.347634763476348,
-      "grad_norm": 0.00396728515625,
+      "grad_norm": 0.005615234375,
       "learning_rate": 1.3087069800613283e-05,
       "loss": 0.0001,
       "step": 3794
     },
     {
       "epoch": 8.34983498349835,
-      "grad_norm": 0.003082275390625,
+      "grad_norm": 0.0069580078125,
       "learning_rate": 1.3052790827675343e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3795
     },
     {
       "epoch": 8.352035203520352,
-      "grad_norm": 0.003021240234375,
+      "grad_norm": 0.003753662109375,
       "learning_rate": 1.3018553672324529e-05,
       "loss": 0.0001,
       "step": 3796
     },
     {
       "epoch": 8.354235423542354,
-      "grad_norm": 0.003173828125,
+      "grad_norm": 0.00909423828125,
       "learning_rate": 1.2984358351027349e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3797
     },
     {
       "epoch": 8.356435643564357,
-      "grad_norm": 0.00372314453125,
+      "grad_norm": 0.0029449462890625,
       "learning_rate": 1.2950204880230187e-05,
       "loss": 0.0001,
       "step": 3798
     },
     {
       "epoch": 8.35863586358636,
-      "grad_norm": 0.003173828125,
+      "grad_norm": 0.004302978515625,
       "learning_rate": 1.291609327635921e-05,
       "loss": 0.0001,
       "step": 3799
     },
     {
       "epoch": 8.36083608360836,
-      "grad_norm": 0.0030975341796875,
+      "grad_norm": 0.0042724609375,
       "learning_rate": 1.28820235558206e-05,
       "loss": 0.0001,
       "step": 3800
     },
     {
       "epoch": 8.363036303630363,
-      "grad_norm": 0.00579833984375,
+      "grad_norm": 0.004364013671875,
       "learning_rate": 1.2847995735000295e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3801
     },
     {
       "epoch": 8.365236523652365,
-      "grad_norm": 0.005706787109375,
+      "grad_norm": 0.0106201171875,
       "learning_rate": 1.2814009830264095e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3802
     },
     {
       "epoch": 8.367436743674368,
-      "grad_norm": 0.0023651123046875,
+      "grad_norm": 0.002166748046875,
       "learning_rate": 1.2780065857957669e-05,
       "loss": 0.0001,
       "step": 3803
     },
     {
       "epoch": 8.369636963696369,
-      "grad_norm": 0.00213623046875,
+      "grad_norm": 0.0021820068359375,
       "learning_rate": 1.2746163834406444e-05,
       "loss": 0.0001,
       "step": 3804
     },
     {
       "epoch": 8.371837183718371,
-      "grad_norm": 0.00213623046875,
+      "grad_norm": 0.0048828125,
       "learning_rate": 1.2712303775915802e-05,
       "loss": 0.0001,
       "step": 3805
     },
     {
       "epoch": 8.374037403740374,
-      "grad_norm": 0.0028533935546875,
+      "grad_norm": 0.002349853515625,
       "learning_rate": 1.2678485698770782e-05,
       "loss": 0.0001,
       "step": 3806
     },
     {
       "epoch": 8.376237623762377,
-      "grad_norm": 0.0031585693359375,
+      "grad_norm": 0.006805419921875,
       "learning_rate": 1.2644709619236383e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3807
     },
     {
       "epoch": 8.37843784378438,
-      "grad_norm": 0.0023040771484375,
+      "grad_norm": 0.00225830078125,
       "learning_rate": 1.2610975553557336e-05,
       "loss": 0.0001,
       "step": 3808
     },
     {
       "epoch": 8.38063806380638,
-      "grad_norm": 0.0018463134765625,
+      "grad_norm": 0.0034942626953125,
       "learning_rate": 1.2577283517958105e-05,
       "loss": 0.0001,
       "step": 3809
     },
     {
       "epoch": 8.382838283828383,
-      "grad_norm": 0.002899169921875,
+      "grad_norm": 0.006134033203125,
       "learning_rate": 1.2543633528643084e-05,
       "loss": 0.0001,
       "step": 3810
     },
     {
       "epoch": 8.385038503850385,
-      "grad_norm": 0.0108642578125,
+      "grad_norm": 0.01348876953125,
       "learning_rate": 1.251002560179627e-05,
       "loss": 0.0003,
       "step": 3811
     },
     {
       "epoch": 8.387238723872388,
-      "grad_norm": 0.00286865234375,
+      "grad_norm": 0.003509521484375,
       "learning_rate": 1.2476459753581604e-05,
       "loss": 0.0001,
       "step": 3812
     },
     {
       "epoch": 8.389438943894389,
-      "grad_norm": 0.00732421875,
+      "grad_norm": 0.00799560546875,
       "learning_rate": 1.244293600014268e-05,
       "loss": 0.0002,
       "step": 3813
     },
     {
       "epoch": 8.391639163916391,
-      "grad_norm": 0.00811767578125,
+      "grad_norm": 0.00567626953125,
       "learning_rate": 1.2409454357602834e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3814
     },
     {
       "epoch": 8.393839383938394,
-      "grad_norm": 0.003265380859375,
+      "grad_norm": 0.0028076171875,
       "learning_rate": 1.2376014842065265e-05,
       "loss": 0.0001,
       "step": 3815
     },
     {
       "epoch": 8.396039603960396,
-      "grad_norm": 0.006927490234375,
+      "grad_norm": 0.0068359375,
       "learning_rate": 1.2342617469612738e-05,
       "loss": 0.0002,
       "step": 3816
     },
     {
       "epoch": 8.398239823982399,
-      "grad_norm": 0.0028533935546875,
+      "grad_norm": 0.0035247802734375,
       "learning_rate": 1.2309262256307918e-05,
       "loss": 0.0001,
       "step": 3817
     },
     {
       "epoch": 8.4004400440044,
-      "grad_norm": 0.00457763671875,
+      "grad_norm": 0.003448486328125,
       "learning_rate": 1.2275949218193129e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3818
     },
     {
       "epoch": 8.402640264026402,
-      "grad_norm": 0.00274658203125,
+      "grad_norm": 0.004180908203125,
       "learning_rate": 1.2242678371290328e-05,
       "loss": 0.0001,
       "step": 3819
     },
     {
       "epoch": 8.404840484048405,
-      "grad_norm": 0.032470703125,
+      "grad_norm": 0.00714111328125,
       "learning_rate": 1.220944973160133e-05,
       "loss": 0.0002,
       "step": 3820
     },
     {
       "epoch": 8.407040704070408,
-      "grad_norm": 0.0037994384765625,
+      "grad_norm": 0.0032501220703125,
       "learning_rate": 1.2176263315107506e-05,
       "loss": 0.0001,
       "step": 3821
     },
     {
       "epoch": 8.409240924092408,
-      "grad_norm": 0.002410888671875,
+      "grad_norm": 0.01214599609375,
       "learning_rate": 1.2143119137770064e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3822
     },
     {
       "epoch": 8.411441144114411,
-      "grad_norm": 0.005767822265625,
+      "grad_norm": 0.004486083984375,
       "learning_rate": 1.2110017215529745e-05,
       "loss": 0.0001,
       "step": 3823
     },
     {
       "epoch": 8.413641364136414,
-      "grad_norm": 0.004241943359375,
+      "grad_norm": 0.00384521484375,
       "learning_rate": 1.207695756430709e-05,
       "loss": 0.0001,
       "step": 3824
     },
     {
       "epoch": 8.415841584158416,
-      "grad_norm": 0.006683349609375,
+      "grad_norm": 0.00823974609375,
       "learning_rate": 1.2043940200002268e-05,
       "loss": 0.0002,
       "step": 3825
     },
     {
       "epoch": 8.418041804180419,
-      "grad_norm": 0.003021240234375,
+      "grad_norm": 0.0032196044921875,
       "learning_rate": 1.2010965138495055e-05,
       "loss": 0.0001,
       "step": 3826
     },
     {
       "epoch": 8.42024202420242,
-      "grad_norm": 0.0030975341796875,
+      "grad_norm": 0.0228271484375,
       "learning_rate": 1.1978032395645e-05,
-      "loss": 0.0001,
+      "loss": 0.0003,
       "step": 3827
     },
     {
       "epoch": 8.422442244224422,
-      "grad_norm": 0.01611328125,
+      "grad_norm": 0.002685546875,
       "learning_rate": 1.1945141987291153e-05,
-      "loss": 0.0003,
+      "loss": 0.0001,
       "step": 3828
     },
     {
       "epoch": 8.424642464246425,
-      "grad_norm": 0.0028228759765625,
+      "grad_norm": 0.00665283203125,
       "learning_rate": 1.1912293929252338e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3829
     },
     {
       "epoch": 8.426842684268427,
-      "grad_norm": 0.0021820068359375,
+      "grad_norm": 0.004852294921875,
       "learning_rate": 1.1879488237326952e-05,
       "loss": 0.0001,
       "step": 3830
     },
     {
       "epoch": 8.429042904290428,
-      "grad_norm": 0.00189971923828125,
+      "grad_norm": 0.005157470703125,
       "learning_rate": 1.1846724927292963e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3831
     },
     {
       "epoch": 8.43124312431243,
-      "grad_norm": 0.00311279296875,
+      "grad_norm": 0.00360107421875,
       "learning_rate": 1.1814004014908076e-05,
       "loss": 0.0001,
       "step": 3832
     },
     {
       "epoch": 8.433443344334433,
-      "grad_norm": 0.0024566650390625,
+      "grad_norm": 0.0025482177734375,
       "learning_rate": 1.1781325515909459e-05,
       "loss": 0.0001,
       "step": 3833
     },
     {
       "epoch": 8.435643564356436,
-      "grad_norm": 0.003509521484375,
+      "grad_norm": 0.00311279296875,
       "learning_rate": 1.1748689446014028e-05,
       "loss": 0.0001,
       "step": 3834
     },
     {
       "epoch": 8.437843784378439,
-      "grad_norm": 0.00311279296875,
+      "grad_norm": 0.0045166015625,
       "learning_rate": 1.1716095820918216e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3835
     },
     {
       "epoch": 8.44004400440044,
-      "grad_norm": 0.0025177001953125,
+      "grad_norm": 0.0036468505859375,
       "learning_rate": 1.1683544656297995e-05,
       "loss": 0.0001,
       "step": 3836
     },
     {
       "epoch": 8.442244224422442,
-      "grad_norm": 0.0030517578125,
+      "grad_norm": 0.00396728515625,
       "learning_rate": 1.1651035967809042e-05,
       "loss": 0.0001,
       "step": 3837
     },
     {
       "epoch": 8.444444444444445,
-      "grad_norm": 0.0027923583984375,
+      "grad_norm": 0.003387451171875,
       "learning_rate": 1.1618569771086463e-05,
       "loss": 0.0001,
       "step": 3838
     },
     {
       "epoch": 8.446644664466447,
-      "grad_norm": 0.00616455078125,
+      "grad_norm": 0.003662109375,
       "learning_rate": 1.1586146081745052e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3839
     },
     {
       "epoch": 8.448844884488448,
-      "grad_norm": 0.003753662109375,
+      "grad_norm": 0.00775146484375,
       "learning_rate": 1.1553764915379095e-05,
       "loss": 0.0001,
       "step": 3840
     },
     {
       "epoch": 8.45104510451045,
-      "grad_norm": 0.0054931640625,
+      "grad_norm": 0.004119873046875,
       "learning_rate": 1.1521426287562397e-05,
       "loss": 0.0001,
       "step": 3841
     },
     {
       "epoch": 8.453245324532453,
-      "grad_norm": 0.0031280517578125,
+      "grad_norm": 0.0034027099609375,
       "learning_rate": 1.148913021384841e-05,
       "loss": 0.0001,
       "step": 3842
     },
     {
       "epoch": 8.455445544554456,
-      "grad_norm": 0.0028076171875,
+      "grad_norm": 0.005950927734375,
       "learning_rate": 1.1456876709769959e-05,
       "loss": 0.0001,
       "step": 3843
     },
     {
       "epoch": 8.457645764576458,
-      "grad_norm": 0.0023345947265625,
+      "grad_norm": 0.00189208984375,
       "learning_rate": 1.1424665790839594e-05,
       "loss": 0.0001,
       "step": 3844
     },
     {
       "epoch": 8.45984598459846,
-      "grad_norm": 0.0037994384765625,
+      "grad_norm": 0.004180908203125,
       "learning_rate": 1.1392497472549202e-05,
       "loss": 0.0001,
       "step": 3845
     },
     {
       "epoch": 8.462046204620462,
-      "grad_norm": 0.00433349609375,
+      "grad_norm": 0.00360107421875,
       "learning_rate": 1.136037177037026e-05,
       "loss": 0.0001,
       "step": 3846
     },
     {
       "epoch": 8.464246424642464,
-      "grad_norm": 0.01019287109375,
+      "grad_norm": 0.0035552978515625,
       "learning_rate": 1.1328288699753808e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3847
     },
     {
       "epoch": 8.466446644664467,
-      "grad_norm": 0.002777099609375,
+      "grad_norm": 0.003173828125,
       "learning_rate": 1.1296248276130239e-05,
       "loss": 0.0001,
       "step": 3848
     },
     {
       "epoch": 8.468646864686468,
-      "grad_norm": 0.0023651123046875,
+      "grad_norm": 0.00421142578125,
       "learning_rate": 1.126425051490958e-05,
       "loss": 0.0001,
       "step": 3849
     },
     {
       "epoch": 8.47084708470847,
-      "grad_norm": 0.0037078857421875,
+      "grad_norm": 0.0098876953125,
       "learning_rate": 1.1232295431481222e-05,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 3850
     },
     {
       "epoch": 8.473047304730473,
-      "grad_norm": 0.0023193359375,
+      "grad_norm": 0.00286865234375,
       "learning_rate": 1.1200383041214113e-05,
       "loss": 0.0001,
       "step": 3851
     },
     {
       "epoch": 8.475247524752476,
-      "grad_norm": 0.0029296875,
+      "grad_norm": 0.0137939453125,
       "learning_rate": 1.1168513359456667e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3852
     },
     {
       "epoch": 8.477447744774478,
-      "grad_norm": 0.0030059814453125,
+      "grad_norm": 0.002899169921875,
       "learning_rate": 1.1136686401536644e-05,
       "loss": 0.0001,
       "step": 3853
     },
     {
       "epoch": 8.479647964796479,
-      "grad_norm": 0.0029754638671875,
+      "grad_norm": 0.0029144287109375,
       "learning_rate": 1.1104902182761434e-05,
       "loss": 0.0001,
       "step": 3854
     },
     {
       "epoch": 8.481848184818482,
-      "grad_norm": 0.0031585693359375,
+      "grad_norm": 0.003265380859375,
       "learning_rate": 1.1073160718417686e-05,
       "loss": 0.0001,
       "step": 3855
     },
     {
       "epoch": 8.484048404840484,
-      "grad_norm": 0.00225830078125,
+      "grad_norm": 0.0025482177734375,
       "learning_rate": 1.1041462023771631e-05,
       "loss": 0.0001,
       "step": 3856
     },
     {
       "epoch": 8.486248624862487,
-      "grad_norm": 0.00714111328125,
+      "grad_norm": 0.0031890869140625,
       "learning_rate": 1.10098061140689e-05,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3857
     },
     {
       "epoch": 8.488448844884488,
-      "grad_norm": 0.0026397705078125,
+      "grad_norm": 0.00341796875,
       "learning_rate": 1.0978193004534432e-05,
       "loss": 0.0001,
       "step": 3858
     },
     {
       "epoch": 8.49064906490649,
-      "grad_norm": 0.0022430419921875,
+      "grad_norm": 0.0025787353515625,
       "learning_rate": 1.0946622710372778e-05,
       "loss": 0.0001,
       "step": 3859
     },
     {
       "epoch": 8.492849284928493,
-      "grad_norm": 0.003082275390625,
+      "grad_norm": 0.0023651123046875,
       "learning_rate": 1.0915095246767692e-05,
       "loss": 0.0001,
       "step": 3860
     },
     {
       "epoch": 8.495049504950495,
-      "grad_norm": 0.00286865234375,
+      "grad_norm": 0.003936767578125,
       "learning_rate": 1.0883610628882513e-05,
       "loss": 0.0001,
       "step": 3861
     },
     {
       "epoch": 8.497249724972498,
-      "grad_norm": 0.002838134765625,
+      "grad_norm": 0.002471923828125,
       "learning_rate": 1.0852168871859825e-05,
       "loss": 0.0001,
       "step": 3862
     },
     {
       "epoch": 8.499449944994499,
-      "grad_norm": 0.00299072265625,
+      "grad_norm": 0.00701904296875,
       "learning_rate": 1.0820769990821655e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3863
     },
     {
       "epoch": 8.501650165016502,
-      "grad_norm": 0.0042724609375,
+      "grad_norm": 0.004302978515625,
       "learning_rate": 1.078941400086948e-05,
       "loss": 0.0001,
       "step": 3864
     },
     {
       "epoch": 8.503850385038504,
-      "grad_norm": 0.003936767578125,
+      "grad_norm": 0.0028076171875,
       "learning_rate": 1.0758100917083991e-05,
       "loss": 0.0001,
       "step": 3865
     },
     {
       "epoch": 8.506050605060507,
-      "grad_norm": 0.0030975341796875,
+      "grad_norm": 0.00201416015625,
       "learning_rate": 1.072683075452543e-05,
       "loss": 0.0001,
       "step": 3866
     },
     {
       "epoch": 8.508250825082508,
-      "grad_norm": 0.0037689208984375,
+      "grad_norm": 0.002532958984375,
       "learning_rate": 1.0695603528233222e-05,
       "loss": 0.0001,
       "step": 3867
     },
     {
       "epoch": 8.51045104510451,
-      "grad_norm": 0.004608154296875,
+      "grad_norm": 0.002471923828125,
       "learning_rate": 1.0664419253226232e-05,
       "loss": 0.0001,
       "step": 3868
     },
     {
       "epoch": 8.512651265126513,
-      "grad_norm": 0.0035552978515625,
+      "grad_norm": 0.01214599609375,
       "learning_rate": 1.0633277944502706e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3869
     },
     {
       "epoch": 8.514851485148515,
-      "grad_norm": 0.0025177001953125,
+      "grad_norm": 0.0019073486328125,
       "learning_rate": 1.0602179617040098e-05,
       "loss": 0.0001,
       "step": 3870
     },
     {
       "epoch": 8.517051705170516,
-      "grad_norm": 0.00244140625,
+      "grad_norm": 0.00311279296875,
       "learning_rate": 1.0571124285795342e-05,
       "loss": 0.0001,
       "step": 3871
     },
     {
       "epoch": 8.519251925192519,
-      "grad_norm": 0.002716064453125,
+      "grad_norm": 0.0032196044921875,
       "learning_rate": 1.0540111965704568e-05,
       "loss": 0.0001,
       "step": 3872
     },
     {
       "epoch": 8.521452145214521,
-      "grad_norm": 0.0021820068359375,
+      "grad_norm": 0.0054931640625,
       "learning_rate": 1.0509142671683248e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3873
     },
     {
       "epoch": 8.523652365236524,
-      "grad_norm": 0.002532958984375,
+      "grad_norm": 0.0037384033203125,
       "learning_rate": 1.0478216418626252e-05,
       "loss": 0.0001,
       "step": 3874
     },
     {
       "epoch": 8.525852585258527,
-      "grad_norm": 0.007720947265625,
+      "grad_norm": 0.006317138671875,
       "learning_rate": 1.0447333221407597e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3875
     },
     {
       "epoch": 8.528052805280527,
-      "grad_norm": 0.003814697265625,
+      "grad_norm": 0.005859375,
       "learning_rate": 1.0416493094880753e-05,
       "loss": 0.0001,
       "step": 3876
     },
     {
       "epoch": 8.528052805280527,
-      "eval_loss": 0.0002406568091828376,
-      "eval_runtime": 10.1919,
-      "eval_samples_per_second": 37.579,
-      "eval_steps_per_second": 4.71,
+      "eval_loss": 0.00035242459853179753,
+      "eval_runtime": 10.061,
+      "eval_samples_per_second": 38.068,
+      "eval_steps_per_second": 4.771,
       "step": 3876
     },
     {
       "epoch": 8.53025302530253,
-      "grad_norm": 0.003173828125,
+      "grad_norm": 0.0032806396484375,
       "learning_rate": 1.0385696053878313e-05,
       "loss": 0.0001,
       "step": 3877
     },
     {
       "epoch": 8.532453245324533,
-      "grad_norm": 0.002410888671875,
+      "grad_norm": 0.00262451171875,
       "learning_rate": 1.0354942113212285e-05,
       "loss": 0.0001,
       "step": 3878
     },
     {
       "epoch": 8.534653465346535,
-      "grad_norm": 0.00445556640625,
+      "grad_norm": 0.003875732421875,
       "learning_rate": 1.0324231287673902e-05,
       "loss": 0.0001,
       "step": 3879
     },
     {
       "epoch": 8.536853685368538,
-      "grad_norm": 0.00579833984375,
+      "grad_norm": 0.0022735595703125,
       "learning_rate": 1.0293563592033595e-05,
       "loss": 0.0001,
       "step": 3880
     },
     {
       "epoch": 8.539053905390539,
-      "grad_norm": 0.0021820068359375,
+      "grad_norm": 0.0027313232421875,
       "learning_rate": 1.0262939041041163e-05,
       "loss": 0.0001,
       "step": 3881
     },
     {
       "epoch": 8.541254125412541,
-      "grad_norm": 0.004241943359375,
+      "grad_norm": 0.003997802734375,
       "learning_rate": 1.023235764942555e-05,
       "loss": 0.0001,
       "step": 3882
     },
     {
       "epoch": 8.543454345434544,
-      "grad_norm": 0.0018310546875,
+      "grad_norm": 0.002655029296875,
       "learning_rate": 1.0201819431895044e-05,
       "loss": 0.0001,
       "step": 3883
     },
     {
       "epoch": 8.545654565456546,
-      "grad_norm": 0.0037841796875,
+      "grad_norm": 0.004241943359375,
       "learning_rate": 1.0171324403137084e-05,
       "loss": 0.0001,
       "step": 3884
     },
     {
       "epoch": 8.547854785478547,
-      "grad_norm": 0.0020904541015625,
+      "grad_norm": 0.00185394287109375,
       "learning_rate": 1.014087257781834e-05,
       "loss": 0.0001,
       "step": 3885
     },
     {
       "epoch": 8.55005500550055,
-      "grad_norm": 0.0030364990234375,
