diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5801 @@
+{
+  "best_metric": 2.344815969467163,
+  "best_model_checkpoint": "outputs/checkpoint-615",
+  "epoch": 1.0,
+  "eval_steps": 205,
+  "global_step": 818,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0012224938875305623,
+      "grad_norm": 8.178321838378906,
+      "learning_rate": 0.001,
+      "loss": 5.6592,
+      "step": 1
+    },
+    {
+      "epoch": 0.0024449877750611247,
+      "grad_norm": 6.750397205352783,
+      "learning_rate": 0.001,
+      "loss": 4.8289,
+      "step": 2
+    },
+    {
+      "epoch": 0.003667481662591687,
+      "grad_norm": 4.012933731079102,
+      "learning_rate": 0.001,
+      "loss": 3.9844,
+      "step": 3
+    },
+    {
+      "epoch": 0.004889975550122249,
+      "grad_norm": 4.729739665985107,
+      "learning_rate": 0.001,
+      "loss": 3.4222,
+      "step": 4
+    },
+    {
+      "epoch": 0.006112469437652812,
+      "grad_norm": 9.694947242736816,
+      "learning_rate": 0.001,
+      "loss": 3.1776,
+      "step": 5
+    },
+    {
+      "epoch": 0.007334963325183374,
+      "grad_norm": 3.631967306137085,
+      "learning_rate": 0.001,
+      "loss": 3.0943,
+      "step": 6
+    },
+    {
+      "epoch": 0.008557457212713936,
+      "grad_norm": 1.8573168516159058,
+      "learning_rate": 0.001,
+      "loss": 3.0438,
+      "step": 7
+    },
+    {
+      "epoch": 0.009779951100244499,
+      "grad_norm": 1.1996679306030273,
+      "learning_rate": 0.001,
+      "loss": 2.9171,
+      "step": 8
+    },
+    {
+      "epoch": 0.011002444987775062,
+      "grad_norm": 6.059974193572998,
+      "learning_rate": 0.001,
+      "loss": 2.896,
+      "step": 9
+    },
+    {
+      "epoch": 0.012224938875305624,
+      "grad_norm": 2.2413384914398193,
+      "learning_rate": 0.001,
+      "loss": 2.8097,
+      "step": 10
+    },
+    {
+      "epoch": 0.013447432762836185,
+      "grad_norm": 3.7154123783111572,
+      "learning_rate": 0.001,
+      "loss": 2.9233,
+      "step": 11
+    },
+    {
+      "epoch": 0.014669926650366748,
+      "grad_norm": 9.796507835388184,
+      "learning_rate": 0.001,
+      "loss": 2.9526,
+      "step": 12
+    },
+    {
+      "epoch": 0.01589242053789731,
+      "grad_norm": 1.3841707706451416,
+      "learning_rate": 0.001,
+      "loss": 2.9088,
+      "step": 13
+    },
+    {
+      "epoch": 0.017114914425427872,
+      "grad_norm": 3.4672791957855225,
+      "learning_rate": 0.001,
+      "loss": 2.7995,
+      "step": 14
+    },
+    {
+      "epoch": 0.018337408312958436,
+      "grad_norm": 0.9586966037750244,
+      "learning_rate": 0.001,
+      "loss": 2.7673,
+      "step": 15
+    },
+    {
+      "epoch": 0.019559902200488997,
+      "grad_norm": 0.747156023979187,
+      "learning_rate": 0.001,
+      "loss": 2.7673,
+      "step": 16
+    },
+    {
+      "epoch": 0.02078239608801956,
+      "grad_norm": 0.7908157110214233,
+      "learning_rate": 0.001,
+      "loss": 2.7809,
+      "step": 17
+    },
+    {
+      "epoch": 0.022004889975550123,
+      "grad_norm": 0.9333175420761108,
+      "learning_rate": 0.001,
+      "loss": 2.738,
+      "step": 18
+    },
+    {
+      "epoch": 0.023227383863080684,
+      "grad_norm": 1.2243553400039673,
+      "learning_rate": 0.001,
+      "loss": 2.6815,
+      "step": 19
+    },
+    {
+      "epoch": 0.02444987775061125,
+      "grad_norm": 0.9989001750946045,
+      "learning_rate": 0.001,
+      "loss": 2.7002,
+      "step": 20
+    },
+    {
+      "epoch": 0.02567237163814181,
+      "grad_norm": 0.664013147354126,
+      "learning_rate": 0.001,
+      "loss": 2.6622,
+      "step": 21
+    },
+    {
+      "epoch": 0.02689486552567237,
+      "grad_norm": 0.8166808485984802,
+      "learning_rate": 0.001,
+      "loss": 2.6829,
+      "step": 22
+    },
+    {
+      "epoch": 0.028117359413202935,
+      "grad_norm": 0.8486875295639038,
+      "learning_rate": 0.001,
+      "loss": 2.7649,
+      "step": 23
+    },
+    {
+      "epoch": 0.029339853300733496,
+      "grad_norm": 0.672136664390564,
+      "learning_rate": 0.001,
+      "loss": 2.5397,
+      "step": 24
+    },
+    {
+      "epoch": 0.030562347188264057,
+      "grad_norm": 0.6852248907089233,
+      "learning_rate": 0.001,
+      "loss": 2.5688,
+      "step": 25
+    },
+    {
+      "epoch": 0.03178484107579462,
+      "grad_norm": 0.6197850704193115,
+      "learning_rate": 0.001,
+      "loss": 2.6037,
+      "step": 26
+    },
+    {
+      "epoch": 0.03300733496332518,
+      "grad_norm": 0.671241044998169,
+      "learning_rate": 0.001,
+      "loss": 2.6554,
+      "step": 27
+    },
+    {
+      "epoch": 0.034229828850855744,
+      "grad_norm": 0.675551176071167,
+      "learning_rate": 0.001,
+      "loss": 2.5556,
+      "step": 28
+    },
+    {
+      "epoch": 0.035452322738386305,
+      "grad_norm": 0.5328904390335083,
+      "learning_rate": 0.001,
+      "loss": 2.5926,
+      "step": 29
+    },
+    {
+      "epoch": 0.03667481662591687,
+      "grad_norm": 0.6622318029403687,
+      "learning_rate": 0.001,
+      "loss": 2.5812,
+      "step": 30
+    },
+    {
+      "epoch": 0.037897310513447434,
+      "grad_norm": 0.7308458685874939,
+      "learning_rate": 0.001,
+      "loss": 2.5887,
+      "step": 31
+    },
+    {
+      "epoch": 0.039119804400977995,
+      "grad_norm": 0.6278252601623535,
+      "learning_rate": 0.001,
+      "loss": 2.6097,
+      "step": 32
+    },
+    {
+      "epoch": 0.040342298288508556,
+      "grad_norm": 0.7257432341575623,
+      "learning_rate": 0.001,
+      "loss": 2.6019,
+      "step": 33
+    },
+    {
+      "epoch": 0.04156479217603912,
+      "grad_norm": 0.5468400120735168,
+      "learning_rate": 0.001,
+      "loss": 2.4808,
+      "step": 34
+    },
+    {
+      "epoch": 0.042787286063569685,
+      "grad_norm": 0.5938531160354614,
+      "learning_rate": 0.001,
+      "loss": 2.6267,
+      "step": 35
+    },
+    {
+      "epoch": 0.044009779951100246,
+      "grad_norm": 0.6380032896995544,
+      "learning_rate": 0.001,
+      "loss": 2.4288,
+      "step": 36
+    },
+    {
+      "epoch": 0.04523227383863081,
+      "grad_norm": 0.5627170205116272,
+      "learning_rate": 0.001,
+      "loss": 2.5154,
+      "step": 37
+    },
+    {
+      "epoch": 0.04645476772616137,
+      "grad_norm": 0.638714075088501,
+      "learning_rate": 0.001,
+      "loss": 2.5662,
+      "step": 38
+    },
+    {
+      "epoch": 0.04767726161369193,
+      "grad_norm": 0.5756145715713501,
+      "learning_rate": 0.001,
+      "loss": 2.5508,
+      "step": 39
+    },
+    {
+      "epoch": 0.0488997555012225,
+      "grad_norm": 0.631705105304718,
+      "learning_rate": 0.001,
+      "loss": 2.4879,
+      "step": 40
+    },
+    {
+      "epoch": 0.05012224938875306,
+      "grad_norm": 0.5856027007102966,
+      "learning_rate": 0.001,
+      "loss": 2.5639,
+      "step": 41
+    },
+    {
+      "epoch": 0.05134474327628362,
+      "grad_norm": 0.5451909303665161,
+      "learning_rate": 0.001,
+      "loss": 2.4391,
+      "step": 42
+    },
+    {
+      "epoch": 0.05256723716381418,
+      "grad_norm": 0.8142032623291016,
+      "learning_rate": 0.001,
+      "loss": 2.6332,
+      "step": 43
+    },
+    {
+      "epoch": 0.05378973105134474,
+      "grad_norm": 0.5957739353179932,
+      "learning_rate": 0.001,
+      "loss": 2.5546,
+      "step": 44
+    },
+    {
+      "epoch": 0.0550122249388753,
+      "grad_norm": 0.5841150283813477,
+      "learning_rate": 0.001,
+      "loss": 2.5132,
+      "step": 45
+    },
+    {
+      "epoch": 0.05623471882640587,
+      "grad_norm": 0.634911298751831,
+      "learning_rate": 0.001,
+      "loss": 2.4678,
+      "step": 46
+    },
+    {
+      "epoch": 0.05745721271393643,
+      "grad_norm": 0.6035284399986267,
+      "learning_rate": 0.001,
+      "loss": 2.5785,
+      "step": 47
+    },
+    {
+      "epoch": 0.05867970660146699,
+      "grad_norm": 0.5425002574920654,
+      "learning_rate": 0.001,
+      "loss": 2.5575,
+      "step": 48
+    },
+    {
+      "epoch": 0.05990220048899755,
+      "grad_norm": 0.6026980876922607,
+      "learning_rate": 0.001,
+      "loss": 2.5258,
+      "step": 49
+    },
+    {
+      "epoch": 0.061124694376528114,
+      "grad_norm": 0.5246846079826355,
+      "learning_rate": 0.001,
+      "loss": 2.4779,
+      "step": 50
+    },
+    {
+      "epoch": 0.06234718826405868,
+      "grad_norm": 0.5958104729652405,
+      "learning_rate": 0.001,
+      "loss": 2.5578,
+      "step": 51
+    },
+    {
+      "epoch": 0.06356968215158924,
+      "grad_norm": 0.5678538084030151,
+      "learning_rate": 0.001,
+      "loss": 2.4066,
+      "step": 52
+    },
+    {
+      "epoch": 0.0647921760391198,
+      "grad_norm": 0.5942307114601135,
+      "learning_rate": 0.001,
+      "loss": 2.4738,
+      "step": 53
+    },
+    {
+      "epoch": 0.06601466992665037,
+      "grad_norm": 0.5238040089607239,
+      "learning_rate": 0.001,
+      "loss": 2.4607,
+      "step": 54
+    },
+    {
+      "epoch": 0.06723716381418093,
+      "grad_norm": 0.511229932308197,
+      "learning_rate": 0.001,
+      "loss": 2.549,
+      "step": 55
+    },
+    {
+      "epoch": 0.06845965770171149,
+      "grad_norm": 0.5474979281425476,
+      "learning_rate": 0.001,
+      "loss": 2.5325,
+      "step": 56
+    },
+    {
+      "epoch": 0.06968215158924206,
+      "grad_norm": 0.5083491206169128,
+      "learning_rate": 0.001,
+      "loss": 2.49,
+      "step": 57
+    },
+    {
+      "epoch": 0.07090464547677261,
+      "grad_norm": 0.4880111813545227,
+      "learning_rate": 0.001,
+      "loss": 2.4353,
+      "step": 58
+    },
+    {
+      "epoch": 0.07212713936430318,
+      "grad_norm": 0.5999699831008911,
+      "learning_rate": 0.001,
+      "loss": 2.4644,
+      "step": 59
+    },
+    {
+      "epoch": 0.07334963325183375,
+      "grad_norm": 0.5478151440620422,
+      "learning_rate": 0.001,
+      "loss": 2.62,
+      "step": 60
+    },
+    {
+      "epoch": 0.0745721271393643,
+      "grad_norm": 0.4989413619041443,
+      "learning_rate": 0.001,
+      "loss": 2.4704,
+      "step": 61
+    },
+    {
+      "epoch": 0.07579462102689487,
+      "grad_norm": 0.6229702830314636,
+      "learning_rate": 0.001,
+      "loss": 2.5979,
+      "step": 62
+    },
+    {
+      "epoch": 0.07701711491442542,
+      "grad_norm": 0.7081549167633057,
+      "learning_rate": 0.001,
+      "loss": 2.4865,
+      "step": 63
+    },
+    {
+      "epoch": 0.07823960880195599,
+      "grad_norm": 0.5195762515068054,
+      "learning_rate": 0.001,
+      "loss": 2.5042,
+      "step": 64
+    },
+    {
+      "epoch": 0.07946210268948656,
+      "grad_norm": 0.5830724835395813,
+      "learning_rate": 0.001,
+      "loss": 2.5001,
+      "step": 65
+    },
+    {
+      "epoch": 0.08068459657701711,
+      "grad_norm": 0.5564406514167786,
+      "learning_rate": 0.001,
+      "loss": 2.5957,
+      "step": 66
+    },
+    {
+      "epoch": 0.08190709046454768,
+      "grad_norm": 0.5549057722091675,
+      "learning_rate": 0.001,
+      "loss": 2.5493,
+      "step": 67
+    },
+    {
+      "epoch": 0.08312958435207823,
+      "grad_norm": 0.5666396617889404,
+      "learning_rate": 0.001,
+      "loss": 2.5333,
+      "step": 68
+    },
+    {
+      "epoch": 0.0843520782396088,
+      "grad_norm": 0.5428187251091003,
+      "learning_rate": 0.001,
+      "loss": 2.5426,
+      "step": 69
+    },
+    {
+      "epoch": 0.08557457212713937,
+      "grad_norm": 0.653217613697052,
+      "learning_rate": 0.001,
+      "loss": 2.4808,
+      "step": 70
+    },
+    {
+      "epoch": 0.08679706601466992,
+      "grad_norm": 0.5372606515884399,
+      "learning_rate": 0.001,
+      "loss": 2.4246,
+      "step": 71
+    },
+    {
+      "epoch": 0.08801955990220049,
+      "grad_norm": 0.6210645437240601,
+      "learning_rate": 0.001,
+      "loss": 2.4826,
+      "step": 72
+    },
+    {
+      "epoch": 0.08924205378973105,
+      "grad_norm": 0.6422852277755737,
+      "learning_rate": 0.001,
+      "loss": 2.5226,
+      "step": 73
+    },
+    {
+      "epoch": 0.09046454767726161,
+      "grad_norm": 0.5775130987167358,
+      "learning_rate": 0.001,
+      "loss": 2.4696,
+      "step": 74
+    },
+    {
+      "epoch": 0.09168704156479218,
+      "grad_norm": 0.5871543884277344,
+      "learning_rate": 0.001,
+      "loss": 2.4852,
+      "step": 75
+    },
+    {
+      "epoch": 0.09290953545232274,
+      "grad_norm": 0.5282509326934814,
+      "learning_rate": 0.001,
+      "loss": 2.4266,
+      "step": 76
+    },
+    {
+      "epoch": 0.0941320293398533,
+      "grad_norm": 0.6395830512046814,
+      "learning_rate": 0.001,
+      "loss": 2.5749,
+      "step": 77
+    },
+    {
+      "epoch": 0.09535452322738386,
+      "grad_norm": 0.5609179139137268,
+      "learning_rate": 0.001,
+      "loss": 2.4805,
+      "step": 78
+    },
+    {
+      "epoch": 0.09657701711491443,
+      "grad_norm": 0.5313189029693604,
+      "learning_rate": 0.001,
+      "loss": 2.4155,
+      "step": 79
+    },
+    {
+      "epoch": 0.097799511002445,
+      "grad_norm": 0.5251710414886475,
+      "learning_rate": 0.001,
+      "loss": 2.4232,
+      "step": 80
+    },
+    {
+      "epoch": 0.09902200488997555,
+      "grad_norm": 0.5261017084121704,
+      "learning_rate": 0.001,
+      "loss": 2.4875,
+      "step": 81
+    },
+    {
+      "epoch": 0.10024449877750612,
+      "grad_norm": 0.5420725345611572,
+      "learning_rate": 0.001,
+      "loss": 2.4924,
+      "step": 82
+    },
+    {
+      "epoch": 0.10146699266503667,
+      "grad_norm": 0.5187302827835083,
+      "learning_rate": 0.001,
+      "loss": 2.3575,
+      "step": 83
+    },
+    {
+      "epoch": 0.10268948655256724,
+      "grad_norm": 0.5224878787994385,
+      "learning_rate": 0.001,
+      "loss": 2.6074,
+      "step": 84
+    },
+    {
+      "epoch": 0.1039119804400978,
+      "grad_norm": 0.5518419146537781,
+      "learning_rate": 0.001,
+      "loss": 2.4174,
+      "step": 85
+    },
+    {
+      "epoch": 0.10513447432762836,
+      "grad_norm": 0.4974496066570282,
+      "learning_rate": 0.001,
+      "loss": 2.425,
+      "step": 86
+    },
+    {
+      "epoch": 0.10635696821515893,
+      "grad_norm": 0.47749748826026917,
+      "learning_rate": 0.001,
+      "loss": 2.468,
+      "step": 87
+    },
+    {
+      "epoch": 0.10757946210268948,
+      "grad_norm": 0.5429526567459106,
+      "learning_rate": 0.001,
+      "loss": 2.4249,
+      "step": 88
+    },
+    {
+      "epoch": 0.10880195599022005,
+      "grad_norm": 0.4692726135253906,
+      "learning_rate": 0.001,
+      "loss": 2.4903,
+      "step": 89
+    },
+    {
+      "epoch": 0.1100244498777506,
+      "grad_norm": 0.5202171206474304,
+      "learning_rate": 0.001,
+      "loss": 2.4415,
+      "step": 90
+    },
+    {
+      "epoch": 0.11124694376528117,
+      "grad_norm": 0.5242786407470703,
+      "learning_rate": 0.001,
+      "loss": 2.5025,
+      "step": 91
+    },
+    {
+      "epoch": 0.11246943765281174,
+      "grad_norm": 0.5316799879074097,
+      "learning_rate": 0.001,
+      "loss": 2.4626,
+      "step": 92
+    },
+    {
+      "epoch": 0.1136919315403423,
+      "grad_norm": 0.5226518511772156,
+      "learning_rate": 0.001,
+      "loss": 2.3977,
+      "step": 93
+    },
+    {
+      "epoch": 0.11491442542787286,
+      "grad_norm": 0.5194711685180664,
+      "learning_rate": 0.001,
+      "loss": 2.4447,
+      "step": 94
+    },
+    {
+      "epoch": 0.11613691931540342,
+      "grad_norm": 0.5524601340293884,
+      "learning_rate": 0.001,
+      "loss": 2.3604,
+      "step": 95
+    },
+    {
+      "epoch": 0.11735941320293398,
+      "grad_norm": 0.4714466333389282,
+      "learning_rate": 0.001,
+      "loss": 2.4618,
+      "step": 96
+    },
+    {
+      "epoch": 0.11858190709046455,
+      "grad_norm": 0.4986712336540222,
+      "learning_rate": 0.001,
+      "loss": 2.551,
+      "step": 97
+    },
+    {
+      "epoch": 0.1198044009779951,
+      "grad_norm": 0.5235204100608826,
+      "learning_rate": 0.001,
+      "loss": 2.5354,
+      "step": 98
+    },
+    {
+      "epoch": 0.12102689486552567,
+      "grad_norm": 0.5750053524971008,
+      "learning_rate": 0.001,
+      "loss": 2.4457,
+      "step": 99
+    },
+    {
+      "epoch": 0.12224938875305623,
+      "grad_norm": 0.5225120782852173,
+      "learning_rate": 0.001,
+      "loss": 2.4616,
+      "step": 100
+    },
+    {
+      "epoch": 0.1234718826405868,
+      "grad_norm": 0.5144570469856262,
+      "learning_rate": 0.001,
+      "loss": 2.4833,
+      "step": 101
+    },
+    {
+      "epoch": 0.12469437652811736,
+      "grad_norm": 0.5372641086578369,
+      "learning_rate": 0.001,
+      "loss": 2.504,
+      "step": 102
+    },
+    {
