diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7223 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.9970559371933267,
+  "eval_steps": 128,
+  "global_step": 1018,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001962708537782139,
+      "grad_norm": 62.355553075278046,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.8187,
+      "step": 1
+    },
+    {
+      "epoch": 0.001962708537782139,
+      "eval_loss": 0.8275482654571533,
+      "eval_runtime": 246.1117,
+      "eval_samples_per_second": 123.33,
+      "eval_steps_per_second": 3.856,
+      "step": 1
+    },
+    {
+      "epoch": 0.003925417075564278,
+      "grad_norm": 62.468487550636745,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.8042,
+      "step": 2
+    },
+    {
+      "epoch": 0.005888125613346418,
+      "grad_norm": 4.2406831230151605,
+      "learning_rate": 3e-06,
+      "loss": 0.638,
+      "step": 3
+    },
+    {
+      "epoch": 0.007850834151128557,
+      "grad_norm": 0.6232773471755494,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.5818,
+      "step": 4
+    },
+    {
+      "epoch": 0.009813542688910697,
+      "grad_norm": 1.2988116476459395,
+      "learning_rate": 5e-06,
+      "loss": 0.5652,
+      "step": 5
+    },
+    {
+      "epoch": 0.011776251226692836,
+      "grad_norm": 19.326748341786118,
+      "learning_rate": 6e-06,
+      "loss": 0.988,
+      "step": 6
+    },
+    {
+      "epoch": 0.013738959764474975,
+      "grad_norm": 1.2669160024303416,
+      "learning_rate": 7e-06,
+      "loss": 0.6037,
+      "step": 7
+    },
+    {
+      "epoch": 0.015701668302257114,
+      "grad_norm": 1.2907800971966077,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.6027,
+      "step": 8
+    },
+    {
+      "epoch": 0.017664376840039256,
+      "grad_norm": 0.7888911759146412,
+      "learning_rate": 9e-06,
+      "loss": 0.5711,
+      "step": 9
+    },
+    {
+      "epoch": 0.019627085377821395,
+      "grad_norm": 0.7260839669652744,
+      "learning_rate": 1e-05,
+      "loss": 0.5624,
+      "step": 10
+    },
+    {
+      "epoch": 0.021589793915603533,
+      "grad_norm": 0.6532838806100564,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.5612,
+      "step": 11
+    },
+    {
+      "epoch": 0.023552502453385672,
+      "grad_norm": 0.7559746737848189,
+      "learning_rate": 1.2e-05,
+      "loss": 0.5493,
+      "step": 12
+    },
+    {
+      "epoch": 0.02551521099116781,
+      "grad_norm": 0.5482270203080323,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 0.5435,
+      "step": 13
+    },
+    {
+      "epoch": 0.02747791952894995,
+      "grad_norm": 0.4970646557650867,
+      "learning_rate": 1.4e-05,
+      "loss": 0.5408,
+      "step": 14
+    },
+    {
+      "epoch": 0.029440628066732092,
+      "grad_norm": 0.391725935764316,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.5241,
+      "step": 15
+    },
+    {
+      "epoch": 0.03140333660451423,
+      "grad_norm": 0.43457047378514496,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.5488,
+      "step": 16
+    },
+    {
+      "epoch": 0.033366045142296366,
+      "grad_norm": 0.3789337362265595,
+      "learning_rate": 1.7e-05,
+      "loss": 0.5266,
+      "step": 17
+    },
+    {
+      "epoch": 0.03532875368007851,
+      "grad_norm": 0.3331484005850679,
+      "learning_rate": 1.8e-05,
+      "loss": 0.5112,
+      "step": 18
+    },
+    {
+      "epoch": 0.03729146221786065,
+      "grad_norm": 0.3444864658048081,
+      "learning_rate": 1.9e-05,
+      "loss": 0.5051,
+      "step": 19
+    },
+    {
+      "epoch": 0.03925417075564279,
+      "grad_norm": 0.2887437422361683,
+      "learning_rate": 2e-05,
+      "loss": 0.5289,
+      "step": 20
+    },
+    {
+      "epoch": 0.04121687929342493,
+      "grad_norm": 0.278825983002472,
+      "learning_rate": 1.9999987858045354e-05,
+      "loss": 0.5289,
+      "step": 21
+    },
+    {
+      "epoch": 0.04317958783120707,
+      "grad_norm": 0.39102343961692493,
+      "learning_rate": 1.9999951432210905e-05,
+      "loss": 0.5076,
+      "step": 22
+    },
+    {
+      "epoch": 0.045142296368989206,
+      "grad_norm": 0.5521017110831627,
+      "learning_rate": 1.9999890722585106e-05,
+      "loss": 0.5108,
+      "step": 23
+    },
+    {
+      "epoch": 0.047105004906771344,
+      "grad_norm": 0.9336591410925016,
+      "learning_rate": 1.9999805729315383e-05,
+      "loss": 0.5316,
+      "step": 24
+    },
+    {
+      "epoch": 0.04906771344455348,
+      "grad_norm": 0.8299578992443886,
+      "learning_rate": 1.9999696452608135e-05,