+      "grad_norm": 0.0033111572265625,
       "learning_rate": 1.0110463970584805e-05,
       "loss": 0.0001,
       "step": 3886
     },
     {
       "epoch": 8.552255225522552,
-      "grad_norm": 0.00482177734375,
+      "grad_norm": 0.00494384765625,
       "learning_rate": 1.0080098596061538e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3887
     },
     {
       "epoch": 8.554455445544555,
-      "grad_norm": 0.002105712890625,
+      "grad_norm": 0.0040283203125,
       "learning_rate": 1.0049776468852956e-05,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3888
     },
     {
       "epoch": 8.556655665566556,
-      "grad_norm": 0.004425048828125,
+      "grad_norm": 0.00640869140625,
       "learning_rate": 1.0019497603542538e-05,
       "loss": 0.0002,
       "step": 3889
     },
     {
       "epoch": 8.558855885588558,
-      "grad_norm": 0.00341796875,
+      "grad_norm": 0.003326416015625,
       "learning_rate": 9.989262014693013e-06,
       "loss": 0.0001,
       "step": 3890
     },
     {
       "epoch": 8.561056105610561,
-      "grad_norm": 0.0062255859375,
+      "grad_norm": 0.01104736328125,
       "learning_rate": 9.95906971684637e-06,
       "loss": 0.0002,
       "step": 3891
     },
     {
       "epoch": 8.563256325632564,
-      "grad_norm": 0.006195068359375,
+      "grad_norm": 0.004974365234375,
       "learning_rate": 9.928920724523627e-06,
       "loss": 0.0001,
       "step": 3892
     },
     {
       "epoch": 8.565456545654566,
-      "grad_norm": 0.002593994140625,
+      "grad_norm": 0.0023193359375,
       "learning_rate": 9.89881505222513e-06,
       "loss": 0.0001,
       "step": 3893
     },
     {
       "epoch": 8.567656765676567,
-      "grad_norm": 0.0027618408203125,
+      "grad_norm": 0.0032501220703125,
       "learning_rate": 9.868752714430263e-06,
       "loss": 0.0001,
       "step": 3894
     },
     {
       "epoch": 8.56985698569857,
-      "grad_norm": 0.0035552978515625,
+      "grad_norm": 0.0028228759765625,
       "learning_rate": 9.838733725597615e-06,
       "loss": 0.0001,
       "step": 3895
     },
     {
       "epoch": 8.572057205720572,
-      "grad_norm": 0.00372314453125,
+      "grad_norm": 0.002593994140625,
       "learning_rate": 9.808758100164994e-06,
       "loss": 0.0001,
       "step": 3896
     },
     {
       "epoch": 8.574257425742575,
-      "grad_norm": 0.0045166015625,
+      "grad_norm": 0.0034942626953125,
       "learning_rate": 9.7788258525492e-06,
       "loss": 0.0002,
       "step": 3897
     },
     {
       "epoch": 8.576457645764577,
-      "grad_norm": 0.0036468505859375,
+      "grad_norm": 0.00799560546875,
       "learning_rate": 9.748936997146351e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3898
     },
     {
       "epoch": 8.578657865786578,
-      "grad_norm": 0.0024566650390625,
+      "grad_norm": 0.002685546875,
       "learning_rate": 9.719091548331549e-06,
       "loss": 0.0001,
       "step": 3899
     },
     {
       "epoch": 8.58085808580858,
-      "grad_norm": 0.0029754638671875,
+      "grad_norm": 0.00732421875,
       "learning_rate": 9.689289520459088e-06,
       "loss": 0.0001,
       "step": 3900
     },
     {
       "epoch": 8.583058305830583,
-      "grad_norm": 0.002655029296875,
+      "grad_norm": 0.0024871826171875,
       "learning_rate": 9.659530927862359e-06,
       "loss": 0.0001,
       "step": 3901
     },
     {
       "epoch": 8.585258525852586,
-      "grad_norm": 0.0072021484375,
+      "grad_norm": 0.0040283203125,
       "learning_rate": 9.629815784853858e-06,
       "loss": 0.0001,
       "step": 3902
     },
     {
       "epoch": 8.587458745874587,
-      "grad_norm": 0.00360107421875,
+      "grad_norm": 0.0032196044921875,
       "learning_rate": 9.60014410572525e-06,
       "loss": 0.0001,
       "step": 3903
     },
     {
       "epoch": 8.58965896589659,
-      "grad_norm": 0.00616455078125,
+      "grad_norm": 0.0035400390625,
       "learning_rate": 9.570515904747157e-06,
       "loss": 0.0001,
       "step": 3904
     },
     {
       "epoch": 8.591859185918592,
-      "grad_norm": 0.0036468505859375,
+      "grad_norm": 0.0023956298828125,
       "learning_rate": 9.540931196169466e-06,
       "loss": 0.0001,
       "step": 3905
     },
     {
       "epoch": 8.594059405940595,
-      "grad_norm": 0.0040283203125,
+      "grad_norm": 0.005767822265625,
       "learning_rate": 9.511389994220987e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3906
     },
     {
       "epoch": 8.596259625962595,
-      "grad_norm": 0.00634765625,
+      "grad_norm": 0.0042724609375,
       "learning_rate": 9.481892313109674e-06,
       "loss": 0.0002,
       "step": 3907
     },
     {
       "epoch": 8.598459845984598,
-      "grad_norm": 0.003173828125,
+      "grad_norm": 0.0021209716796875,
       "learning_rate": 9.452438167022593e-06,
       "loss": 0.0001,
       "step": 3908
     },
     {
       "epoch": 8.6006600660066,
-      "grad_norm": 0.0028076171875,
+      "grad_norm": 0.0025177001953125,
       "learning_rate": 9.423027570125776e-06,
       "loss": 0.0001,
       "step": 3909
     },
     {
       "epoch": 8.602860286028603,
-      "grad_norm": 0.004150390625,
+      "grad_norm": 0.002197265625,
       "learning_rate": 9.393660536564408e-06,
       "loss": 0.0001,
       "step": 3910
     },
     {
       "epoch": 8.605060506050606,
-      "grad_norm": 0.0021209716796875,
+      "grad_norm": 0.00531005859375,
       "learning_rate": 9.364337080462637e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3911
     },
     {
       "epoch": 8.607260726072607,
-      "grad_norm": 0.005157470703125,
+      "grad_norm": 0.0054931640625,
       "learning_rate": 9.335057215923681e-06,
       "loss": 0.0002,
       "step": 3912
     },
     {
       "epoch": 8.60946094609461,
-      "grad_norm": 0.0038299560546875,
+      "grad_norm": 0.0024566650390625,
       "learning_rate": 9.305820957029853e-06,
       "loss": 0.0001,
       "step": 3913
     },
     {
       "epoch": 8.611661166116612,
-      "grad_norm": 0.0098876953125,
+      "grad_norm": 0.0057373046875,
       "learning_rate": 9.276628317842385e-06,
       "loss": 0.0001,
       "step": 3914
     },
     {
       "epoch": 8.613861386138614,
-      "grad_norm": 0.0018768310546875,
+      "grad_norm": 0.007568359375,
       "learning_rate": 9.247479312401641e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3915
     },
     {
       "epoch": 8.616061606160617,
-      "grad_norm": 0.0034637451171875,
+      "grad_norm": 0.005615234375,
       "learning_rate": 9.2183739547269e-06,
       "loss": 0.0001,
       "step": 3916
     },
     {
       "epoch": 8.618261826182618,
-      "grad_norm": 0.005523681640625,
+      "grad_norm": 0.002960205078125,
       "learning_rate": 9.189312258816495e-06,
       "loss": 0.0001,
       "step": 3917
     },
     {
       "epoch": 8.62046204620462,
-      "grad_norm": 0.0034942626953125,
+      "grad_norm": 0.00299072265625,
       "learning_rate": 9.160294238647804e-06,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3918
     },
     {
       "epoch": 8.622662266226623,
-      "grad_norm": 0.0028228759765625,
+      "grad_norm": 0.00555419921875,
       "learning_rate": 9.131319908177083e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3919
     },
     {
       "epoch": 8.624862486248626,
-      "grad_norm": 0.0040283203125,
+      "grad_norm": 0.00897216796875,
       "learning_rate": 9.102389281339718e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3920
     },
     {
       "epoch": 8.627062706270626,
-      "grad_norm": 0.003692626953125,
+      "grad_norm": 0.003662109375,
       "learning_rate": 9.073502372049946e-06,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3921
     },
     {
       "epoch": 8.629262926292629,
-      "grad_norm": 0.004364013671875,
+      "grad_norm": 0.002593994140625,
       "learning_rate": 9.044659194201054e-06,
       "loss": 0.0001,
       "step": 3922
     },
     {
       "epoch": 8.631463146314632,
-      "grad_norm": 0.00335693359375,
+      "grad_norm": 0.007659912109375,
       "learning_rate": 9.015859761665291e-06,
       "loss": 0.0001,
       "step": 3923
     },
     {
       "epoch": 8.633663366336634,
-      "grad_norm": 0.002655029296875,
+      "grad_norm": 0.00286865234375,
       "learning_rate": 8.987104088293818e-06,
       "loss": 0.0001,
       "step": 3924
     },
     {
       "epoch": 8.635863586358635,
-      "grad_norm": 0.0034332275390625,
+      "grad_norm": 0.0025787353515625,
       "learning_rate": 8.958392187916841e-06,
       "loss": 0.0001,
       "step": 3925
     },
     {
       "epoch": 8.638063806380638,
-      "grad_norm": 0.0027008056640625,
+      "grad_norm": 0.0026702880859375,
       "learning_rate": 8.929724074343403e-06,
       "loss": 0.0001,
       "step": 3926
     },
     {
       "epoch": 8.64026402640264,
-      "grad_norm": 0.006439208984375,
+      "grad_norm": 0.006072998046875,
       "learning_rate": 8.901099761361564e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3927
     },
     {
       "epoch": 8.642464246424643,
-      "grad_norm": 0.005218505859375,
+      "grad_norm": 0.004058837890625,
       "learning_rate": 8.872519262738287e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3928
     },
     {
       "epoch": 8.644664466446645,
-      "grad_norm": 0.002899169921875,
+      "grad_norm": 0.00616455078125,
       "learning_rate": 8.843982592219458e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3929
     },
     {
       "epoch": 8.646864686468646,
-      "grad_norm": 0.003509521484375,
+      "grad_norm": 0.00341796875,
       "learning_rate": 8.815489763529938e-06,
       "loss": 0.0001,
       "step": 3930
     },
     {
       "epoch": 8.649064906490649,
-      "grad_norm": 0.00323486328125,
+      "grad_norm": 0.00433349609375,
       "learning_rate": 8.787040790373413e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3931
     },
     {
       "epoch": 8.651265126512651,
-      "grad_norm": 0.013427734375,
+      "grad_norm": 0.003631591796875,
       "learning_rate": 8.75863568643257e-06,
       "loss": 0.0001,
       "step": 3932
     },
     {
       "epoch": 8.653465346534654,
-      "grad_norm": 0.00469970703125,
+      "grad_norm": 0.004150390625,
       "learning_rate": 8.730274465368914e-06,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3933
     },
     {
       "epoch": 8.655665566556655,
-      "grad_norm": 0.0025482177734375,
+      "grad_norm": 0.0035400390625,
       "learning_rate": 8.70195714082287e-06,
       "loss": 0.0001,
       "step": 3934
     },
     {
       "epoch": 8.657865786578657,
-      "grad_norm": 0.0028228759765625,
+      "grad_norm": 0.003753662109375,
       "learning_rate": 8.673683726413827e-06,
       "loss": 0.0001,
       "step": 3935
     },
     {
       "epoch": 8.66006600660066,
-      "grad_norm": 0.0031585693359375,
+      "grad_norm": 0.00860595703125,
       "learning_rate": 8.645454235739903e-06,
       "loss": 0.0002,
       "step": 3936
     },
     {
       "epoch": 8.662266226622663,
-      "grad_norm": 0.0023651123046875,
+      "grad_norm": 0.00250244140625,
       "learning_rate": 8.617268682378266e-06,
       "loss": 0.0001,
       "step": 3937
     },
     {
       "epoch": 8.664466446644665,
-      "grad_norm": 0.0021820068359375,
+      "grad_norm": 0.003875732421875,
       "learning_rate": 8.589127079884796e-06,
       "loss": 0.0001,
       "step": 3938
     },
     {
       "epoch": 8.666666666666666,
-      "grad_norm": 0.002471923828125,
+      "grad_norm": 0.00262451171875,
       "learning_rate": 8.561029441794299e-06,
       "loss": 0.0001,
       "step": 3939
     },
     {
       "epoch": 8.668866886688669,
-      "grad_norm": 0.0025634765625,
+      "grad_norm": 0.00469970703125,
       "learning_rate": 8.532975781620512e-06,
       "loss": 0.0001,
       "step": 3940
     },
     {
       "epoch": 8.671067106710671,
-      "grad_norm": 0.0030364990234375,
+      "grad_norm": 0.0023956298828125,
       "learning_rate": 8.504966112855872e-06,
       "loss": 0.0001,
       "step": 3941
     },
     {
       "epoch": 8.673267326732674,
-      "grad_norm": 0.002838134765625,
+      "grad_norm": 0.00482177734375,
       "learning_rate": 8.477000448971794e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3942
     },
     {
       "epoch": 8.675467546754675,
-      "grad_norm": 0.00244140625,
+      "grad_norm": 0.002899169921875,
       "learning_rate": 8.44907880341842e-06,
       "loss": 0.0001,
       "step": 3943
     },
     {
       "epoch": 8.677667766776677,
-      "grad_norm": 0.0022125244140625,
+      "grad_norm": 0.002532958984375,
       "learning_rate": 8.421201189624794e-06,
       "loss": 0.0001,
       "step": 3944
     },
     {
       "epoch": 8.67986798679868,
-      "grad_norm": 0.00201416015625,
+      "grad_norm": 0.004425048828125,
       "learning_rate": 8.39336762099876e-06,
       "loss": 0.0001,
       "step": 3945
     },
     {
       "epoch": 8.682068206820682,
-      "grad_norm": 0.002166748046875,
+      "grad_norm": 0.01025390625,
       "learning_rate": 8.365578110926953e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3946
     },
     {
       "epoch": 8.684268426842685,
-      "grad_norm": 0.003753662109375,
+      "grad_norm": 0.01153564453125,
       "learning_rate": 8.337832672774903e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3947
     },
     {
       "epoch": 8.686468646864686,
-      "grad_norm": 0.00286865234375,
+      "grad_norm": 0.0034637451171875,
       "learning_rate": 8.310131319886816e-06,
       "loss": 0.0001,
       "step": 3948
     },
     {
       "epoch": 8.688668866886688,
-      "grad_norm": 0.0025482177734375,
+      "grad_norm": 0.00787353515625,
       "learning_rate": 8.282474065585788e-06,
       "loss": 0.0001,
       "step": 3949
     },
     {
       "epoch": 8.690869086908691,
-      "grad_norm": 0.0027923583984375,
+      "grad_norm": 0.0019683837890625,
       "learning_rate": 8.254860923173691e-06,
       "loss": 0.0001,
       "step": 3950
     },
     {
       "epoch": 8.693069306930694,
-      "grad_norm": 0.0034637451171875,
+      "grad_norm": 0.0032958984375,
       "learning_rate": 8.227291905931135e-06,
       "loss": 0.0001,
       "step": 3951
     },
     {
       "epoch": 8.695269526952695,
-      "grad_norm": 0.0020294189453125,
+      "grad_norm": 0.00335693359375,
       "learning_rate": 8.1997670271176e-06,
       "loss": 0.0001,
       "step": 3952
     },
     {
       "epoch": 8.697469746974697,
-      "grad_norm": 0.0023193359375,
+      "grad_norm": 0.0025787353515625,
       "learning_rate": 8.172286299971233e-06,
       "loss": 0.0001,
       "step": 3953
     },
     {
       "epoch": 8.6996699669967,
-      "grad_norm": 0.0036468505859375,
+      "grad_norm": 0.0026702880859375,
       "learning_rate": 8.144849737709005e-06,
       "loss": 0.0001,
       "step": 3954
     },
     {
       "epoch": 8.701870187018702,
-      "grad_norm": 0.0035858154296875,
+      "grad_norm": 0.005859375,
       "learning_rate": 8.117457353526625e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3955
     },
     {
       "epoch": 8.704070407040705,
-      "grad_norm": 0.0020904541015625,
+      "grad_norm": 0.00238037109375,
       "learning_rate": 8.090109160598569e-06,
       "loss": 0.0001,
       "step": 3956
     },
     {
       "epoch": 8.706270627062706,
-      "grad_norm": 0.0029296875,
+      "grad_norm": 0.0025177001953125,
       "learning_rate": 8.062805172078081e-06,
       "loss": 0.0001,
       "step": 3957
     },
     {
       "epoch": 8.708470847084708,
-      "grad_norm": 0.00250244140625,
+      "grad_norm": 0.00396728515625,
       "learning_rate": 8.035545401097066e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3958
     },
     {
       "epoch": 8.710671067106711,
-      "grad_norm": 0.003570556640625,
+      "grad_norm": 0.00433349609375,
       "learning_rate": 8.008329860766272e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3959
     },
     {
       "epoch": 8.712871287128714,
-      "grad_norm": 0.00194549560546875,
+      "grad_norm": 0.004547119140625,
       "learning_rate": 7.981158564175072e-06,
       "loss": 0.0001,
       "step": 3960
     },
     {
       "epoch": 8.715071507150714,
-      "grad_norm": 0.006622314453125,
+      "grad_norm": 0.0035247802734375,
       "learning_rate": 7.954031524391615e-06,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 3961
     },
     {
       "epoch": 8.717271727172717,
-      "grad_norm": 0.00567626953125,
+      "grad_norm": 0.005523681640625,
       "learning_rate": 7.92694875446275e-06,
       "loss": 0.0002,
       "step": 3962
     },
     {
       "epoch": 8.71947194719472,
-      "grad_norm": 0.00360107421875,
+      "grad_norm": 0.00299072265625,
       "learning_rate": 7.899910267414023e-06,
       "loss": 0.0001,
       "step": 3963
     },
     {
       "epoch": 8.721672167216722,
-      "grad_norm": 0.0040283203125,
+      "grad_norm": 0.005401611328125,
       "learning_rate": 7.872916076249748e-06,
       "loss": 0.0002,
       "step": 3964
     },
     {
       "epoch": 8.723872387238725,
-      "grad_norm": 0.003997802734375,
+      "grad_norm": 0.00677490234375,
       "learning_rate": 7.845966193952824e-06,
       "loss": 0.0002,
       "step": 3965
     },
     {
       "epoch": 8.726072607260726,
-      "grad_norm": 0.003173828125,
+      "grad_norm": 0.00396728515625,
       "learning_rate": 7.819060633484931e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3966
     },
     {
       "epoch": 8.728272827282728,
-      "grad_norm": 0.0027923583984375,
+      "grad_norm": 0.002960205078125,
       "learning_rate": 7.792199407786383e-06,
       "loss": 0.0001,
       "step": 3967
     },
     {
       "epoch": 8.73047304730473,
-      "grad_norm": 0.0022735595703125,
+      "grad_norm": 0.003936767578125,
       "learning_rate": 7.765382529776189e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3968
     },
     {
       "epoch": 8.732673267326733,
-      "grad_norm": 0.007781982421875,
+      "grad_norm": 0.004974365234375,
       "learning_rate": 7.738610012352054e-06,
       "loss": 0.0001,
       "step": 3969
     },
     {
       "epoch": 8.734873487348734,
-      "grad_norm": 0.002105712890625,
+      "grad_norm": 0.003082275390625,
       "learning_rate": 7.711881868390291e-06,
       "loss": 0.0001,
       "step": 3970
     },
     {
       "epoch": 8.737073707370737,
-      "grad_norm": 0.002960205078125,
+      "grad_norm": 0.003570556640625,
       "learning_rate": 7.685198110745906e-06,
       "loss": 0.0001,
       "step": 3971
     },
     {
       "epoch": 8.73927392739274,
-      "grad_norm": 0.00274658203125,
+      "grad_norm": 0.0030975341796875,
       "learning_rate": 7.658558752252564e-06,
       "loss": 0.0001,
       "step": 3972
     },
     {
       "epoch": 8.741474147414742,
-      "grad_norm": 0.003997802734375,
+      "grad_norm": 0.00518798828125,
       "learning_rate": 7.63196380572253e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3973
     },
     {
       "epoch": 8.743674367436743,
-      "grad_norm": 0.0091552734375,
+      "grad_norm": 0.0033416748046875,
       "learning_rate": 7.605413283946816e-06,
       "loss": 0.0001,
       "step": 3974
     },
     {
       "epoch": 8.745874587458745,
-      "grad_norm": 0.004638671875,
+      "grad_norm": 0.0023956298828125,
       "learning_rate": 7.578907199694929e-06,
       "loss": 0.0001,
       "step": 3975
     },
     {
       "epoch": 8.748074807480748,
-      "grad_norm": 0.0024566650390625,
+      "grad_norm": 0.0027313232421875,
       "learning_rate": 7.552445565715094e-06,
       "loss": 0.0001,
       "step": 3976
     },
     {
       "epoch": 8.75027502750275,
-      "grad_norm": 0.00579833984375,
+      "grad_norm": 0.00506591796875,
       "learning_rate": 7.526028394734119e-06,
       "loss": 0.0002,
       "step": 3977
     },
     {
       "epoch": 8.752475247524753,
-      "grad_norm": 0.0023040771484375,
+      "grad_norm": 0.0037384033203125,
       "learning_rate": 7.499655699457453e-06,
       "loss": 0.0001,
       "step": 3978
     },
     {
       "epoch": 8.754675467546754,
-      "grad_norm": 0.0033416748046875,
+      "grad_norm": 0.0031280517578125,
       "learning_rate": 7.47332749256916e-06,
       "loss": 0.0002,
       "step": 3979
     },
     {
       "epoch": 8.756875687568757,
-      "grad_norm": 0.00543212890625,
+      "grad_norm": 0.0024871826171875,
       "learning_rate": 7.447043786731866e-06,