+      "epoch": 0.12591687041564792,
+      "grad_norm": 0.5242504477500916,
+      "learning_rate": 0.001,
+      "loss": 2.4074,
+      "step": 103
+    },
+    {
+      "epoch": 0.1271393643031785,
+      "grad_norm": 0.5856664180755615,
+      "learning_rate": 0.001,
+      "loss": 2.5,
+      "step": 104
+    },
+    {
+      "epoch": 0.12836185819070906,
+      "grad_norm": 0.5015215873718262,
+      "learning_rate": 0.001,
+      "loss": 2.4654,
+      "step": 105
+    },
+    {
+      "epoch": 0.1295843520782396,
+      "grad_norm": 0.5485916137695312,
+      "learning_rate": 0.001,
+      "loss": 2.4536,
+      "step": 106
+    },
+    {
+      "epoch": 0.13080684596577016,
+      "grad_norm": 0.4952264726161957,
+      "learning_rate": 0.001,
+      "loss": 2.449,
+      "step": 107
+    },
+    {
+      "epoch": 0.13202933985330073,
+      "grad_norm": 0.5012094378471375,
+      "learning_rate": 0.001,
+      "loss": 2.5035,
+      "step": 108
+    },
+    {
+      "epoch": 0.1332518337408313,
+      "grad_norm": 0.5141774415969849,
+      "learning_rate": 0.001,
+      "loss": 2.403,
+      "step": 109
+    },
+    {
+      "epoch": 0.13447432762836187,
+      "grad_norm": 0.5255733728408813,
+      "learning_rate": 0.001,
+      "loss": 2.385,
+      "step": 110
+    },
+    {
+      "epoch": 0.1356968215158924,
+      "grad_norm": 0.5168445706367493,
+      "learning_rate": 0.001,
+      "loss": 2.4347,
+      "step": 111
+    },
+    {
+      "epoch": 0.13691931540342298,
+      "grad_norm": 0.5361398458480835,
+      "learning_rate": 0.001,
+      "loss": 2.4392,
+      "step": 112
+    },
+    {
+      "epoch": 0.13814180929095354,
+      "grad_norm": 0.48438146710395813,
+      "learning_rate": 0.001,
+      "loss": 2.5366,
+      "step": 113
+    },
+    {
+      "epoch": 0.1393643031784841,
+      "grad_norm": 0.5347726941108704,
+      "learning_rate": 0.001,
+      "loss": 2.5162,
+      "step": 114
+    },
+    {
+      "epoch": 0.14058679706601468,
+      "grad_norm": 0.5615862011909485,
+      "learning_rate": 0.001,
+      "loss": 2.4167,
+      "step": 115
+    },
+    {
+      "epoch": 0.14180929095354522,
+      "grad_norm": 0.4931885004043579,
+      "learning_rate": 0.001,
+      "loss": 2.414,
+      "step": 116
+    },
+    {
+      "epoch": 0.1430317848410758,
+      "grad_norm": 0.5923001766204834,
+      "learning_rate": 0.001,
+      "loss": 2.568,
+      "step": 117
+    },
+    {
+      "epoch": 0.14425427872860636,
+      "grad_norm": 0.5969946980476379,
+      "learning_rate": 0.001,
+      "loss": 2.424,
+      "step": 118
+    },
+    {
+      "epoch": 0.14547677261613692,
+      "grad_norm": 0.5627629160881042,
+      "learning_rate": 0.001,
+      "loss": 2.4501,
+      "step": 119
+    },
+    {
+      "epoch": 0.1466992665036675,
+      "grad_norm": 0.5513372421264648,
+      "learning_rate": 0.001,
+      "loss": 2.4966,
+      "step": 120
+    },
+    {
+      "epoch": 0.14792176039119803,
+      "grad_norm": 0.5943652391433716,
+      "learning_rate": 0.001,
+      "loss": 2.5369,
+      "step": 121
+    },
+    {
+      "epoch": 0.1491442542787286,
+      "grad_norm": 0.5254310369491577,
+      "learning_rate": 0.001,
+      "loss": 2.5201,
+      "step": 122
+    },
+    {
+      "epoch": 0.15036674816625917,
+      "grad_norm": 0.5547128319740295,
+      "learning_rate": 0.001,
+      "loss": 2.4959,
+      "step": 123
+    },
+    {
+      "epoch": 0.15158924205378974,
+      "grad_norm": 0.48801299929618835,
+      "learning_rate": 0.001,
+      "loss": 2.4478,
+      "step": 124
+    },
+    {
+      "epoch": 0.1528117359413203,
+      "grad_norm": 0.5650225877761841,
+      "learning_rate": 0.001,
+      "loss": 2.4411,
+      "step": 125
+    },
+    {
+      "epoch": 0.15403422982885084,
+      "grad_norm": 0.5800032615661621,
+      "learning_rate": 0.001,
+      "loss": 2.3194,
+      "step": 126
+    },
+    {
+      "epoch": 0.1552567237163814,
+      "grad_norm": 0.517168402671814,
+      "learning_rate": 0.001,
+      "loss": 2.4967,
+      "step": 127
+    },
+    {
+      "epoch": 0.15647921760391198,
+      "grad_norm": 0.5031977295875549,
+      "learning_rate": 0.001,
+      "loss": 2.5614,
+      "step": 128
+    },
+    {
+      "epoch": 0.15770171149144255,
+      "grad_norm": 0.5302630066871643,
+      "learning_rate": 0.001,
+      "loss": 2.4458,
+      "step": 129
+    },
+    {
+      "epoch": 0.15892420537897312,
+      "grad_norm": 0.47516071796417236,
+      "learning_rate": 0.001,
+      "loss": 2.4412,
+      "step": 130
+    },
+    {
+      "epoch": 0.16014669926650366,
+      "grad_norm": 0.4758307635784149,
+      "learning_rate": 0.001,
+      "loss": 2.4896,
+      "step": 131
+    },
+    {
+      "epoch": 0.16136919315403422,
+      "grad_norm": 0.5344830751419067,
+      "learning_rate": 0.001,
+      "loss": 2.4147,
+      "step": 132
+    },
+    {
+      "epoch": 0.1625916870415648,
+      "grad_norm": 0.528596043586731,
+      "learning_rate": 0.001,
+      "loss": 2.5358,
+      "step": 133
+    },
+    {
+      "epoch": 0.16381418092909536,
+      "grad_norm": 0.47966817021369934,
+      "learning_rate": 0.001,
+      "loss": 2.4143,
+      "step": 134
+    },
+    {
+      "epoch": 0.16503667481662593,
+      "grad_norm": 0.6001800298690796,
+      "learning_rate": 0.001,
+      "loss": 2.363,
+      "step": 135
+    },
+    {
+      "epoch": 0.16625916870415647,
+      "grad_norm": 0.5011780261993408,
+      "learning_rate": 0.001,
+      "loss": 2.4008,
+      "step": 136
+    },
+    {
+      "epoch": 0.16748166259168704,
+      "grad_norm": 0.5605891346931458,
+      "learning_rate": 0.001,
+      "loss": 2.3927,
+      "step": 137
+    },
+    {
+      "epoch": 0.1687041564792176,
+      "grad_norm": 0.48609134554862976,
+      "learning_rate": 0.001,
+      "loss": 2.4455,
+      "step": 138
+    },
+    {
+      "epoch": 0.16992665036674817,
+      "grad_norm": 0.5741631388664246,
+      "learning_rate": 0.001,
+      "loss": 2.4371,
+      "step": 139
+    },
+    {
+      "epoch": 0.17114914425427874,
+      "grad_norm": 0.49061235785484314,
+      "learning_rate": 0.001,
+      "loss": 2.495,
+      "step": 140
+    },
+    {
+      "epoch": 0.17237163814180928,
+      "grad_norm": 0.48857757449150085,
+      "learning_rate": 0.001,
+      "loss": 2.4461,
+      "step": 141
+    },
+    {
+      "epoch": 0.17359413202933985,
+      "grad_norm": 0.6730465292930603,
+      "learning_rate": 0.001,
+      "loss": 2.5839,
+      "step": 142
+    },
+    {
+      "epoch": 0.17481662591687042,
+      "grad_norm": 0.5764455795288086,
+      "learning_rate": 0.001,
+      "loss": 2.4067,
+      "step": 143
+    },
+    {
+      "epoch": 0.17603911980440098,
+      "grad_norm": 0.5319432020187378,
+      "learning_rate": 0.001,
+      "loss": 2.4074,
+      "step": 144
+    },
+    {
+      "epoch": 0.17726161369193155,
+      "grad_norm": 0.5144186019897461,
+      "learning_rate": 0.001,
+      "loss": 2.3889,
+      "step": 145
+    },
+    {
+      "epoch": 0.1784841075794621,
+      "grad_norm": 0.6696956157684326,
+      "learning_rate": 0.001,
+      "loss": 2.4969,
+      "step": 146
+    },
+    {
+      "epoch": 0.17970660146699266,
+      "grad_norm": 0.5184170007705688,
+      "learning_rate": 0.001,
+      "loss": 2.3741,
+      "step": 147
+    },
+    {
+      "epoch": 0.18092909535452323,
+      "grad_norm": 0.4999801218509674,
+      "learning_rate": 0.001,
+      "loss": 2.3909,
+      "step": 148
+    },
+    {
+      "epoch": 0.1821515892420538,
+      "grad_norm": 0.5894871950149536,
+      "learning_rate": 0.001,
+      "loss": 2.4628,
+      "step": 149
+    },
+    {
+      "epoch": 0.18337408312958436,
+      "grad_norm": 0.5019111633300781,
+      "learning_rate": 0.001,
+      "loss": 2.5056,
+      "step": 150
+    },
+    {
+      "epoch": 0.1845965770171149,
+      "grad_norm": 0.636766791343689,
+      "learning_rate": 0.001,
+      "loss": 2.4687,
+      "step": 151
+    },
+    {
+      "epoch": 0.18581907090464547,
+      "grad_norm": 0.5625051259994507,
+      "learning_rate": 0.001,
+      "loss": 2.5754,
+      "step": 152
+    },
+    {
+      "epoch": 0.18704156479217604,
+      "grad_norm": 0.5557712912559509,
+      "learning_rate": 0.001,
+      "loss": 2.4302,
+      "step": 153
+    },
+    {
+      "epoch": 0.1882640586797066,
+      "grad_norm": 0.5352074503898621,
+      "learning_rate": 0.001,
+      "loss": 2.3118,
+      "step": 154
+    },
+    {
+      "epoch": 0.18948655256723718,
+      "grad_norm": 0.5917800664901733,
+      "learning_rate": 0.001,
+      "loss": 2.4828,
+      "step": 155
+    },
+    {
+      "epoch": 0.19070904645476772,
+      "grad_norm": 0.49687594175338745,
+      "learning_rate": 0.001,
+      "loss": 2.4071,
+      "step": 156
+    },
+    {
+      "epoch": 0.19193154034229828,
+      "grad_norm": 0.5303874611854553,
+      "learning_rate": 0.001,
+      "loss": 2.4907,
+      "step": 157
+    },
+    {
+      "epoch": 0.19315403422982885,
+      "grad_norm": 0.5479294061660767,
+      "learning_rate": 0.001,
+      "loss": 2.4308,
+      "step": 158
+    },
+    {
+      "epoch": 0.19437652811735942,
+      "grad_norm": 0.56096351146698,
+      "learning_rate": 0.001,
+      "loss": 2.4961,
+      "step": 159
+    },
+    {
+      "epoch": 0.19559902200489,
+      "grad_norm": 0.5233320593833923,
+      "learning_rate": 0.001,
+      "loss": 2.4017,
+      "step": 160
+    },
+    {
+      "epoch": 0.19682151589242053,
+      "grad_norm": 0.540540337562561,
+      "learning_rate": 0.001,
+      "loss": 2.3942,
+      "step": 161
+    },
+    {
+      "epoch": 0.1980440097799511,
+      "grad_norm": 0.5374789834022522,
+      "learning_rate": 0.001,
+      "loss": 2.4976,
+      "step": 162
+    },
+    {
+      "epoch": 0.19926650366748166,
+      "grad_norm": 0.5012461543083191,
+      "learning_rate": 0.001,
+      "loss": 2.4017,
+      "step": 163
+    },
+    {
+      "epoch": 0.20048899755501223,
+      "grad_norm": 0.5607964396476746,
+      "learning_rate": 0.001,
+      "loss": 2.3732,
+      "step": 164
+    },
+    {
+      "epoch": 0.2017114914425428,
+      "grad_norm": 0.48764991760253906,
+      "learning_rate": 0.001,
+      "loss": 2.4003,
+      "step": 165
+    },
+    {
+      "epoch": 0.20293398533007334,
+      "grad_norm": 0.5440587997436523,
+      "learning_rate": 0.001,
+      "loss": 2.4653,
+      "step": 166
+    },
+    {
+      "epoch": 0.2041564792176039,
+      "grad_norm": 0.5021042227745056,
+      "learning_rate": 0.001,
+      "loss": 2.3921,
+      "step": 167
+    },
+    {
+      "epoch": 0.20537897310513448,
+      "grad_norm": 0.5661401748657227,
+      "learning_rate": 0.001,
+      "loss": 2.471,
+      "step": 168
+    },
+    {
+      "epoch": 0.20660146699266504,
+      "grad_norm": 0.5360035300254822,
+      "learning_rate": 0.001,
+      "loss": 2.5091,
+      "step": 169
+    },
+    {
+      "epoch": 0.2078239608801956,
+      "grad_norm": 0.5328943133354187,
+      "learning_rate": 0.001,
+      "loss": 2.3627,
+      "step": 170
+    },
+    {
+      "epoch": 0.20904645476772615,
+      "grad_norm": 0.5662642121315002,
+      "learning_rate": 0.001,
+      "loss": 2.4411,
+      "step": 171
+    },
+    {
+      "epoch": 0.21026894865525672,
+      "grad_norm": 0.4912776052951813,
+      "learning_rate": 0.001,
+      "loss": 2.3941,
+      "step": 172
+    },
+    {
+      "epoch": 0.2114914425427873,
+      "grad_norm": 0.5288503766059875,
+      "learning_rate": 0.001,
+      "loss": 2.4138,
+      "step": 173
+    },
+    {
+      "epoch": 0.21271393643031786,
+      "grad_norm": 0.5616488456726074,
+      "learning_rate": 0.001,
+      "loss": 2.5105,
+      "step": 174
+    },
+    {
+      "epoch": 0.2139364303178484,
+      "grad_norm": 0.5832340717315674,
+      "learning_rate": 0.001,
+      "loss": 2.5503,
+      "step": 175
+    },
+    {
+      "epoch": 0.21515892420537897,
+      "grad_norm": 0.49610635638237,
+      "learning_rate": 0.001,
+      "loss": 2.4249,
+      "step": 176
+    },
+    {
+      "epoch": 0.21638141809290953,
+      "grad_norm": 0.5436179637908936,
+      "learning_rate": 0.001,
+      "loss": 2.3545,
+      "step": 177
+    },
+    {
+      "epoch": 0.2176039119804401,
+      "grad_norm": 0.586430549621582,
+      "learning_rate": 0.001,
+      "loss": 2.418,
+      "step": 178
+    },
+    {
+      "epoch": 0.21882640586797067,
+      "grad_norm": 0.5396440029144287,
+      "learning_rate": 0.001,
+      "loss": 2.3565,
+      "step": 179
+    },
+    {
+      "epoch": 0.2200488997555012,
+      "grad_norm": 0.5660167336463928,
+      "learning_rate": 0.001,
+      "loss": 2.4284,
+      "step": 180
+    },
+    {
+      "epoch": 0.22127139364303178,
+      "grad_norm": 0.5947809219360352,
+      "learning_rate": 0.001,
+      "loss": 2.3546,
+      "step": 181
+    },
+    {
+      "epoch": 0.22249388753056235,
+      "grad_norm": 0.5506652593612671,
+      "learning_rate": 0.001,
+      "loss": 2.3467,
+      "step": 182
+    },
+    {
+      "epoch": 0.2237163814180929,
+      "grad_norm": 0.6263591647148132,
+      "learning_rate": 0.001,
+      "loss": 2.4552,
+      "step": 183
+    },
+    {
+      "epoch": 0.22493887530562348,
+      "grad_norm": 0.5154738426208496,
+      "learning_rate": 0.001,
+      "loss": 2.4508,
+      "step": 184
+    },
+    {
+      "epoch": 0.22616136919315402,
+      "grad_norm": 0.6300072073936462,
+      "learning_rate": 0.001,
+      "loss": 2.4095,
+      "step": 185
+    },
+    {
+      "epoch": 0.2273838630806846,
+      "grad_norm": 0.5022745132446289,
+      "learning_rate": 0.001,
+      "loss": 2.423,
+      "step": 186
+    },
+    {
+      "epoch": 0.22860635696821516,
+      "grad_norm": 0.599410355091095,
+      "learning_rate": 0.001,
+      "loss": 2.3954,
+      "step": 187
+    },
+    {
+      "epoch": 0.22982885085574573,
+      "grad_norm": 0.5577519536018372,
+      "learning_rate": 0.001,
+      "loss": 2.3529,
+      "step": 188
+    },
+    {
+      "epoch": 0.2310513447432763,
+      "grad_norm": 0.5671312808990479,
+      "learning_rate": 0.001,
+      "loss": 2.4023,
+      "step": 189
+    },
+    {
+      "epoch": 0.23227383863080683,
+      "grad_norm": 0.5201452970504761,
+      "learning_rate": 0.001,
+      "loss": 2.4185,
+      "step": 190
+    },
+    {
+      "epoch": 0.2334963325183374,
+      "grad_norm": 0.5384425520896912,
+      "learning_rate": 0.001,
+      "loss": 2.4863,
+      "step": 191
+    },
+    {
+      "epoch": 0.23471882640586797,
+      "grad_norm": 0.4962243139743805,
+      "learning_rate": 0.001,
+      "loss": 2.5203,
+      "step": 192
+    },
+    {
+      "epoch": 0.23594132029339854,
+      "grad_norm": 0.5636964440345764,
+      "learning_rate": 0.001,
+      "loss": 2.4688,
+      "step": 193
+    },
+    {
+      "epoch": 0.2371638141809291,
+      "grad_norm": 0.538328230381012,
+      "learning_rate": 0.001,
+      "loss": 2.3808,
+      "step": 194
+    },
+    {
+      "epoch": 0.23838630806845965,
+      "grad_norm": 0.5091468691825867,
+      "learning_rate": 0.001,
+      "loss": 2.3791,
+      "step": 195
+    },
+    {
+      "epoch": 0.2396088019559902,
+      "grad_norm": 0.5237581729888916,
+      "learning_rate": 0.001,
+      "loss": 2.3898,
+      "step": 196
+    },
+    {
+      "epoch": 0.24083129584352078,
+      "grad_norm": 0.510797381401062,
+      "learning_rate": 0.001,
+      "loss": 2.4203,
+      "step": 197
+    },
+    {
+      "epoch": 0.24205378973105135,
+      "grad_norm": 0.5363763570785522,
+      "learning_rate": 0.001,
+      "loss": 2.4296,
+      "step": 198
+    },
+    {
+      "epoch": 0.24327628361858192,
+      "grad_norm": 0.5146409869194031,
+      "learning_rate": 0.001,
+      "loss": 2.5014,
+      "step": 199
+    },
+    {
+      "epoch": 0.24449877750611246,
+      "grad_norm": 0.4853112995624542,
+      "learning_rate": 0.001,
+      "loss": 2.4751,
+      "step": 200
+    },
+    {
+      "epoch": 0.24572127139364303,
+      "grad_norm": 0.5601246356964111,
+      "learning_rate": 0.001,
+      "loss": 2.4001,
+      "step": 201
+    },
+    {
+      "epoch": 0.2469437652811736,
+      "grad_norm": 0.5031394362449646,
+      "learning_rate": 0.001,
+      "loss": 2.3763,
+      "step": 202
+    },
+    {
+      "epoch": 0.24816625916870416,
+      "grad_norm": 0.5693891644477844,
+      "learning_rate": 0.001,
+      "loss": 2.3937,
+      "step": 203
+    },
+    {
+      "epoch": 0.24938875305623473,
+      "grad_norm": 0.5505186319351196,
+      "learning_rate": 0.001,
+      "loss": 2.3517,
+      "step": 204
+    },
+    {
+      "epoch": 0.2506112469437653,
+      "grad_norm": 0.5346199870109558,
+      "learning_rate": 0.001,
+      "loss": 2.4259,
+      "step": 205
+    },
+    {
+      "epoch": 0.2506112469437653,