       "loss": 0.0001,
       "step": 3980
     },
     {
       "epoch": 8.75907590759076,
-      "grad_norm": 0.003631591796875,
+      "grad_norm": 0.0035400390625,
       "learning_rate": 7.4208045945868275e-06,
       "loss": 0.0001,
       "step": 3981
     },
     {
       "epoch": 8.761276127612762,
-      "grad_norm": 0.0021820068359375,
+      "grad_norm": 0.0054931640625,
       "learning_rate": 7.394609928753882e-06,
       "loss": 0.0001,
       "step": 3982
     },
     {
       "epoch": 8.763476347634764,
-      "grad_norm": 0.005859375,
+      "grad_norm": 0.0050048828125,
       "learning_rate": 7.368459801831451e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3983
     },
     {
       "epoch": 8.765676567656765,
-      "grad_norm": 0.0024566650390625,
+      "grad_norm": 0.004119873046875,
       "learning_rate": 7.342354226396531e-06,
       "loss": 0.0001,
       "step": 3984
     },
     {
       "epoch": 8.767876787678768,
-      "grad_norm": 0.00213623046875,
+      "grad_norm": 0.0037078857421875,
       "learning_rate": 7.3162932150046885e-06,
       "loss": 0.0001,
       "step": 3985
     },
     {
       "epoch": 8.77007700770077,
-      "grad_norm": 0.0023651123046875,
+      "grad_norm": 0.007232666015625,
       "learning_rate": 7.2902767801901084e-06,
       "loss": 0.0001,
       "step": 3986
     },
     {
       "epoch": 8.772277227722773,
-      "grad_norm": 0.007171630859375,
+      "grad_norm": 0.01025390625,
       "learning_rate": 7.264304934465449e-06,
       "loss": 0.0002,
       "step": 3987
     },
     {
       "epoch": 8.774477447744774,
-      "grad_norm": 0.0022430419921875,
+      "grad_norm": 0.0036163330078125,
       "learning_rate": 7.2383776903219845e-06,
       "loss": 0.0001,
       "step": 3988
     },
     {
       "epoch": 8.776677667766776,
-      "grad_norm": 0.0024261474609375,
+      "grad_norm": 0.0035400390625,
       "learning_rate": 7.212495060229507e-06,
       "loss": 0.0001,
       "step": 3989
     },
     {
       "epoch": 8.778877887788779,
-      "grad_norm": 0.003662109375,
+      "grad_norm": 0.00390625,
       "learning_rate": 7.18665705663637e-06,
       "loss": 0.0001,
       "step": 3990
     },
     {
       "epoch": 8.778877887788779,
-      "eval_loss": 0.00024147164367605,
-      "eval_runtime": 10.2677,
-      "eval_samples_per_second": 37.301,
-      "eval_steps_per_second": 4.675,
+      "eval_loss": 0.0003478019789326936,
+      "eval_runtime": 10.1885,
+      "eval_samples_per_second": 37.591,
+      "eval_steps_per_second": 4.711,
       "step": 3990
     },
     {
       "epoch": 8.781078107810782,
-      "grad_norm": 0.005157470703125,
+      "grad_norm": 0.003936767578125,
       "learning_rate": 7.1608636919694885e-06,
       "loss": 0.0001,
       "step": 3991
     },
     {
       "epoch": 8.783278327832782,
-      "grad_norm": 0.002960205078125,
+      "grad_norm": 0.0023193359375,
       "learning_rate": 7.135114978634238e-06,
       "loss": 0.0001,
       "step": 3992
     },
     {
       "epoch": 8.785478547854785,
-      "grad_norm": 0.0033416748046875,
+      "grad_norm": 0.006195068359375,
       "learning_rate": 7.109410929014582e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3993
     },
     {
       "epoch": 8.787678767876788,
-      "grad_norm": 0.0029449462890625,
+      "grad_norm": 0.00347900390625,
       "learning_rate": 7.0837515554729616e-06,
       "loss": 0.0001,
       "step": 3994
     },
     {
       "epoch": 8.78987898789879,
-      "grad_norm": 0.0089111328125,
+      "grad_norm": 0.00238037109375,
       "learning_rate": 7.05813687035034e-06,
       "loss": 0.0001,
       "step": 3995
     },
     {
       "epoch": 8.792079207920793,
-      "grad_norm": 0.0031585693359375,
+      "grad_norm": 0.0031280517578125,
       "learning_rate": 7.032566885966263e-06,
       "loss": 0.0001,
       "step": 3996
@@ -28277,1894 +28277,1894 @@
     },
     {
       "epoch": 8.796479647964796,
-      "grad_norm": 0.0022125244140625,
+      "grad_norm": 0.00616455078125,
       "learning_rate": 6.981561068583964e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 3998
     },
     {
       "epoch": 8.798679867986799,
-      "grad_norm": 0.0027313232421875,
+      "grad_norm": 0.0025177001953125,
       "learning_rate": 6.956125260117208e-06,
       "loss": 0.0001,
       "step": 3999
     },
     {
       "epoch": 8.800880088008801,
-      "grad_norm": 0.0027008056640625,
+      "grad_norm": 0.002899169921875,
       "learning_rate": 6.930734201451816e-06,
       "loss": 0.0001,
       "step": 4000
     },
     {
       "epoch": 8.803080308030804,
-      "grad_norm": 0.00244140625,
+      "grad_norm": 0.00531005859375,
       "learning_rate": 6.905387904799709e-06,
       "loss": 0.0001,
       "step": 4001
     },
     {
       "epoch": 8.805280528052805,
-      "grad_norm": 0.0027008056640625,
+      "grad_norm": 0.002288818359375,
       "learning_rate": 6.880086382351303e-06,
       "loss": 0.0001,
       "step": 4002
     },
     {
       "epoch": 8.807480748074807,
-      "grad_norm": 0.002838134765625,
+      "grad_norm": 0.00665283203125,
       "learning_rate": 6.854829646275429e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4003
     },
     {
       "epoch": 8.80968096809681,
-      "grad_norm": 0.002532958984375,
+      "grad_norm": 0.0028839111328125,
       "learning_rate": 6.829617708719438e-06,
       "loss": 0.0001,
       "step": 4004
     },
     {
       "epoch": 8.811881188118813,
-      "grad_norm": 0.002471923828125,
+      "grad_norm": 0.004241943359375,
       "learning_rate": 6.804450581809085e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4005
     },
     {
       "epoch": 8.814081408140813,
-      "grad_norm": 0.0030975341796875,
+      "grad_norm": 0.0028533935546875,
       "learning_rate": 6.779328277648611e-06,
       "loss": 0.0001,
       "step": 4006
     },
     {
       "epoch": 8.816281628162816,
-      "grad_norm": 0.00274658203125,
+      "grad_norm": 0.004302978515625,
       "learning_rate": 6.754250808320672e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4007
     },
     {
       "epoch": 8.818481848184819,
-      "grad_norm": 0.004364013671875,
+      "grad_norm": 0.00408935546875,
       "learning_rate": 6.729218185886377e-06,
       "loss": 0.0001,
       "step": 4008
     },
     {
       "epoch": 8.820682068206821,
-      "grad_norm": 0.0031280517578125,
+      "grad_norm": 0.0027923583984375,
       "learning_rate": 6.704230422385249e-06,
       "loss": 0.0001,
       "step": 4009
     },
     {
       "epoch": 8.822882288228822,
-      "grad_norm": 0.002471923828125,
+      "grad_norm": 0.00250244140625,
       "learning_rate": 6.679287529835266e-06,
       "loss": 0.0001,
       "step": 4010
     },
     {
       "epoch": 8.825082508250825,
-      "grad_norm": 0.002105712890625,
+      "grad_norm": 0.0040283203125,
       "learning_rate": 6.6543895202327844e-06,
       "loss": 0.0001,
       "step": 4011
     },
     {
       "epoch": 8.827282728272827,
-      "grad_norm": 0.0029449462890625,
+      "grad_norm": 0.004119873046875,
       "learning_rate": 6.629536405552595e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4012
     },
     {
       "epoch": 8.82948294829483,
-      "grad_norm": 0.002593994140625,
+      "grad_norm": 0.0027923583984375,
       "learning_rate": 6.604728197747945e-06,
       "loss": 0.0001,
       "step": 4013
     },
     {
       "epoch": 8.831683168316832,
-      "grad_norm": 0.004425048828125,
+      "grad_norm": 0.002777099609375,
       "learning_rate": 6.5799649087503686e-06,
       "loss": 0.0001,
       "step": 4014
     },
     {
       "epoch": 8.833883388338833,
-      "grad_norm": 0.0025634765625,
+      "grad_norm": 0.0033416748046875,
       "learning_rate": 6.555246550469907e-06,
       "loss": 0.0001,
       "step": 4015
     },
     {
       "epoch": 8.836083608360836,
-      "grad_norm": 0.002716064453125,
+      "grad_norm": 0.0028228759765625,
       "learning_rate": 6.530573134794915e-06,
       "loss": 0.0001,
       "step": 4016
     },
     {
       "epoch": 8.838283828382838,
-      "grad_norm": 0.01416015625,
+      "grad_norm": 0.00421142578125,
       "learning_rate": 6.505944673592179e-06,
       "loss": 0.0002,
       "step": 4017
     },
     {
       "epoch": 8.840484048404841,
-      "grad_norm": 0.0025177001953125,
+      "grad_norm": 0.004486083984375,
       "learning_rate": 6.481361178706879e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4018
     },
     {
       "epoch": 8.842684268426842,
-      "grad_norm": 0.002166748046875,
+      "grad_norm": 0.004974365234375,
       "learning_rate": 6.456822661962503e-06,
       "loss": 0.0001,
       "step": 4019
     },
     {
       "epoch": 8.844884488448844,
-      "grad_norm": 0.0052490234375,
+      "grad_norm": 0.005889892578125,
       "learning_rate": 6.432329135160953e-06,
       "loss": 0.0002,
       "step": 4020
     },
     {
       "epoch": 8.847084708470847,
-      "grad_norm": 0.0042724609375,
+      "grad_norm": 0.0030517578125,
       "learning_rate": 6.407880610082484e-06,
       "loss": 0.0001,
       "step": 4021
     },
     {
       "epoch": 8.84928492849285,
-      "grad_norm": 0.00341796875,
+      "grad_norm": 0.005950927734375,
       "learning_rate": 6.383477098485702e-06,
       "loss": 0.0001,
       "step": 4022
     },
     {
       "epoch": 8.851485148514852,
-      "grad_norm": 0.0033721923828125,
+      "grad_norm": 0.005859375,
       "learning_rate": 6.359118612107562e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4023
     },
     {
       "epoch": 8.853685368536853,
-      "grad_norm": 0.00701904296875,
+      "grad_norm": 0.0034637451171875,
       "learning_rate": 6.3348051626633934e-06,
       "loss": 0.0001,
       "step": 4024
     },
     {
       "epoch": 8.855885588558856,
-      "grad_norm": 0.002899169921875,
+      "grad_norm": 0.002288818359375,
       "learning_rate": 6.310536761846819e-06,
       "loss": 0.0001,
       "step": 4025
     },
     {
       "epoch": 8.858085808580858,
-      "grad_norm": 0.00262451171875,
+      "grad_norm": 0.0035400390625,
       "learning_rate": 6.286313421329826e-06,
       "loss": 0.0001,
       "step": 4026
     },
     {
       "epoch": 8.86028602860286,
-      "grad_norm": 0.0026092529296875,
+      "grad_norm": 0.0027618408203125,
       "learning_rate": 6.262135152762716e-06,
       "loss": 0.0001,
       "step": 4027
     },
     {
       "epoch": 8.862486248624862,
-      "grad_norm": 0.004150390625,
+      "grad_norm": 0.00830078125,
       "learning_rate": 6.238001967774121e-06,
       "loss": 0.0002,
       "step": 4028
     },
     {
       "epoch": 8.864686468646864,
-      "grad_norm": 0.002410888671875,
+      "grad_norm": 0.002716064453125,
       "learning_rate": 6.213913877971001e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4029
     },
     {
       "epoch": 8.866886688668867,
-      "grad_norm": 0.00244140625,
+      "grad_norm": 0.003204345703125,
       "learning_rate": 6.189870894938587e-06,
       "loss": 0.0001,
       "step": 4030
     },
     {
       "epoch": 8.86908690869087,
-      "grad_norm": 0.0023040771484375,
+      "grad_norm": 0.00439453125,
       "learning_rate": 6.165873030240454e-06,
       "loss": 0.0001,
       "step": 4031
     },
     {
       "epoch": 8.871287128712872,
-      "grad_norm": 0.004241943359375,
+      "grad_norm": 0.00506591796875,
       "learning_rate": 6.141920295418479e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4032
     },
     {
       "epoch": 8.873487348734873,
-      "grad_norm": 0.0057373046875,
+      "grad_norm": 0.00689697265625,
       "learning_rate": 6.118012701992804e-06,
       "loss": 0.0002,
       "step": 4033
     },
     {
       "epoch": 8.875687568756875,
-      "grad_norm": 0.0067138671875,
+      "grad_norm": 0.005401611328125,
       "learning_rate": 6.094150261461873e-06,
       "loss": 0.0002,
       "step": 4034
     },
     {
       "epoch": 8.877887788778878,
-      "grad_norm": 0.006622314453125,
+      "grad_norm": 0.004913330078125,
       "learning_rate": 6.0703329853024185e-06,
       "loss": 0.0002,
       "step": 4035
     },
     {
       "epoch": 8.88008800880088,
-      "grad_norm": 0.02099609375,
+      "grad_norm": 0.0038909912109375,
       "learning_rate": 6.0465608849694655e-06,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 4036
     },
     {
       "epoch": 8.882288228822881,
-      "grad_norm": 0.0035400390625,
+      "grad_norm": 0.0027008056640625,
       "learning_rate": 6.02283397189628e-06,
       "loss": 0.0001,
       "step": 4037
     },
     {
       "epoch": 8.884488448844884,
-      "grad_norm": 0.003997802734375,
+      "grad_norm": 0.0029754638671875,
       "learning_rate": 5.9991522574944115e-06,
       "loss": 0.0001,
       "step": 4038
     },
     {
       "epoch": 8.886688668866887,
-      "grad_norm": 0.004791259765625,
+      "grad_norm": 0.0037841796875,
       "learning_rate": 5.975515753153671e-06,
       "loss": 0.0001,
       "step": 4039
     },
     {
       "epoch": 8.88888888888889,
-      "grad_norm": 0.0029449462890625,
+      "grad_norm": 0.0026092529296875,
       "learning_rate": 5.951924470242121e-06,
       "loss": 0.0001,
       "step": 4040
     },
     {
       "epoch": 8.891089108910892,
-      "grad_norm": 0.007720947265625,
+      "grad_norm": 0.0037384033203125,
       "learning_rate": 5.928378420106084e-06,
       "loss": 0.0001,
       "step": 4041
     },
     {
       "epoch": 8.893289328932893,
-      "grad_norm": 0.003204345703125,
+      "grad_norm": 0.0045166015625,
       "learning_rate": 5.904877614070114e-06,
       "loss": 0.0002,
       "step": 4042
     },
     {
       "epoch": 8.895489548954895,
-      "grad_norm": 0.003814697265625,
+      "grad_norm": 0.00640869140625,
       "learning_rate": 5.881422063437014e-06,
       "loss": 0.0001,
       "step": 4043
     },
     {
       "epoch": 8.897689768976898,
-      "grad_norm": 0.004425048828125,
+      "grad_norm": 0.004486083984375,
       "learning_rate": 5.858011779487827e-06,
       "loss": 0.0002,
       "step": 4044
     },
     {
       "epoch": 8.8998899889989,
-      "grad_norm": 0.00396728515625,
+      "grad_norm": 0.003204345703125,
       "learning_rate": 5.834646773481811e-06,
       "loss": 0.0001,
       "step": 4045
     },
     {
       "epoch": 8.902090209020901,
-      "grad_norm": 0.007232666015625,
+      "grad_norm": 0.003997802734375,
       "learning_rate": 5.811327056656457e-06,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 4046
     },
     {
       "epoch": 8.904290429042904,
-      "grad_norm": 0.0030670166015625,
+      "grad_norm": 0.00421142578125,
       "learning_rate": 5.7880526402274705e-06,
       "loss": 0.0001,
       "step": 4047
     },
     {
       "epoch": 8.906490649064907,
-      "grad_norm": 0.00469970703125,
+      "grad_norm": 0.00408935546875,
       "learning_rate": 5.764823535388763e-06,
       "loss": 0.0001,
       "step": 4048
     },
     {
       "epoch": 8.908690869086909,
-      "grad_norm": 0.0029754638671875,
+      "grad_norm": 0.0022125244140625,
       "learning_rate": 5.741639753312455e-06,
       "loss": 0.0001,
       "step": 4049
     },
     {
       "epoch": 8.910891089108912,
-      "grad_norm": 0.0025634765625,
+      "grad_norm": 0.002349853515625,
       "learning_rate": 5.7185013051488935e-06,
       "loss": 0.0001,
       "step": 4050
     },
     {
       "epoch": 8.913091309130913,
-      "grad_norm": 0.0027008056640625,
+      "grad_norm": 0.0029144287109375,
       "learning_rate": 5.695408202026586e-06,
       "loss": 0.0001,
       "step": 4051
     },
     {
       "epoch": 8.915291529152915,
-      "grad_norm": 0.0023956298828125,
+      "grad_norm": 0.0020904541015625,
       "learning_rate": 5.67236045505225e-06,
       "loss": 0.0001,
       "step": 4052
     },
     {
       "epoch": 8.917491749174918,
-      "grad_norm": 0.00250244140625,
+      "grad_norm": 0.0025634765625,
       "learning_rate": 5.649358075310784e-06,
       "loss": 0.0001,
       "step": 4053
     },
     {
       "epoch": 8.91969196919692,
-      "grad_norm": 0.0037384033203125,
+      "grad_norm": 0.00390625,
       "learning_rate": 5.626401073865284e-06,
       "loss": 0.0001,
       "step": 4054
     },
     {
       "epoch": 8.921892189218921,
-      "grad_norm": 0.003143310546875,
+      "grad_norm": 0.00830078125,
       "learning_rate": 5.603489461756984e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4055
     },
     {
       "epoch": 8.924092409240924,
-      "grad_norm": 0.0233154296875,
+      "grad_norm": 0.01434326171875,
       "learning_rate": 5.5806232500053344e-06,
-      "loss": 0.0005,
+      "loss": 0.0002,
       "step": 4056
     },
     {
       "epoch": 8.926292629262926,
-      "grad_norm": 0.003265380859375,
+      "grad_norm": 0.00341796875,
       "learning_rate": 5.557802449607907e-06,
       "loss": 0.0001,
       "step": 4057
     },
     {
       "epoch": 8.928492849284929,
-      "grad_norm": 0.002655029296875,
+      "grad_norm": 0.002532958984375,
       "learning_rate": 5.535027071540455e-06,
       "loss": 0.0001,
       "step": 4058
     },
     {
       "epoch": 8.930693069306932,
-      "grad_norm": 0.0032501220703125,
+      "grad_norm": 0.0078125,
       "learning_rate": 5.512297126756882e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4059
     },
     {
       "epoch": 8.932893289328932,
-      "grad_norm": 0.0037078857421875,
+      "grad_norm": 0.005401611328125,
       "learning_rate": 5.489612626189245e-06,
       "loss": 0.0001,
       "step": 4060
     },
     {
       "epoch": 8.935093509350935,
-      "grad_norm": 0.00469970703125,
+      "grad_norm": 0.002899169921875,
       "learning_rate": 5.466973580747736e-06,
       "loss": 0.0001,
       "step": 4061
     },
     {
       "epoch": 8.937293729372938,
-      "grad_norm": 0.002716064453125,
+      "grad_norm": 0.003509521484375,
       "learning_rate": 5.444380001320692e-06,
       "loss": 0.0001,
       "step": 4062
     },
     {
       "epoch": 8.93949394939494,
-      "grad_norm": 0.0034942626953125,
+      "grad_norm": 0.005615234375,
       "learning_rate": 5.421831898774577e-06,
       "loss": 0.0001,
       "step": 4063
     },
     {
       "epoch": 8.941694169416941,
-      "grad_norm": 0.0026092529296875,
+      "grad_norm": 0.0035552978515625,
       "learning_rate": 5.399329283953991e-06,
       "loss": 0.0001,
       "step": 4064
     },
     {
       "epoch": 8.943894389438944,
-      "grad_norm": 0.007049560546875,
+      "grad_norm": 0.0033721923828125,
       "learning_rate": 5.376872167681635e-06,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 4065
     },
     {
       "epoch": 8.946094609460946,
-      "grad_norm": 0.00323486328125,
+      "grad_norm": 0.00421142578125,
       "learning_rate": 5.3544605607583675e-06,
       "loss": 0.0001,
       "step": 4066
     },
     {
       "epoch": 8.948294829482949,
-      "grad_norm": 0.003448486328125,
+      "grad_norm": 0.0033416748046875,
       "learning_rate": 5.332094473963123e-06,
       "loss": 0.0001,
       "step": 4067
     },
     {
       "epoch": 8.950495049504951,
-      "grad_norm": 0.0064697265625,
+      "grad_norm": 0.0038604736328125,
       "learning_rate": 5.3097739180529514e-06,
       "loss": 0.0001,
       "step": 4068
     },
     {
       "epoch": 8.952695269526952,
-      "grad_norm": 0.005615234375,
+      "grad_norm": 0.0034027099609375,
       "learning_rate": 5.28749890376301e-06,
       "loss": 0.0001,
       "step": 4069
     },
     {
       "epoch": 8.954895489548955,
-      "grad_norm": 0.00213623046875,
+      "grad_norm": 0.0038299560546875,
       "learning_rate": 5.265269441806564e-06,
       "loss": 0.0001,
       "step": 4070
     },
     {
       "epoch": 8.957095709570957,
-      "grad_norm": 0.004425048828125,
+      "grad_norm": 0.00531005859375,
       "learning_rate": 5.243085542874937e-06,
       "loss": 0.0001,
       "step": 4071
     },
     {
       "epoch": 8.95929592959296,
-      "grad_norm": 0.0048828125,
+      "grad_norm": 0.003997802734375,
       "learning_rate": 5.220947217637562e-06,
       "loss": 0.0001,
       "step": 4072
     },
     {
       "epoch": 8.96149614961496,
-      "grad_norm": 0.003204345703125,
+      "grad_norm": 0.0025177001953125,
       "learning_rate": 5.198854476741965e-06,
       "loss": 0.0001,
       "step": 4073
     },
     {
       "epoch": 8.963696369636963,
-      "grad_norm": 0.005279541015625,