+      "eval_loss": 2.4109721183776855,
+      "eval_runtime": 44.1933,
+      "eval_samples_per_second": 236.801,
+      "eval_steps_per_second": 29.62,
+      "step": 205
+    },
+    {
+      "epoch": 0.25183374083129584,
+      "grad_norm": 0.5326516032218933,
+      "learning_rate": 0.001,
+      "loss": 2.3483,
+      "step": 206
+    },
+    {
+      "epoch": 0.2530562347188264,
+      "grad_norm": 0.522074818611145,
+      "learning_rate": 0.001,
+      "loss": 2.3972,
+      "step": 207
+    },
+    {
+      "epoch": 0.254278728606357,
+      "grad_norm": 0.5074703097343445,
+      "learning_rate": 0.001,
+      "loss": 2.4478,
+      "step": 208
+    },
+    {
+      "epoch": 0.2555012224938875,
+      "grad_norm": 0.5311232805252075,
+      "learning_rate": 0.001,
+      "loss": 2.5704,
+      "step": 209
+    },
+    {
+      "epoch": 0.2567237163814181,
+      "grad_norm": 0.6050538420677185,
+      "learning_rate": 0.001,
+      "loss": 2.4128,
+      "step": 210
+    },
+    {
+      "epoch": 0.25794621026894865,
+      "grad_norm": 0.5808090567588806,
+      "learning_rate": 0.001,
+      "loss": 2.4556,
+      "step": 211
+    },
+    {
+      "epoch": 0.2591687041564792,
+      "grad_norm": 0.6892769932746887,
+      "learning_rate": 0.001,
+      "loss": 2.4149,
+      "step": 212
+    },
+    {
+      "epoch": 0.2603911980440098,
+      "grad_norm": 0.5970401167869568,
+      "learning_rate": 0.001,
+      "loss": 2.4635,
+      "step": 213
+    },
+    {
+      "epoch": 0.2616136919315403,
+      "grad_norm": 0.5463938117027283,
+      "learning_rate": 0.001,
+      "loss": 2.4833,
+      "step": 214
+    },
+    {
+      "epoch": 0.2628361858190709,
+      "grad_norm": 0.5739268660545349,
+      "learning_rate": 0.001,
+      "loss": 2.3873,
+      "step": 215
+    },
+    {
+      "epoch": 0.26405867970660146,
+      "grad_norm": 0.5724747776985168,
+      "learning_rate": 0.001,
+      "loss": 2.4843,
+      "step": 216
+    },
+    {
+      "epoch": 0.265281173594132,
+      "grad_norm": 0.6063271760940552,
+      "learning_rate": 0.001,
+      "loss": 2.4753,
+      "step": 217
+    },
+    {
+      "epoch": 0.2665036674816626,
+      "grad_norm": 0.5265262722969055,
+      "learning_rate": 0.001,
+      "loss": 2.4289,
+      "step": 218
+    },
+    {
+      "epoch": 0.26772616136919314,
+      "grad_norm": 0.6319820880889893,
+      "learning_rate": 0.001,
+      "loss": 2.3428,
+      "step": 219
+    },
+    {
+      "epoch": 0.26894865525672373,
+      "grad_norm": 0.6463683247566223,
+      "learning_rate": 0.001,
+      "loss": 2.4524,
+      "step": 220
+    },
+    {
+      "epoch": 0.2701711491442543,
+      "grad_norm": 0.5888187885284424,
+      "learning_rate": 0.001,
+      "loss": 2.4852,
+      "step": 221
+    },
+    {
+      "epoch": 0.2713936430317848,
+      "grad_norm": 0.6894919276237488,
+      "learning_rate": 0.001,
+      "loss": 2.4537,
+      "step": 222
+    },
+    {
+      "epoch": 0.2726161369193154,
+      "grad_norm": 0.5876211524009705,
+      "learning_rate": 0.001,
+      "loss": 2.3938,
+      "step": 223
+    },
+    {
+      "epoch": 0.27383863080684595,
+      "grad_norm": 0.5431022644042969,
+      "learning_rate": 0.001,
+      "loss": 2.3515,
+      "step": 224
+    },
+    {
+      "epoch": 0.27506112469437655,
+      "grad_norm": 0.5700949430465698,
+      "learning_rate": 0.001,
+      "loss": 2.2915,
+      "step": 225
+    },
+    {
+      "epoch": 0.2762836185819071,
+      "grad_norm": 0.5603176355361938,
+      "learning_rate": 0.001,
+      "loss": 2.3743,
+      "step": 226
+    },
+    {
+      "epoch": 0.2775061124694376,
+      "grad_norm": 0.6375773549079895,
+      "learning_rate": 0.001,
+      "loss": 2.5753,
+      "step": 227
+    },
+    {
+      "epoch": 0.2787286063569682,
+      "grad_norm": 0.5627942085266113,
+      "learning_rate": 0.001,
+      "loss": 2.308,
+      "step": 228
+    },
+    {
+      "epoch": 0.27995110024449876,
+      "grad_norm": 0.5948708653450012,
+      "learning_rate": 0.001,
+      "loss": 2.3451,
+      "step": 229
+    },
+    {
+      "epoch": 0.28117359413202936,
+      "grad_norm": 0.48628291487693787,
+      "learning_rate": 0.001,
+      "loss": 2.4578,
+      "step": 230
+    },
+    {
+      "epoch": 0.2823960880195599,
+      "grad_norm": 0.651255190372467,
+      "learning_rate": 0.001,
+      "loss": 2.4205,
+      "step": 231
+    },
+    {
+      "epoch": 0.28361858190709044,
+      "grad_norm": 0.5403560400009155,
+      "learning_rate": 0.001,
+      "loss": 2.4456,
+      "step": 232
+    },
+    {
+      "epoch": 0.28484107579462103,
+      "grad_norm": 0.5174692869186401,
+      "learning_rate": 0.001,
+      "loss": 2.3616,
+      "step": 233
+    },
+    {
+      "epoch": 0.2860635696821516,
+      "grad_norm": 0.6177205443382263,
+      "learning_rate": 0.001,
+      "loss": 2.5222,
+      "step": 234
+    },
+    {
+      "epoch": 0.28728606356968217,
+      "grad_norm": 0.5286775231361389,
+      "learning_rate": 0.001,
+      "loss": 2.4493,
+      "step": 235
+    },
+    {
+      "epoch": 0.2885085574572127,
+      "grad_norm": 0.5577534437179565,
+      "learning_rate": 0.001,
+      "loss": 2.4203,
+      "step": 236
+    },
+    {
+      "epoch": 0.28973105134474325,
+      "grad_norm": 0.5461423397064209,
+      "learning_rate": 0.001,
+      "loss": 2.4714,
+      "step": 237
+    },
+    {
+      "epoch": 0.29095354523227385,
+      "grad_norm": 0.5145041346549988,
+      "learning_rate": 0.001,
+      "loss": 2.4164,
+      "step": 238
+    },
+    {
+      "epoch": 0.2921760391198044,
+      "grad_norm": 0.5773872137069702,
+      "learning_rate": 0.001,
+      "loss": 2.4079,
+      "step": 239
+    },
+    {
+      "epoch": 0.293398533007335,
+      "grad_norm": 0.5076879858970642,
+      "learning_rate": 0.001,
+      "loss": 2.4656,
+      "step": 240
+    },
+    {
+      "epoch": 0.2946210268948655,
+      "grad_norm": 0.5377123951911926,
+      "learning_rate": 0.001,
+      "loss": 2.4747,
+      "step": 241
+    },
+    {
+      "epoch": 0.29584352078239606,
+      "grad_norm": 0.537397563457489,
+      "learning_rate": 0.001,
+      "loss": 2.4301,
+      "step": 242
+    },
+    {
+      "epoch": 0.29706601466992666,
+      "grad_norm": 0.5414688587188721,
+      "learning_rate": 0.001,
+      "loss": 2.4186,
+      "step": 243
+    },
+    {
+      "epoch": 0.2982885085574572,
+      "grad_norm": 0.5407987833023071,
+      "learning_rate": 0.001,
+      "loss": 2.4053,
+      "step": 244
+    },
+    {
+      "epoch": 0.2995110024449878,
+      "grad_norm": 0.6226968169212341,
+      "learning_rate": 0.001,
+      "loss": 2.2887,
+      "step": 245
+    },
+    {
+      "epoch": 0.30073349633251834,
+      "grad_norm": 0.5672168135643005,
+      "learning_rate": 0.001,
+      "loss": 2.4579,
+      "step": 246
+    },
+    {
+      "epoch": 0.3019559902200489,
+      "grad_norm": 0.5122395157814026,
+      "learning_rate": 0.001,
+      "loss": 2.3413,
+      "step": 247
+    },
+    {
+      "epoch": 0.30317848410757947,
+      "grad_norm": 0.5855796337127686,
+      "learning_rate": 0.001,
+      "loss": 2.3533,
+      "step": 248
+    },
+    {
+      "epoch": 0.30440097799511,
+      "grad_norm": 0.5746926069259644,
+      "learning_rate": 0.001,
+      "loss": 2.4499,
+      "step": 249
+    },
+    {
+      "epoch": 0.3056234718826406,
+      "grad_norm": 0.5400431752204895,
+      "learning_rate": 0.001,
+      "loss": 2.3914,
+      "step": 250
+    },
+    {
+      "epoch": 0.30684596577017115,
+      "grad_norm": 0.571844756603241,
+      "learning_rate": 0.001,
+      "loss": 2.3883,
+      "step": 251
+    },
+    {
+      "epoch": 0.3080684596577017,
+      "grad_norm": 0.5428387522697449,
+      "learning_rate": 0.001,
+      "loss": 2.3792,
+      "step": 252
+    },
+    {
+      "epoch": 0.3092909535452323,
+      "grad_norm": 0.5149421095848083,
+      "learning_rate": 0.001,
+      "loss": 2.3698,
+      "step": 253
+    },
+    {
+      "epoch": 0.3105134474327628,
+      "grad_norm": 0.8051120638847351,
+      "learning_rate": 0.001,
+      "loss": 2.4721,
+      "step": 254
+    },
+    {
+      "epoch": 0.3117359413202934,
+      "grad_norm": 0.5592311024665833,
+      "learning_rate": 0.001,
+      "loss": 2.4215,
+      "step": 255
+    },
+    {
+      "epoch": 0.31295843520782396,
+      "grad_norm": 0.6142417788505554,
+      "learning_rate": 0.001,
+      "loss": 2.4274,
+      "step": 256
+    },
+    {
+      "epoch": 0.3141809290953545,
+      "grad_norm": 0.5631799697875977,
+      "learning_rate": 0.001,
+      "loss": 2.4201,
+      "step": 257
+    },
+    {
+      "epoch": 0.3154034229828851,
+      "grad_norm": 0.5501461029052734,
+      "learning_rate": 0.001,
+      "loss": 2.481,
+      "step": 258
+    },
+    {
+      "epoch": 0.31662591687041564,
+      "grad_norm": 0.5580623149871826,
+      "learning_rate": 0.001,
+      "loss": 2.4626,
+      "step": 259
+    },
+    {
+      "epoch": 0.31784841075794623,
+      "grad_norm": 0.5429579019546509,
+      "learning_rate": 0.001,
+      "loss": 2.4345,
+      "step": 260
+    },
+    {
+      "epoch": 0.31907090464547677,
+      "grad_norm": 0.5621469020843506,
+      "learning_rate": 0.001,
+      "loss": 2.4714,
+      "step": 261
+    },
+    {
+      "epoch": 0.3202933985330073,
+      "grad_norm": 0.6367106437683105,
+      "learning_rate": 0.001,
+      "loss": 2.4289,
+      "step": 262
+    },
+    {
+      "epoch": 0.3215158924205379,
+      "grad_norm": 0.5402144193649292,
+      "learning_rate": 0.001,
+      "loss": 2.4761,
+      "step": 263
+    },
+    {
+      "epoch": 0.32273838630806845,
+      "grad_norm": 0.5153480172157288,
+      "learning_rate": 0.001,
+      "loss": 2.3916,
+      "step": 264
+    },
+    {
+      "epoch": 0.32396088019559904,
+      "grad_norm": 0.554441511631012,
+      "learning_rate": 0.001,
+      "loss": 2.4412,
+      "step": 265
+    },
+    {
+      "epoch": 0.3251833740831296,
+      "grad_norm": 0.5480029582977295,
+      "learning_rate": 0.001,
+      "loss": 2.3595,
+      "step": 266
+    },
+    {
+      "epoch": 0.3264058679706601,
+      "grad_norm": 0.6597166657447815,
+      "learning_rate": 0.001,
+      "loss": 2.434,
+      "step": 267
+    },
+    {
+      "epoch": 0.3276283618581907,
+      "grad_norm": 0.6232181787490845,
+      "learning_rate": 0.001,
+      "loss": 2.438,
+      "step": 268
+    },
+    {
+      "epoch": 0.32885085574572126,
+      "grad_norm": 0.5220692157745361,
+      "learning_rate": 0.001,
+      "loss": 2.3218,
+      "step": 269
+    },
+    {
+      "epoch": 0.33007334963325186,
+      "grad_norm": 0.5414378046989441,
+      "learning_rate": 0.001,
+      "loss": 2.449,
+      "step": 270
+    },
+    {
+      "epoch": 0.3312958435207824,
+      "grad_norm": 0.5560607314109802,
+      "learning_rate": 0.001,
+      "loss": 2.4541,
+      "step": 271
+    },
+    {
+      "epoch": 0.33251833740831294,
+      "grad_norm": 0.6442712545394897,
+      "learning_rate": 0.001,
+      "loss": 2.4726,
+      "step": 272
+    },
+    {
+      "epoch": 0.33374083129584353,
+      "grad_norm": 0.5586593151092529,
+      "learning_rate": 0.001,
+      "loss": 2.4483,
+      "step": 273
+    },
+    {
+      "epoch": 0.33496332518337407,
+      "grad_norm": 0.5124000906944275,
+      "learning_rate": 0.001,
+      "loss": 2.3708,
+      "step": 274
+    },
+    {
+      "epoch": 0.33618581907090467,
+      "grad_norm": 0.5178530812263489,
+      "learning_rate": 0.001,
+      "loss": 2.4624,
+      "step": 275
+    },
+    {
+      "epoch": 0.3374083129584352,
+      "grad_norm": 0.5543733835220337,
+      "learning_rate": 0.001,
+      "loss": 2.4235,
+      "step": 276
+    },
+    {
+      "epoch": 0.33863080684596575,
+      "grad_norm": 0.4837280213832855,
+      "learning_rate": 0.001,
+      "loss": 2.3998,
+      "step": 277
+    },
+    {
+      "epoch": 0.33985330073349634,
+      "grad_norm": 0.5489963293075562,
+      "learning_rate": 0.001,
+      "loss": 2.3875,
+      "step": 278
+    },
+    {
+      "epoch": 0.3410757946210269,
+      "grad_norm": 0.4998355209827423,
+      "learning_rate": 0.001,
+      "loss": 2.3404,
+      "step": 279
+    },
+    {
+      "epoch": 0.3422982885085575,
+      "grad_norm": 0.5590275526046753,
+      "learning_rate": 0.001,
+      "loss": 2.4129,
+      "step": 280
+    },
+    {
+      "epoch": 0.343520782396088,
+      "grad_norm": 0.5230002403259277,
+      "learning_rate": 0.001,
+      "loss": 2.4659,
+      "step": 281
+    },
+    {
+      "epoch": 0.34474327628361856,
+      "grad_norm": 0.5316391587257385,
+      "learning_rate": 0.001,
+      "loss": 2.4093,
+      "step": 282
+    },
+    {
+      "epoch": 0.34596577017114916,
+      "grad_norm": 0.5131670832633972,
+      "learning_rate": 0.001,
+      "loss": 2.4951,
+      "step": 283
+    },
+    {
+      "epoch": 0.3471882640586797,
+      "grad_norm": 0.6025881767272949,
+      "learning_rate": 0.001,
+      "loss": 2.3711,
+      "step": 284
+    },
+    {
+      "epoch": 0.3484107579462103,
+      "grad_norm": 0.5566193461418152,
+      "learning_rate": 0.001,
+      "loss": 2.4284,
+      "step": 285
+    },
+    {
+      "epoch": 0.34963325183374083,
+      "grad_norm": 0.5089775323867798,
+      "learning_rate": 0.001,
+      "loss": 2.3437,
+      "step": 286
+    },
+    {
+      "epoch": 0.3508557457212714,
+      "grad_norm": 0.5217143297195435,
+      "learning_rate": 0.001,
+      "loss": 2.3794,
+      "step": 287
+    },
+    {
+      "epoch": 0.35207823960880197,
+      "grad_norm": 0.5089473128318787,
+      "learning_rate": 0.001,
+      "loss": 2.4371,
+      "step": 288
+    },
+    {
+      "epoch": 0.3533007334963325,
+      "grad_norm": 0.6327386498451233,
+      "learning_rate": 0.001,
+      "loss": 2.4685,
+      "step": 289
+    },
+    {
+      "epoch": 0.3545232273838631,
+      "grad_norm": 0.5239629745483398,
+      "learning_rate": 0.001,
+      "loss": 2.3723,
+      "step": 290
+    },
+    {
+      "epoch": 0.35574572127139364,
+      "grad_norm": 0.5306011438369751,
+      "learning_rate": 0.001,
+      "loss": 2.3622,
+      "step": 291
+    },
+    {
+      "epoch": 0.3569682151589242,
+      "grad_norm": 0.5014710426330566,
+      "learning_rate": 0.001,
+      "loss": 2.3371,
+      "step": 292
+    },
+    {
+      "epoch": 0.3581907090464548,
+      "grad_norm": 0.5310878157615662,
+      "learning_rate": 0.001,
+      "loss": 2.4087,
+      "step": 293
+    },
+    {
+      "epoch": 0.3594132029339853,
+      "grad_norm": 0.5552276968955994,
+      "learning_rate": 0.001,
+      "loss": 2.4021,
+      "step": 294
+    },
+    {
+      "epoch": 0.3606356968215159,
+      "grad_norm": 0.5808470845222473,
+      "learning_rate": 0.001,
+      "loss": 2.4378,
+      "step": 295
+    },
+    {
+      "epoch": 0.36185819070904646,
+      "grad_norm": 0.5333114266395569,
+      "learning_rate": 0.001,
+      "loss": 2.5238,
+      "step": 296
+    },
+    {
+      "epoch": 0.363080684596577,
+      "grad_norm": 0.5774756073951721,
+      "learning_rate": 0.001,
+      "loss": 2.48,
+      "step": 297
+    },
+    {
+      "epoch": 0.3643031784841076,
+      "grad_norm": 0.5121151804924011,
+      "learning_rate": 0.001,
+      "loss": 2.3687,
+      "step": 298
+    },
+    {
+      "epoch": 0.36552567237163813,
+      "grad_norm": 0.5714420676231384,
+      "learning_rate": 0.001,
+      "loss": 2.4214,
+      "step": 299
+    },
+    {
+      "epoch": 0.36674816625916873,
+      "grad_norm": 0.5754179358482361,
+      "learning_rate": 0.001,
+      "loss": 2.4247,
+      "step": 300
+    },
+    {
+      "epoch": 0.36797066014669927,
+      "grad_norm": 0.7446674704551697,
+      "learning_rate": 0.001,
+      "loss": 2.4925,
+      "step": 301
+    },
+    {
+      "epoch": 0.3691931540342298,
+      "grad_norm": 0.5484544038772583,
+      "learning_rate": 0.001,
+      "loss": 2.3883,
+      "step": 302
+    },
+    {
+      "epoch": 0.3704156479217604,
+      "grad_norm": 0.5274435877799988,
+      "learning_rate": 0.001,
+      "loss": 2.4063,
+      "step": 303
+    },
+    {
+      "epoch": 0.37163814180929094,
+      "grad_norm": 0.5567039847373962,
+      "learning_rate": 0.001,
+      "loss": 2.4273,
+      "step": 304
+    },
+    {
+      "epoch": 0.37286063569682154,
+      "grad_norm": 0.5994822978973389,
+      "learning_rate": 0.001,
+      "loss": 2.5011,
+      "step": 305
+    },
+    {
+      "epoch": 0.3740831295843521,
+      "grad_norm": 0.5415587425231934,
+      "learning_rate": 0.001,
+      "loss": 2.3553,
+      "step": 306
+    },
+    {
+      "epoch": 0.3753056234718826,
+      "grad_norm": 0.6581827998161316,
+      "learning_rate": 0.001,
+      "loss": 2.4837,
+      "step": 307
+    },
+    {
+      "epoch": 0.3765281173594132,
+      "grad_norm": 0.6114663481712341,
+      "learning_rate": 0.001,
+      "loss": 2.4078,
+      "step": 308
+    },
+    {
+      "epoch": 0.37775061124694376,
+      "grad_norm": 0.5403485894203186,
+      "learning_rate": 0.001,
+      "loss": 2.3983,
+      "step": 309
+    },
+    {
+      "epoch": 0.37897310513447435,
+      "grad_norm": 0.5375344157218933,
+      "learning_rate": 0.001,
+      "loss": 2.392,
+      "step": 310
+    },
+    {
+      "epoch": 0.3801955990220049,
+      "grad_norm": 0.5954229831695557,
+      "learning_rate": 0.001,
+      "loss": 2.3351,
+      "step": 311
+    },
+    {
+      "epoch": 0.38141809290953543,
+      "grad_norm": 0.6141201257705688,
+      "learning_rate": 0.001,
+      "loss": 2.4802,
+      "step": 312
+    },
+    {
+      "epoch": 0.38264058679706603,
+      "grad_norm": 0.5843420028686523,
+      "learning_rate": 0.001,
+      "loss": 2.3837,
+      "step": 313
+    },
+    {
+      "epoch": 0.38386308068459657,
+      "grad_norm": 0.5854500532150269,
+      "learning_rate": 0.001,
+      "loss": 2.4691,
+      "step": 314
+    },
+    {
+      "epoch": 0.38508557457212717,
+      "grad_norm": 0.49812009930610657,
+      "learning_rate": 0.001,
+      "loss": 2.2868,
+      "step": 315
+    },
+    {
+      "epoch": 0.3863080684596577,
+      "grad_norm": 0.5608780980110168,
+      "learning_rate": 0.001,
+      "loss": 2.3501,
+      "step": 316
+    },
+    {
+      "epoch": 0.38753056234718825,
+      "grad_norm": 0.6470975279808044,
+      "learning_rate": 0.001,
+      "loss": 2.3845,
+      "step": 317
+    },
+    {
+      "epoch": 0.38875305623471884,
+      "grad_norm": 0.5103499293327332,
+      "learning_rate": 0.001,
+      "loss": 2.3544,
+      "step": 318
+    },
+    {
+      "epoch": 0.3899755501222494,
+      "grad_norm": 0.5402243733406067,
+      "learning_rate": 0.001,
+      "loss": 2.494,
+      "step": 319
+    },
+    {
+      "epoch": 0.39119804400978,
+      "grad_norm": 0.5098109245300293,
+      "learning_rate": 0.001,
+      "loss": 2.3771,
+      "step": 320
+    },
+    {
+      "epoch": 0.3924205378973105,
+      "grad_norm": 0.5166260600090027,
+      "learning_rate": 0.001,
+      "loss": 2.4165,
+      "step": 321
+    },
+    {
+      "epoch": 0.39364303178484106,
+      "grad_norm": 0.47534164786338806,
+      "learning_rate": 0.001,
+      "loss": 2.376,
+      "step": 322
+    },
+    {
+      "epoch": 0.39486552567237165,
+      "grad_norm": 0.5538984537124634,
+      "learning_rate": 0.001,
+      "loss": 2.377,
+      "step": 323
+    },
+    {
+      "epoch": 0.3960880195599022,
+      "grad_norm": 0.516322910785675,
+      "learning_rate": 0.001,
+      "loss": 2.4509,
+      "step": 324
+    },
+    {
+      "epoch": 0.3973105134474328,
+      "grad_norm": 0.6167128682136536,
+      "learning_rate": 0.001,
+      "loss": 2.3846,
+      "step": 325
+    },
+    {
+      "epoch": 0.39853300733496333,
+      "grad_norm": 0.5723311305046082,
+      "learning_rate": 0.001,
+      "loss": 2.3715,
+      "step": 326
+    },
+    {
+      "epoch": 0.39975550122249387,
+      "grad_norm": 0.5759738087654114,
+      "learning_rate": 0.001,
+      "loss": 2.3472,
+      "step": 327
+    },
+    {
+      "epoch": 0.40097799511002447,
+      "grad_norm": 0.6038229465484619,
+      "learning_rate": 0.001,
+      "loss": 2.4291,
+      "step": 328
+    },
+    {
+      "epoch": 0.402200488997555,
+      "grad_norm": 0.534037709236145,
+      "learning_rate": 0.001,
+      "loss": 2.4265,
+      "step": 329
+    },
+    {
+      "epoch": 0.4034229828850856,
+      "grad_norm": 0.59047931432724,
+      "learning_rate": 0.001,
+      "loss": 2.421,
+      "step": 330
+    },
+    {
+      "epoch": 0.40464547677261614,
+      "grad_norm": 0.5716297626495361,
+      "learning_rate": 0.001,
+      "loss": 2.4336,
+      "step": 331
+    },
+    {
+      "epoch": 0.4058679706601467,
+      "grad_norm": 0.6059308052062988,
+      "learning_rate": 0.001,
+      "loss": 2.4722,
+      "step": 332
+    },
+    {
+      "epoch": 0.4070904645476773,
+      "grad_norm": 0.5313076376914978,
+      "learning_rate": 0.001,
+      "loss": 2.4054,
+      "step": 333
+    },
+    {
+      "epoch": 0.4083129584352078,
+      "grad_norm": 0.5419294834136963,
+      "learning_rate": 0.001,
+      "loss": 2.3978,
+      "step": 334
+    },
+    {
+      "epoch": 0.4095354523227384,
+      "grad_norm": 0.5434341430664062,
+      "learning_rate": 0.001,
+      "loss": 2.423,
+      "step": 335
+    },
+    {
+      "epoch": 0.41075794621026895,
+      "grad_norm": 0.5907472968101501,
+      "learning_rate": 0.001,
+      "loss": 2.399,
+      "step": 336
+    },
+    {
+      "epoch": 0.4119804400977995,
+      "grad_norm": 0.5154129266738892,
+      "learning_rate": 0.001,
+      "loss": 2.4192,
+      "step": 337
+    },
+    {
+      "epoch": 0.4132029339853301,
+      "grad_norm": 0.5305936336517334,
+      "learning_rate": 0.001,
+      "loss": 2.3631,
+      "step": 338
+    },
+    {
+      "epoch": 0.41442542787286063,
+      "grad_norm": 0.5667030215263367,
+      "learning_rate": 0.001,
+      "loss": 2.4021,
+      "step": 339
+    },
+    {
+      "epoch": 0.4156479217603912,
+      "grad_norm": 0.602198600769043,
+      "learning_rate": 0.001,
+      "loss": 2.3771,
+      "step": 340
+    },
+    {
+      "epoch": 0.41687041564792177,
+      "grad_norm": 0.5672926902770996,
+      "learning_rate": 0.001,
+      "loss": 2.4006,
+      "step": 341
+    },
+    {
+      "epoch": 0.4180929095354523,
+      "grad_norm": 0.5844594836235046,
+      "learning_rate": 0.001,
+      "loss": 2.3617,
+      "step": 342
+    },
+    {
+      "epoch": 0.4193154034229829,
+      "grad_norm": 0.5684323906898499,
+      "learning_rate": 0.001,
+      "loss": 2.4159,
+      "step": 343
+    },
+    {
+      "epoch": 0.42053789731051344,
+      "grad_norm": 0.6692149639129639,
+      "learning_rate": 0.001,
+      "loss": 2.4121,
+      "step": 344
+    },
+    {
+      "epoch": 0.421760391198044,
+      "grad_norm": 0.5522905588150024,
+      "learning_rate": 0.001,
+      "loss": 2.5351,
+      "step": 345
+    },
+    {
+      "epoch": 0.4229828850855746,
+      "grad_norm": 0.5908128619194031,
+      "learning_rate": 0.001,
+      "loss": 2.303,
+      "step": 346
+    },
+    {
+      "epoch": 0.4242053789731051,
+      "grad_norm": 0.6019754409790039,
+      "learning_rate": 0.001,
+      "loss": 2.4212,
+      "step": 347
+    },
+    {
+      "epoch": 0.4254278728606357,
+      "grad_norm": 0.5244648456573486,
+      "learning_rate": 0.001,
+      "loss": 2.4506,
+      "step": 348
+    },
+    {
+      "epoch": 0.42665036674816625,
+      "grad_norm": 0.6195573806762695,
+      "learning_rate": 0.001,
+      "loss": 2.4397,
+      "step": 349
+    },
+    {
+      "epoch": 0.4278728606356968,
+      "grad_norm": 0.5599300265312195,
+      "learning_rate": 0.001,
+      "loss": 2.4369,
+      "step": 350
+    },
+    {
+      "epoch": 0.4290953545232274,
+      "grad_norm": 0.5785163640975952,
+      "learning_rate": 0.001,
+      "loss": 2.2845,
+      "step": 351
+    },
+    {
+      "epoch": 0.43031784841075793,
+      "grad_norm": 0.5526220202445984,
+      "learning_rate": 0.001,
+      "loss": 2.3835,
+      "step": 352
+    },
+    {
+      "epoch": 0.4315403422982885,
+      "grad_norm": 0.520535409450531,
+      "learning_rate": 0.001,
+      "loss": 2.4043,
+      "step": 353
+    },
+    {
+      "epoch": 0.43276283618581907,
+      "grad_norm": 0.5346786975860596,
+      "learning_rate": 0.001,
+      "loss": 2.4634,
+      "step": 354
+    },
+    {
+      "epoch": 0.4339853300733496,
+      "grad_norm": 0.5633550882339478,
+      "learning_rate": 0.001,
+      "loss": 2.4424,
+      "step": 355
+    },
+    {
+      "epoch": 0.4352078239608802,
+      "grad_norm": 0.5584668517112732,
+      "learning_rate": 0.001,
+      "loss": 2.4128,
+      "step": 356
+    },
+    {
+      "epoch": 0.43643031784841074,
+      "grad_norm": 0.5257803201675415,
+      "learning_rate": 0.001,
+      "loss": 2.3762,
+      "step": 357
+    },
+    {
+      "epoch": 0.43765281173594134,
+      "grad_norm": 0.5566312670707703,
+      "learning_rate": 0.001,
+      "loss": 2.3918,
+      "step": 358
+    },
+    {
+      "epoch": 0.4388753056234719,
+      "grad_norm": 0.514863908290863,
+      "learning_rate": 0.001,
+      "loss": 2.3162,
+      "step": 359
+    },
+    {
+      "epoch": 0.4400977995110024,
+      "grad_norm": 0.5604429841041565,
+      "learning_rate": 0.001,
+      "loss": 2.4515,
+      "step": 360
+    },
+    {
+      "epoch": 0.441320293398533,
+      "grad_norm": 0.5394683480262756,
+      "learning_rate": 0.001,
+      "loss": 2.3196,
+      "step": 361
+    },
+    {
+      "epoch": 0.44254278728606355,
+      "grad_norm": 0.6114275455474854,
+      "learning_rate": 0.001,
+      "loss": 2.3253,
+      "step": 362
+    },
+    {
+      "epoch": 0.44376528117359415,
+      "grad_norm": 0.55677330493927,
+      "learning_rate": 0.001,
+      "loss": 2.3898,
+      "step": 363
+    },
+    {
+      "epoch": 0.4449877750611247,
+      "grad_norm": 0.6122755408287048,
+      "learning_rate": 0.001,
+      "loss": 2.4332,
+      "step": 364
+    },
+    {
+      "epoch": 0.44621026894865523,
+      "grad_norm": 0.5454927682876587,
+      "learning_rate": 0.001,
+      "loss": 2.3705,
+      "step": 365
+    },
+    {
+      "epoch": 0.4474327628361858,
+      "grad_norm": 0.6068079471588135,
+      "learning_rate": 0.001,
+      "loss": 2.2946,
+      "step": 366
+    },
+    {
+      "epoch": 0.44865525672371637,
+      "grad_norm": 0.542259693145752,
+      "learning_rate": 0.001,
+      "loss": 2.3949,
+      "step": 367
+    },
+    {
+      "epoch": 0.44987775061124696,
+      "grad_norm": 0.5911988615989685,
+      "learning_rate": 0.001,
+      "loss": 2.479,
+      "step": 368
+    },
+    {
+      "epoch": 0.4511002444987775,
+      "grad_norm": 0.5436906814575195,
+      "learning_rate": 0.001,
+      "loss": 2.4562,
+      "step": 369
+    },
+    {
+      "epoch": 0.45232273838630804,
+      "grad_norm": 0.6720252633094788,
+      "learning_rate": 0.001,
+      "loss": 2.3401,
+      "step": 370
+    },
+    {
+      "epoch": 0.45354523227383864,
+      "grad_norm": 0.594752848148346,
+      "learning_rate": 0.001,
+      "loss": 2.3364,
+      "step": 371
+    },
+    {
+      "epoch": 0.4547677261613692,
+      "grad_norm": 0.5525220036506653,
+      "learning_rate": 0.001,
+      "loss": 2.3593,
+      "step": 372
+    },
+    {
+      "epoch": 0.4559902200488998,
+      "grad_norm": 0.5798346996307373,
+      "learning_rate": 0.001,
+      "loss": 2.3677,
+      "step": 373
+    },
+    {
+      "epoch": 0.4572127139364303,
+      "grad_norm": 0.5627158880233765,
+      "learning_rate": 0.001,
+      "loss": 2.3556,
+      "step": 374
+    },
+    {
+      "epoch": 0.45843520782396086,
+      "grad_norm": 0.5486916303634644,
+      "learning_rate": 0.001,
+      "loss": 2.3581,
+      "step": 375
+    },
+    {
+      "epoch": 0.45965770171149145,
+      "grad_norm": 0.515575110912323,
+      "learning_rate": 0.001,
+      "loss": 2.2755,
+      "step": 376
+    },
+    {
+      "epoch": 0.460880195599022,
+      "grad_norm": 0.5921189785003662,
+      "learning_rate": 0.001,
+      "loss": 2.3344,
+      "step": 377
+    },
+    {
+      "epoch": 0.4621026894865526,
+      "grad_norm": 0.5230167508125305,
+      "learning_rate": 0.001,
+      "loss": 2.3135,
+      "step": 378
+    },
+    {
+      "epoch": 0.4633251833740831,
+      "grad_norm": 0.5792455673217773,
+      "learning_rate": 0.001,
+      "loss": 2.3844,
+      "step": 379
+    },
+    {
+      "epoch": 0.46454767726161367,
+      "grad_norm": 0.5649365186691284,
+      "learning_rate": 0.001,
+      "loss": 2.4243,
+      "step": 380
+    },
+    {
+      "epoch": 0.46577017114914426,
+      "grad_norm": 0.5909812450408936,
+      "learning_rate": 0.001,
+      "loss": 2.4008,
+      "step": 381
+    },
+    {
+      "epoch": 0.4669926650366748,
+      "grad_norm": 0.6039749383926392,
+      "learning_rate": 0.001,
+      "loss": 2.4104,
+      "step": 382
+    },
+    {
+      "epoch": 0.4682151589242054,
+      "grad_norm": 0.6151771545410156,
+      "learning_rate": 0.001,
+      "loss": 2.4914,
+      "step": 383
+    },
+    {
+      "epoch": 0.46943765281173594,
+      "grad_norm": 0.6254723072052002,
+      "learning_rate": 0.001,
+      "loss": 2.3747,
+      "step": 384
+    },
+    {
+      "epoch": 0.4706601466992665,
+      "grad_norm": 0.5194308161735535,
+      "learning_rate": 0.001,
+      "loss": 2.3765,
+      "step": 385
+    },
+    {
+      "epoch": 0.4718826405867971,
+      "grad_norm": 0.5091795325279236,
+      "learning_rate": 0.001,
+      "loss": 2.4028,
+      "step": 386
+    },
+    {
+      "epoch": 0.4731051344743276,
+      "grad_norm": 0.5679914951324463,
+      "learning_rate": 0.001,
+      "loss": 2.4091,
+      "step": 387
+    },
+    {
+      "epoch": 0.4743276283618582,
+      "grad_norm": 0.6551725268363953,
+      "learning_rate": 0.001,
+      "loss": 2.4814,
+      "step": 388
+    },
+    {
+      "epoch": 0.47555012224938875,
+      "grad_norm": 0.5748315453529358,
+      "learning_rate": 0.001,
+      "loss": 2.4518,
+      "step": 389
+    },
+    {
+      "epoch": 0.4767726161369193,
+      "grad_norm": 0.5252829194068909,
+      "learning_rate": 0.001,
+      "loss": 2.5021,
+      "step": 390
+    },
+    {
+      "epoch": 0.4779951100244499,
+      "grad_norm": 0.6099783182144165,
+      "learning_rate": 0.001,
+      "loss": 2.4139,
+      "step": 391
+    },
+    {
+      "epoch": 0.4792176039119804,
+      "grad_norm": 0.5559167265892029,
+      "learning_rate": 0.001,
+      "loss": 2.3817,
+      "step": 392
+    },
+    {
+      "epoch": 0.480440097799511,
+      "grad_norm": 0.5587165951728821,
+      "learning_rate": 0.001,
+      "loss": 2.4002,
+      "step": 393
+    },
+    {
+      "epoch": 0.48166259168704156,
+      "grad_norm": 0.5515249967575073,
+      "learning_rate": 0.001,
+      "loss": 2.4173,
+      "step": 394
+    },
+    {
+      "epoch": 0.4828850855745721,
+      "grad_norm": 0.5275049209594727,
+      "learning_rate": 0.001,
+      "loss": 2.4235,
+      "step": 395
+    },
+    {
+      "epoch": 0.4841075794621027,
+      "grad_norm": 0.519877552986145,
+      "learning_rate": 0.001,
+      "loss": 2.3802,
+      "step": 396
+    },
+    {
+      "epoch": 0.48533007334963324,
+      "grad_norm": 0.5368367433547974,
+      "learning_rate": 0.001,
+      "loss": 2.4164,
+      "step": 397
+    },
+    {
+      "epoch": 0.48655256723716384,
+      "grad_norm": 0.5479859113693237,
+      "learning_rate": 0.001,
+      "loss": 2.4451,
+      "step": 398
+    },
+    {
+      "epoch": 0.4877750611246944,
+      "grad_norm": 0.524018406867981,
+      "learning_rate": 0.001,
+      "loss": 2.2618,
+      "step": 399
+    },
+    {
+      "epoch": 0.4889975550122249,
+      "grad_norm": 0.5853855609893799,
+      "learning_rate": 0.001,
+      "loss": 2.4095,
+      "step": 400
+    },
+    {
+      "epoch": 0.4902200488997555,
+      "grad_norm": 0.5248914957046509,
+      "learning_rate": 0.001,
+      "loss": 2.3441,
+      "step": 401
+    },
+    {
+      "epoch": 0.49144254278728605,
+      "grad_norm": 0.5809511542320251,
+      "learning_rate": 0.001,
+      "loss": 2.4463,
+      "step": 402
+    },
+    {
+      "epoch": 0.49266503667481665,
+      "grad_norm": 0.5517978668212891,
+      "learning_rate": 0.001,
+      "loss": 2.4001,
+      "step": 403
+    },
+    {
+      "epoch": 0.4938875305623472,
+      "grad_norm": 0.5410706400871277,
+      "learning_rate": 0.001,
+      "loss": 2.3972,
+      "step": 404
+    },
+    {
+      "epoch": 0.49511002444987773,
+      "grad_norm": 0.5287546515464783,
+      "learning_rate": 0.001,
+      "loss": 2.3371,
+      "step": 405
+    },
+    {
+      "epoch": 0.4963325183374083,
+      "grad_norm": 0.5337257981300354,
+      "learning_rate": 0.001,
+      "loss": 2.3417,
+      "step": 406
+    },
+    {
+      "epoch": 0.49755501222493886,
+      "grad_norm": 0.5736370086669922,
+      "learning_rate": 0.001,
+      "loss": 2.3409,
+      "step": 407
+    },
+    {
+      "epoch": 0.49877750611246946,
+      "grad_norm": 0.5621830224990845,
+      "learning_rate": 0.001,
+      "loss": 2.3743,
+      "step": 408
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5526315569877625,
+      "learning_rate": 0.001,
+      "loss": 2.4956,
+      "step": 409
+    },
+    {
+      "epoch": 0.5012224938875306,
+      "grad_norm": 0.513525128364563,
+      "learning_rate": 0.001,
+      "loss": 2.3512,
+      "step": 410
+    },
+    {
+      "epoch": 0.5012224938875306,
+      "eval_loss": 2.371687889099121,
+      "eval_runtime": 44.3741,