+      "grad_norm": 0.00836181640625,
       "learning_rate": 5.176807330813715e-06,
       "loss": 0.0002,
       "step": 4074
     },
     {
       "epoch": 8.965896589658966,
-      "grad_norm": 0.00262451171875,
+      "grad_norm": 0.005645751953125,
       "learning_rate": 5.154805790456485e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4075
     },
     {
       "epoch": 8.968096809680969,
-      "grad_norm": 0.0027618408203125,
+      "grad_norm": 0.003082275390625,
       "learning_rate": 5.132849866251977e-06,
       "loss": 0.0001,
       "step": 4076
     },
     {
       "epoch": 8.97029702970297,
-      "grad_norm": 0.004058837890625,
+      "grad_norm": 0.006072998046875,
       "learning_rate": 5.110939568759998e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4077
     },
     {
       "epoch": 8.972497249724972,
-      "grad_norm": 0.003173828125,
+      "grad_norm": 0.00341796875,
       "learning_rate": 5.089074908518388e-06,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 4078
     },
     {
       "epoch": 8.974697469746975,
-      "grad_norm": 0.003387451171875,
+      "grad_norm": 0.0025634765625,
       "learning_rate": 5.0672558960430105e-06,
       "loss": 0.0001,
       "step": 4079
     },
     {
       "epoch": 8.976897689768977,
-      "grad_norm": 0.00286865234375,
+      "grad_norm": 0.004547119140625,
       "learning_rate": 5.0454825418278285e-06,
       "loss": 0.0001,
       "step": 4080
     },
     {
       "epoch": 8.97909790979098,
-      "grad_norm": 0.002471923828125,
+      "grad_norm": 0.0062255859375,
       "learning_rate": 5.023754856344809e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4081
     },
     {
       "epoch": 8.98129812981298,
-      "grad_norm": 0.002349853515625,
+      "grad_norm": 0.004119873046875,
       "learning_rate": 5.002072850043971e-06,
       "loss": 0.0001,
       "step": 4082
     },
     {
       "epoch": 8.983498349834983,
-      "grad_norm": 0.012451171875,
+      "grad_norm": 0.006805419921875,
       "learning_rate": 4.980436533353339e-06,
       "loss": 0.0002,
       "step": 4083
     },
     {
       "epoch": 8.985698569856986,
-      "grad_norm": 0.004150390625,
+      "grad_norm": 0.01141357421875,
       "learning_rate": 4.958845916679e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4084
     },
     {
       "epoch": 8.987898789878988,
-      "grad_norm": 0.0027008056640625,
+      "grad_norm": 0.0024566650390625,
       "learning_rate": 4.937301010405038e-06,
       "loss": 0.0001,
       "step": 4085
     },
     {
       "epoch": 8.990099009900991,
-      "grad_norm": 0.0028228759765625,
+      "grad_norm": 0.0024261474609375,
       "learning_rate": 4.91580182489354e-06,
       "loss": 0.0001,
       "step": 4086
     },
     {
       "epoch": 8.992299229922992,
-      "grad_norm": 0.0023345947265625,
+      "grad_norm": 0.002716064453125,
       "learning_rate": 4.8943483704846475e-06,
       "loss": 0.0001,
       "step": 4087
     },
     {
       "epoch": 8.994499449944994,
-      "grad_norm": 0.00335693359375,
+      "grad_norm": 0.003662109375,
       "learning_rate": 4.872940657496461e-06,
       "loss": 0.0001,
       "step": 4088
     },
     {
       "epoch": 8.996699669966997,
-      "grad_norm": 0.00299072265625,
+      "grad_norm": 0.0029449462890625,
       "learning_rate": 4.851578696225101e-06,
       "loss": 0.0001,
       "step": 4089
     },
     {
       "epoch": 8.998899889989,
-      "grad_norm": 0.0024261474609375,
+      "grad_norm": 0.0032958984375,
       "learning_rate": 4.830262496944693e-06,
       "loss": 0.0001,
       "step": 4090
     },
     {
       "epoch": 9.001100110011,
-      "grad_norm": 0.00390625,
+      "grad_norm": 0.0032958984375,
       "learning_rate": 4.8089920699073366e-06,
       "loss": 0.0001,
       "step": 4091
     },
     {
       "epoch": 9.003300330033003,
-      "grad_norm": 0.002960205078125,
+      "grad_norm": 0.00811767578125,
       "learning_rate": 4.787767425343126e-06,
       "loss": 0.0001,
       "step": 4092
     },
     {
       "epoch": 9.005500550055006,
-      "grad_norm": 0.002471923828125,
+      "grad_norm": 0.00250244140625,
       "learning_rate": 4.766588573460129e-06,
       "loss": 0.0001,
       "step": 4093
     },
     {
       "epoch": 9.007700770077008,
-      "grad_norm": 0.002410888671875,
+      "grad_norm": 0.00634765625,
       "learning_rate": 4.745455524444409e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4094
     },
     {
       "epoch": 9.009900990099009,
-      "grad_norm": 0.007293701171875,
+      "grad_norm": 0.00244140625,
       "learning_rate": 4.724368288459957e-06,
       "loss": 0.0001,
       "step": 4095
     },
     {
       "epoch": 9.012101210121012,
-      "grad_norm": 0.0022430419921875,
+      "grad_norm": 0.002288818359375,
       "learning_rate": 4.703326875648795e-06,
       "loss": 0.0001,
       "step": 4096
     },
     {
       "epoch": 9.014301430143014,
-      "grad_norm": 0.0020751953125,
+      "grad_norm": 0.0034942626953125,
       "learning_rate": 4.682331296130837e-06,
       "loss": 0.0001,
       "step": 4097
     },
     {
       "epoch": 9.016501650165017,
-      "grad_norm": 0.0023956298828125,
+      "grad_norm": 0.0034637451171875,
       "learning_rate": 4.661381560003997e-06,
       "loss": 0.0001,
       "step": 4098
     },
     {
       "epoch": 9.01870187018702,
-      "grad_norm": 0.0030364990234375,
+      "grad_norm": 0.014404296875,
       "learning_rate": 4.640477677344135e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4099
     },
     {
       "epoch": 9.02090209020902,
-      "grad_norm": 0.0026397705078125,
+      "grad_norm": 0.005096435546875,
       "learning_rate": 4.6196196582050545e-06,
       "loss": 0.0001,
       "step": 4100
     },
     {
       "epoch": 9.023102310231023,
-      "grad_norm": 0.007415771484375,
+      "grad_norm": 0.0032196044921875,
       "learning_rate": 4.598807512618486e-06,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 4101
     },
     {
       "epoch": 9.025302530253025,
-      "grad_norm": 0.00738525390625,
+      "grad_norm": 0.0054931640625,
       "learning_rate": 4.578041250594112e-06,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 4102
     },
     {
       "epoch": 9.027502750275028,
-      "grad_norm": 0.0047607421875,
+      "grad_norm": 0.004547119140625,
       "learning_rate": 4.557320882119553e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4103
     },
     {
       "epoch": 9.029702970297029,
-      "grad_norm": 0.0037078857421875,
+      "grad_norm": 0.0062255859375,
       "learning_rate": 4.536646417160351e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4104
     },
     {
       "epoch": 9.029702970297029,
-      "eval_loss": 0.0002391614980297163,
-      "eval_runtime": 10.5173,
-      "eval_samples_per_second": 36.416,
-      "eval_steps_per_second": 4.564,
+      "eval_loss": 0.00034816397237591445,
+      "eval_runtime": 10.1906,
+      "eval_samples_per_second": 37.584,
+      "eval_steps_per_second": 4.71,
       "step": 4104
     },
     {
       "epoch": 9.031903190319031,
-      "grad_norm": 0.0030517578125,
+      "grad_norm": 0.0019073486328125,
       "learning_rate": 4.516017865659949e-06,
       "loss": 0.0001,
       "step": 4105
     },
     {
       "epoch": 9.034103410341034,
-      "grad_norm": 0.00262451171875,
+      "grad_norm": 0.0072021484375,
       "learning_rate": 4.495435237539735e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4106
     },
     {
       "epoch": 9.036303630363037,
-      "grad_norm": 0.00335693359375,
+      "grad_norm": 0.00244140625,
       "learning_rate": 4.474898542699002e-06,
       "loss": 0.0001,
       "step": 4107
     },
     {
       "epoch": 9.03850385038504,
-      "grad_norm": 0.0030517578125,
+      "grad_norm": 0.0029754638671875,
       "learning_rate": 4.454407791014959e-06,
       "loss": 0.0001,
       "step": 4108
     },
     {
       "epoch": 9.04070407040704,
-      "grad_norm": 0.005615234375,
+      "grad_norm": 0.00347900390625,
       "learning_rate": 4.433962992342699e-06,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 4109
     },
     {
       "epoch": 9.042904290429043,
-      "grad_norm": 0.0021820068359375,
+      "grad_norm": 0.0028839111328125,
       "learning_rate": 4.413564156515226e-06,
       "loss": 0.0001,
       "step": 4110
     },
     {
       "epoch": 9.045104510451045,
-      "grad_norm": 0.0021820068359375,
+      "grad_norm": 0.0027618408203125,
       "learning_rate": 4.393211293343446e-06,
       "loss": 0.0001,
       "step": 4111
     },
     {
       "epoch": 9.047304730473048,
-      "grad_norm": 0.0023345947265625,
+      "grad_norm": 0.0022125244140625,
       "learning_rate": 4.3729044126161415e-06,
       "loss": 0.0001,
       "step": 4112
     },
     {
       "epoch": 9.049504950495049,
-      "grad_norm": 0.004180908203125,
+      "grad_norm": 0.0025787353515625,
       "learning_rate": 4.352643524099975e-06,
       "loss": 0.0001,
       "step": 4113
     },
     {
       "epoch": 9.051705170517051,
-      "grad_norm": 0.005218505859375,
+      "grad_norm": 0.003936767578125,
       "learning_rate": 4.332428637539499e-06,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 4114
     },
     {
       "epoch": 9.053905390539054,
-      "grad_norm": 0.00390625,
+      "grad_norm": 0.0027313232421875,
       "learning_rate": 4.312259762657145e-06,
       "loss": 0.0001,
       "step": 4115
     },
     {
       "epoch": 9.056105610561056,
-      "grad_norm": 0.002593994140625,
+      "grad_norm": 0.0031890869140625,
       "learning_rate": 4.292136909153199e-06,
       "loss": 0.0001,
       "step": 4116
     },
     {
       "epoch": 9.058305830583059,
-      "grad_norm": 0.0037689208984375,
+      "grad_norm": 0.0034027099609375,
       "learning_rate": 4.272060086705842e-06,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 4117
     },
     {
       "epoch": 9.06050605060506,
-      "grad_norm": 0.0017852783203125,
+      "grad_norm": 0.0029296875,
       "learning_rate": 4.252029304971083e-06,
       "loss": 0.0001,
       "step": 4118
     },
     {
       "epoch": 9.062706270627062,
-      "grad_norm": 0.002288818359375,
+      "grad_norm": 0.00640869140625,
       "learning_rate": 4.232044573582794e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4119
     },
     {
       "epoch": 9.064906490649065,
-      "grad_norm": 0.0026092529296875,
+      "grad_norm": 0.0035247802734375,
       "learning_rate": 4.21210590215273e-06,
       "loss": 0.0001,
       "step": 4120
     },
     {
       "epoch": 9.067106710671068,
-      "grad_norm": 0.0021209716796875,
+      "grad_norm": 0.0027923583984375,
       "learning_rate": 4.192213300270453e-06,
       "loss": 0.0001,
       "step": 4121
     },
     {
       "epoch": 9.069306930693068,
-      "grad_norm": 0.00238037109375,
+      "grad_norm": 0.0025787353515625,
       "learning_rate": 4.172366777503389e-06,
       "loss": 0.0001,
       "step": 4122
     },
     {
       "epoch": 9.071507150715071,
-      "grad_norm": 0.001983642578125,
+      "grad_norm": 0.0022430419921875,
       "learning_rate": 4.152566343396791e-06,
       "loss": 0.0001,
       "step": 4123
     },
     {
       "epoch": 9.073707370737074,
-      "grad_norm": 0.004425048828125,
+      "grad_norm": 0.003936767578125,
       "learning_rate": 4.132812007473774e-06,
       "loss": 0.0002,
       "step": 4124
     },
     {
       "epoch": 9.075907590759076,
-      "grad_norm": 0.003021240234375,
+      "grad_norm": 0.0027313232421875,
       "learning_rate": 4.113103779235228e-06,
       "loss": 0.0001,
       "step": 4125
     },
     {
       "epoch": 9.078107810781079,
-      "grad_norm": 0.0027008056640625,
+      "grad_norm": 0.006500244140625,
       "learning_rate": 4.093441668159925e-06,
       "loss": 0.0001,
       "step": 4126
     },
     {
       "epoch": 9.08030803080308,
-      "grad_norm": 0.0026092529296875,
+      "grad_norm": 0.0027923583984375,
       "learning_rate": 4.073825683704424e-06,
       "loss": 0.0001,
       "step": 4127
     },
     {
       "epoch": 9.082508250825082,
-      "grad_norm": 0.00201416015625,
+      "grad_norm": 0.004180908203125,
       "learning_rate": 4.0542558353031e-06,
       "loss": 0.0001,
       "step": 4128
     },
     {
       "epoch": 9.084708470847085,
-      "grad_norm": 0.007720947265625,
+      "grad_norm": 0.004486083984375,
       "learning_rate": 4.0347321323681576e-06,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 4129
     },
     {
       "epoch": 9.086908690869087,
-      "grad_norm": 0.0027313232421875,
+      "grad_norm": 0.0027008056640625,
       "learning_rate": 4.015254584289585e-06,
       "loss": 0.0001,
       "step": 4130
     },
     {
       "epoch": 9.089108910891088,
-      "grad_norm": 0.002288818359375,
+      "grad_norm": 0.0023345947265625,
       "learning_rate": 3.995823200435178e-06,
       "loss": 0.0001,
       "step": 4131
     },
     {
       "epoch": 9.091309130913091,
-      "grad_norm": 0.002838134765625,
+      "grad_norm": 0.006011962890625,
       "learning_rate": 3.97643799015055e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4132
     },
     {
       "epoch": 9.093509350935093,
-      "grad_norm": 0.002685546875,
+      "grad_norm": 0.00262451171875,
       "learning_rate": 3.957098962759076e-06,
       "loss": 0.0001,
       "step": 4133
     },
     {
       "epoch": 9.095709570957096,
-      "grad_norm": 0.0023040771484375,
+      "grad_norm": 0.00250244140625,
       "learning_rate": 3.937806127561949e-06,
       "loss": 0.0001,
       "step": 4134
     },
     {
       "epoch": 9.097909790979099,
-      "grad_norm": 0.006011962890625,
+      "grad_norm": 0.0026702880859375,
       "learning_rate": 3.918559493838114e-06,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 4135
     },
     {
       "epoch": 9.1001100110011,
-      "grad_norm": 0.0037689208984375,
+      "grad_norm": 0.00335693359375,
       "learning_rate": 3.899359070844311e-06,
       "loss": 0.0001,
       "step": 4136
     },
     {
       "epoch": 9.102310231023102,
-      "grad_norm": 0.0037078857421875,
+      "grad_norm": 0.003082275390625,
       "learning_rate": 3.880204867815074e-06,
       "loss": 0.0001,
       "step": 4137
     },
     {
       "epoch": 9.104510451045105,
-      "grad_norm": 0.0032958984375,
+      "grad_norm": 0.0035552978515625,
       "learning_rate": 3.861096893962679e-06,
       "loss": 0.0001,
       "step": 4138
     },
     {
       "epoch": 9.106710671067107,
-      "grad_norm": 0.002655029296875,
+      "grad_norm": 0.003753662109375,
       "learning_rate": 3.842035158477175e-06,
       "loss": 0.0001,
       "step": 4139
     },
     {
       "epoch": 9.108910891089108,
-      "grad_norm": 0.003570556640625,
+      "grad_norm": 0.004791259765625,
       "learning_rate": 3.8230196705263734e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4140
     },
     {
       "epoch": 9.11111111111111,
-      "grad_norm": 0.0028839111328125,
+      "grad_norm": 0.0033721923828125,
       "learning_rate": 3.8040504392558463e-06,
       "loss": 0.0001,
       "step": 4141
     },
     {
       "epoch": 9.113311331133113,
-      "grad_norm": 0.0032196044921875,
+      "grad_norm": 0.0042724609375,
       "learning_rate": 3.78512747378893e-06,
       "loss": 0.0001,
       "step": 4142
     },
     {
       "epoch": 9.115511551155116,
-      "grad_norm": 0.00543212890625,
+      "grad_norm": 0.00433349609375,
       "learning_rate": 3.7662507832266767e-06,
       "loss": 0.0001,
       "step": 4143
     },
     {
       "epoch": 9.117711771177119,
-      "grad_norm": 0.00347900390625,
+      "grad_norm": 0.0033111572265625,
       "learning_rate": 3.7474203766479143e-06,
       "loss": 0.0001,
       "step": 4144
     },
     {
       "epoch": 9.11991199119912,
-      "grad_norm": 0.005096435546875,
+      "grad_norm": 0.0029754638671875,
       "learning_rate": 3.7286362631091975e-06,
       "loss": 0.0001,
       "step": 4145
     },
     {
       "epoch": 9.122112211221122,
-      "grad_norm": 0.0035552978515625,
+      "grad_norm": 0.002227783203125,
       "learning_rate": 3.709898451644811e-06,
       "loss": 0.0001,
       "step": 4146
     },
     {
       "epoch": 9.124312431243125,
-      "grad_norm": 0.00238037109375,
+      "grad_norm": 0.004058837890625,
       "learning_rate": 3.6912069512667678e-06,
       "loss": 0.0001,
       "step": 4147
     },
     {
       "epoch": 9.126512651265127,
-      "grad_norm": 0.0036163330078125,
+      "grad_norm": 0.005340576171875,
       "learning_rate": 3.672561770964833e-06,
       "loss": 0.0001,
       "step": 4148
     },
     {
       "epoch": 9.128712871287128,
-      "grad_norm": 0.002838134765625,
+      "grad_norm": 0.0029144287109375,
       "learning_rate": 3.653962919706466e-06,
       "loss": 0.0001,
       "step": 4149
     },
     {
       "epoch": 9.13091309130913,
-      "grad_norm": 0.002838134765625,
+      "grad_norm": 0.005706787109375,
       "learning_rate": 3.6354104064368566e-06,
       "loss": 0.0001,
       "step": 4150
     },
     {
       "epoch": 9.133113311331133,
-      "grad_norm": 0.00323486328125,
+      "grad_norm": 0.0030517578125,
       "learning_rate": 3.6169042400789e-06,
       "loss": 0.0001,
       "step": 4151
     },
     {
       "epoch": 9.135313531353136,
-      "grad_norm": 0.0031280517578125,
+      "grad_norm": 0.0032196044921875,
       "learning_rate": 3.598444429533221e-06,
       "loss": 0.0001,
       "step": 4152
     },
     {
       "epoch": 9.137513751375138,
-      "grad_norm": 0.002471923828125,
+      "grad_norm": 0.006927490234375,
       "learning_rate": 3.5800309836781175e-06,
       "loss": 0.0001,
       "step": 4153
     },
     {
       "epoch": 9.13971397139714,
-      "grad_norm": 0.001708984375,
+      "grad_norm": 0.005035400390625,
       "learning_rate": 3.561663911369617e-06,
       "loss": 0.0001,
       "step": 4154
     },
     {
       "epoch": 9.141914191419142,
-      "grad_norm": 0.006439208984375,
+      "grad_norm": 0.00323486328125,
       "learning_rate": 3.543343221441431e-06,
       "loss": 0.0001,
       "step": 4155
     },
     {
       "epoch": 9.144114411441144,
-      "grad_norm": 0.00433349609375,
+      "grad_norm": 0.0048828125,
       "learning_rate": 3.525068922704966e-06,
       "loss": 0.0001,
       "step": 4156
     },
     {
       "epoch": 9.146314631463147,
-      "grad_norm": 0.00390625,
+      "grad_norm": 0.0054931640625,
       "learning_rate": 3.506841023949292e-06,
       "loss": 0.0001,
       "step": 4157
     },
     {
       "epoch": 9.148514851485148,
-      "grad_norm": 0.003570556640625,
+      "grad_norm": 0.004180908203125,
       "learning_rate": 3.4886595339412186e-06,
       "loss": 0.0001,
       "step": 4158
     },
     {
       "epoch": 9.15071507150715,
-      "grad_norm": 0.0026397705078125,
+      "grad_norm": 0.0028076171875,
       "learning_rate": 3.470524461425184e-06,
       "loss": 0.0001,
       "step": 4159
     },
     {
       "epoch": 9.152915291529153,
-      "grad_norm": 0.002532958984375,
+      "grad_norm": 0.00390625,
       "learning_rate": 3.4524358151233227e-06,
       "loss": 0.0001,
       "step": 4160
     },
     {
       "epoch": 9.155115511551156,
-      "grad_norm": 0.002288818359375,
+      "grad_norm": 0.0020294189453125,
       "learning_rate": 3.4343936037354527e-06,
       "loss": 0.0001,
       "step": 4161
     },
     {
       "epoch": 9.157315731573158,
-      "grad_norm": 0.00482177734375,
+      "grad_norm": 0.004730224609375,
       "learning_rate": 3.4163978359390003e-06,
       "loss": 0.0002,
       "step": 4162
     },
     {
       "epoch": 9.159515951595159,
-      "grad_norm": 0.0026397705078125,
+      "grad_norm": 0.00299072265625,
       "learning_rate": 3.3984485203891415e-06,
       "loss": 0.0001,
       "step": 4163
     },
     {
       "epoch": 9.161716171617162,
-      "grad_norm": 0.0042724609375,
+      "grad_norm": 0.01129150390625,
       "learning_rate": 3.3805456657186376e-06,
       "loss": 0.0002,
       "step": 4164
     },
     {
       "epoch": 9.163916391639164,
-      "grad_norm": 0.0022125244140625,
+      "grad_norm": 0.0081787109375,
       "learning_rate": 3.3626892805379562e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4165
     },
     {
       "epoch": 9.166116611661167,
-      "grad_norm": 0.00238037109375,
+      "grad_norm": 0.0035400390625,
       "learning_rate": 3.3448793734351835e-06,
       "loss": 0.0001,
       "step": 4166
     },