+      "eval_samples_per_second": 235.836,
+      "eval_steps_per_second": 29.499,
+      "step": 410
+    },
+    {
+      "epoch": 0.5024449877750611,
+      "grad_norm": 0.5600430965423584,
+      "learning_rate": 0.001,
+      "loss": 2.3617,
+      "step": 411
+    },
+    {
+      "epoch": 0.5036674816625917,
+      "grad_norm": 0.5299474000930786,
+      "learning_rate": 0.001,
+      "loss": 2.3518,
+      "step": 412
+    },
+    {
+      "epoch": 0.5048899755501223,
+      "grad_norm": 0.6220374703407288,
+      "learning_rate": 0.001,
+      "loss": 2.4092,
+      "step": 413
+    },
+    {
+      "epoch": 0.5061124694376528,
+      "grad_norm": 0.5975641012191772,
+      "learning_rate": 0.001,
+      "loss": 2.4272,
+      "step": 414
+    },
+    {
+      "epoch": 0.5073349633251834,
+      "grad_norm": 0.57210773229599,
+      "learning_rate": 0.001,
+      "loss": 2.4094,
+      "step": 415
+    },
+    {
+      "epoch": 0.508557457212714,
+      "grad_norm": 0.5716942548751831,
+      "learning_rate": 0.001,
+      "loss": 2.3184,
+      "step": 416
+    },
+    {
+      "epoch": 0.5097799511002445,
+      "grad_norm": 0.5893986225128174,
+      "learning_rate": 0.001,
+      "loss": 2.3627,
+      "step": 417
+    },
+    {
+      "epoch": 0.511002444987775,
+      "grad_norm": 0.5796381235122681,
+      "learning_rate": 0.001,
+      "loss": 2.4797,
+      "step": 418
+    },
+    {
+      "epoch": 0.5122249388753056,
+      "grad_norm": 0.6783269047737122,
+      "learning_rate": 0.001,
+      "loss": 2.4106,
+      "step": 419
+    },
+    {
+      "epoch": 0.5134474327628362,
+      "grad_norm": 0.6270769238471985,
+      "learning_rate": 0.001,
+      "loss": 2.4528,
+      "step": 420
+    },
+    {
+      "epoch": 0.5146699266503667,
+      "grad_norm": 0.5378878712654114,
+      "learning_rate": 0.001,
+      "loss": 2.3958,
+      "step": 421
+    },
+    {
+      "epoch": 0.5158924205378973,
+      "grad_norm": 0.6691122651100159,
+      "learning_rate": 0.001,
+      "loss": 2.3864,
+      "step": 422
+    },
+    {
+      "epoch": 0.5171149144254279,
+      "grad_norm": 0.6228435039520264,
+      "learning_rate": 0.001,
+      "loss": 2.4361,
+      "step": 423
+    },
+    {
+      "epoch": 0.5183374083129584,
+      "grad_norm": 0.5937249660491943,
+      "learning_rate": 0.001,
+      "loss": 2.403,
+      "step": 424
+    },
+    {
+      "epoch": 0.519559902200489,
+      "grad_norm": 0.7730590105056763,
+      "learning_rate": 0.001,
+      "loss": 2.3516,
+      "step": 425
+    },
+    {
+      "epoch": 0.5207823960880196,
+      "grad_norm": 0.5874658823013306,
+      "learning_rate": 0.001,
+      "loss": 2.3768,
+      "step": 426
+    },
+    {
+      "epoch": 0.5220048899755502,
+      "grad_norm": 0.610345184803009,
+      "learning_rate": 0.001,
+      "loss": 2.3741,
+      "step": 427
+    },
+    {
+      "epoch": 0.5232273838630807,
+      "grad_norm": 0.6153433322906494,
+      "learning_rate": 0.001,
+      "loss": 2.3821,
+      "step": 428
+    },
+    {
+      "epoch": 0.5244498777506112,
+      "grad_norm": 0.553920567035675,
+      "learning_rate": 0.001,
+      "loss": 2.4033,
+      "step": 429
+    },
+    {
+      "epoch": 0.5256723716381418,
+      "grad_norm": 0.6375141143798828,
+      "learning_rate": 0.001,
+      "loss": 2.3904,
+      "step": 430
+    },
+    {
+      "epoch": 0.5268948655256723,
+      "grad_norm": 0.6468214988708496,
+      "learning_rate": 0.001,
+      "loss": 2.422,
+      "step": 431
+    },
+    {
+      "epoch": 0.5281173594132029,
+      "grad_norm": 0.502112090587616,
+      "learning_rate": 0.001,
+      "loss": 2.4082,
+      "step": 432
+    },
+    {
+      "epoch": 0.5293398533007335,
+      "grad_norm": 0.63483726978302,
+      "learning_rate": 0.001,
+      "loss": 2.3971,
+      "step": 433
+    },
+    {
+      "epoch": 0.530562347188264,
+      "grad_norm": 0.6903530359268188,
+      "learning_rate": 0.001,
+      "loss": 2.5567,
+      "step": 434
+    },
+    {
+      "epoch": 0.5317848410757946,
+      "grad_norm": 0.590471088886261,
+      "learning_rate": 0.001,
+      "loss": 2.3714,
+      "step": 435
+    },
+    {
+      "epoch": 0.5330073349633252,
+      "grad_norm": 0.5665633678436279,
+      "learning_rate": 0.001,
+      "loss": 2.51,
+      "step": 436
+    },
+    {
+      "epoch": 0.5342298288508558,
+      "grad_norm": 0.5448029637336731,
+      "learning_rate": 0.001,
+      "loss": 2.4756,
+      "step": 437
+    },
+    {
+      "epoch": 0.5354523227383863,
+      "grad_norm": 0.5371047258377075,
+      "learning_rate": 0.001,
+      "loss": 2.4703,
+      "step": 438
+    },
+    {
+      "epoch": 0.5366748166259169,
+      "grad_norm": 0.6409993171691895,
+      "learning_rate": 0.001,
+      "loss": 2.5327,
+      "step": 439
+    },
+    {
+      "epoch": 0.5378973105134475,
+      "grad_norm": 0.5884791612625122,
+      "learning_rate": 0.001,
+      "loss": 2.4883,
+      "step": 440
+    },
+    {
+      "epoch": 0.539119804400978,
+      "grad_norm": 0.5778174996376038,
+      "learning_rate": 0.001,
+      "loss": 2.3682,
+      "step": 441
+    },
+    {
+      "epoch": 0.5403422982885085,
+      "grad_norm": 0.6152352094650269,
+      "learning_rate": 0.001,
+      "loss": 2.5577,
+      "step": 442
+    },
+    {
+      "epoch": 0.5415647921760391,
+      "grad_norm": 0.5729324221611023,
+      "learning_rate": 0.001,
+      "loss": 2.4113,
+      "step": 443
+    },
+    {
+      "epoch": 0.5427872860635696,
+      "grad_norm": 0.5118098258972168,
+      "learning_rate": 0.001,
+      "loss": 2.3661,
+      "step": 444
+    },
+    {
+      "epoch": 0.5440097799511002,
+      "grad_norm": 0.5968663692474365,
+      "learning_rate": 0.001,
+      "loss": 2.3956,
+      "step": 445
+    },
+    {
+      "epoch": 0.5452322738386308,
+      "grad_norm": 0.5854983329772949,
+      "learning_rate": 0.001,
+      "loss": 2.3685,
+      "step": 446
+    },
+    {
+      "epoch": 0.5464547677261614,
+      "grad_norm": 0.5025555491447449,
+      "learning_rate": 0.001,
+      "loss": 2.4087,
+      "step": 447
+    },
+    {
+      "epoch": 0.5476772616136919,
+      "grad_norm": 0.5501389503479004,
+      "learning_rate": 0.001,
+      "loss": 2.464,
+      "step": 448
+    },
+    {
+      "epoch": 0.5488997555012225,
+      "grad_norm": 0.5119859576225281,
+      "learning_rate": 0.001,
+      "loss": 2.4211,
+      "step": 449
+    },
+    {
+      "epoch": 0.5501222493887531,
+      "grad_norm": 0.5670396089553833,
+      "learning_rate": 0.001,
+      "loss": 2.3828,
+      "step": 450
+    },
+    {
+      "epoch": 0.5513447432762836,
+      "grad_norm": 0.6066712141036987,
+      "learning_rate": 0.001,
+      "loss": 2.398,
+      "step": 451
+    },
+    {
+      "epoch": 0.5525672371638142,
+      "grad_norm": 0.6207214593887329,
+      "learning_rate": 0.001,
+      "loss": 2.4942,
+      "step": 452
+    },
+    {
+      "epoch": 0.5537897310513448,
+      "grad_norm": 0.5594907402992249,
+      "learning_rate": 0.001,
+      "loss": 2.4766,
+      "step": 453
+    },
+    {
+      "epoch": 0.5550122249388753,
+      "grad_norm": 0.5723139643669128,
+      "learning_rate": 0.001,
+      "loss": 2.3692,
+      "step": 454
+    },
+    {
+      "epoch": 0.5562347188264058,
+      "grad_norm": 0.5299721956253052,
+      "learning_rate": 0.001,
+      "loss": 2.3481,
+      "step": 455
+    },
+    {
+      "epoch": 0.5574572127139364,
+      "grad_norm": 0.5901131629943848,
+      "learning_rate": 0.001,
+      "loss": 2.4067,
+      "step": 456
+    },
+    {
+      "epoch": 0.558679706601467,
+      "grad_norm": 0.6540071368217468,
+      "learning_rate": 0.001,
+      "loss": 2.4758,
+      "step": 457
+    },
+    {
+      "epoch": 0.5599022004889975,
+      "grad_norm": 0.6255263686180115,
+      "learning_rate": 0.001,
+      "loss": 2.4349,
+      "step": 458
+    },
+    {
+      "epoch": 0.5611246943765281,
+      "grad_norm": 0.6452546715736389,
+      "learning_rate": 0.001,
+      "loss": 2.3834,
+      "step": 459
+    },
+    {
+      "epoch": 0.5623471882640587,
+      "grad_norm": 0.6318612098693848,
+      "learning_rate": 0.001,
+      "loss": 2.38,
+      "step": 460
+    },
+    {
+      "epoch": 0.5635696821515892,
+      "grad_norm": 0.518966794013977,
+      "learning_rate": 0.001,
+      "loss": 2.3138,
+      "step": 461
+    },
+    {
+      "epoch": 0.5647921760391198,
+      "grad_norm": 0.6043367385864258,
+      "learning_rate": 0.001,
+      "loss": 2.512,
+      "step": 462
+    },
+    {
+      "epoch": 0.5660146699266504,
+      "grad_norm": 0.49966487288475037,
+      "learning_rate": 0.001,
+      "loss": 2.3268,
+      "step": 463
+    },
+    {
+      "epoch": 0.5672371638141809,
+      "grad_norm": 0.5622513890266418,
+      "learning_rate": 0.001,
+      "loss": 2.4228,
+      "step": 464
+    },
+    {
+      "epoch": 0.5684596577017115,
+      "grad_norm": 0.539992094039917,
+      "learning_rate": 0.001,
+      "loss": 2.4627,
+      "step": 465
+    },
+    {
+      "epoch": 0.5696821515892421,
+      "grad_norm": 0.5305809378623962,
+      "learning_rate": 0.001,
+      "loss": 2.3507,
+      "step": 466
+    },
+    {
+      "epoch": 0.5709046454767727,
+      "grad_norm": 0.5940194725990295,
+      "learning_rate": 0.001,
+      "loss": 2.3989,
+      "step": 467
+    },
+    {
+      "epoch": 0.5721271393643031,
+      "grad_norm": 0.765998125076294,
+      "learning_rate": 0.001,
+      "loss": 2.3981,
+      "step": 468
+    },
+    {
+      "epoch": 0.5733496332518337,
+      "grad_norm": 0.6094310283660889,
+      "learning_rate": 0.001,
+      "loss": 2.451,
+      "step": 469
+    },
+    {
+      "epoch": 0.5745721271393643,
+      "grad_norm": 0.523844838142395,
+      "learning_rate": 0.001,
+      "loss": 2.3669,
+      "step": 470
+    },
+    {
+      "epoch": 0.5757946210268948,
+      "grad_norm": 0.6060677170753479,
+      "learning_rate": 0.001,
+      "loss": 2.3594,
+      "step": 471
+    },
+    {
+      "epoch": 0.5770171149144254,
+      "grad_norm": 0.6068428158760071,
+      "learning_rate": 0.001,
+      "loss": 2.3889,
+      "step": 472
+    },
+    {
+      "epoch": 0.578239608801956,
+      "grad_norm": 0.6003894805908203,
+      "learning_rate": 0.001,
+      "loss": 2.3811,
+      "step": 473
+    },
+    {
+      "epoch": 0.5794621026894865,
+      "grad_norm": 0.6235598921775818,
+      "learning_rate": 0.001,
+      "loss": 2.4453,
+      "step": 474
+    },
+    {
+      "epoch": 0.5806845965770171,
+      "grad_norm": 0.6744781732559204,
+      "learning_rate": 0.001,
+      "loss": 2.3894,
+      "step": 475
+    },
+    {
+      "epoch": 0.5819070904645477,
+      "grad_norm": 0.6238363981246948,
+      "learning_rate": 0.001,
+      "loss": 2.3208,
+      "step": 476
+    },
+    {
+      "epoch": 0.5831295843520783,
+      "grad_norm": 0.6921210289001465,
+      "learning_rate": 0.001,
+      "loss": 2.4008,
+      "step": 477
+    },
+    {
+      "epoch": 0.5843520782396088,
+      "grad_norm": 0.6112361550331116,
+      "learning_rate": 0.001,
+      "loss": 2.5079,
+      "step": 478
+    },
+    {
+      "epoch": 0.5855745721271394,
+      "grad_norm": 0.6528260707855225,
+      "learning_rate": 0.001,
+      "loss": 2.4417,
+      "step": 479
+    },
+    {
+      "epoch": 0.58679706601467,
+      "grad_norm": 0.5697323679924011,
+      "learning_rate": 0.001,
+      "loss": 2.3566,
+      "step": 480
+    },
+    {
+      "epoch": 0.5880195599022005,
+      "grad_norm": 0.5770347714424133,
+      "learning_rate": 0.001,
+      "loss": 2.3936,
+      "step": 481
+    },
+    {
+      "epoch": 0.589242053789731,
+      "grad_norm": 0.620617151260376,
+      "learning_rate": 0.001,
+      "loss": 2.3786,
+      "step": 482
+    },
+    {
+      "epoch": 0.5904645476772616,
+      "grad_norm": 0.5476077198982239,
+      "learning_rate": 0.001,
+      "loss": 2.439,
+      "step": 483
+    },
+    {
+      "epoch": 0.5916870415647921,
+      "grad_norm": 0.5232786536216736,
+      "learning_rate": 0.001,
+      "loss": 2.38,
+      "step": 484
+    },
+    {
+      "epoch": 0.5929095354523227,
+      "grad_norm": 0.552232563495636,
+      "learning_rate": 0.001,
+      "loss": 2.4613,
+      "step": 485
+    },
+    {
+      "epoch": 0.5941320293398533,
+      "grad_norm": 0.5390328764915466,
+      "learning_rate": 0.001,
+      "loss": 2.3949,
+      "step": 486
+    },
+    {
+      "epoch": 0.5953545232273839,
+      "grad_norm": 0.5667278170585632,
+      "learning_rate": 0.001,
+      "loss": 2.3934,
+      "step": 487
+    },
+    {
+      "epoch": 0.5965770171149144,
+      "grad_norm": 0.5378979444503784,
+      "learning_rate": 0.001,
+      "loss": 2.328,
+      "step": 488
+    },
+    {
+      "epoch": 0.597799511002445,
+      "grad_norm": 0.520291805267334,
+      "learning_rate": 0.001,
+      "loss": 2.4101,
+      "step": 489
+    },
+    {
+      "epoch": 0.5990220048899756,
+      "grad_norm": 0.5654510855674744,
+      "learning_rate": 0.001,
+      "loss": 2.4033,
+      "step": 490
+    },
+    {
+      "epoch": 0.6002444987775061,
+      "grad_norm": 0.5622351765632629,
+      "learning_rate": 0.001,
+      "loss": 2.4365,
+      "step": 491
+    },
+    {
+      "epoch": 0.6014669926650367,
+      "grad_norm": 0.5403612852096558,
+      "learning_rate": 0.001,
+      "loss": 2.3842,
+      "step": 492
+    },
+    {
+      "epoch": 0.6026894865525673,
+      "grad_norm": 0.5934503674507141,
+      "learning_rate": 0.001,
+      "loss": 2.4396,
+      "step": 493
+    },
+    {
+      "epoch": 0.6039119804400978,
+      "grad_norm": 0.5867724418640137,
+      "learning_rate": 0.001,
+      "loss": 2.4173,
+      "step": 494
+    },
+    {
+      "epoch": 0.6051344743276283,
+      "grad_norm": 0.5405483841896057,
+      "learning_rate": 0.001,
+      "loss": 2.4576,
+      "step": 495
+    },
+    {
+      "epoch": 0.6063569682151589,
+      "grad_norm": 0.5979167819023132,
+      "learning_rate": 0.001,
+      "loss": 2.4444,
+      "step": 496
+    },
+    {
+      "epoch": 0.6075794621026895,
+      "grad_norm": 0.55023193359375,
+      "learning_rate": 0.001,
+      "loss": 2.4416,
+      "step": 497
+    },
+    {
+      "epoch": 0.60880195599022,
+      "grad_norm": 0.5640861392021179,
+      "learning_rate": 0.001,
+      "loss": 2.4057,
+      "step": 498
+    },
+    {
+      "epoch": 0.6100244498777506,
+      "grad_norm": 0.6123327016830444,
+      "learning_rate": 0.001,
+      "loss": 2.3173,
+      "step": 499
+    },
+    {
+      "epoch": 0.6112469437652812,
+      "grad_norm": 0.5508670806884766,
+      "learning_rate": 0.001,
+      "loss": 2.3907,
+      "step": 500
+    },
+    {
+      "epoch": 0.6124694376528117,
+      "grad_norm": 0.5583838224411011,
+      "learning_rate": 0.001,
+      "loss": 2.3548,
+      "step": 501
+    },
+    {
+      "epoch": 0.6136919315403423,
+      "grad_norm": 0.5637821555137634,
+      "learning_rate": 0.001,
+      "loss": 2.3176,
+      "step": 502
+    },
+    {
+      "epoch": 0.6149144254278729,
+      "grad_norm": 0.5943612456321716,
+      "learning_rate": 0.001,
+      "loss": 2.388,
+      "step": 503
+    },
+    {
+      "epoch": 0.6161369193154034,
+      "grad_norm": 0.717356264591217,
+      "learning_rate": 0.001,
+      "loss": 2.3649,
+      "step": 504
+    },
+    {
+      "epoch": 0.617359413202934,
+      "grad_norm": 0.5627100467681885,
+      "learning_rate": 0.001,
+      "loss": 2.3938,
+      "step": 505
+    },
+    {
+      "epoch": 0.6185819070904646,
+      "grad_norm": 0.5377146601676941,
+      "learning_rate": 0.001,
+      "loss": 2.369,
+      "step": 506
+    },
+    {
+      "epoch": 0.6198044009779952,
+      "grad_norm": 0.5633614659309387,
+      "learning_rate": 0.001,
+      "loss": 2.4127,
+      "step": 507
+    },
+    {
+      "epoch": 0.6210268948655256,
+      "grad_norm": 0.5996368527412415,
+      "learning_rate": 0.001,
+      "loss": 2.3729,
+      "step": 508
+    },
+    {
+      "epoch": 0.6222493887530562,
+      "grad_norm": 0.6659756898880005,
+      "learning_rate": 0.001,
+      "loss": 2.3896,
+      "step": 509
+    },
+    {
+      "epoch": 0.6234718826405868,
+      "grad_norm": 0.4918464124202728,
+      "learning_rate": 0.001,
+      "loss": 2.3871,
+      "step": 510
+    },
+    {
+      "epoch": 0.6246943765281173,
+      "grad_norm": 0.5441576838493347,
+      "learning_rate": 0.001,
+      "loss": 2.3487,
+      "step": 511
+    },
+    {
+      "epoch": 0.6259168704156479,
+      "grad_norm": 0.5248016715049744,
+      "learning_rate": 0.001,
+      "loss": 2.4756,
+      "step": 512
+    },
+    {
+      "epoch": 0.6271393643031785,
+      "grad_norm": 0.5138648152351379,
+      "learning_rate": 0.001,
+      "loss": 2.4046,
+      "step": 513
+    },
+    {
+      "epoch": 0.628361858190709,
+      "grad_norm": 0.5615628361701965,
+      "learning_rate": 0.001,
+      "loss": 2.434,
+      "step": 514
+    },
+    {
+      "epoch": 0.6295843520782396,
+      "grad_norm": 0.5971861481666565,
+      "learning_rate": 0.001,
+      "loss": 2.3772,
+      "step": 515
+    },
+    {
+      "epoch": 0.6308068459657702,
+      "grad_norm": 0.6121475100517273,
+      "learning_rate": 0.001,
+      "loss": 2.3834,
+      "step": 516
+    },
+    {
+      "epoch": 0.6320293398533008,
+      "grad_norm": 0.5112194418907166,
+      "learning_rate": 0.001,
+      "loss": 2.3088,
+      "step": 517
+    },
+    {
+      "epoch": 0.6332518337408313,
+      "grad_norm": 0.5220752358436584,
+      "learning_rate": 0.001,
+      "loss": 2.3559,
+      "step": 518
+    },
+    {
+      "epoch": 0.6344743276283619,
+      "grad_norm": 0.5204265713691711,
+      "learning_rate": 0.001,
+      "loss": 2.496,
+      "step": 519
+    },
+    {
+      "epoch": 0.6356968215158925,
+      "grad_norm": 0.5493486523628235,
+      "learning_rate": 0.001,
+      "loss": 2.3303,
+      "step": 520
+    },
+    {
+      "epoch": 0.636919315403423,
+      "grad_norm": 0.5837205052375793,
+      "learning_rate": 0.001,
+      "loss": 2.3196,
+      "step": 521
+    },
+    {
+      "epoch": 0.6381418092909535,
+      "grad_norm": 0.5629017353057861,
+      "learning_rate": 0.001,
+      "loss": 2.458,
+      "step": 522
+    },
+    {
+      "epoch": 0.6393643031784841,
+      "grad_norm": 0.5637010931968689,
+      "learning_rate": 0.001,
+      "loss": 2.3734,
+      "step": 523
+    },
+    {
+      "epoch": 0.6405867970660146,
+      "grad_norm": 0.5641871094703674,
+      "learning_rate": 0.001,
+      "loss": 2.353,
+      "step": 524
+    },
+    {
+      "epoch": 0.6418092909535452,
+      "grad_norm": 0.5671785473823547,
+      "learning_rate": 0.001,
+      "loss": 2.4907,
+      "step": 525
+    },
+    {
+      "epoch": 0.6430317848410758,
+      "grad_norm": 0.5508561730384827,
+      "learning_rate": 0.001,
+      "loss": 2.3804,
+      "step": 526
+    },
+    {
+      "epoch": 0.6442542787286064,
+      "grad_norm": 0.5103657245635986,
+      "learning_rate": 0.001,
+      "loss": 2.4565,
+      "step": 527
+    },
+    {
+      "epoch": 0.6454767726161369,
+      "grad_norm": 0.5293738842010498,
+      "learning_rate": 0.001,
+      "loss": 2.4496,
+      "step": 528
+    },
+    {
+      "epoch": 0.6466992665036675,
+      "grad_norm": 0.5606434941291809,
+      "learning_rate": 0.001,
+      "loss": 2.2755,
+      "step": 529
+    },
+    {
+      "epoch": 0.6479217603911981,
+      "grad_norm": 0.5806160569190979,
+      "learning_rate": 0.001,
+      "loss": 2.4226,
+      "step": 530
+    },
+    {
+      "epoch": 0.6491442542787286,
+      "grad_norm": 0.6059505343437195,
+      "learning_rate": 0.001,
+      "loss": 2.3963,
+      "step": 531
+    },
+    {
+      "epoch": 0.6503667481662592,
+      "grad_norm": 0.6261471509933472,
+      "learning_rate": 0.001,
+      "loss": 2.3978,
+      "step": 532
+    },
+    {
+      "epoch": 0.6515892420537898,
+      "grad_norm": 0.5536214113235474,
+      "learning_rate": 0.001,
+      "loss": 2.3348,
+      "step": 533
+    },
+    {
+      "epoch": 0.6528117359413202,
+      "grad_norm": 0.5901937484741211,
+      "learning_rate": 0.001,
+      "loss": 2.4816,
+      "step": 534
+    },
+    {
+      "epoch": 0.6540342298288508,
+      "grad_norm": 0.5521092414855957,
+      "learning_rate": 0.001,
+      "loss": 2.3598,
+      "step": 535
+    },
+    {
+      "epoch": 0.6552567237163814,
+      "grad_norm": 0.526321291923523,
+      "learning_rate": 0.001,
+      "loss": 2.358,
+      "step": 536
+    },
+    {
+      "epoch": 0.656479217603912,
+      "grad_norm": 0.5706471800804138,
+      "learning_rate": 0.001,
+      "loss": 2.324,
+      "step": 537
+    },
+    {
+      "epoch": 0.6577017114914425,
+      "grad_norm": 0.5586467981338501,
+      "learning_rate": 0.001,
+      "loss": 2.3881,
+      "step": 538
+    },
+    {
+      "epoch": 0.6589242053789731,
+      "grad_norm": 0.5118228793144226,
+      "learning_rate": 0.001,
+      "loss": 2.3338,
+      "step": 539
+    },
+    {
+      "epoch": 0.6601466992665037,
+      "grad_norm": 0.6145654320716858,
+      "learning_rate": 0.001,
+      "loss": 2.3243,
+      "step": 540
+    },
+    {
+      "epoch": 0.6613691931540342,
+      "grad_norm": 0.5887059569358826,
+      "learning_rate": 0.001,
+      "loss": 2.335,
+      "step": 541
+    },
+    {
+      "epoch": 0.6625916870415648,
+      "grad_norm": 0.5634593367576599,
+      "learning_rate": 0.001,
+      "loss": 2.3915,
+      "step": 542
+    },
+    {
+      "epoch": 0.6638141809290954,
+      "grad_norm": 0.5326385498046875,
+      "learning_rate": 0.001,
+      "loss": 2.3112,
+      "step": 543
+    },
+    {
+      "epoch": 0.6650366748166259,
+      "grad_norm": 0.5700005888938904,
+      "learning_rate": 0.001,
+      "loss": 2.3378,
+      "step": 544
+    },
+    {
+      "epoch": 0.6662591687041565,
+      "grad_norm": 0.5960122346878052,
+      "learning_rate": 0.001,
+      "loss": 2.4213,
+      "step": 545
+    },
+    {
+      "epoch": 0.6674816625916871,
+      "grad_norm": 0.6229601502418518,
+      "learning_rate": 0.001,
+      "loss": 2.4157,
+      "step": 546
+    },
+    {
+      "epoch": 0.6687041564792175,
+      "grad_norm": 0.588280975818634,
+      "learning_rate": 0.001,
+      "loss": 2.3102,
+      "step": 547
+    },
+    {
+      "epoch": 0.6699266503667481,
+      "grad_norm": 0.509946346282959,
+      "learning_rate": 0.001,
+      "loss": 2.4242,
+      "step": 548
+    },
+    {
+      "epoch": 0.6711491442542787,
+      "grad_norm": 0.5596705675125122,
+      "learning_rate": 0.001,
+      "loss": 2.4396,
+      "step": 549
+    },
+    {
+      "epoch": 0.6723716381418093,
+      "grad_norm": 0.5317002534866333,
+      "learning_rate": 0.001,
+      "loss": 2.3731,
+      "step": 550
+    },
+    {
+      "epoch": 0.6735941320293398,
+      "grad_norm": 0.5407710671424866,
+      "learning_rate": 0.001,
+      "loss": 2.3861,
+      "step": 551
+    },
+    {
+      "epoch": 0.6748166259168704,
+      "grad_norm": 0.5366936922073364,
+      "learning_rate": 0.001,
+      "loss": 2.3498,
+      "step": 552
+    },
+    {
+      "epoch": 0.676039119804401,
+      "grad_norm": 0.5394434332847595,
+      "learning_rate": 0.001,
+      "loss": 2.4139,
+      "step": 553
+    },
+    {
+      "epoch": 0.6772616136919315,
+      "grad_norm": 0.6320561766624451,
+      "learning_rate": 0.001,
+      "loss": 2.3204,
+      "step": 554
+    },
+    {
+      "epoch": 0.6784841075794621,
+      "grad_norm": 0.5851796865463257,
+      "learning_rate": 0.001,
+      "loss": 2.5296,
+      "step": 555
+    },
+    {
+      "epoch": 0.6797066014669927,
+      "grad_norm": 0.7752942442893982,
+      "learning_rate": 0.001,
+      "loss": 2.3896,
+      "step": 556
+    },
+    {
+      "epoch": 0.6809290953545232,
+      "grad_norm": 0.6305093765258789,
+      "learning_rate": 0.001,
+      "loss": 2.4496,
+      "step": 557
+    },
+    {
+      "epoch": 0.6821515892420538,
+      "grad_norm": 0.563012957572937,
+      "learning_rate": 0.001,
+      "loss": 2.3887,
+      "step": 558
+    },
+    {
+      "epoch": 0.6833740831295844,
+      "grad_norm": 0.676686704158783,
+      "learning_rate": 0.001,
+      "loss": 2.4618,
+      "step": 559
+    },
+    {
+      "epoch": 0.684596577017115,
+      "grad_norm": 0.5765560269355774,
+      "learning_rate": 0.001,
+      "loss": 2.3942,
+      "step": 560
+    },
+    {
+      "epoch": 0.6858190709046454,
+      "grad_norm": 0.5478991866111755,
+      "learning_rate": 0.001,
+      "loss": 2.4521,
+      "step": 561
+    },
+    {
+      "epoch": 0.687041564792176,
+      "grad_norm": 0.6409326791763306,
+      "learning_rate": 0.001,
+      "loss": 2.5013,
+      "step": 562
+    },
+    {
+      "epoch": 0.6882640586797066,
+      "grad_norm": 0.49936145544052124,
+      "learning_rate": 0.001,
+      "loss": 2.3794,
+      "step": 563
+    },
+    {
+      "epoch": 0.6894865525672371,
+      "grad_norm": 0.5707321763038635,
+      "learning_rate": 0.001,
+      "loss": 2.3323,
+      "step": 564
+    },
+    {
+      "epoch": 0.6907090464547677,
+      "grad_norm": 0.5072236657142639,
+      "learning_rate": 0.001,
+      "loss": 2.4768,
+      "step": 565
+    },
+    {
+      "epoch": 0.6919315403422983,
+      "grad_norm": 0.5288402438163757,
+      "learning_rate": 0.001,
+      "loss": 2.227,
+      "step": 566
+    },
+    {
+      "epoch": 0.6931540342298288,
+      "grad_norm": 0.5510168671607971,
+      "learning_rate": 0.001,
+      "loss": 2.4368,
+      "step": 567
+    },
+    {
+      "epoch": 0.6943765281173594,
+      "grad_norm": 0.5543464422225952,
+      "learning_rate": 0.001,
+      "loss": 2.5117,
+      "step": 568
+    },
+    {
+      "epoch": 0.69559902200489,
+      "grad_norm": 0.6391542553901672,
+      "learning_rate": 0.001,
+      "loss": 2.3455,
+      "step": 569
+    },
+    {
+      "epoch": 0.6968215158924206,
+      "grad_norm": 0.5454332232475281,
+      "learning_rate": 0.001,
+      "loss": 2.4398,
+      "step": 570
+    },
+    {
+      "epoch": 0.6980440097799511,
+      "grad_norm": 0.5730705261230469,
+      "learning_rate": 0.001,
+      "loss": 2.4693,
+      "step": 571
+    },
+    {
+      "epoch": 0.6992665036674817,
+      "grad_norm": 0.6171644330024719,
+      "learning_rate": 0.001,
+      "loss": 2.5105,
+      "step": 572
+    },
+    {
+      "epoch": 0.7004889975550123,
+      "grad_norm": 0.5583361387252808,
+      "learning_rate": 0.001,
+      "loss": 2.3169,
+      "step": 573
+    },
+    {
+      "epoch": 0.7017114914425427,
+      "grad_norm": 0.7579666972160339,
+      "learning_rate": 0.001,
+      "loss": 2.4683,
+      "step": 574
+    },
+    {
+      "epoch": 0.7029339853300733,
+      "grad_norm": 0.5713133215904236,
+      "learning_rate": 0.001,
+      "loss": 2.3683,
+      "step": 575
+    },
+    {
+      "epoch": 0.7041564792176039,
+      "grad_norm": 0.5870834589004517,
+      "learning_rate": 0.001,
+      "loss": 2.4675,
+      "step": 576
+    },
+    {
+      "epoch": 0.7053789731051344,
+      "grad_norm": 0.631624162197113,
+      "learning_rate": 0.001,
+      "loss": 2.3634,
+      "step": 577
+    },
+    {
+      "epoch": 0.706601466992665,
+      "grad_norm": 0.6153734922409058,
+      "learning_rate": 0.001,
+      "loss": 2.3948,
+      "step": 578
+    },
+    {
+      "epoch": 0.7078239608801956,
+      "grad_norm": 0.5715532302856445,
+      "learning_rate": 0.001,
+      "loss": 2.4551,
+      "step": 579
+    },
+    {
+      "epoch": 0.7090464547677262,
+      "grad_norm": 0.5857019424438477,
+      "learning_rate": 0.001,
+      "loss": 2.4935,
+      "step": 580
+    },
+    {
+      "epoch": 0.7102689486552567,
+      "grad_norm": 0.6221391558647156,
+      "learning_rate": 0.001,
+      "loss": 2.4721,
+      "step": 581
+    },
+    {
+      "epoch": 0.7114914425427873,
+      "grad_norm": 0.6044909358024597,
+      "learning_rate": 0.001,
+      "loss": 2.2741,
+      "step": 582
+    },
+    {
+      "epoch": 0.7127139364303179,
+      "grad_norm": 0.6020300388336182,
+      "learning_rate": 0.001,
+      "loss": 2.3917,
+      "step": 583
+    },
+    {
+      "epoch": 0.7139364303178484,
+      "grad_norm": 0.5730435252189636,
+      "learning_rate": 0.001,
+      "loss": 2.4477,
+      "step": 584
+    },
+    {
+      "epoch": 0.715158924205379,
+      "grad_norm": 0.5461198687553406,
+      "learning_rate": 0.001,
+      "loss": 2.4284,
+      "step": 585
+    },
+    {
+      "epoch": 0.7163814180929096,
+      "grad_norm": 0.5864646434783936,
+      "learning_rate": 0.001,
+      "loss": 2.3605,
+      "step": 586
+    },
+    {
+      "epoch": 0.71760391198044,
+      "grad_norm": 0.6093571782112122,
+      "learning_rate": 0.001,
+      "loss": 2.4669,
+      "step": 587
+    },
+    {
+      "epoch": 0.7188264058679706,
+      "grad_norm": 0.6475409865379333,
+      "learning_rate": 0.001,
+      "loss": 2.4698,
+      "step": 588
+    },
+    {
+      "epoch": 0.7200488997555012,
+      "grad_norm": 0.6631804704666138,
+      "learning_rate": 0.001,
+      "loss": 2.3759,
+      "step": 589
+    },
+    {
+      "epoch": 0.7212713936430318,
+      "grad_norm": 0.5505629777908325,
+      "learning_rate": 0.001,
+      "loss": 2.3559,
+      "step": 590
+    },
+    {
+      "epoch": 0.7224938875305623,
+      "grad_norm": 0.6232132315635681,
+      "learning_rate": 0.001,
+      "loss": 2.3618,
+      "step": 591
+    },
+    {
+      "epoch": 0.7237163814180929,
+      "grad_norm": 0.6157463788986206,
+      "learning_rate": 0.001,
+      "loss": 2.3225,
+      "step": 592
+    },
+    {
+      "epoch": 0.7249388753056235,
+      "grad_norm": 0.6107804179191589,
+      "learning_rate": 0.001,
+      "loss": 2.4047,
+      "step": 593
+    },
+    {
+      "epoch": 0.726161369193154,
+      "grad_norm": 0.6085322499275208,
+      "learning_rate": 0.001,
+      "loss": 2.4066,
+      "step": 594
+    },
+    {
+      "epoch": 0.7273838630806846,
+      "grad_norm": 0.5943775177001953,
+      "learning_rate": 0.001,
+      "loss": 2.4326,
+      "step": 595
+    },
+    {
+      "epoch": 0.7286063569682152,
+      "grad_norm": 0.5640223026275635,
+      "learning_rate": 0.001,
+      "loss": 2.3126,
+      "step": 596
+    },
+    {
+      "epoch": 0.7298288508557457,
+      "grad_norm": 0.5814293026924133,
+      "learning_rate": 0.001,
+      "loss": 2.4832,
+      "step": 597
+    },
+    {
+      "epoch": 0.7310513447432763,
+      "grad_norm": 0.5499405264854431,
+      "learning_rate": 0.001,
+      "loss": 2.3934,
+      "step": 598
+    },
+    {
+      "epoch": 0.7322738386308069,
+      "grad_norm": 0.5558736324310303,
+      "learning_rate": 0.001,
+      "loss": 2.4986,
+      "step": 599
+    },
+    {
+      "epoch": 0.7334963325183375,
+      "grad_norm": 0.5608446598052979,
+      "learning_rate": 0.001,
+      "loss": 2.3605,
+      "step": 600
+    },
+    {
+      "epoch": 0.7347188264058679,
+      "grad_norm": 0.5901703238487244,
+      "learning_rate": 0.001,
+      "loss": 2.4288,
+      "step": 601
+    },
+    {
+      "epoch": 0.7359413202933985,
+      "grad_norm": 0.6036423444747925,
+      "learning_rate": 0.001,
+      "loss": 2.3991,
+      "step": 602
+    },
+    {
+      "epoch": 0.7371638141809291,
+      "grad_norm": 0.5142916440963745,
+      "learning_rate": 0.001,
+      "loss": 2.3906,
+      "step": 603
+    },
+    {
+      "epoch": 0.7383863080684596,
+      "grad_norm": 0.627358078956604,
+      "learning_rate": 0.001,
+      "loss": 2.374,
+      "step": 604
+    },
+    {
+      "epoch": 0.7396088019559902,
+      "grad_norm": 0.546642541885376,
+      "learning_rate": 0.001,
+      "loss": 2.3263,
+      "step": 605
+    },
+    {
+      "epoch": 0.7408312958435208,
+      "grad_norm": 0.5707128643989563,
+      "learning_rate": 0.001,
+      "loss": 2.3437,
+      "step": 606
+    },
+    {
+      "epoch": 0.7420537897310513,
+      "grad_norm": 0.6058228015899658,
+      "learning_rate": 0.001,
+      "loss": 2.4895,
+      "step": 607
+    },
+    {
+      "epoch": 0.7432762836185819,
+      "grad_norm": 0.5580143928527832,
+      "learning_rate": 0.001,
+      "loss": 2.3298,
+      "step": 608
+    },
+    {
+      "epoch": 0.7444987775061125,
+      "grad_norm": 0.6101015210151672,
+      "learning_rate": 0.001,
+      "loss": 2.3968,
+      "step": 609
+    },
+    {
+      "epoch": 0.7457212713936431,
+      "grad_norm": 0.6657527089118958,
+      "learning_rate": 0.001,
+      "loss": 2.447,
+      "step": 610
+    },
+    {
+      "epoch": 0.7469437652811736,
+      "grad_norm": 0.5650172829627991,
+      "learning_rate": 0.001,
+      "loss": 2.3942,
+      "step": 611
+    },
+    {
+      "epoch": 0.7481662591687042,
+      "grad_norm": 0.6829528212547302,
+      "learning_rate": 0.001,
+      "loss": 2.4355,
+      "step": 612
+    },
+    {
+      "epoch": 0.7493887530562348,
+      "grad_norm": 0.646058976650238,
+      "learning_rate": 0.001,
+      "loss": 2.3792,
+      "step": 613
+    },
+    {
+      "epoch": 0.7506112469437652,
+      "grad_norm": 0.6063521504402161,
+      "learning_rate": 0.001,
+      "loss": 2.3637,
+      "step": 614
+    },
+    {
+      "epoch": 0.7518337408312958,
+      "grad_norm": 0.5286898612976074,
+      "learning_rate": 0.001,
+      "loss": 2.4182,
+      "step": 615
+    },
+    {
+      "epoch": 0.7518337408312958,
+      "eval_loss": 2.344815969467163,
+      "eval_runtime": 104.078,
+      "eval_samples_per_second": 100.55,
+      "eval_steps_per_second": 12.577,
+      "step": 615
+    },
+    {
+      "epoch": 0.7530562347188264,