     {
       "epoch": 9.168316831683168,
-      "grad_norm": 0.002716064453125,
+      "grad_norm": 0.004791259765625,
       "learning_rate": 3.3271159529760675e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4167
     },
     {
       "epoch": 9.17051705170517,
-      "grad_norm": 0.0035552978515625,
+      "grad_norm": 0.00396728515625,
       "learning_rate": 3.309399027703974e-06,
       "loss": 0.0001,
       "step": 4168
     },
     {
       "epoch": 9.172717271727173,
-      "grad_norm": 0.0037078857421875,
+      "grad_norm": 0.005035400390625,
       "learning_rate": 3.2917286061399434e-06,
       "loss": 0.0002,
       "step": 4169
     },
     {
       "epoch": 9.174917491749175,
-      "grad_norm": 0.004791259765625,
+      "grad_norm": 0.00396728515625,
       "learning_rate": 3.2741046967826205e-06,
       "loss": 0.0001,
       "step": 4170
     },
     {
       "epoch": 9.177117711771178,
-      "grad_norm": 0.0024566650390625,
+      "grad_norm": 0.0030670166015625,
       "learning_rate": 3.2565273081083038e-06,
       "loss": 0.0001,
       "step": 4171
     },
     {
       "epoch": 9.179317931793179,
-      "grad_norm": 0.003448486328125,
+      "grad_norm": 0.006439208984375,
       "learning_rate": 3.238996448570886e-06,
       "loss": 0.0001,
       "step": 4172
     },
     {
       "epoch": 9.181518151815181,
-      "grad_norm": 0.0030364990234375,
+      "grad_norm": 0.00323486328125,
       "learning_rate": 3.221512126601911e-06,
       "loss": 0.0001,
       "step": 4173
     },
     {
       "epoch": 9.183718371837184,
-      "grad_norm": 0.00921630859375,
+      "grad_norm": 0.008544921875,
       "learning_rate": 3.204074350610531e-06,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 4174
     },
     {
       "epoch": 9.185918591859187,
-      "grad_norm": 0.0025634765625,
+      "grad_norm": 0.0031280517578125,
       "learning_rate": 3.186683128983503e-06,
       "loss": 0.0001,
       "step": 4175
     },
     {
       "epoch": 9.188118811881187,
-      "grad_norm": 0.0025177001953125,
+      "grad_norm": 0.004730224609375,
       "learning_rate": 3.1693384700852148e-06,
       "loss": 0.0001,
       "step": 4176
     },
     {
       "epoch": 9.19031903190319,
-      "grad_norm": 0.0027618408203125,
+      "grad_norm": 0.0030517578125,
       "learning_rate": 3.152040382257626e-06,
       "loss": 0.0001,
       "step": 4177
     },
     {
       "epoch": 9.192519251925193,
-      "grad_norm": 0.0026397705078125,
+      "grad_norm": 0.00201416015625,
       "learning_rate": 3.134788873820338e-06,
       "loss": 0.0001,
       "step": 4178
     },
     {
       "epoch": 9.194719471947195,
-      "grad_norm": 0.002349853515625,
+      "grad_norm": 0.0024261474609375,
       "learning_rate": 3.117583953070491e-06,
       "loss": 0.0001,
       "step": 4179
     },
     {
       "epoch": 9.196919691969198,
-      "grad_norm": 0.00360107421875,
+      "grad_norm": 0.004547119140625,
       "learning_rate": 3.100425628282899e-06,
       "loss": 0.0002,
       "step": 4180
     },
     {
       "epoch": 9.199119911991199,
-      "grad_norm": 0.00445556640625,
+      "grad_norm": 0.0025787353515625,
       "learning_rate": 3.083313907709895e-06,
       "loss": 0.0001,
       "step": 4181
     },
     {
       "epoch": 9.201320132013201,
-      "grad_norm": 0.0024261474609375,
+      "grad_norm": 0.005126953125,
       "learning_rate": 3.0662487995814505e-06,
       "loss": 0.0001,
       "step": 4182
     },
     {
       "epoch": 9.203520352035204,
-      "grad_norm": 0.00469970703125,
+      "grad_norm": 0.00262451171875,
       "learning_rate": 3.049230312105078e-06,
       "loss": 0.0001,
       "step": 4183
     },
     {
       "epoch": 9.205720572057206,
-      "grad_norm": 0.002105712890625,
+      "grad_norm": 0.00323486328125,
       "learning_rate": 3.0322584534658636e-06,
       "loss": 0.0001,
       "step": 4184
     },
     {
       "epoch": 9.207920792079207,
-      "grad_norm": 0.004150390625,
+      "grad_norm": 0.0037384033203125,
       "learning_rate": 3.0153332318265226e-06,
       "loss": 0.0002,
       "step": 4185
     },
     {
       "epoch": 9.21012101210121,
-      "grad_norm": 0.004486083984375,
+      "grad_norm": 0.005859375,
       "learning_rate": 2.9984546553272873e-06,
       "loss": 0.0001,
       "step": 4186
     },
     {
       "epoch": 9.212321232123212,
-      "grad_norm": 0.00311279296875,
+      "grad_norm": 0.0023651123046875,
       "learning_rate": 2.981622732085965e-06,
       "loss": 0.0001,
       "step": 4187
     },
     {
       "epoch": 9.214521452145215,
-      "grad_norm": 0.0022735595703125,
+      "grad_norm": 0.0024871826171875,
       "learning_rate": 2.9648374701979564e-06,
       "loss": 0.0001,
       "step": 4188
     },
     {
       "epoch": 9.216721672167218,
-      "grad_norm": 0.004241943359375,
+      "grad_norm": 0.00341796875,
       "learning_rate": 2.948098877736161e-06,
       "loss": 0.0001,
       "step": 4189
     },
     {
       "epoch": 9.218921892189218,
-      "grad_norm": 0.0034332275390625,
+      "grad_norm": 0.00543212890625,
       "learning_rate": 2.9314069627511044e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4190
     },
     {
       "epoch": 9.221122112211221,
-      "grad_norm": 0.002777099609375,
+      "grad_norm": 0.005859375,
       "learning_rate": 2.914761733270799e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4191
     },
     {
       "epoch": 9.223322332233224,
-      "grad_norm": 0.00933837890625,
+      "grad_norm": 0.002960205078125,
       "learning_rate": 2.8981631973008405e-06,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 4192
     },
     {
       "epoch": 9.225522552255226,
-      "grad_norm": 0.0029144287109375,
+      "grad_norm": 0.0030975341796875,
       "learning_rate": 2.8816113628243545e-06,
       "loss": 0.0001,
       "step": 4193
     },
     {
       "epoch": 9.227722772277227,
-      "grad_norm": 0.0181884765625,
+      "grad_norm": 0.0040283203125,
       "learning_rate": 2.8651062378020178e-06,
-      "loss": 0.0004,
+      "loss": 0.0002,
       "step": 4194
     },
     {
       "epoch": 9.22992299229923,
-      "grad_norm": 0.00921630859375,
+      "grad_norm": 0.0037689208984375,
       "learning_rate": 2.848647830172024e-06,
       "loss": 0.0002,
       "step": 4195
     },
     {
       "epoch": 9.232123212321232,
-      "grad_norm": 0.0030670166015625,
+      "grad_norm": 0.00384521484375,
       "learning_rate": 2.832236147850109e-06,
       "loss": 0.0001,
       "step": 4196
     },
     {
       "epoch": 9.234323432343235,
-      "grad_norm": 0.00408935546875,
+      "grad_norm": 0.0037841796875,
       "learning_rate": 2.815871198729525e-06,
       "loss": 0.0001,
       "step": 4197
     },
     {
       "epoch": 9.236523652365236,
-      "grad_norm": 0.0023193359375,
+      "grad_norm": 0.003448486328125,
       "learning_rate": 2.7995529906810647e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4198
     },
     {
       "epoch": 9.238723872387238,
-      "grad_norm": 0.0028076171875,
+      "grad_norm": 0.0052490234375,
       "learning_rate": 2.783281531553028e-06,
       "loss": 0.0001,
       "step": 4199
     },
     {
       "epoch": 9.24092409240924,
-      "grad_norm": 0.0037078857421875,
+      "grad_norm": 0.00396728515625,
       "learning_rate": 2.767056829171255e-06,
       "loss": 0.0001,
       "step": 4200
     },
     {
       "epoch": 9.243124312431243,
-      "grad_norm": 0.004486083984375,
+      "grad_norm": 0.004364013671875,
       "learning_rate": 2.750878891339026e-06,
       "loss": 0.0001,
       "step": 4201
     },
     {
       "epoch": 9.245324532453246,
-      "grad_norm": 0.0027313232421875,
+      "grad_norm": 0.005828857421875,
       "learning_rate": 2.7347477258372278e-06,
       "loss": 0.0001,
       "step": 4202
     },
     {
       "epoch": 9.247524752475247,
-      "grad_norm": 0.0030517578125,
+      "grad_norm": 0.01055908203125,
       "learning_rate": 2.718663340424188e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4203
     },
     {
       "epoch": 9.24972497249725,
-      "grad_norm": 0.006622314453125,
+      "grad_norm": 0.00433349609375,
       "learning_rate": 2.7026257428357515e-06,
       "loss": 0.0002,
       "step": 4204
     },
     {
       "epoch": 9.251925192519252,
-      "grad_norm": 0.00335693359375,
+      "grad_norm": 0.00433349609375,
       "learning_rate": 2.686634940785282e-06,
       "loss": 0.0001,
       "step": 4205
     },
     {
       "epoch": 9.254125412541255,
-      "grad_norm": 0.005157470703125,
+      "grad_norm": 0.0098876953125,
       "learning_rate": 2.6706909419635717e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4206
     },
     {
       "epoch": 9.256325632563255,
-      "grad_norm": 0.0024871826171875,
+      "grad_norm": 0.004425048828125,
       "learning_rate": 2.6547937540389758e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4207
     },
     {
       "epoch": 9.258525852585258,
-      "grad_norm": 0.01116943359375,
+      "grad_norm": 0.0036163330078125,
       "learning_rate": 2.638943384657322e-06,
       "loss": 0.0001,
       "step": 4208
     },
     {
       "epoch": 9.26072607260726,
-      "grad_norm": 0.004486083984375,
+      "grad_norm": 0.005584716796875,
       "learning_rate": 2.623139841441868e-06,
       "loss": 0.0002,
       "step": 4209
     },
     {
       "epoch": 9.262926292629263,
-      "grad_norm": 0.0223388671875,
+      "grad_norm": 0.0078125,
       "learning_rate": 2.607383131993424e-06,
-      "loss": 0.0005,
+      "loss": 0.0002,
       "step": 4210
     },
     {
       "epoch": 9.265126512651266,
-      "grad_norm": 0.00250244140625,
+      "grad_norm": 0.00830078125,
       "learning_rate": 2.5916732638902043e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4211
     },
     {
       "epoch": 9.267326732673267,
-      "grad_norm": 0.00439453125,
+      "grad_norm": 0.00567626953125,
       "learning_rate": 2.576010244687954e-06,
       "loss": 0.0001,
       "step": 4212
     },
     {
       "epoch": 9.26952695269527,
-      "grad_norm": 0.0027313232421875,
+      "grad_norm": 0.0107421875,
       "learning_rate": 2.560394081919848e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4213
     },
     {
       "epoch": 9.271727172717272,
-      "grad_norm": 0.006134033203125,
+      "grad_norm": 0.007171630859375,
       "learning_rate": 2.5448247830965444e-06,
       "loss": 0.0002,
       "step": 4214
     },
     {
       "epoch": 9.273927392739274,
-      "grad_norm": 0.0032806396484375,
+      "grad_norm": 0.00714111328125,
       "learning_rate": 2.5293023557061644e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4215
     },
     {
       "epoch": 9.276127612761275,
-      "grad_norm": 0.0033111572265625,
+      "grad_norm": 0.00482177734375,
       "learning_rate": 2.513826807214248e-06,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 4216
     },
     {
       "epoch": 9.278327832783278,
-      "grad_norm": 0.003326416015625,
+      "grad_norm": 0.004364013671875,
       "learning_rate": 2.498398145063863e-06,
       "loss": 0.0001,
       "step": 4217
     },
     {
       "epoch": 9.28052805280528,
-      "grad_norm": 0.0027008056640625,
+      "grad_norm": 0.0023040771484375,
       "learning_rate": 2.48301637667544e-06,
       "loss": 0.0001,
       "step": 4218
     },
     {
       "epoch": 9.28052805280528,
-      "eval_loss": 0.00024145781935658306,
-      "eval_runtime": 10.3787,
-      "eval_samples_per_second": 36.903,
-      "eval_steps_per_second": 4.625,
+      "eval_loss": 0.0003449491341598332,
+      "eval_runtime": 10.3104,
+      "eval_samples_per_second": 37.147,
+      "eval_steps_per_second": 4.655,
       "step": 4218
     },
     {
       "epoch": 9.282728272827283,
-      "grad_norm": 0.0040283203125,
+      "grad_norm": 0.003326416015625,
       "learning_rate": 2.4676815094469173e-06,
       "loss": 0.0001,
       "step": 4219
     },
     {
       "epoch": 9.284928492849286,
-      "grad_norm": 0.004302978515625,
+      "grad_norm": 0.004058837890625,
       "learning_rate": 2.452393550753662e-06,
       "loss": 0.0002,
       "step": 4220
     },
     {
       "epoch": 9.287128712871286,
-      "grad_norm": 0.0032501220703125,
+      "grad_norm": 0.00494384765625,
       "learning_rate": 2.437152507948459e-06,
       "loss": 0.0001,
       "step": 4221
     },
     {
       "epoch": 9.289328932893289,
-      "grad_norm": 0.00311279296875,
+      "grad_norm": 0.0035858154296875,
       "learning_rate": 2.421958388361578e-06,
       "loss": 0.0001,
       "step": 4222
     },
     {
       "epoch": 9.291529152915292,
-      "grad_norm": 0.0037078857421875,
+      "grad_norm": 0.005584716796875,
       "learning_rate": 2.4068111993006404e-06,
       "loss": 0.0002,
       "step": 4223
     },
     {
       "epoch": 9.293729372937294,
-      "grad_norm": 0.002532958984375,
+      "grad_norm": 0.00347900390625,
       "learning_rate": 2.391710948050774e-06,
       "loss": 0.0001,
       "step": 4224
     },
     {
       "epoch": 9.295929592959295,
-      "grad_norm": 0.004058837890625,
+      "grad_norm": 0.004119873046875,
       "learning_rate": 2.3766576418745022e-06,
       "loss": 0.0001,
       "step": 4225
     },
     {
       "epoch": 9.298129812981298,
-      "grad_norm": 0.002777099609375,
+      "grad_norm": 0.00341796875,
       "learning_rate": 2.361651288011757e-06,
       "loss": 0.0001,
       "step": 4226
     },
     {
       "epoch": 9.3003300330033,
-      "grad_norm": 0.00714111328125,
+      "grad_norm": 0.0030975341796875,
       "learning_rate": 2.3466918936799197e-06,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 4227
     },
     {
       "epoch": 9.302530253025303,
-      "grad_norm": 0.00506591796875,
+      "grad_norm": 0.007049560546875,
       "learning_rate": 2.3317794660737356e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4228
     },
     {
       "epoch": 9.304730473047305,
-      "grad_norm": 0.00311279296875,
+      "grad_norm": 0.0037384033203125,
       "learning_rate": 2.3169140123654234e-06,
       "loss": 0.0001,
       "step": 4229
     },
     {
       "epoch": 9.306930693069306,
-      "grad_norm": 0.00390625,
+      "grad_norm": 0.01422119140625,
       "learning_rate": 2.302095539704563e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4230
     },
     {
       "epoch": 9.309130913091309,
-      "grad_norm": 0.0035858154296875,
+      "grad_norm": 0.01470947265625,
       "learning_rate": 2.287324055218165e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4231
     },
     {
       "epoch": 9.311331133113312,
-      "grad_norm": 0.004180908203125,
+      "grad_norm": 0.00531005859375,
       "learning_rate": 2.2725995660106247e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4232
     },
     {
       "epoch": 9.313531353135314,
-      "grad_norm": 0.002960205078125,
+      "grad_norm": 0.0032501220703125,
       "learning_rate": 2.257922079163721e-06,
       "loss": 0.0001,
       "step": 4233
     },
     {
       "epoch": 9.315731573157315,
-      "grad_norm": 0.00634765625,
+      "grad_norm": 0.01141357421875,
       "learning_rate": 2.2432916017366744e-06,
       "loss": 0.0002,
       "step": 4234
     },
     {
       "epoch": 9.317931793179318,
-      "grad_norm": 0.002716064453125,
+      "grad_norm": 0.003662109375,
       "learning_rate": 2.228708140766067e-06,
       "loss": 0.0001,
       "step": 4235
     },
     {
       "epoch": 9.32013201320132,
-      "grad_norm": 0.0036468505859375,
+      "grad_norm": 0.003204345703125,
       "learning_rate": 2.2141717032658547e-06,
       "loss": 0.0001,
       "step": 4236
     },
     {
       "epoch": 9.322332233223323,
-      "grad_norm": 0.01007080078125,
+      "grad_norm": 0.01611328125,
       "learning_rate": 2.1996822962274123e-06,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 4237
     },
     {
       "epoch": 9.324532453245325,
-      "grad_norm": 0.002960205078125,
+      "grad_norm": 0.0027008056640625,
       "learning_rate": 2.1852399266194314e-06,
       "loss": 0.0001,
       "step": 4238
     },
     {
       "epoch": 9.326732673267326,
-      "grad_norm": 0.0023651123046875,
+      "grad_norm": 0.004852294921875,
       "learning_rate": 2.170844601388078e-06,
       "loss": 0.0001,
       "step": 4239
     },
     {
       "epoch": 9.328932893289329,
-      "grad_norm": 0.0023956298828125,
+      "grad_norm": 0.002960205078125,
       "learning_rate": 2.1564963274568027e-06,
       "loss": 0.0001,
       "step": 4240
     },
     {
       "epoch": 9.331133113311331,
-      "grad_norm": 0.002197265625,
+      "grad_norm": 0.0032958984375,
       "learning_rate": 2.142195111726464e-06,
       "loss": 0.0001,
       "step": 4241
     },
     {
       "epoch": 9.333333333333334,
-      "grad_norm": 0.00421142578125,
+      "grad_norm": 0.0162353515625,
       "learning_rate": 2.127940961075314e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4242
     },
     {
       "epoch": 9.335533553355335,
-      "grad_norm": 0.002349853515625,
+      "grad_norm": 0.00323486328125,
       "learning_rate": 2.1137338823589036e-06,
       "loss": 0.0001,
       "step": 4243
     },
     {
       "epoch": 9.337733773377337,
-      "grad_norm": 0.0029449462890625,
+      "grad_norm": 0.00286865234375,
       "learning_rate": 2.0995738824101995e-06,
       "loss": 0.0001,
       "step": 4244
     },
     {
       "epoch": 9.33993399339934,
-      "grad_norm": 0.0048828125,
+      "grad_norm": 0.00286865234375,
       "learning_rate": 2.085460968039499e-06,
       "loss": 0.0001,
       "step": 4245
     },
     {
       "epoch": 9.342134213421343,
-      "grad_norm": 0.003173828125,
+      "grad_norm": 0.00457763671875,
       "learning_rate": 2.0713951460344714e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4246
     },
     {
       "epoch": 9.344334433443345,
-      "grad_norm": 0.006256103515625,
+      "grad_norm": 0.003082275390625,
       "learning_rate": 2.0573764231601177e-06,
       "loss": 0.0001,
       "step": 4247
     },
     {
       "epoch": 9.346534653465346,
-      "grad_norm": 0.0032501220703125,
+      "grad_norm": 0.00738525390625,
       "learning_rate": 2.0434048061588107e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4248
     },
     {
       "epoch": 9.348734873487349,
-      "grad_norm": 0.0019683837890625,
+      "grad_norm": 0.0034332275390625,
       "learning_rate": 2.029480301750242e-06,
       "loss": 0.0001,
       "step": 4249
     },
     {
       "epoch": 9.350935093509351,
-      "grad_norm": 0.0037841796875,
+      "grad_norm": 0.0021209716796875,
       "learning_rate": 2.0156029166314315e-06,
       "loss": 0.0001,
       "step": 4250
     },
     {
       "epoch": 9.353135313531354,
-      "grad_norm": 0.0030364990234375,
+      "grad_norm": 0.0027313232421875,
       "learning_rate": 2.001772657476797e-06,
       "loss": 0.0001,
       "step": 4251
     },
     {
       "epoch": 9.355335533553355,
-      "grad_norm": 0.004547119140625,
+      "grad_norm": 0.002655029296875,
       "learning_rate": 1.987989530938028e-06,
       "loss": 0.0001,
       "step": 4252
     },
     {
       "epoch": 9.357535753575357,
-      "grad_norm": 0.00347900390625,
+      "grad_norm": 0.0028839111328125,
       "learning_rate": 1.974253543644189e-06,
       "loss": 0.0001,
       "step": 4253
     },
     {
       "epoch": 9.35973597359736,
-      "grad_norm": 0.0027008056640625,
+      "grad_norm": 0.005523681640625,
       "learning_rate": 1.9605647022016506e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4254
     },
     {
       "epoch": 9.361936193619362,
-      "grad_norm": 0.0068359375,
+      "grad_norm": 0.0042724609375,
       "learning_rate": 1.9469230131940907e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4255
     },
     {
       "epoch": 9.364136413641365,
-      "grad_norm": 0.004150390625,
+      "grad_norm": 0.006591796875,
       "learning_rate": 1.9333284831825727e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4256
     },
     {
       "epoch": 9.366336633663366,
-      "grad_norm": 0.00182342529296875,
+      "grad_norm": 0.00213623046875,
       "learning_rate": 1.9197811187053884e-06,
       "loss": 0.0001,
       "step": 4257
     },
     {
       "epoch": 9.368536853685368,
-      "grad_norm": 0.00653076171875,
+      "grad_norm": 0.00439453125,
       "learning_rate": 1.9062809262782255e-06,
       "loss": 0.0001,
       "step": 4258
     },
     {
       "epoch": 9.370737073707371,
-      "grad_norm": 0.003875732421875,
+      "grad_norm": 0.0028533935546875,
       "learning_rate": 1.8928279123940572e-06,
       "loss": 0.0001,
       "step": 4259
     },
     {
       "epoch": 9.372937293729374,
-      "grad_norm": 0.0072021484375,
+      "grad_norm": 0.0024871826171875,
       "learning_rate": 1.8794220835231412e-06,
       "loss": 0.0001,
       "step": 4260
     },
     {
       "epoch": 9.375137513751374,
-      "grad_norm": 0.002410888671875,
+      "grad_norm": 0.0032958984375,
       "learning_rate": 1.8660634461130755e-06,
       "loss": 0.0001,
       "step": 4261
     },
     {
       "epoch": 9.377337733773377,
-      "grad_norm": 0.002685546875,
+      "grad_norm": 0.00262451171875,
       "learning_rate": 1.8527520065887206e-06,
       "loss": 0.0001,
       "step": 4262
     },
     {
       "epoch": 9.37953795379538,
-      "grad_norm": 0.00238037109375,
+      "grad_norm": 0.00640869140625,
       "learning_rate": 1.8394877713522996e-06,
       "loss": 0.0001,
       "step": 4263
     },
     {
       "epoch": 9.381738173817382,
-      "grad_norm": 0.0032196044921875,
+      "grad_norm": 0.00311279296875,
       "learning_rate": 1.8262707467832873e-06,
       "loss": 0.0001,
       "step": 4264
     },
     {
       "epoch": 9.383938393839385,
-      "grad_norm": 0.004486083984375,
+      "grad_norm": 0.0081787109375,
       "learning_rate": 1.8131009392384323e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4265
     },
     {
@@ -30176,659 +30176,659 @@
     },
     {
       "epoch": 9.388338833883388,
-      "grad_norm": 0.001861572265625,
+      "grad_norm": 0.00299072265625,
       "learning_rate": 1.7869030005348343e-06,
       "loss": 0.0001,
       "step": 4267
     },
     {
       "epoch": 9.39053905390539,
-      "grad_norm": 0.003387451171875,
+      "grad_norm": 0.008056640625,
       "learning_rate": 1.7738748819760898e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4268
     },
     {
       "epoch": 9.392739273927393,
-      "grad_norm": 0.019287109375,
+      "grad_norm": 0.002471923828125,
       "learning_rate": 1.7608940056415113e-06,
-      "loss": 0.0004,
+      "loss": 0.0001,
       "step": 4269
     },
     {
       "epoch": 9.394939493949394,
-      "grad_norm": 0.00225830078125,
+      "grad_norm": 0.00244140625,
       "learning_rate": 1.7479603777742938e-06,
       "loss": 0.0001,
       "step": 4270
     },
     {
       "epoch": 9.397139713971397,
-      "grad_norm": 0.0023956298828125,
+      "grad_norm": 0.0089111328125,
       "learning_rate": 1.7350740045949386e-06,
       "loss": 0.0001,
       "step": 4271
     },
     {
       "epoch": 9.3993399339934,
-      "grad_norm": 0.005462646484375,
+      "grad_norm": 0.0057373046875,
       "learning_rate": 1.7222348923011555e-06,
       "loss": 0.0001,
       "step": 4272
     },
     {
       "epoch": 9.401540154015402,
-      "grad_norm": 0.0031890869140625,
+      "grad_norm": 0.00830078125,
       "learning_rate": 1.7094430470679933e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4273
     },
     {
       "epoch": 9.403740374037405,
-      "grad_norm": 0.002899169921875,
+      "grad_norm": 0.0029144287109375,
       "learning_rate": 1.6966984750477422e-06,
       "loss": 0.0001,
       "step": 4274
     },
     {
       "epoch": 9.405940594059405,
-      "grad_norm": 0.0023193359375,
+      "grad_norm": 0.0017547607421875,
       "learning_rate": 1.684001182369932e-06,
       "loss": 0.0001,
       "step": 4275
     },
     {
       "epoch": 9.408140814081408,
-      "grad_norm": 0.0054931640625,
+      "grad_norm": 0.004730224609375,
       "learning_rate": 1.671351175141389e-06,
       "loss": 0.0002,
       "step": 4276
     },
     {
       "epoch": 9.41034103410341,
-      "grad_norm": 0.00262451171875,
+      "grad_norm": 0.003692626953125,
       "learning_rate": 1.658748459446169e-06,
       "loss": 0.0001,
       "step": 4277
     },
     {
       "epoch": 9.412541254125413,
-      "grad_norm": 0.004119873046875,
+      "grad_norm": 0.004638671875,
       "learning_rate": 1.6461930413456117e-06,
       "loss": 0.0002,
       "step": 4278
     },
     {
       "epoch": 9.414741474147414,
-      "grad_norm": 0.005706787109375,
+      "grad_norm": 0.004180908203125,
       "learning_rate": 1.633684926878265e-06,
       "loss": 0.0002,
       "step": 4279
     },
     {
       "epoch": 9.416941694169417,
-      "grad_norm": 0.00775146484375,
+      "grad_norm": 0.00830078125,
       "learning_rate": 1.6212241220599834e-06,
       "loss": 0.0002,
       "step": 4280
     },
     {
       "epoch": 9.41914191419142,
-      "grad_norm": 0.003173828125,
+      "grad_norm": 0.0037689208984375,
       "learning_rate": 1.6088106328838283e-06,
       "loss": 0.0001,
       "step": 4281
     },
     {
       "epoch": 9.421342134213422,
-      "grad_norm": 0.003143310546875,
+      "grad_norm": 0.004852294921875,
       "learning_rate": 1.5964444653201017e-06,
       "loss": 0.0002,
       "step": 4282
     },
     {
       "epoch": 9.423542354235423,
-      "grad_norm": 0.0024871826171875,
+      "grad_norm": 0.00286865234375,
       "learning_rate": 1.5841256253163794e-06,
       "loss": 0.0001,
       "step": 4283
     },
     {
       "epoch": 9.425742574257425,
-      "grad_norm": 0.0036773681640625,
+      "grad_norm": 0.0030670166015625,
       "learning_rate": 1.5718541187974334e-06,
       "loss": 0.0001,
       "step": 4284
     },
     {
       "epoch": 9.427942794279428,
-      "grad_norm": 0.003204345703125,
+      "grad_norm": 0.0035400390625,
       "learning_rate": 1.559629951665298e-06,
       "loss": 0.0001,
       "step": 4285
     },
     {
       "epoch": 9.43014301430143,
-      "grad_norm": 0.00628662109375,
+      "grad_norm": 0.0045166015625,
       "learning_rate": 1.547453129799248e-06,
       "loss": 0.0001,
       "step": 4286
     },
     {
       "epoch": 9.432343234323433,
-      "grad_norm": 0.00567626953125,
+      "grad_norm": 0.0028076171875,
       "learning_rate": 1.535323659055743e-06,
       "loss": 0.0001,
       "step": 4287
     },
     {
       "epoch": 9.434543454345434,
-      "grad_norm": 0.004180908203125,
+      "grad_norm": 0.0052490234375,
       "learning_rate": 1.5232415452685278e-06,
       "loss": 0.0002,
       "step": 4288
     },
     {
       "epoch": 9.436743674367436,
-      "grad_norm": 0.00677490234375,
+      "grad_norm": 0.004669189453125,
       "learning_rate": 1.5112067942485098e-06,
       "loss": 0.0002,
       "step": 4289
     },
     {
       "epoch": 9.438943894389439,
-      "grad_norm": 0.002197265625,
+      "grad_norm": 0.004486083984375,
       "learning_rate": 1.4992194117838699e-06,
       "loss": 0.0001,
       "step": 4290
     },
     {
       "epoch": 9.441144114411442,
-      "grad_norm": 0.011474609375,
+      "grad_norm": 0.0037078857421875,
       "learning_rate": 1.4872794036399961e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4291
     },
     {
       "epoch": 9.443344334433444,
-      "grad_norm": 0.004119873046875,
+      "grad_norm": 0.0036468505859375,
       "learning_rate": 1.4753867755594287e-06,
       "loss": 0.0002,
       "step": 4292
     },
     {
       "epoch": 9.445544554455445,
-      "grad_norm": 0.002838134765625,
+      "grad_norm": 0.00518798828125,
       "learning_rate": 1.463541533262036e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4293
     },
     {
       "epoch": 9.447744774477448,
-      "grad_norm": 0.0035400390625,
+      "grad_norm": 0.004638671875,
       "learning_rate": 1.4517436824447838e-06,
       "loss": 0.0001,
       "step": 4294
     },
     {
       "epoch": 9.44994499449945,
-      "grad_norm": 0.003143310546875,
+      "grad_norm": 0.004791259765625,
       "learning_rate": 1.4399932287819107e-06,
       "loss": 0.0001,
       "step": 4295
     },
     {
       "epoch": 9.452145214521453,
-      "grad_norm": 0.00714111328125,
+      "grad_norm": 0.0027618408203125,
       "learning_rate": 1.4282901779248513e-06,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 4296
     },
     {
       "epoch": 9.454345434543454,
-      "grad_norm": 0.003448486328125,
+      "grad_norm": 0.0037689208984375,
       "learning_rate": 1.4166345355022037e-06,
       "loss": 0.0001,
       "step": 4297
     },
     {
       "epoch": 9.456545654565456,
-      "grad_norm": 0.0029296875,
+      "grad_norm": 0.002593994140625,
       "learning_rate": 1.4050263071198278e-06,
       "loss": 0.0001,
       "step": 4298
     },
     {
       "epoch": 9.458745874587459,
-      "grad_norm": 0.0024261474609375,
+      "grad_norm": 0.0032806396484375,
       "learning_rate": 1.393465498360713e-06,
       "loss": 0.0001,
       "step": 4299
     },
     {
       "epoch": 9.460946094609461,
-      "grad_norm": 0.0028839111328125,
+      "grad_norm": 0.00823974609375,
       "learning_rate": 1.3819521147851123e-06,
       "loss": 0.0001,
       "step": 4300
     },
     {
       "epoch": 9.463146314631462,
-      "grad_norm": 0.0023345947265625,
+      "grad_norm": 0.0024871826171875,
       "learning_rate": 1.3704861619303843e-06,
       "loss": 0.0001,
       "step": 4301
     },
     {
       "epoch": 9.465346534653465,
-      "grad_norm": 0.0026092529296875,
+      "grad_norm": 0.009521484375,
       "learning_rate": 1.3590676453111517e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4302
     },
     {
       "epoch": 9.467546754675467,
-      "grad_norm": 0.00238037109375,
+      "grad_norm": 0.0126953125,
       "learning_rate": 1.347696570419199e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4303
     },
     {
       "epoch": 9.46974697469747,
-      "grad_norm": 0.004241943359375,
+      "grad_norm": 0.002471923828125,
       "learning_rate": 1.3363729427234516e-06,
       "loss": 0.0001,
       "step": 4304
     },
     {
       "epoch": 9.471947194719473,
-      "grad_norm": 0.0019683837890625,
+      "grad_norm": 0.002685546875,
       "learning_rate": 1.3250967676700976e-06,
       "loss": 0.0001,
       "step": 4305
     },
     {
       "epoch": 9.474147414741473,
-      "grad_norm": 0.002105712890625,
+      "grad_norm": 0.004486083984375,
       "learning_rate": 1.3138680506824097e-06,
       "loss": 0.0001,
       "step": 4306
     },
     {
       "epoch": 9.476347634763476,
-      "grad_norm": 0.003814697265625,
+      "grad_norm": 0.007476806640625,
       "learning_rate": 1.3026867971609124e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4307
     },
     {
       "epoch": 9.478547854785479,
-      "grad_norm": 0.00335693359375,
+      "grad_norm": 0.0038604736328125,
       "learning_rate": 1.291553012483271e-06,
       "loss": 0.0001,
       "step": 4308
     },
     {
       "epoch": 9.480748074807481,
-      "grad_norm": 0.0028533935546875,
+      "grad_norm": 0.013916015625,
       "learning_rate": 1.2804667020043014e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4309
     },
     {
       "epoch": 9.482948294829482,
-      "grad_norm": 0.005462646484375,
+      "grad_norm": 0.003021240234375,
       "learning_rate": 1.2694278710560281e-06,
       "loss": 0.0001,
       "step": 4310
     },
     {
       "epoch": 9.485148514851485,
-      "grad_norm": 0.0023193359375,
+      "grad_norm": 0.0037689208984375,
       "learning_rate": 1.2584365249476037e-06,
       "loss": 0.0001,
       "step": 4311
     },
     {
       "epoch": 9.487348734873487,
-      "grad_norm": 0.0036163330078125,
+      "grad_norm": 0.00433349609375,
       "learning_rate": 1.2474926689653554e-06,
       "loss": 0.0001,
       "step": 4312
     },
     {
       "epoch": 9.48954895489549,
-      "grad_norm": 0.00341796875,
+      "grad_norm": 0.003326416015625,
       "learning_rate": 1.236596308372795e-06,
       "loss": 0.0001,
       "step": 4313
     },
     {
       "epoch": 9.491749174917492,
-      "grad_norm": 0.0027313232421875,
+      "grad_norm": 0.00689697265625,
       "learning_rate": 1.2257474484105413e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4314
     },
     {
       "epoch": 9.493949394939493,
-      "grad_norm": 0.0028228759765625,
+      "grad_norm": 0.005523681640625,
       "learning_rate": 1.2149460942964098e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4315
     },
     {
       "epoch": 9.496149614961496,
-      "grad_norm": 0.002105712890625,
+      "grad_norm": 0.0036163330078125,
       "learning_rate": 1.2041922512253334e-06,
       "loss": 0.0001,
       "step": 4316
     },
     {
       "epoch": 9.498349834983498,
-      "grad_norm": 0.0033721923828125,
+      "grad_norm": 0.004058837890625,
       "learning_rate": 1.1934859243694417e-06,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 4317
     },
     {
       "epoch": 9.500550055005501,
-      "grad_norm": 0.002197265625,
+      "grad_norm": 0.0042724609375,
       "learning_rate": 1.1828271188779715e-06,
       "loss": 0.0001,
       "step": 4318
     },
     {
       "epoch": 9.502750275027502,
-      "grad_norm": 0.0020751953125,
+      "grad_norm": 0.0025787353515625,
       "learning_rate": 1.1722158398772887e-06,
       "loss": 0.0001,
       "step": 4319
     },
     {
       "epoch": 9.504950495049505,
-      "grad_norm": 0.0035552978515625,
+      "grad_norm": 0.005889892578125,
       "learning_rate": 1.1616520924709773e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4320
     },
     {
       "epoch": 9.507150715071507,
-      "grad_norm": 0.0037994384765625,
+      "grad_norm": 0.0026702880859375,
       "learning_rate": 1.1511358817396733e-06,
       "loss": 0.0001,
       "step": 4321
     },
     {
       "epoch": 9.50935093509351,
-      "grad_norm": 0.00421142578125,
+      "grad_norm": 0.0031890869140625,
       "learning_rate": 1.1406672127412089e-06,
       "loss": 0.0001,
       "step": 4322
     },
     {
       "epoch": 9.511551155115512,
-      "grad_norm": 0.0054931640625,
+      "grad_norm": 0.004364013671875,
       "learning_rate": 1.1302460905105117e-06,
       "loss": 0.0002,
       "step": 4323
     },
     {
       "epoch": 9.513751375137513,
-      "grad_norm": 0.0024871826171875,
+      "grad_norm": 0.002899169921875,
       "learning_rate": 1.1198725200596615e-06,
       "loss": 0.0001,
       "step": 4324
     },
     {
       "epoch": 9.515951595159516,
-      "grad_norm": 0.0023040771484375,
+      "grad_norm": 0.00439453125,
       "learning_rate": 1.1095465063778898e-06,
       "loss": 0.0001,
       "step": 4325
     },
     {
       "epoch": 9.518151815181518,
-      "grad_norm": 0.002899169921875,
+      "grad_norm": 0.002960205078125,
       "learning_rate": 1.0992680544315014e-06,
       "loss": 0.0001,
       "step": 4326
     },
     {
       "epoch": 9.520352035203521,
-      "grad_norm": 0.004547119140625,
+      "grad_norm": 0.002288818359375,
       "learning_rate": 1.0890371691639868e-06,
       "loss": 0.0001,
       "step": 4327
     },
     {
       "epoch": 9.522552255225522,
-      "grad_norm": 0.0035247802734375,
+      "grad_norm": 0.008056640625,
       "learning_rate": 1.0788538554958871e-06,
       "loss": 0.0002,
       "step": 4328
     },
     {
       "epoch": 9.524752475247524,
-      "grad_norm": 0.0023345947265625,
+      "grad_norm": 0.00555419921875,
       "learning_rate": 1.0687181183249518e-06,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4329
     },
     {
       "epoch": 9.526952695269527,
-      "grad_norm": 0.002410888671875,
+      "grad_norm": 0.003753662109375,
       "learning_rate": 1.05862996252597e-06,
       "loss": 0.0001,
       "step": 4330
     },
     {
       "epoch": 9.52915291529153,
-      "grad_norm": 0.01025390625,
+      "grad_norm": 0.0024566650390625,
       "learning_rate": 1.0485893929508939e-06,
       "loss": 0.0001,
       "step": 4331
     },
     {
       "epoch": 9.531353135313532,
-      "grad_norm": 0.0029449462890625,
+      "grad_norm": 0.003814697265625,
       "learning_rate": 1.038596414428772e-06,
       "loss": 0.0001,
       "step": 4332
     },
     {
       "epoch": 9.531353135313532,
-      "eval_loss": 0.00024176519946195185,
-      "eval_runtime": 10.2358,
-      "eval_samples_per_second": 37.418,
-      "eval_steps_per_second": 4.689,
+      "eval_loss": 0.000350463786162436,
+      "eval_runtime": 10.1923,
+      "eval_samples_per_second": 37.577,
+      "eval_steps_per_second": 4.709,
       "step": 4332
     },
     {
       "epoch": 9.533553355335533,
-      "grad_norm": 0.0054931640625,
+      "grad_norm": 0.005706787109375,
       "learning_rate": 1.0286510317657482e-06,
       "loss": 0.0002,
       "step": 4333
     },
     {
       "epoch": 9.535753575357536,
-      "grad_norm": 0.0028533935546875,
+      "grad_norm": 0.00299072265625,
       "learning_rate": 1.018753249745108e-06,
       "loss": 0.0001,
       "step": 4334
     },
     {
       "epoch": 9.537953795379538,
-      "grad_norm": 0.0078125,
+      "grad_norm": 0.003631591796875,
       "learning_rate": 1.008903073127232e-06,
       "loss": 0.0001,
       "step": 4335
     },
     {
       "epoch": 9.54015401540154,
-      "grad_norm": 0.00225830078125,
+      "grad_norm": 0.0025634765625,
       "learning_rate": 9.991005066495752e-07,
       "loss": 0.0001,
       "step": 4336
     },
     {
       "epoch": 9.542354235423542,
-      "grad_norm": 0.0028228759765625,
+      "grad_norm": 0.003082275390625,
       "learning_rate": 9.893455550267438e-07,
       "loss": 0.0001,
       "step": 4337
     },
     {
       "epoch": 9.544554455445544,
-      "grad_norm": 0.006134033203125,
+      "grad_norm": 0.007781982421875,
       "learning_rate": 9.796382229504075e-07,
       "loss": 0.0002,
       "step": 4338
     },
     {
       "epoch": 9.546754675467547,
-      "grad_norm": 0.0026702880859375,
+      "grad_norm": 0.0045166015625,
       "learning_rate": 9.699785150893426e-07,
       "loss": 0.0001,
       "step": 4339
     },
     {
       "epoch": 9.54895489548955,
-      "grad_norm": 0.00250244140625,
+      "grad_norm": 0.0047607421875,
       "learning_rate": 9.603664360894327e-07,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4340
     },
     {
       "epoch": 9.551155115511552,
-      "grad_norm": 0.003997802734375,
+      "grad_norm": 0.003936767578125,
       "learning_rate": 9.508019905736242e-07,
       "loss": 0.0001,
       "step": 4341
     },
     {
       "epoch": 9.553355335533553,
-      "grad_norm": 0.002410888671875,
+      "grad_norm": 0.002685546875,
       "learning_rate": 9.412851831419934e-07,
       "loss": 0.0001,
       "step": 4342
     },
     {
       "epoch": 9.555555555555555,
-      "grad_norm": 0.0031280517578125,
+      "grad_norm": 0.01025390625,
       "learning_rate": 9.318160183716674e-07,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4343
     },
     {
       "epoch": 9.557755775577558,
-      "grad_norm": 0.004058837890625,
+      "grad_norm": 0.004669189453125,
       "learning_rate": 9.223945008169033e-07,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4344
     },
     {
       "epoch": 9.55995599559956,
-      "grad_norm": 0.0026702880859375,
+      "grad_norm": 0.005950927734375,
       "learning_rate": 9.130206350089765e-07,
       "loss": 0.0001,
       "step": 4345
     },
     {
       "epoch": 9.562156215621561,
-      "grad_norm": 0.00384521484375,
+      "grad_norm": 0.003387451171875,
       "learning_rate": 9.036944254563029e-07,
       "loss": 0.0001,
       "step": 4346
     },
     {
       "epoch": 9.564356435643564,
-      "grad_norm": 0.003448486328125,
+      "grad_norm": 0.0025482177734375,
       "learning_rate": 8.944158766443722e-07,
       "loss": 0.0001,
       "step": 4347
     },
     {
       "epoch": 9.566556655665567,
-      "grad_norm": 0.00250244140625,
+      "grad_norm": 0.004608154296875,
       "learning_rate": 8.851849930357148e-07,
       "loss": 0.0001,
       "step": 4348
     },
     {
       "epoch": 9.56875687568757,
-      "grad_norm": 0.0022125244140625,
+      "grad_norm": 0.003509521484375,
       "learning_rate": 8.760017790699793e-07,
       "loss": 0.0001,
       "step": 4349
     },
     {
       "epoch": 9.570957095709572,
-      "grad_norm": 0.00347900390625,
+      "grad_norm": 0.002593994140625,
       "learning_rate": 8.668662391638437e-07,
       "loss": 0.0001,
       "step": 4350
     },
     {
       "epoch": 9.573157315731573,
-      "grad_norm": 0.003936767578125,
+      "grad_norm": 0.00311279296875,
       "learning_rate": 8.577783777110937e-07,
       "loss": 0.0001,
       "step": 4351
     },
     {
       "epoch": 9.575357535753575,
-      "grad_norm": 0.0034942626953125,
+      "grad_norm": 0.002288818359375,