+      "grad_norm": 0.5201755166053772,
+      "learning_rate": 0.001,
+      "loss": 2.3792,
+      "step": 616
+    },
+    {
+      "epoch": 0.7542787286063569,
+      "grad_norm": 0.5472811460494995,
+      "learning_rate": 0.001,
+      "loss": 2.3532,
+      "step": 617
+    },
+    {
+      "epoch": 0.7555012224938875,
+      "grad_norm": 0.5708293914794922,
+      "learning_rate": 0.001,
+      "loss": 2.4103,
+      "step": 618
+    },
+    {
+      "epoch": 0.7567237163814181,
+      "grad_norm": 0.5900188088417053,
+      "learning_rate": 0.001,
+      "loss": 2.4371,
+      "step": 619
+    },
+    {
+      "epoch": 0.7579462102689487,
+      "grad_norm": 0.6372197866439819,
+      "learning_rate": 0.001,
+      "loss": 2.3237,
+      "step": 620
+    },
+    {
+      "epoch": 0.7591687041564792,
+      "grad_norm": 0.5709636807441711,
+      "learning_rate": 0.001,
+      "loss": 2.3432,
+      "step": 621
+    },
+    {
+      "epoch": 0.7603911980440098,
+      "grad_norm": 0.6171260476112366,
+      "learning_rate": 0.001,
+      "loss": 2.3663,
+      "step": 622
+    },
+    {
+      "epoch": 0.7616136919315404,
+      "grad_norm": 0.6021854877471924,
+      "learning_rate": 0.001,
+      "loss": 2.3367,
+      "step": 623
+    },
+    {
+      "epoch": 0.7628361858190709,
+      "grad_norm": 0.6226273775100708,
+      "learning_rate": 0.001,
+      "loss": 2.336,
+      "step": 624
+    },
+    {
+      "epoch": 0.7640586797066015,
+      "grad_norm": 0.6232383251190186,
+      "learning_rate": 0.001,
+      "loss": 2.3983,
+      "step": 625
+    },
+    {
+      "epoch": 0.7652811735941321,
+      "grad_norm": 0.5813260674476624,
+      "learning_rate": 0.001,
+      "loss": 2.4898,
+      "step": 626
+    },
+    {
+      "epoch": 0.7665036674816625,
+      "grad_norm": 0.620922863483429,
+      "learning_rate": 0.001,
+      "loss": 2.4339,
+      "step": 627
+    },
+    {
+      "epoch": 0.7677261613691931,
+      "grad_norm": 0.5710574388504028,
+      "learning_rate": 0.001,
+      "loss": 2.3937,
+      "step": 628
+    },
+    {
+      "epoch": 0.7689486552567237,
+      "grad_norm": 0.5268881916999817,
+      "learning_rate": 0.001,
+      "loss": 2.4622,
+      "step": 629
+    },
+    {
+      "epoch": 0.7701711491442543,
+      "grad_norm": 0.5241808295249939,
+      "learning_rate": 0.001,
+      "loss": 2.3836,
+      "step": 630
+    },
+    {
+      "epoch": 0.7713936430317848,
+      "grad_norm": 0.5768167972564697,
+      "learning_rate": 0.001,
+      "loss": 2.3474,
+      "step": 631
+    },
+    {
+      "epoch": 0.7726161369193154,
+      "grad_norm": 0.5535464286804199,
+      "learning_rate": 0.001,
+      "loss": 2.4747,
+      "step": 632
+    },
+    {
+      "epoch": 0.773838630806846,
+      "grad_norm": 0.5237547159194946,
+      "learning_rate": 0.001,
+      "loss": 2.4178,
+      "step": 633
+    },
+    {
+      "epoch": 0.7750611246943765,
+      "grad_norm": 0.6961969137191772,
+      "learning_rate": 0.001,
+      "loss": 2.3371,
+      "step": 634
+    },
+    {
+      "epoch": 0.7762836185819071,
+      "grad_norm": 0.6474157571792603,
+      "learning_rate": 0.001,
+      "loss": 2.448,
+      "step": 635
+    },
+    {
+      "epoch": 0.7775061124694377,
+      "grad_norm": 0.5978342294692993,
+      "learning_rate": 0.001,
+      "loss": 2.3929,
+      "step": 636
+    },
+    {
+      "epoch": 0.7787286063569682,
+      "grad_norm": 0.5999522805213928,
+      "learning_rate": 0.001,
+      "loss": 2.3485,
+      "step": 637
+    },
+    {
+      "epoch": 0.7799511002444988,
+      "grad_norm": 0.5849292278289795,
+      "learning_rate": 0.001,
+      "loss": 2.487,
+      "step": 638
+    },
+    {
+      "epoch": 0.7811735941320294,
+      "grad_norm": 0.6249788999557495,
+      "learning_rate": 0.001,
+      "loss": 2.4578,
+      "step": 639
+    },
+    {
+      "epoch": 0.78239608801956,
+      "grad_norm": 0.6195868253707886,
+      "learning_rate": 0.001,
+      "loss": 2.5804,
+      "step": 640
+    },
+    {
+      "epoch": 0.7836185819070904,
+      "grad_norm": 0.5440888404846191,
+      "learning_rate": 0.001,
+      "loss": 2.4577,
+      "step": 641
+    },
+    {
+      "epoch": 0.784841075794621,
+      "grad_norm": 0.6838685870170593,
+      "learning_rate": 0.001,
+      "loss": 2.3788,
+      "step": 642
+    },
+    {
+      "epoch": 0.7860635696821516,
+      "grad_norm": 0.6188467741012573,
+      "learning_rate": 0.001,
+      "loss": 2.4285,
+      "step": 643
+    },
+    {
+      "epoch": 0.7872860635696821,
+      "grad_norm": 0.5791029930114746,
+      "learning_rate": 0.001,
+      "loss": 2.4295,
+      "step": 644
+    },
+    {
+      "epoch": 0.7885085574572127,
+      "grad_norm": 0.5324996709823608,
+      "learning_rate": 0.001,
+      "loss": 2.2872,
+      "step": 645
+    },
+    {
+      "epoch": 0.7897310513447433,
+      "grad_norm": 0.5715858936309814,
+      "learning_rate": 0.001,
+      "loss": 2.3472,
+      "step": 646
+    },
+    {
+      "epoch": 0.7909535452322738,
+      "grad_norm": 0.5822765231132507,
+      "learning_rate": 0.001,
+      "loss": 2.4128,
+      "step": 647
+    },
+    {
+      "epoch": 0.7921760391198044,
+      "grad_norm": 0.6102983951568604,
+      "learning_rate": 0.001,
+      "loss": 2.4367,
+      "step": 648
+    },
+    {
+      "epoch": 0.793398533007335,
+      "grad_norm": 0.5699944496154785,
+      "learning_rate": 0.001,
+      "loss": 2.4394,
+      "step": 649
+    },
+    {
+      "epoch": 0.7946210268948656,
+      "grad_norm": 0.5703781843185425,
+      "learning_rate": 0.001,
+      "loss": 2.3587,
+      "step": 650
+    },
+    {
+      "epoch": 0.7958435207823961,
+      "grad_norm": 0.6027846336364746,
+      "learning_rate": 0.001,
+      "loss": 2.4859,
+      "step": 651
+    },
+    {
+      "epoch": 0.7970660146699267,
+      "grad_norm": 0.5607686638832092,
+      "learning_rate": 0.001,
+      "loss": 2.3393,
+      "step": 652
+    },
+    {
+      "epoch": 0.7982885085574573,
+      "grad_norm": 0.5783531665802002,
+      "learning_rate": 0.001,
+      "loss": 2.4062,
+      "step": 653
+    },
+    {
+      "epoch": 0.7995110024449877,
+      "grad_norm": 0.5493980050086975,
+      "learning_rate": 0.001,
+      "loss": 2.3283,
+      "step": 654
+    },
+    {
+      "epoch": 0.8007334963325183,
+      "grad_norm": 0.5421778559684753,
+      "learning_rate": 0.001,
+      "loss": 2.4083,
+      "step": 655
+    },
+    {
+      "epoch": 0.8019559902200489,
+      "grad_norm": 0.539010763168335,
+      "learning_rate": 0.001,
+      "loss": 2.3952,
+      "step": 656
+    },
+    {
+      "epoch": 0.8031784841075794,
+      "grad_norm": 0.5817722082138062,
+      "learning_rate": 0.001,
+      "loss": 2.4124,
+      "step": 657
+    },
+    {
+      "epoch": 0.80440097799511,
+      "grad_norm": 0.5872360467910767,
+      "learning_rate": 0.001,
+      "loss": 2.4038,
+      "step": 658
+    },
+    {
+      "epoch": 0.8056234718826406,
+      "grad_norm": 0.6575203537940979,
+      "learning_rate": 0.001,
+      "loss": 2.4344,
+      "step": 659
+    },
+    {
+      "epoch": 0.8068459657701712,
+      "grad_norm": 0.5091464519500732,
+      "learning_rate": 0.001,
+      "loss": 2.3349,
+      "step": 660
+    },
+    {
+      "epoch": 0.8080684596577017,
+      "grad_norm": 0.5787954330444336,
+      "learning_rate": 0.001,
+      "loss": 2.3587,
+      "step": 661
+    },
+    {
+      "epoch": 0.8092909535452323,
+      "grad_norm": 0.6609556078910828,
+      "learning_rate": 0.001,
+      "loss": 2.4197,
+      "step": 662
+    },
+    {
+      "epoch": 0.8105134474327629,
+      "grad_norm": 0.5616955161094666,
+      "learning_rate": 0.001,
+      "loss": 2.5327,
+      "step": 663
+    },
+    {
+      "epoch": 0.8117359413202934,
+      "grad_norm": 0.5685104727745056,
+      "learning_rate": 0.001,
+      "loss": 2.3639,
+      "step": 664
+    },
+    {
+      "epoch": 0.812958435207824,
+      "grad_norm": 0.6164966225624084,
+      "learning_rate": 0.001,
+      "loss": 2.3692,
+      "step": 665
+    },
+    {
+      "epoch": 0.8141809290953546,
+      "grad_norm": 0.6559178829193115,
+      "learning_rate": 0.001,
+      "loss": 2.3249,
+      "step": 666
+    },
+    {
+      "epoch": 0.815403422982885,
+      "grad_norm": 0.5554947257041931,
+      "learning_rate": 0.001,
+      "loss": 2.3248,
+      "step": 667
+    },
+    {
+      "epoch": 0.8166259168704156,
+      "grad_norm": 0.5663809180259705,
+      "learning_rate": 0.001,
+      "loss": 2.4098,
+      "step": 668
+    },
+    {
+      "epoch": 0.8178484107579462,
+      "grad_norm": 0.5895231366157532,
+      "learning_rate": 0.001,
+      "loss": 2.3251,
+      "step": 669
+    },
+    {
+      "epoch": 0.8190709046454768,
+      "grad_norm": 0.5413554906845093,
+      "learning_rate": 0.001,
+      "loss": 2.3619,
+      "step": 670
+    },
+    {
+      "epoch": 0.8202933985330073,
+      "grad_norm": 0.5487458109855652,
+      "learning_rate": 0.001,
+      "loss": 2.404,
+      "step": 671
+    },
+    {
+      "epoch": 0.8215158924205379,
+      "grad_norm": 0.5602384805679321,
+      "learning_rate": 0.001,
+      "loss": 2.3715,
+      "step": 672
+    },
+    {
+      "epoch": 0.8227383863080685,
+      "grad_norm": 0.5400522947311401,
+      "learning_rate": 0.001,
+      "loss": 2.3271,
+      "step": 673
+    },
+    {
+      "epoch": 0.823960880195599,
+      "grad_norm": 0.5108867883682251,
+      "learning_rate": 0.001,
+      "loss": 2.4399,
+      "step": 674
+    },
+    {
+      "epoch": 0.8251833740831296,
+      "grad_norm": 0.5453303456306458,
+      "learning_rate": 0.001,
+      "loss": 2.4627,
+      "step": 675
+    },
+    {
+      "epoch": 0.8264058679706602,
+      "grad_norm": 0.6444577574729919,
+      "learning_rate": 0.001,
+      "loss": 2.3977,
+      "step": 676
+    },
+    {
+      "epoch": 0.8276283618581907,
+      "grad_norm": 0.5792852640151978,
+      "learning_rate": 0.001,
+      "loss": 2.4131,
+      "step": 677
+    },
+    {
+      "epoch": 0.8288508557457213,
+      "grad_norm": 0.738950252532959,
+      "learning_rate": 0.001,
+      "loss": 2.2458,
+      "step": 678
+    },
+    {
+      "epoch": 0.8300733496332519,
+      "grad_norm": 0.5346181988716125,
+      "learning_rate": 0.001,
+      "loss": 2.3125,
+      "step": 679
+    },
+    {
+      "epoch": 0.8312958435207825,
+      "grad_norm": 0.5854019522666931,
+      "learning_rate": 0.001,
+      "loss": 2.389,
+      "step": 680
+    },
+    {
+      "epoch": 0.8325183374083129,
+      "grad_norm": 0.6267631649971008,
+      "learning_rate": 0.001,
+      "loss": 2.5181,
+      "step": 681
+    },
+    {
+      "epoch": 0.8337408312958435,
+      "grad_norm": 0.6359632015228271,
+      "learning_rate": 0.001,
+      "loss": 2.3248,
+      "step": 682
+    },
+    {
+      "epoch": 0.8349633251833741,
+      "grad_norm": 0.5600630640983582,
+      "learning_rate": 0.001,
+      "loss": 2.3173,
+      "step": 683
+    },
+    {
+      "epoch": 0.8361858190709046,
+      "grad_norm": 0.5631881952285767,
+      "learning_rate": 0.001,
+      "loss": 2.4153,
+      "step": 684
+    },
+    {
+      "epoch": 0.8374083129584352,
+      "grad_norm": 0.5842933654785156,
+      "learning_rate": 0.001,
+      "loss": 2.4056,
+      "step": 685
+    },
+    {
+      "epoch": 0.8386308068459658,
+      "grad_norm": 0.5515209436416626,
+      "learning_rate": 0.001,
+      "loss": 2.373,
+      "step": 686
+    },
+    {
+      "epoch": 0.8398533007334963,
+      "grad_norm": 0.6496553421020508,
+      "learning_rate": 0.001,
+      "loss": 2.3827,
+      "step": 687
+    },
+    {
+      "epoch": 0.8410757946210269,
+      "grad_norm": 0.6773370504379272,
+      "learning_rate": 0.001,
+      "loss": 2.4588,
+      "step": 688
+    },
+    {
+      "epoch": 0.8422982885085575,
+      "grad_norm": 0.5586926937103271,
+      "learning_rate": 0.001,
+      "loss": 2.3336,
+      "step": 689
+    },
+    {
+      "epoch": 0.843520782396088,
+      "grad_norm": 0.576976478099823,
+      "learning_rate": 0.001,
+      "loss": 2.3248,
+      "step": 690
+    },
+    {
+      "epoch": 0.8447432762836186,
+      "grad_norm": 0.5717596411705017,
+      "learning_rate": 0.001,
+      "loss": 2.3765,
+      "step": 691
+    },
+    {
+      "epoch": 0.8459657701711492,
+      "grad_norm": 0.878169059753418,
+      "learning_rate": 0.001,
+      "loss": 2.3114,
+      "step": 692
+    },
+    {
+      "epoch": 0.8471882640586798,
+      "grad_norm": 0.5785524845123291,
+      "learning_rate": 0.001,
+      "loss": 2.4307,
+      "step": 693
+    },
+    {
+      "epoch": 0.8484107579462102,
+      "grad_norm": 0.5417555570602417,
+      "learning_rate": 0.001,
+      "loss": 2.3569,
+      "step": 694
+    },
+    {
+      "epoch": 0.8496332518337408,
+      "grad_norm": 0.5829872488975525,
+      "learning_rate": 0.001,
+      "loss": 2.3458,
+      "step": 695
+    },
+    {
+      "epoch": 0.8508557457212714,
+      "grad_norm": 0.5894091129302979,
+      "learning_rate": 0.001,
+      "loss": 2.2858,
+      "step": 696
+    },
+    {
+      "epoch": 0.8520782396088019,
+      "grad_norm": 0.601750910282135,
+      "learning_rate": 0.001,
+      "loss": 2.2999,
+      "step": 697
+    },
+    {
+      "epoch": 0.8533007334963325,
+      "grad_norm": 0.6677522659301758,
+      "learning_rate": 0.001,
+      "loss": 2.4061,
+      "step": 698
+    },
+    {
+      "epoch": 0.8545232273838631,
+      "grad_norm": 0.616008460521698,
+      "learning_rate": 0.001,
+      "loss": 2.4467,
+      "step": 699
+    },
+    {
+      "epoch": 0.8557457212713936,
+      "grad_norm": 0.6218184232711792,
+      "learning_rate": 0.001,
+      "loss": 2.4714,
+      "step": 700
+    },
+    {
+      "epoch": 0.8569682151589242,
+      "grad_norm": 0.6129118204116821,
+      "learning_rate": 0.001,
+      "loss": 2.3915,
+      "step": 701
+    },
+    {
+      "epoch": 0.8581907090464548,
+      "grad_norm": 0.5880963802337646,
+      "learning_rate": 0.001,
+      "loss": 2.3564,
+      "step": 702
+    },
+    {
+      "epoch": 0.8594132029339854,
+      "grad_norm": 0.5852961540222168,
+      "learning_rate": 0.001,
+      "loss": 2.3743,
+      "step": 703
+    },
+    {
+      "epoch": 0.8606356968215159,
+      "grad_norm": 0.595486581325531,
+      "learning_rate": 0.001,
+      "loss": 2.2989,
+      "step": 704
+    },
+    {
+      "epoch": 0.8618581907090465,
+      "grad_norm": 0.5296614766120911,
+      "learning_rate": 0.001,
+      "loss": 2.3549,
+      "step": 705
+    },
+    {
+      "epoch": 0.863080684596577,
+      "grad_norm": 0.5757498145103455,
+      "learning_rate": 0.001,
+      "loss": 2.4028,
+      "step": 706
+    },
+    {
+      "epoch": 0.8643031784841075,
+      "grad_norm": 0.537127673625946,
+      "learning_rate": 0.001,
+      "loss": 2.3948,
+      "step": 707
+    },
+    {
+      "epoch": 0.8655256723716381,
+      "grad_norm": 0.5664181113243103,
+      "learning_rate": 0.001,
+      "loss": 2.4288,
+      "step": 708
+    },
+    {
+      "epoch": 0.8667481662591687,
+      "grad_norm": 0.5783668160438538,
+      "learning_rate": 0.001,
+      "loss": 2.3615,
+      "step": 709
+    },
+    {
+      "epoch": 0.8679706601466992,
+      "grad_norm": 0.5113952159881592,
+      "learning_rate": 0.001,
+      "loss": 2.3418,
+      "step": 710
+    },
+    {
+      "epoch": 0.8691931540342298,
+      "grad_norm": 0.7263479232788086,
+      "learning_rate": 0.001,
+      "loss": 2.4439,
+      "step": 711
+    },
+    {
+      "epoch": 0.8704156479217604,
+      "grad_norm": 0.5564237833023071,
+      "learning_rate": 0.001,
+      "loss": 2.4239,
+      "step": 712
+    },
+    {
+      "epoch": 0.871638141809291,
+      "grad_norm": 0.6223247051239014,
+      "learning_rate": 0.001,
+      "loss": 2.3884,
+      "step": 713
+    },
+    {
+      "epoch": 0.8728606356968215,
+      "grad_norm": 0.592124342918396,
+      "learning_rate": 0.001,
+      "loss": 2.3621,
+      "step": 714
+    },
+    {
+      "epoch": 0.8740831295843521,
+      "grad_norm": 0.5872472524642944,
+      "learning_rate": 0.001,
+      "loss": 2.4129,
+      "step": 715
+    },
+    {
+      "epoch": 0.8753056234718827,
+      "grad_norm": 0.7134682536125183,
+      "learning_rate": 0.001,
+      "loss": 2.4038,
+      "step": 716
+    },
+    {
+      "epoch": 0.8765281173594132,
+      "grad_norm": 0.7206189632415771,
+      "learning_rate": 0.001,
+      "loss": 2.3734,
+      "step": 717
+    },
+    {
+      "epoch": 0.8777506112469438,
+      "grad_norm": 0.6114242076873779,
+      "learning_rate": 0.001,
+      "loss": 2.4213,
+      "step": 718
+    },
+    {
+      "epoch": 0.8789731051344744,
+      "grad_norm": 0.6361877918243408,
+      "learning_rate": 0.001,
+      "loss": 2.399,
+      "step": 719
+    },
+    {
+      "epoch": 0.8801955990220048,
+      "grad_norm": 0.6251677870750427,
+      "learning_rate": 0.001,
+      "loss": 2.422,
+      "step": 720
+    },
+    {
+      "epoch": 0.8814180929095354,
+      "grad_norm": 0.6224179863929749,
+      "learning_rate": 0.001,
+      "loss": 2.4036,
+      "step": 721
+    },
+    {
+      "epoch": 0.882640586797066,
+      "grad_norm": 0.6215149164199829,
+      "learning_rate": 0.001,
+      "loss": 2.3502,
+      "step": 722
+    },
+    {
+      "epoch": 0.8838630806845966,
+      "grad_norm": 0.6429134011268616,
+      "learning_rate": 0.001,
+      "loss": 2.5216,
+      "step": 723
+    },
+    {
+      "epoch": 0.8850855745721271,
+      "grad_norm": 0.5223509073257446,
+      "learning_rate": 0.001,
+      "loss": 2.3913,
+      "step": 724
+    },
+    {
+      "epoch": 0.8863080684596577,
+      "grad_norm": 0.6248764395713806,
+      "learning_rate": 0.001,
+      "loss": 2.352,
+      "step": 725
+    },
+    {
+      "epoch": 0.8875305623471883,
+      "grad_norm": 0.5972513556480408,
+      "learning_rate": 0.001,
+      "loss": 2.3474,
+      "step": 726
+    },
+    {
+      "epoch": 0.8887530562347188,
+      "grad_norm": 0.5999692678451538,
+      "learning_rate": 0.001,
+      "loss": 2.3817,
+      "step": 727
+    },
+    {
+      "epoch": 0.8899755501222494,
+      "grad_norm": 0.6036021709442139,
+      "learning_rate": 0.001,
+      "loss": 2.4853,
+      "step": 728
+    },
+    {
+      "epoch": 0.89119804400978,
+      "grad_norm": 0.6136165261268616,
+      "learning_rate": 0.001,
+      "loss": 2.4091,
+      "step": 729
+    },
+    {
+      "epoch": 0.8924205378973105,
+      "grad_norm": 0.6014193296432495,
+      "learning_rate": 0.001,
+      "loss": 2.3446,
+      "step": 730
+    },
+    {
+      "epoch": 0.8936430317848411,
+      "grad_norm": 0.5973393321037292,
+      "learning_rate": 0.001,
+      "loss": 2.2264,
+      "step": 731
+    },
+    {
+      "epoch": 0.8948655256723717,
+      "grad_norm": 0.5201925039291382,
+      "learning_rate": 0.001,
+      "loss": 2.3571,
+      "step": 732
+    },
+    {
+      "epoch": 0.8960880195599022,
+      "grad_norm": 0.6981174945831299,
+      "learning_rate": 0.001,
+      "loss": 2.3528,
+      "step": 733
+    },
+    {
+      "epoch": 0.8973105134474327,
+      "grad_norm": 0.6088836789131165,
+      "learning_rate": 0.001,
+      "loss": 2.3604,
+      "step": 734
+    },
+    {
+      "epoch": 0.8985330073349633,
+      "grad_norm": 0.5979176759719849,
+      "learning_rate": 0.001,
+      "loss": 2.3966,
+      "step": 735
+    },
+    {
+      "epoch": 0.8997555012224939,
+      "grad_norm": 0.5857772827148438,
+      "learning_rate": 0.001,
+      "loss": 2.4123,
+      "step": 736
+    },
+    {
+      "epoch": 0.9009779951100244,
+      "grad_norm": 0.5723506212234497,
+      "learning_rate": 0.001,
+      "loss": 2.3642,
+      "step": 737
+    },
+    {
+      "epoch": 0.902200488997555,
+      "grad_norm": 0.6450780034065247,
+      "learning_rate": 0.001,
+      "loss": 2.3892,
+      "step": 738
+    },
+    {
+      "epoch": 0.9034229828850856,
+      "grad_norm": 0.5753511190414429,
+      "learning_rate": 0.001,
+      "loss": 2.3683,
+      "step": 739
+    },
+    {
+      "epoch": 0.9046454767726161,
+      "grad_norm": 0.5976255536079407,
+      "learning_rate": 0.001,
+      "loss": 2.4115,
+      "step": 740
+    },
+    {
+      "epoch": 0.9058679706601467,
+      "grad_norm": 0.603735625743866,
+      "learning_rate": 0.001,
+      "loss": 2.3854,
+      "step": 741
+    },
+    {
+      "epoch": 0.9070904645476773,
+      "grad_norm": 0.5742889642715454,
+      "learning_rate": 0.001,
+      "loss": 2.461,
+      "step": 742
+    },
+    {
+      "epoch": 0.9083129584352079,
+      "grad_norm": 0.5430094003677368,
+      "learning_rate": 0.001,
+      "loss": 2.4041,
+      "step": 743
+    },
+    {
+      "epoch": 0.9095354523227384,
+      "grad_norm": 0.6404337286949158,
+      "learning_rate": 0.001,
+      "loss": 2.498,
+      "step": 744
+    },
+    {
+      "epoch": 0.910757946210269,
+      "grad_norm": 0.6157411932945251,
+      "learning_rate": 0.001,
+      "loss": 2.3859,
+      "step": 745
+    },
+    {
+      "epoch": 0.9119804400977995,
+      "grad_norm": 0.5406906008720398,
+      "learning_rate": 0.001,
+      "loss": 2.4062,
+      "step": 746
+    },
+    {
+      "epoch": 0.91320293398533,
+      "grad_norm": 0.5866022109985352,
+      "learning_rate": 0.001,
+      "loss": 2.5032,
+      "step": 747
+    },
+    {
+      "epoch": 0.9144254278728606,
+      "grad_norm": 0.5461808443069458,
+      "learning_rate": 0.001,
+      "loss": 2.3162,
+      "step": 748
+    },
+    {
+      "epoch": 0.9156479217603912,
+      "grad_norm": 0.6297031044960022,
+      "learning_rate": 0.001,
+      "loss": 2.4081,
+      "step": 749
+    },
+    {
+      "epoch": 0.9168704156479217,
+      "grad_norm": 0.5485174059867859,
+      "learning_rate": 0.001,
+      "loss": 2.3578,
+      "step": 750
+    },
+    {
+      "epoch": 0.9180929095354523,
+      "grad_norm": 0.6516104340553284,
+      "learning_rate": 0.001,
+      "loss": 2.4404,
+      "step": 751
+    },
+    {
+      "epoch": 0.9193154034229829,
+      "grad_norm": 0.6003808379173279,
+      "learning_rate": 0.001,
+      "loss": 2.3506,
+      "step": 752
+    },
+    {
+      "epoch": 0.9205378973105135,
+      "grad_norm": 0.583228349685669,
+      "learning_rate": 0.001,
+      "loss": 2.4624,
+      "step": 753
+    },
+    {
+      "epoch": 0.921760391198044,
+      "grad_norm": 0.6453356742858887,
+      "learning_rate": 0.001,
+      "loss": 2.4198,
+      "step": 754
+    },
+    {
+      "epoch": 0.9229828850855746,
+      "grad_norm": 0.5110669136047363,
+      "learning_rate": 0.001,
+      "loss": 2.3705,
+      "step": 755
+    },
+    {
+      "epoch": 0.9242053789731052,
+      "grad_norm": 0.5734696984291077,
+      "learning_rate": 0.001,
+      "loss": 2.3523,
+      "step": 756
+    },
+    {
+      "epoch": 0.9254278728606357,
+      "grad_norm": 0.5642021894454956,
+      "learning_rate": 0.001,
+      "loss": 2.2984,
+      "step": 757
+    },
+    {
+      "epoch": 0.9266503667481663,
+      "grad_norm": 0.5395249724388123,
+      "learning_rate": 0.001,
+      "loss": 2.2935,
+      "step": 758
+    },
+    {
+      "epoch": 0.9278728606356969,
+      "grad_norm": 0.5964503288269043,
+      "learning_rate": 0.001,
+      "loss": 2.4717,
+      "step": 759
+    },
+    {
+      "epoch": 0.9290953545232273,
+      "grad_norm": 0.530672550201416,
+      "learning_rate": 0.001,
+      "loss": 2.4752,
+      "step": 760
+    },
+    {
+      "epoch": 0.9303178484107579,
+      "grad_norm": 0.5743510127067566,
+      "learning_rate": 0.001,
+      "loss": 2.3707,
+      "step": 761
+    },
+    {
+      "epoch": 0.9315403422982885,
+      "grad_norm": 0.6197736263275146,
+      "learning_rate": 0.001,
+      "loss": 2.4328,
+      "step": 762
+    },
+    {
+      "epoch": 0.9327628361858191,
+      "grad_norm": 0.5292767286300659,
+      "learning_rate": 0.001,
+      "loss": 2.3821,
+      "step": 763
+    },
+    {
+      "epoch": 0.9339853300733496,
+      "grad_norm": 0.6877421736717224,
+      "learning_rate": 0.001,
+      "loss": 2.3904,
+      "step": 764
+    },
+    {
+      "epoch": 0.9352078239608802,
+      "grad_norm": 0.5813961029052734,
+      "learning_rate": 0.001,
+      "loss": 2.4042,
+      "step": 765
+    },
+    {
+      "epoch": 0.9364303178484108,
+      "grad_norm": 0.6664645075798035,
+      "learning_rate": 0.001,
+      "loss": 2.407,
+      "step": 766
+    },
+    {
+      "epoch": 0.9376528117359413,
+      "grad_norm": 0.6104966402053833,
+      "learning_rate": 0.001,
+      "loss": 2.4558,
+      "step": 767
+    },
+    {
+      "epoch": 0.9388753056234719,
+      "grad_norm": 0.7187817692756653,
+      "learning_rate": 0.001,
+      "loss": 2.4366,
+      "step": 768
+    },
+    {
+      "epoch": 0.9400977995110025,
+      "grad_norm": 0.7083351612091064,
+      "learning_rate": 0.001,
+      "loss": 2.3937,
+      "step": 769
+    },
+    {
+      "epoch": 0.941320293398533,
+      "grad_norm": 0.5205307006835938,
+      "learning_rate": 0.001,
+      "loss": 2.333,
+      "step": 770
+    },
+    {
+      "epoch": 0.9425427872860636,
+      "grad_norm": 0.7257646918296814,
+      "learning_rate": 0.001,
+      "loss": 2.3803,
+      "step": 771
+    },
+    {
+      "epoch": 0.9437652811735942,
+      "grad_norm": 0.5259858965873718,
+      "learning_rate": 0.001,
+      "loss": 2.361,
+      "step": 772
+    },
+    {
+      "epoch": 0.9449877750611247,
+      "grad_norm": 0.7148779630661011,
+      "learning_rate": 0.001,
+      "loss": 2.3112,
+      "step": 773
+    },
+    {
+      "epoch": 0.9462102689486552,
+      "grad_norm": 0.644771158695221,
+      "learning_rate": 0.001,
+      "loss": 2.3839,
+      "step": 774
+    },
+    {
+      "epoch": 0.9474327628361858,
+      "grad_norm": 0.5165205597877502,
+      "learning_rate": 0.001,
+      "loss": 2.4143,
+      "step": 775
+    },
+    {
+      "epoch": 0.9486552567237164,
+      "grad_norm": 0.600974977016449,
+      "learning_rate": 0.001,
+      "loss": 2.389,
+      "step": 776
+    },
+    {
+      "epoch": 0.9498777506112469,
+      "grad_norm": 0.6595199108123779,
+      "learning_rate": 0.001,
+      "loss": 2.3992,
+      "step": 777
+    },
+    {
+      "epoch": 0.9511002444987775,
+      "grad_norm": 0.556730329990387,
+      "learning_rate": 0.001,
+      "loss": 2.4006,
+      "step": 778
+    },
+    {
+      "epoch": 0.9523227383863081,
+      "grad_norm": 0.6529319882392883,
+      "learning_rate": 0.001,
+      "loss": 2.3843,
+      "step": 779
+    },
+    {
+      "epoch": 0.9535452322738386,
+      "grad_norm": 0.5654332637786865,
+      "learning_rate": 0.001,
+      "loss": 2.3503,
+      "step": 780
+    },
+    {
+      "epoch": 0.9547677261613692,
+      "grad_norm": 0.5724945664405823,
+      "learning_rate": 0.001,
+      "loss": 2.4135,
+      "step": 781
+    },
+    {
+      "epoch": 0.9559902200488998,
+      "grad_norm": 0.5892456769943237,
+      "learning_rate": 0.001,
+      "loss": 2.5017,
+      "step": 782
+    },
+    {
+      "epoch": 0.9572127139364304,
+      "grad_norm": 0.6245690584182739,
+      "learning_rate": 0.001,
+      "loss": 2.3421,
+      "step": 783
+    },
+    {
+      "epoch": 0.9584352078239609,
+      "grad_norm": 0.5814808011054993,
+      "learning_rate": 0.001,
+      "loss": 2.4081,
+      "step": 784
+    },
+    {
+      "epoch": 0.9596577017114915,
+      "grad_norm": 0.6698013544082642,
+      "learning_rate": 0.001,
+      "loss": 2.4073,
+      "step": 785
+    },
+    {
+      "epoch": 0.960880195599022,
+      "grad_norm": 0.5794418454170227,
+      "learning_rate": 0.001,
+      "loss": 2.4799,
+      "step": 786
+    },
+    {
+      "epoch": 0.9621026894865525,
+      "grad_norm": 0.5708101987838745,
+      "learning_rate": 0.001,
+      "loss": 2.4533,
+      "step": 787
+    },
+    {
+      "epoch": 0.9633251833740831,
+      "grad_norm": 0.6697220802307129,
+      "learning_rate": 0.001,
+      "loss": 2.5029,
+      "step": 788
+    },
+    {
+      "epoch": 0.9645476772616137,
+      "grad_norm": 0.5752573609352112,
+      "learning_rate": 0.001,
+      "loss": 2.402,
+      "step": 789
+    },
+    {
+      "epoch": 0.9657701711491442,
+      "grad_norm": 0.5376598834991455,
+      "learning_rate": 0.001,
+      "loss": 2.3827,
+      "step": 790
+    },
+    {
+      "epoch": 0.9669926650366748,
+      "grad_norm": 0.6070631146430969,
+      "learning_rate": 0.001,
+      "loss": 2.4628,
+      "step": 791
+    },
+    {
+      "epoch": 0.9682151589242054,
+      "grad_norm": 0.5507974624633789,
+      "learning_rate": 0.001,
+      "loss": 2.4281,
+      "step": 792
+    },
+    {
+      "epoch": 0.969437652811736,
+      "grad_norm": 0.6238107085227966,
+      "learning_rate": 0.001,
+      "loss": 2.403,
+      "step": 793
+    },
+    {
+      "epoch": 0.9706601466992665,
+      "grad_norm": 0.5629689693450928,
+      "learning_rate": 0.001,
+      "loss": 2.2755,
+      "step": 794
+    },
+    {
+      "epoch": 0.9718826405867971,
+      "grad_norm": 0.5647727847099304,
+      "learning_rate": 0.001,
+      "loss": 2.3897,
+      "step": 795
+    },
+    {
+      "epoch": 0.9731051344743277,
+      "grad_norm": 0.5957530736923218,
+      "learning_rate": 0.001,
+      "loss": 2.2812,
+      "step": 796
+    },
+    {
+      "epoch": 0.9743276283618582,
+      "grad_norm": 0.5573234558105469,
+      "learning_rate": 0.001,
+      "loss": 2.3699,
+      "step": 797
+    },
+    {
+      "epoch": 0.9755501222493888,
+      "grad_norm": 0.5734822750091553,
+      "learning_rate": 0.001,
+      "loss": 2.3682,
+      "step": 798
+    },
+    {
+      "epoch": 0.9767726161369193,
+      "grad_norm": 0.5806570649147034,
+      "learning_rate": 0.001,
+      "loss": 2.2825,
+      "step": 799
+    },
+    {
+      "epoch": 0.9779951100244498,
+      "grad_norm": 0.6516204476356506,
+      "learning_rate": 0.001,
+      "loss": 2.3894,
+      "step": 800
+    },
+    {
+      "epoch": 0.9792176039119804,
+      "grad_norm": 0.6128764748573303,
+      "learning_rate": 0.001,
+      "loss": 2.3822,
+      "step": 801
+    },
+    {
+      "epoch": 0.980440097799511,
+      "grad_norm": 0.5696424841880798,
+      "learning_rate": 0.001,
+      "loss": 2.4311,
+      "step": 802
+    },
+    {
+      "epoch": 0.9816625916870416,
+      "grad_norm": 0.5665503740310669,
+      "learning_rate": 0.001,
+      "loss": 2.4249,
+      "step": 803
+    },
+    {
+      "epoch": 0.9828850855745721,
+      "grad_norm": 0.6103429198265076,
+      "learning_rate": 0.001,
+      "loss": 2.3657,
+      "step": 804
+    },
+    {
+      "epoch": 0.9841075794621027,
+      "grad_norm": 0.6361311078071594,
+      "learning_rate": 0.001,
+      "loss": 2.368,
+      "step": 805
+    },
+    {
+      "epoch": 0.9853300733496333,
+      "grad_norm": 0.5717911720275879,
+      "learning_rate": 0.001,
+      "loss": 2.3035,
+      "step": 806
+    },
+    {
+      "epoch": 0.9865525672371638,
+      "grad_norm": 0.6038938164710999,
+      "learning_rate": 0.001,
+      "loss": 2.4214,
+      "step": 807
+    },
+    {
+      "epoch": 0.9877750611246944,
+      "grad_norm": 0.5755453705787659,
+      "learning_rate": 0.001,
+      "loss": 2.3541,
+      "step": 808
+    },
+    {
+      "epoch": 0.988997555012225,
+      "grad_norm": 0.5793865323066711,
+      "learning_rate": 0.001,
+      "loss": 2.3962,
+      "step": 809
+    },
+    {
+      "epoch": 0.9902200488997555,
+      "grad_norm": 0.5786888599395752,
+      "learning_rate": 0.001,
+      "loss": 2.3091,
+      "step": 810
+    },
+    {
+      "epoch": 0.991442542787286,
+      "grad_norm": 0.5730612277984619,
+      "learning_rate": 0.001,
+      "loss": 2.2928,
+      "step": 811
+    },
+    {
+      "epoch": 0.9926650366748166,
+      "grad_norm": 0.5544816255569458,
+      "learning_rate": 0.001,
+      "loss": 2.4369,
+      "step": 812
+    },
+    {
+      "epoch": 0.9938875305623472,
+      "grad_norm": 0.6146392226219177,
+      "learning_rate": 0.001,
+      "loss": 2.3794,
+      "step": 813
+    },
+    {
+      "epoch": 0.9951100244498777,
+      "grad_norm": 0.5923306941986084,
+      "learning_rate": 0.001,
+      "loss": 2.3317,
+      "step": 814
+    },
+    {
+      "epoch": 0.9963325183374083,
+      "grad_norm": 0.531193733215332,
+      "learning_rate": 0.001,
+      "loss": 2.3787,
+      "step": 815
+    },
+    {
+      "epoch": 0.9975550122249389,
+      "grad_norm": 0.5377926826477051,
+      "learning_rate": 0.001,
+      "loss": 2.3605,
+      "step": 816
+    },
+    {
+      "epoch": 0.9987775061124694,
+      "grad_norm": 0.5309790372848511,
+      "learning_rate": 0.001,
+      "loss": 2.386,
+      "step": 817
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.753987193107605,
+      "learning_rate": 0.001,
+      "loss": 2.4023,
+      "step": 818
+    },
+    {
+      "epoch": 1.0,
+      "step": 818,
+      "total_flos": 5.575689983066112e+16,
+      "train_loss": 2.4359203247685888,
+      "train_runtime": 747.848,
+      "train_samples_per_second": 139.932,
+      "train_steps_per_second": 1.094
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 818,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 205,
+  "stateful_callbacks": {
+    "EarlyStoppingCallback": {
+      "args": {
+        "early_stopping_patience": 5,
+        "early_stopping_threshold": 0.0
+      },
+      "attributes": {
+        "early_stopping_patience_counter": 0
+      }
+    },
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 5.575689983066112e+16,
+  "train_batch_size": 128,
+  "trial_name": null,
+  "trial_params": null
+}