       "learning_rate": 8.487381990825771e-07,
       "loss": 0.0001,
       "step": 4352
     },
     {
       "epoch": 9.577557755775578,
-      "grad_norm": 0.005157470703125,
+      "grad_norm": 0.00689697265625,
       "learning_rate": 8.397457076261716e-07,
       "loss": 0.0002,
       "step": 4353
     },
     {
       "epoch": 9.57975797579758,
-      "grad_norm": 0.002197265625,
+      "grad_norm": 0.002960205078125,
       "learning_rate": 8.308009076668955e-07,
       "loss": 0.0001,
       "step": 4354
     },
     {
       "epoch": 9.581958195819581,
-      "grad_norm": 0.0025482177734375,
+      "grad_norm": 0.00726318359375,
       "learning_rate": 8.219038035067517e-07,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4355
     },
     {
       "epoch": 9.584158415841584,
-      "grad_norm": 0.0021514892578125,
+      "grad_norm": 0.00274658203125,
       "learning_rate": 8.130543994248507e-07,
       "loss": 0.0001,
       "step": 4356
     },
     {
       "epoch": 9.586358635863586,
-      "grad_norm": 0.0023345947265625,
+      "grad_norm": 0.00262451171875,
       "learning_rate": 8.042526996773326e-07,
       "loss": 0.0001,
       "step": 4357
     },
     {
       "epoch": 9.588558855885589,
-      "grad_norm": 0.003326416015625,
+      "grad_norm": 0.00421142578125,
       "learning_rate": 7.954987084974108e-07,
       "loss": 0.0001,
       "step": 4358
     },
     {
       "epoch": 9.590759075907592,
-      "grad_norm": 0.00775146484375,
+      "grad_norm": 0.006103515625,
       "learning_rate": 7.867924300953621e-07,
       "loss": 0.0002,
       "step": 4359
@@ -30842,820 +30842,820 @@
     },
     {
       "epoch": 9.595159515951595,
-      "grad_norm": 0.002777099609375,
+      "grad_norm": 0.0035400390625,
       "learning_rate": 7.695230283511823e-07,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4361
     },
     {
       "epoch": 9.597359735973598,
-      "grad_norm": 0.0029754638671875,
+      "grad_norm": 0.00408935546875,
       "learning_rate": 7.609599133148293e-07,
       "loss": 0.0001,
       "step": 4362
     },
     {
       "epoch": 9.5995599559956,
-      "grad_norm": 0.0021514892578125,
+      "grad_norm": 0.003143310546875,
       "learning_rate": 7.52444527667906e-07,
       "loss": 0.0001,
       "step": 4363
     },
     {
       "epoch": 9.601760176017601,
-      "grad_norm": 0.003143310546875,
+      "grad_norm": 0.002227783203125,
       "learning_rate": 7.439768755059251e-07,
       "loss": 0.0001,
       "step": 4364
     },
     {
       "epoch": 9.603960396039604,
-      "grad_norm": 0.003265380859375,
+      "grad_norm": 0.004547119140625,
       "learning_rate": 7.35556960901429e-07,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4365
     },
     {
       "epoch": 9.606160616061606,
-      "grad_norm": 0.003753662109375,
+      "grad_norm": 0.009765625,
       "learning_rate": 7.271847879040117e-07,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4366
     },
     {
       "epoch": 9.608360836083609,
-      "grad_norm": 0.0034027099609375,
+      "grad_norm": 0.0033721923828125,
       "learning_rate": 7.188603605402855e-07,
       "loss": 0.0001,
       "step": 4367
     },
     {
       "epoch": 9.61056105610561,
-      "grad_norm": 0.0032806396484375,
+      "grad_norm": 0.0027618408203125,
       "learning_rate": 7.105836828139367e-07,
       "loss": 0.0001,
       "step": 4368
     },
     {
       "epoch": 9.612761276127612,
-      "grad_norm": 0.0032501220703125,
+      "grad_norm": 0.00567626953125,
       "learning_rate": 7.023547587056478e-07,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4369
     },
     {
       "epoch": 9.614961496149615,
-      "grad_norm": 0.0037841796875,
+      "grad_norm": 0.0142822265625,
       "learning_rate": 6.94173592173164e-07,
       "loss": 0.0002,
       "step": 4370
     },
     {
       "epoch": 9.617161716171617,
-      "grad_norm": 0.003997802734375,
+      "grad_norm": 0.0036468505859375,
       "learning_rate": 6.860401871512379e-07,
       "loss": 0.0001,
       "step": 4371
     },
     {
       "epoch": 9.61936193619362,
-      "grad_norm": 0.0027008056640625,
+      "grad_norm": 0.002532958984375,
       "learning_rate": 6.779545475516735e-07,
       "loss": 0.0001,
       "step": 4372
     },
     {
       "epoch": 9.62156215621562,
-      "grad_norm": 0.00408935546875,
+      "grad_norm": 0.003997802734375,
       "learning_rate": 6.699166772632715e-07,
       "loss": 0.0001,
       "step": 4373
     },
     {
       "epoch": 9.623762376237623,
-      "grad_norm": 0.005035400390625,
+      "grad_norm": 0.005645751953125,
       "learning_rate": 6.619265801518947e-07,
       "loss": 0.0002,
       "step": 4374
     },
     {
       "epoch": 9.625962596259626,
-      "grad_norm": 0.002044677734375,
+      "grad_norm": 0.002471923828125,
       "learning_rate": 6.539842600603918e-07,
       "loss": 0.0001,
       "step": 4375
     },
     {
       "epoch": 9.628162816281629,
-      "grad_norm": 0.0037689208984375,
+      "grad_norm": 0.0032806396484375,
       "learning_rate": 6.460897208086625e-07,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 4376
     },
     {
       "epoch": 9.630363036303631,
-      "grad_norm": 0.005279541015625,
+      "grad_norm": 0.004302978515625,
       "learning_rate": 6.38242966193614e-07,
       "loss": 0.0002,
       "step": 4377
     },
     {
       "epoch": 9.632563256325632,
-      "grad_norm": 0.0027313232421875,
+      "grad_norm": 0.003631591796875,
       "learning_rate": 6.304439999891832e-07,
       "loss": 0.0001,
       "step": 4378
     },
     {
       "epoch": 9.634763476347635,
-      "grad_norm": 0.0024871826171875,
+      "grad_norm": 0.0028839111328125,
       "learning_rate": 6.226928259462917e-07,
       "loss": 0.0001,
       "step": 4379
     },
     {
       "epoch": 9.636963696369637,
-      "grad_norm": 0.002777099609375,
+      "grad_norm": 0.003204345703125,
       "learning_rate": 6.149894477928908e-07,
       "loss": 0.0001,
       "step": 4380
     },
     {
       "epoch": 9.63916391639164,
-      "grad_norm": 0.005767822265625,
+      "grad_norm": 0.0023040771484375,
       "learning_rate": 6.073338692339836e-07,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 4381
     },
     {
       "epoch": 9.64136413641364,
-      "grad_norm": 0.002349853515625,
+      "grad_norm": 0.00506591796875,
       "learning_rate": 5.997260939515248e-07,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4382
     },
     {
       "epoch": 9.643564356435643,
-      "grad_norm": 0.00311279296875,
+      "grad_norm": 0.005340576171875,
       "learning_rate": 5.921661256045097e-07,
       "loss": 0.0001,
       "step": 4383
     },
     {
       "epoch": 9.645764576457646,
-      "grad_norm": 0.0025482177734375,
+      "grad_norm": 0.0042724609375,
       "learning_rate": 5.846539678289298e-07,
       "loss": 0.0001,
       "step": 4384
     },
     {
       "epoch": 9.647964796479648,
-      "grad_norm": 0.0027923583984375,
+      "grad_norm": 0.0022735595703125,
       "learning_rate": 5.771896242377727e-07,
       "loss": 0.0001,
       "step": 4385
     },
     {
       "epoch": 9.65016501650165,
-      "grad_norm": 0.002349853515625,
+      "grad_norm": 0.002593994140625,
       "learning_rate": 5.69773098421078e-07,
       "loss": 0.0001,
       "step": 4386
     },
     {
       "epoch": 9.652365236523652,
-      "grad_norm": 0.0062255859375,
+      "grad_norm": 0.0027313232421875,
       "learning_rate": 5.624043939458145e-07,
       "loss": 0.0001,
       "step": 4387
     },
     {
       "epoch": 9.654565456545654,
-      "grad_norm": 0.00201416015625,
+      "grad_norm": 0.0031280517578125,
       "learning_rate": 5.55083514356014e-07,
       "loss": 0.0001,
       "step": 4388
     },
     {
       "epoch": 9.656765676567657,
-      "grad_norm": 0.0026397705078125,
+      "grad_norm": 0.0025634765625,
       "learning_rate": 5.478104631726711e-07,
       "loss": 0.0001,
       "step": 4389
     },
     {
       "epoch": 9.65896589658966,
-      "grad_norm": 0.0040283203125,
+      "grad_norm": 0.0026092529296875,
       "learning_rate": 5.405852438937764e-07,
       "loss": 0.0001,
       "step": 4390
     },
     {
       "epoch": 9.66116611661166,
-      "grad_norm": 0.0029449462890625,
+      "grad_norm": 0.00457763671875,
       "learning_rate": 5.334078599943504e-07,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4391
     },
     {
       "epoch": 9.663366336633663,
-      "grad_norm": 0.0034027099609375,
+      "grad_norm": 0.004302978515625,
       "learning_rate": 5.262783149263539e-07,
       "loss": 0.0001,
       "step": 4392
     },
     {
       "epoch": 9.665566556655666,
-      "grad_norm": 0.003631591796875,
+      "grad_norm": 0.00567626953125,
       "learning_rate": 5.191966121187997e-07,
       "loss": 0.0001,
       "step": 4393
     },
     {
       "epoch": 9.667766776677668,
-      "grad_norm": 0.00396728515625,
+      "grad_norm": 0.005157470703125,
       "learning_rate": 5.121627549776409e-07,
       "loss": 0.0001,
       "step": 4394
     },
     {
       "epoch": 9.66996699669967,
-      "grad_norm": 0.00335693359375,
+      "grad_norm": 0.00445556640625,
       "learning_rate": 5.05176746885827e-07,
       "loss": 0.0001,
       "step": 4395
     },
     {
       "epoch": 9.672167216721672,
-      "grad_norm": 0.0147705078125,
+      "grad_norm": 0.003173828125,
       "learning_rate": 4.982385912033261e-07,
-      "loss": 0.0003,
+      "loss": 0.0001,
       "step": 4396
     },
     {
       "epoch": 9.674367436743674,
-      "grad_norm": 0.004638671875,
+      "grad_norm": 0.01025390625,
       "learning_rate": 4.913482912670687e-07,
       "loss": 0.0002,
       "step": 4397
     },
     {
       "epoch": 9.676567656765677,
-      "grad_norm": 0.00347900390625,
+      "grad_norm": 0.00433349609375,
       "learning_rate": 4.845058503909594e-07,
       "loss": 0.0001,
       "step": 4398
     },
     {
       "epoch": 9.67876787678768,
-      "grad_norm": 0.005279541015625,
+      "grad_norm": 0.0027008056640625,
       "learning_rate": 4.777112718658993e-07,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 4399
     },
     {
       "epoch": 9.68096809680968,
-      "grad_norm": 0.005279541015625,
+      "grad_norm": 0.0022125244140625,
       "learning_rate": 4.7096455895976334e-07,
       "loss": 0.0001,
       "step": 4400
     },
     {
       "epoch": 9.683168316831683,
-      "grad_norm": 0.004302978515625,
+      "grad_norm": 0.0020599365234375,
       "learning_rate": 4.642657149174223e-07,
       "loss": 0.0001,
       "step": 4401
     },
     {
       "epoch": 9.685368536853685,
-      "grad_norm": 0.0032806396484375,
+      "grad_norm": 0.0027313232421875,
       "learning_rate": 4.5761474296068807e-07,
       "loss": 0.0001,
       "step": 4402
     },
     {
       "epoch": 9.687568756875688,
-      "grad_norm": 0.0030059814453125,
+      "grad_norm": 0.005279541015625,
       "learning_rate": 4.5101164628840175e-07,
       "loss": 0.0001,
       "step": 4403
     },
     {
       "epoch": 9.689768976897689,
-      "grad_norm": 0.005859375,
+      "grad_norm": 0.0038604736328125,
       "learning_rate": 4.4445642807631196e-07,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 4404
     },
     {
       "epoch": 9.691969196919691,
-      "grad_norm": 0.005523681640625,
+      "grad_norm": 0.004058837890625,
       "learning_rate": 4.3794909147720773e-07,
       "loss": 0.0002,
       "step": 4405
     },
     {
       "epoch": 9.694169416941694,
-      "grad_norm": 0.008544921875,
+      "grad_norm": 0.0027008056640625,
       "learning_rate": 4.314896396207968e-07,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 4406
     },
     {
       "epoch": 9.696369636963697,
-      "grad_norm": 0.0027923583984375,
+      "grad_norm": 0.006866455078125,
       "learning_rate": 4.2507807561377177e-07,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4407
     },
     {
       "epoch": 9.6985698569857,
-      "grad_norm": 0.002227783203125,
+      "grad_norm": 0.0024261474609375,
       "learning_rate": 4.187144025398215e-07,
       "loss": 0.0001,
       "step": 4408
     },
     {
       "epoch": 9.7007700770077,
-      "grad_norm": 0.0026092529296875,
+      "grad_norm": 0.004974365234375,
       "learning_rate": 4.123986234595534e-07,
       "loss": 0.0001,
       "step": 4409
     },
     {
       "epoch": 9.702970297029703,
-      "grad_norm": 0.0024566650390625,
+      "grad_norm": 0.0052490234375,
       "learning_rate": 4.0613074141059307e-07,
       "loss": 0.0001,
       "step": 4410
     },
     {
       "epoch": 9.705170517051705,
-      "grad_norm": 0.004241943359375,
+      "grad_norm": 0.0032806396484375,
       "learning_rate": 3.999107594074736e-07,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 4411
     },
     {
       "epoch": 9.707370737073708,
-      "grad_norm": 0.0019989013671875,
+      "grad_norm": 0.0029144287109375,
       "learning_rate": 3.9373868044173536e-07,
       "loss": 0.0001,
       "step": 4412
     },
     {
       "epoch": 9.70957095709571,
-      "grad_norm": 0.00726318359375,
+      "grad_norm": 0.0126953125,
       "learning_rate": 3.8761450748187043e-07,
       "loss": 0.0002,
       "step": 4413
     },
     {
       "epoch": 9.711771177117711,
-      "grad_norm": 0.0032958984375,
+      "grad_norm": 0.008544921875,
       "learning_rate": 3.815382434732895e-07,
       "loss": 0.0002,
       "step": 4414
     },
     {
       "epoch": 9.713971397139714,
-      "grad_norm": 0.00323486328125,
+      "grad_norm": 0.005096435546875,
       "learning_rate": 3.755098913384325e-07,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4415
     },
     {
       "epoch": 9.716171617161717,
-      "grad_norm": 0.002960205078125,
+      "grad_norm": 0.0036773681640625,
       "learning_rate": 3.6952945397664696e-07,
       "loss": 0.0001,
       "step": 4416
     },
     {
       "epoch": 9.718371837183719,
-      "grad_norm": 0.0023345947265625,
+      "grad_norm": 0.002471923828125,
       "learning_rate": 3.6359693426423203e-07,
       "loss": 0.0001,
       "step": 4417
     },
     {
       "epoch": 9.72057205720572,
-      "grad_norm": 0.0022125244140625,
+      "grad_norm": 0.0027008056640625,
       "learning_rate": 3.577123350544609e-07,
       "loss": 0.0001,
       "step": 4418
     },
     {
       "epoch": 9.722772277227723,
-      "grad_norm": 0.00372314453125,
+      "grad_norm": 0.002105712890625,
       "learning_rate": 3.518756591775585e-07,
       "loss": 0.0001,
       "step": 4419
     },
     {
       "epoch": 9.724972497249725,
-      "grad_norm": 0.00244140625,
+      "grad_norm": 0.0026397705078125,
       "learning_rate": 3.4608690944071263e-07,
       "loss": 0.0001,
       "step": 4420
     },
     {
       "epoch": 9.727172717271728,
-      "grad_norm": 0.0057373046875,
+      "grad_norm": 0.0028533935546875,
       "learning_rate": 3.4034608862800743e-07,
       "loss": 0.0001,
       "step": 4421
     },
     {
       "epoch": 9.729372937293729,
-      "grad_norm": 0.0021514892578125,
+      "grad_norm": 0.005279541015625,
       "learning_rate": 3.346531995005453e-07,
       "loss": 0.0001,
       "step": 4422
     },
     {
       "epoch": 9.731573157315731,
-      "grad_norm": 0.01220703125,
+      "grad_norm": 0.005126953125,
       "learning_rate": 3.29008244796325e-07,
       "loss": 0.0002,
       "step": 4423
     },
     {
       "epoch": 9.733773377337734,
-      "grad_norm": 0.003814697265625,
+      "grad_norm": 0.002838134765625,
       "learning_rate": 3.2341122723030806e-07,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4424
     },
     {
       "epoch": 9.735973597359736,
-      "grad_norm": 0.004638671875,
+      "grad_norm": 0.0045166015625,
       "learning_rate": 3.1786214949441893e-07,
       "loss": 0.0002,
       "step": 4425
     },
     {
       "epoch": 9.738173817381739,
-      "grad_norm": 0.002838134765625,
+      "grad_norm": 0.002685546875,
       "learning_rate": 3.1236101425747843e-07,
       "loss": 0.0001,
       "step": 4426
     },
     {
       "epoch": 9.74037403740374,
-      "grad_norm": 0.006744384765625,
+      "grad_norm": 0.003753662109375,
       "learning_rate": 3.069078241653034e-07,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4427
     },
     {
       "epoch": 9.742574257425742,
-      "grad_norm": 0.00567626953125,
+      "grad_norm": 0.003204345703125,
       "learning_rate": 3.015025818406181e-07,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 4428
     },
     {
       "epoch": 9.744774477447745,
-      "grad_norm": 0.0038604736328125,
+      "grad_norm": 0.006378173828125,
       "learning_rate": 2.961452898830874e-07,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4429
     },
     {
       "epoch": 9.746974697469748,
-      "grad_norm": 0.0020599365234375,
+      "grad_norm": 0.00201416015625,
       "learning_rate": 2.9083595086933924e-07,
       "loss": 0.0001,
       "step": 4430
     },
     {
       "epoch": 9.749174917491748,
-      "grad_norm": 0.00244140625,
+      "grad_norm": 0.0030975341796875,
       "learning_rate": 2.8557456735289757e-07,
       "loss": 0.0001,
       "step": 4431
     },
     {
       "epoch": 9.751375137513751,
-      "grad_norm": 0.0020904541015625,
+      "grad_norm": 0.00274658203125,
       "learning_rate": 2.8036114186424933e-07,
       "loss": 0.0001,
       "step": 4432
     },
     {
       "epoch": 9.753575357535754,
-      "grad_norm": 0.002288818359375,
+      "grad_norm": 0.00494384765625,
       "learning_rate": 2.7519567691083323e-07,
       "loss": 0.0001,
       "step": 4433
     },
     {
       "epoch": 9.755775577557756,
-      "grad_norm": 0.0022125244140625,
+      "grad_norm": 0.0022735595703125,
       "learning_rate": 2.700781749769732e-07,
       "loss": 0.0001,
       "step": 4434
     },
     {
       "epoch": 9.757975797579759,
-      "grad_norm": 0.006439208984375,
+      "grad_norm": 0.005615234375,
       "learning_rate": 2.6500863852395584e-07,
       "loss": 0.0002,
       "step": 4435
     },
     {
       "epoch": 9.76017601760176,
-      "grad_norm": 0.002960205078125,
+      "grad_norm": 0.007659912109375,
       "learning_rate": 2.5998706998999755e-07,
       "loss": 0.0001,
       "step": 4436
     },
     {
       "epoch": 9.762376237623762,
-      "grad_norm": 0.002655029296875,
+      "grad_norm": 0.004608154296875,
       "learning_rate": 2.5501347179025527e-07,
       "loss": 0.0001,
       "step": 4437
     },
     {
       "epoch": 9.764576457645765,
-      "grad_norm": 0.00274658203125,
+      "grad_norm": 0.00390625,
       "learning_rate": 2.5008784631677106e-07,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4438
     },
     {
       "epoch": 9.766776677667767,
-      "grad_norm": 0.004150390625,
+      "grad_norm": 0.0035247802734375,
       "learning_rate": 2.452101959385611e-07,
       "loss": 0.0001,
       "step": 4439
     },
     {
       "epoch": 9.768976897689768,
-      "grad_norm": 0.0024871826171875,
+      "grad_norm": 0.002685546875,
       "learning_rate": 2.403805230015488e-07,
       "loss": 0.0001,
       "step": 4440
     },
     {
       "epoch": 9.77117711771177,
-      "grad_norm": 0.0167236328125,
+      "grad_norm": 0.0037841796875,
       "learning_rate": 2.3559882982857605e-07,
-      "loss": 0.0004,
+      "loss": 0.0002,
       "step": 4441
     },
     {
       "epoch": 9.773377337733773,
-      "grad_norm": 0.005767822265625,
+      "grad_norm": 0.0034027099609375,
       "learning_rate": 2.3086511871943662e-07,
       "loss": 0.0001,
       "step": 4442
     },
     {
       "epoch": 9.775577557755776,
-      "grad_norm": 0.003021240234375,
+      "grad_norm": 0.0025177001953125,
       "learning_rate": 2.2617939195079818e-07,
       "loss": 0.0001,
       "step": 4443
     },
     {
       "epoch": 9.777777777777779,
-      "grad_norm": 0.0023345947265625,
+      "grad_norm": 0.0032806396484375,
       "learning_rate": 2.215416517762914e-07,
       "loss": 0.0001,
       "step": 4444
     },
     {
       "epoch": 9.77997799779978,
-      "grad_norm": 0.00372314453125,
+      "grad_norm": 0.00531005859375,
       "learning_rate": 2.169519004264542e-07,
       "loss": 0.0001,
       "step": 4445
     },
     {
       "epoch": 9.782178217821782,
-      "grad_norm": 0.004119873046875,
+      "grad_norm": 0.006439208984375,
       "learning_rate": 2.1241014010874306e-07,
       "loss": 0.0001,
       "step": 4446
     },
     {
       "epoch": 9.782178217821782,
-      "eval_loss": 0.0002402140380581841,
-      "eval_runtime": 10.2298,
-      "eval_samples_per_second": 37.44,
-      "eval_steps_per_second": 4.692,
+      "eval_loss": 0.0003497041470836848,
+      "eval_runtime": 11.1132,
+      "eval_samples_per_second": 34.464,
+      "eval_steps_per_second": 4.319,
       "step": 4446
     },
     {
       "epoch": 9.784378437843785,
-      "grad_norm": 0.005340576171875,
+      "grad_norm": 0.00689697265625,
       "learning_rate": 2.0791637300754396e-07,
       "loss": 0.0001,
       "step": 4447
     },
     {
       "epoch": 9.786578657865787,
-      "grad_norm": 0.0027008056640625,
+      "grad_norm": 0.0152587890625,
       "learning_rate": 2.034706012841503e-07,
-      "loss": 0.0001,
+      "loss": 0.0003,
       "step": 4448
     },
     {
       "epoch": 9.788778877887788,
-      "grad_norm": 0.0028076171875,
+      "grad_norm": 0.00250244140625,
       "learning_rate": 1.9907282707675167e-07,
       "loss": 0.0001,
       "step": 4449
     },
     {
       "epoch": 9.79097909790979,
-      "grad_norm": 0.0032806396484375,
+      "grad_norm": 0.0020599365234375,
       "learning_rate": 1.947230525005006e-07,
       "loss": 0.0001,
       "step": 4450
     },
     {
       "epoch": 9.793179317931793,
-      "grad_norm": 0.0024566650390625,
+      "grad_norm": 0.00518798828125,
       "learning_rate": 1.9042127964741252e-07,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4451
     },
     {
       "epoch": 9.795379537953796,
-      "grad_norm": 0.003173828125,
+      "grad_norm": 0.0026397705078125,
       "learning_rate": 1.8616751058647685e-07,
       "loss": 0.0001,
       "step": 4452
     },
     {
       "epoch": 9.797579757975798,
-      "grad_norm": 0.042724609375,
+      "grad_norm": 0.003082275390625,
       "learning_rate": 1.8196174736352378e-07,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 4453
     },
     {
       "epoch": 9.7997799779978,
-      "grad_norm": 0.0033111572265625,
+      "grad_norm": 0.004638671875,
       "learning_rate": 1.7780399200134635e-07,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4454
     },
     {
       "epoch": 9.801980198019802,
-      "grad_norm": 0.0045166015625,
+      "grad_norm": 0.0033721923828125,
       "learning_rate": 1.7369424649962274e-07,
       "loss": 0.0001,
       "step": 4455
     },
     {
       "epoch": 9.804180418041804,
-      "grad_norm": 0.0030517578125,
+      "grad_norm": 0.0030975341796875,
       "learning_rate": 1.6963251283497183e-07,
       "loss": 0.0001,
       "step": 4456
     },
     {
       "epoch": 9.806380638063807,
-      "grad_norm": 0.00390625,
+      "grad_norm": 0.00433349609375,
       "learning_rate": 1.6561879296088657e-07,
       "loss": 0.0002,
       "step": 4457
     },
     {
       "epoch": 9.808580858085808,
-      "grad_norm": 0.0096435546875,
+      "grad_norm": 0.01055908203125,
       "learning_rate": 1.616530888077672e-07,
       "loss": 0.0002,
       "step": 4458
     },
     {
       "epoch": 9.81078107810781,
-      "grad_norm": 0.00323486328125,
+      "grad_norm": 0.0030364990234375,
       "learning_rate": 1.577354022829658e-07,
       "loss": 0.0001,
       "step": 4459
     },
     {
       "epoch": 9.812981298129813,
-      "grad_norm": 0.0029449462890625,
+      "grad_norm": 0.0026702880859375,
       "learning_rate": 1.5386573527067515e-07,
       "loss": 0.0001,
       "step": 4460
     },
     {
       "epoch": 9.815181518151816,
-      "grad_norm": 0.0026397705078125,
+      "grad_norm": 0.005157470703125,
       "learning_rate": 1.5004408963205098e-07,
       "loss": 0.0001,
       "step": 4461
     },
     {
       "epoch": 9.817381738173818,
-      "grad_norm": 0.00628662109375,
+      "grad_norm": 0.0036773681640625,
       "learning_rate": 1.4627046720511183e-07,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 4462
     },
     {
       "epoch": 9.819581958195819,
-      "grad_norm": 0.002227783203125,
+      "grad_norm": 0.003448486328125,
       "learning_rate": 1.4254486980480596e-07,
       "loss": 0.0001,
       "step": 4463
     },
     {
       "epoch": 9.821782178217822,
-      "grad_norm": 0.00335693359375,
+      "grad_norm": 0.005157470703125,
       "learning_rate": 1.388672992229667e-07,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4464
     },
     {
       "epoch": 9.823982398239824,
-      "grad_norm": 0.0086669921875,
+      "grad_norm": 0.0025787353515625,
       "learning_rate": 1.3523775722834587e-07,
       "loss": 0.0001,
       "step": 4465
     },
     {
       "epoch": 9.826182618261827,
-      "grad_norm": 0.0030364990234375,
+      "grad_norm": 0.0050048828125,
       "learning_rate": 1.3165624556656931e-07,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4466
     },
     {
       "epoch": 9.828382838283828,
-      "grad_norm": 0.0030670166015625,
+      "grad_norm": 0.006103515625,
       "learning_rate": 1.2812276596019246e-07,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4467
     },
     {
       "epoch": 9.83058305830583,
-      "grad_norm": 0.002777099609375,
+      "grad_norm": 0.0033111572265625,
       "learning_rate": 1.2463732010864482e-07,
       "loss": 0.0001,
       "step": 4468
     },
     {
       "epoch": 9.832783278327833,
-      "grad_norm": 0.0031585693359375,
+      "grad_norm": 0.00592041015625,
       "learning_rate": 1.2119990968828543e-07,
       "loss": 0.0001,
       "step": 4469
     },
     {
       "epoch": 9.834983498349835,
-      "grad_norm": 0.00537109375,
+      "grad_norm": 0.00433349609375,
       "learning_rate": 1.178105363523252e-07,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4470
     },
     {
       "epoch": 9.837183718371836,
-      "grad_norm": 0.00174713134765625,
+      "grad_norm": 0.0035247802734375,
       "learning_rate": 1.1446920173089349e-07,
       "loss": 0.0001,
       "step": 4471
     },
     {
       "epoch": 9.839383938393839,
-      "grad_norm": 0.00634765625,
+      "grad_norm": 0.0037689208984375,
       "learning_rate": 1.1117590743104923e-07,
       "loss": 0.0001,
       "step": 4472
     },
     {
       "epoch": 9.841584158415841,
-      "grad_norm": 0.002655029296875,
+      "grad_norm": 0.00994873046875,
       "learning_rate": 1.0793065503668098e-07,
-      "loss": 0.0001,
+      "loss": 0.0003,
       "step": 4473
     },
     {
       "epoch": 9.843784378437844,
-      "grad_norm": 0.0026092529296875,
+      "grad_norm": 0.00238037109375,
       "learning_rate": 1.0473344610862912e-07,
       "loss": 0.0001,
       "step": 4474
     },
     {
       "epoch": 9.845984598459847,
-      "grad_norm": 0.003753662109375,
+      "grad_norm": 0.002716064453125,
       "learning_rate": 1.0158428218458583e-07,
       "loss": 0.0001,
       "step": 4475
     },
     {
       "epoch": 9.848184818481847,
-      "grad_norm": 0.00347900390625,
+      "grad_norm": 0.0038299560546875,
       "learning_rate": 9.848316477916175e-08,
       "loss": 0.0001,
       "step": 4476
@@ -31669,191 +31669,191 @@
     },
     {
       "epoch": 9.852585258525853,
-      "grad_norm": 0.0037689208984375,
+      "grad_norm": 0.00439453125,
       "learning_rate": 9.242507546703972e-08,
       "loss": 0.0001,
       "step": 4478
     },
     {
       "epoch": 9.854785478547855,
-      "grad_norm": 0.00189971923828125,
+      "grad_norm": 0.0045166015625,
       "learning_rate": 8.946810647399995e-08,
       "loss": 0.0001,
       "step": 4479
     },
     {
       "epoch": 9.856985698569858,
-      "grad_norm": 0.00225830078125,
+      "grad_norm": 0.0028533935546875,
       "learning_rate": 8.655918982689581e-08,
       "loss": 0.0001,
       "step": 4480
     },
     {
       "epoch": 9.859185918591859,
-      "grad_norm": 0.005218505859375,
+      "grad_norm": 0.00909423828125,
       "learning_rate": 8.369832692478597e-08,
       "loss": 0.0002,
       "step": 4481
     },
     {
       "epoch": 9.861386138613861,
-      "grad_norm": 0.00262451171875,
+      "grad_norm": 0.0026092529296875,
       "learning_rate": 8.08855191436142e-08,
       "loss": 0.0001,
       "step": 4482
     },
     {
       "epoch": 9.863586358635864,
-      "grad_norm": 0.00244140625,
+      "grad_norm": 0.0025482177734375,
       "learning_rate": 7.812076783620947e-08,
       "loss": 0.0001,
       "step": 4483
     },
     {
       "epoch": 9.865786578657866,
-      "grad_norm": 0.003570556640625,
+      "grad_norm": 0.003143310546875,
       "learning_rate": 7.540407433228591e-08,
       "loss": 0.0001,
       "step": 4484
     },
     {
       "epoch": 9.867986798679867,
-      "grad_norm": 0.004241943359375,
+      "grad_norm": 0.0028076171875,
       "learning_rate": 7.273543993845389e-08,
       "loss": 0.0001,
       "step": 4485
     },
     {
       "epoch": 9.87018701870187,
-      "grad_norm": 0.0027008056640625,
+      "grad_norm": 0.0035247802734375,
       "learning_rate": 7.011486593818672e-08,
       "loss": 0.0001,
       "step": 4486
     },
     {
       "epoch": 9.872387238723872,
-      "grad_norm": 0.0022125244140625,
+      "grad_norm": 0.0031280517578125,
       "learning_rate": 6.754235359188732e-08,
       "loss": 0.0001,
       "step": 4487
     },
     {
       "epoch": 9.874587458745875,
-      "grad_norm": 0.0023345947265625,
+      "grad_norm": 0.0022125244140625,
       "learning_rate": 6.501790413678821e-08,
       "loss": 0.0001,
       "step": 4488
     },
     {
       "epoch": 9.876787678767876,
-      "grad_norm": 0.0024566650390625,
+      "grad_norm": 0.006622314453125,
       "learning_rate": 6.25415187870515e-08,
       "loss": 0.0001,
       "step": 4489
     },
     {
       "epoch": 9.878987898789878,
-      "grad_norm": 0.00341796875,
+      "grad_norm": 0.003997802734375,
       "learning_rate": 6.011319873370225e-08,
       "loss": 0.0001,
       "step": 4490
     },
     {
       "epoch": 9.881188118811881,
-      "grad_norm": 0.0030364990234375,
+      "grad_norm": 0.005279541015625,
       "learning_rate": 5.773294514462846e-08,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4491
     },
     {
       "epoch": 9.883388338833884,
-      "grad_norm": 0.00250244140625,
+      "grad_norm": 0.0032196044921875,
       "learning_rate": 5.540075916464771e-08,
       "loss": 0.0001,
       "step": 4492
     },
     {
       "epoch": 9.885588558855886,
-      "grad_norm": 0.003021240234375,
+      "grad_norm": 0.001983642578125,
       "learning_rate": 5.311664191542942e-08,
       "loss": 0.0001,
       "step": 4493
     },
     {
       "epoch": 9.887788778877887,
-      "grad_norm": 0.005126953125,
+      "grad_norm": 0.0034027099609375,
       "learning_rate": 5.0880594495517074e-08,
       "loss": 0.0001,
       "step": 4494
     },
     {
       "epoch": 9.88998899889989,
-      "grad_norm": 0.0059814453125,
+      "grad_norm": 0.00830078125,
       "learning_rate": 4.8692617980350406e-08,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4495
     },
     {
       "epoch": 9.892189218921892,
-      "grad_norm": 0.003814697265625,
+      "grad_norm": 0.003936767578125,
       "learning_rate": 4.655271342225431e-08,
       "loss": 0.0001,
       "step": 4496
     },
     {
       "epoch": 9.894389438943895,
-      "grad_norm": 0.0021209716796875,
+      "grad_norm": 0.0033721923828125,
       "learning_rate": 4.4460881850405537e-08,
       "loss": 0.0001,
       "step": 4497
     },
     {
       "epoch": 9.896589658965897,
-      "grad_norm": 0.0034637451171875,
+      "grad_norm": 0.003692626953125,
       "learning_rate": 4.2417124270899277e-08,
       "loss": 0.0001,
       "step": 4498
     },
     {
       "epoch": 9.898789878987898,
-      "grad_norm": 0.0037994384765625,
+      "grad_norm": 0.004638671875,
       "learning_rate": 4.04214416666715e-08,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4499
     },
     {
       "epoch": 9.900990099009901,
-      "grad_norm": 0.003570556640625,
+      "grad_norm": 0.003997802734375,
       "learning_rate": 3.8473834997565516e-08,
       "loss": 0.0001,
       "step": 4500
     },
     {
       "epoch": 9.903190319031903,
-      "grad_norm": 0.005859375,
+      "grad_norm": 0.006591796875,
       "learning_rate": 3.657430520028759e-08,
       "loss": 0.0002,
       "step": 4501
     },
     {
       "epoch": 9.905390539053906,
-      "grad_norm": 0.005767822265625,
+      "grad_norm": 0.01080322265625,
       "learning_rate": 3.4722853188406957e-08,
       "loss": 0.0002,
       "step": 4502
     },
     {
       "epoch": 9.907590759075907,
-      "grad_norm": 0.00537109375,
+      "grad_norm": 0.005279541015625,
       "learning_rate": 3.2919479852411285e-08,
       "loss": 0.0002,
       "step": 4503
     },
     {
       "epoch": 9.90979097909791,
-      "grad_norm": 0.0076904296875,
+      "grad_norm": 0.00592041015625,
       "learning_rate": 3.116418605962901e-08,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4504
     },
     {
@@ -31865,245 +31865,245 @@
     },
     {
       "epoch": 9.914191419141915,
-      "grad_norm": 0.00262451171875,
+      "grad_norm": 0.002777099609375,
       "learning_rate": 2.7797840457433056e-08,
       "loss": 0.0001,
       "step": 4506
     },
     {
       "epoch": 9.916391639163916,
-      "grad_norm": 0.0029144287109375,
+      "grad_norm": 0.0037841796875,
       "learning_rate": 2.618679026707982e-08,
       "loss": 0.0001,
       "step": 4507
     },
     {
       "epoch": 9.918591859185918,
-      "grad_norm": 0.0029754638671875,
+      "grad_norm": 0.00360107421875,
       "learning_rate": 2.462382285804976e-08,
       "loss": 0.0001,
       "step": 4508
     },
     {
       "epoch": 9.92079207920792,
-      "grad_norm": 0.0086669921875,
+      "grad_norm": 0.0022125244140625,
       "learning_rate": 2.310893898206379e-08,
       "loss": 0.0001,
       "step": 4509
     },
     {
       "epoch": 9.922992299229923,
-      "grad_norm": 0.002593994140625,
+      "grad_norm": 0.0029144287109375,
       "learning_rate": 2.164213936770576e-08,
       "loss": 0.0001,
       "step": 4510
     },
     {
       "epoch": 9.925192519251926,
-      "grad_norm": 0.004730224609375,
+      "grad_norm": 0.0025482177734375,
       "learning_rate": 2.0223424720444695e-08,
       "loss": 0.0001,
       "step": 4511
     },
     {
       "epoch": 9.927392739273927,
-      "grad_norm": 0.0026397705078125,
+      "grad_norm": 0.0029144287109375,
       "learning_rate": 1.8852795722601456e-08,
       "loss": 0.0001,
       "step": 4512
     },
     {
       "epoch": 9.92959295929593,
-      "grad_norm": 0.0050048828125,
+      "grad_norm": 0.002899169921875,
       "learning_rate": 1.7530253033404276e-08,
       "loss": 0.0001,
       "step": 4513
     },
     {
       "epoch": 9.931793179317932,
-      "grad_norm": 0.00183868408203125,
+      "grad_norm": 0.00262451171875,
       "learning_rate": 1.6255797288922126e-08,
       "loss": 0.0001,
       "step": 4514
     },
     {
       "epoch": 9.933993399339935,
-      "grad_norm": 0.002899169921875,
+      "grad_norm": 0.0023651123046875,
       "learning_rate": 1.502942910212024e-08,
       "loss": 0.0001,
       "step": 4515
     },
     {
       "epoch": 9.936193619361937,
-      "grad_norm": 0.002349853515625,
+      "grad_norm": 0.002838134765625,
       "learning_rate": 1.3851149062815704e-08,
       "loss": 0.0001,
       "step": 4516
     },
     {
       "epoch": 9.938393839383938,
-      "grad_norm": 0.00390625,
+      "grad_norm": 0.0030364990234375,
       "learning_rate": 1.2720957737710759e-08,
       "loss": 0.0001,
       "step": 4517
     },
     {
       "epoch": 9.94059405940594,
-      "grad_norm": 0.003570556640625,
+      "grad_norm": 0.003204345703125,
       "learning_rate": 1.1638855670370597e-08,
       "loss": 0.0001,
       "step": 4518
     },
     {
       "epoch": 9.942794279427943,
-      "grad_norm": 0.003875732421875,
+      "grad_norm": 0.00372314453125,
       "learning_rate": 1.0604843381245567e-08,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4519
     },
     {
       "epoch": 9.944994499449946,
-      "grad_norm": 0.004058837890625,
+      "grad_norm": 0.010498046875,
       "learning_rate": 9.618921367637867e-09,
-      "loss": 0.0001,
+      "loss": 0.0003,
       "step": 4520
     },
     {
       "epoch": 9.947194719471947,
-      "grad_norm": 0.006927490234375,
+      "grad_norm": 0.0030364990234375,
       "learning_rate": 8.681090103734857e-09,
       "loss": 0.0001,
       "step": 4521
     },
     {
       "epoch": 9.94939493949395,
-      "grad_norm": 0.00274658203125,
+      "grad_norm": 0.005157470703125,
       "learning_rate": 7.791350040586842e-09,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4522
     },
     {
       "epoch": 9.951595159515952,
-      "grad_norm": 0.002593994140625,
+      "grad_norm": 0.0030059814453125,
       "learning_rate": 6.949701606118186e-09,
       "loss": 0.0001,
       "step": 4523
     },
     {
       "epoch": 9.953795379537954,
-      "grad_norm": 0.0030670166015625,
+      "grad_norm": 0.0034027099609375,
       "learning_rate": 6.156145205127306e-09,
       "loss": 0.0001,
       "step": 4524
     },
     {
       "epoch": 9.955995599559955,
-      "grad_norm": 0.0032196044921875,
+      "grad_norm": 0.0027313232421875,
       "learning_rate": 5.410681219286673e-09,
       "loss": 0.0001,
       "step": 4525
     },
     {
       "epoch": 9.958195819581958,
-      "grad_norm": 0.0023345947265625,
+      "grad_norm": 0.002166748046875,
       "learning_rate": 4.713310007109506e-09,
       "loss": 0.0001,
       "step": 4526
     },
     {
       "epoch": 9.96039603960396,
-      "grad_norm": 0.005584716796875,
+      "grad_norm": 0.0079345703125,
       "learning_rate": 4.064031904005283e-09,
       "loss": 0.0002,
       "step": 4527
     },
     {
       "epoch": 9.962596259625963,
-      "grad_norm": 0.0045166015625,
+      "grad_norm": 0.0062255859375,
       "learning_rate": 3.462847222257537e-09,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4528
     },
     {
       "epoch": 9.964796479647966,
-      "grad_norm": 0.0054931640625,
+      "grad_norm": 0.005950927734375,
       "learning_rate": 2.9097562510016497e-09,
       "loss": 0.0001,
       "step": 4529
     },
     {
       "epoch": 9.966996699669966,
-      "grad_norm": 0.0027618408203125,
+      "grad_norm": 0.00518798828125,
       "learning_rate": 2.404759256247058e-09,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4530
     },
     {
       "epoch": 9.969196919691969,
-      "grad_norm": 0.0028533935546875,
+      "grad_norm": 0.001922607421875,
       "learning_rate": 1.9478564808661503e-09,
       "loss": 0.0001,
       "step": 4531
     },
     {
       "epoch": 9.971397139713972,
-      "grad_norm": 0.005462646484375,
+      "grad_norm": 0.004119873046875,
       "learning_rate": 1.539048144627575e-09,
       "loss": 0.0001,
       "step": 4532
     },
     {
       "epoch": 9.973597359735974,
-      "grad_norm": 0.0036773681640625,
+      "grad_norm": 0.007110595703125,
       "learning_rate": 1.1783344441407273e-09,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4533
     },
     {
       "epoch": 9.975797579757975,
-      "grad_norm": 0.003173828125,
+      "grad_norm": 0.00384521484375,
       "learning_rate": 8.657155528779548e-10,
-      "loss": 0.0001,
+      "loss": 0.0002,
       "step": 4534
     },
     {
       "epoch": 9.977997799779978,
-      "grad_norm": 0.0048828125,
+      "grad_norm": 0.00628662109375,
       "learning_rate": 6.011916212189661e-10,
       "loss": 0.0002,
       "step": 4535
     },
     {
       "epoch": 9.98019801980198,
-      "grad_norm": 0.0030975341796875,
+      "grad_norm": 0.00274658203125,
       "learning_rate": 3.847627763731154e-10,
       "loss": 0.0001,
       "step": 4536
     },
     {
       "epoch": 9.982398239823983,
-      "grad_norm": 0.00469970703125,
+      "grad_norm": 0.0037078857421875,
       "learning_rate": 2.164291224238113e-10,
       "loss": 0.0001,
       "step": 4537
     },
     {
       "epoch": 9.984598459845985,
-      "grad_norm": 0.002197265625,
+      "grad_norm": 0.0020599365234375,
       "learning_rate": 9.619074035072117e-11,
       "loss": 0.0001,
       "step": 4538
     },
     {
       "epoch": 9.986798679867986,
-      "grad_norm": 0.0036773681640625,
+      "grad_norm": 0.003448486328125,
       "learning_rate": 2.404768798536239e-11,
-      "loss": 0.0002,
+      "loss": 0.0001,
       "step": 4539
     },
     {
       "epoch": 9.988998899889989,
-      "grad_norm": 0.00592041015625,
+      "grad_norm": 0.005340576171875,
       "learning_rate": 0.0,
       "loss": 0.0002,
       "step": 4540