+      "loss": 0.5334,
+      "step": 25
+    },
+    {
+      "epoch": 0.05103042198233562,
+      "grad_norm": 4.751183353378357,
+      "learning_rate": 1.999956289272873e-05,
+      "loss": 0.5285,
+      "step": 26
+    },
+    {
+      "epoch": 0.05299313052011776,
+      "grad_norm": 0.860325982956048,
+      "learning_rate": 1.99994050500015e-05,
+      "loss": 0.539,
+      "step": 27
+    },
+    {
+      "epoch": 0.0549558390578999,
+      "grad_norm": 0.4897569644596324,
+      "learning_rate": 1.999922292480975e-05,
+      "loss": 0.5358,
+      "step": 28
+    },
+    {
+      "epoch": 0.05691854759568204,
+      "grad_norm": 0.5398108233225083,
+      "learning_rate": 1.9999016517595752e-05,
+      "loss": 0.5351,
+      "step": 29
+    },
+    {
+      "epoch": 0.058881256133464184,
+      "grad_norm": 0.4780297052699839,
+      "learning_rate": 1.9998785828860744e-05,
+      "loss": 0.5185,
+      "step": 30
+    },
+    {
+      "epoch": 0.06084396467124632,
+      "grad_norm": 0.33693365999492547,
+      "learning_rate": 1.9998530859164926e-05,
+      "loss": 0.5104,
+      "step": 31
+    },
+    {
+      "epoch": 0.06280667320902845,
+      "grad_norm": 0.4261145101597809,
+      "learning_rate": 1.9998251609127465e-05,
+      "loss": 0.5005,
+      "step": 32
+    },
+    {
+      "epoch": 0.0647693817468106,
+      "grad_norm": 0.3067512941150541,
+      "learning_rate": 1.999794807942649e-05,
+      "loss": 0.508,
+      "step": 33
+    },
+    {
+      "epoch": 0.06673209028459273,
+      "grad_norm": 0.3756361155229969,
+      "learning_rate": 1.999762027079909e-05,
+      "loss": 0.5021,
+      "step": 34
+    },
+    {
+      "epoch": 0.06869479882237488,
+      "grad_norm": 0.3065423142817771,
+      "learning_rate": 1.9997268184041318e-05,
+      "loss": 0.5242,
+      "step": 35
+    },
+    {
+      "epoch": 0.07065750736015702,
+      "grad_norm": 0.27326520045279085,
+      "learning_rate": 1.9996891820008165e-05,
+      "loss": 0.5133,
+      "step": 36
+    },
+    {
+      "epoch": 0.07262021589793916,
+      "grad_norm": 0.2964526747556581,
+      "learning_rate": 1.9996491179613597e-05,
+      "loss": 0.492,
+      "step": 37
+    },
+    {
+      "epoch": 0.0745829244357213,
+      "grad_norm": 0.2446010040256801,
+      "learning_rate": 1.9996066263830533e-05,
+      "loss": 0.4952,
+      "step": 38
+    },
+    {
+      "epoch": 0.07654563297350343,
+      "grad_norm": 0.22448559939725016,
+      "learning_rate": 1.999561707369082e-05,
+      "loss": 0.4917,
+      "step": 39
+    },
+    {
+      "epoch": 0.07850834151128558,
+      "grad_norm": 0.22417659964598924,
+      "learning_rate": 1.9995143610285275e-05,
+      "loss": 0.4978,
+      "step": 40
+    },
+    {
+      "epoch": 0.08047105004906771,
+      "grad_norm": 0.24556309982732719,
+      "learning_rate": 1.9994645874763657e-05,
+      "loss": 0.5123,
+      "step": 41
+    },
+    {
+      "epoch": 0.08243375858684986,
+      "grad_norm": 0.2078136882784064,
+      "learning_rate": 1.9994123868334655e-05,
+      "loss": 0.5008,
+      "step": 42
+    },
+    {
+      "epoch": 0.08439646712463199,
+      "grad_norm": 0.2126670968682465,
+      "learning_rate": 1.999357759226591e-05,
+      "loss": 0.5028,
+      "step": 43
+    },
+    {
+      "epoch": 0.08635917566241413,
+      "grad_norm": 0.19399581978765076,
+      "learning_rate": 1.9993007047883988e-05,
+      "loss": 0.4937,
+      "step": 44
+    },
+    {
+      "epoch": 0.08832188420019627,
+      "grad_norm": 0.1873229933356675,
+      "learning_rate": 1.9992412236574396e-05,
+      "loss": 0.4798,
+      "step": 45
+    },
+    {
+      "epoch": 0.09028459273797841,
+      "grad_norm": 0.19589361661093946,
+      "learning_rate": 1.999179315978157e-05,
+      "loss": 0.4794,
+      "step": 46
+    },
+    {
+      "epoch": 0.09224730127576054,
+      "grad_norm": 0.174199073484297,
+      "learning_rate": 1.999114981900887e-05,
+      "loss": 0.4735,
+      "step": 47
+    },
+    {
+      "epoch": 0.09421000981354269,
+      "grad_norm": 0.16944247667745355,
+      "learning_rate": 1.999048221581858e-05,
+      "loss": 0.4843,
+      "step": 48
+    },
+    {
+      "epoch": 0.09617271835132483,
+      "grad_norm": 0.1854184220985238,
+      "learning_rate": 1.9989790351831898e-05,
+      "loss": 0.4774,
+      "step": 49
+    },
+    {
+      "epoch": 0.09813542688910697,
+      "grad_norm": 0.16797468279664923,
+      "learning_rate": 1.9989074228728942e-05,
+      "loss": 0.4684,
+      "step": 50
+    },
+    {
+      "epoch": 0.10009813542688911,
+      "grad_norm": 0.16773676945215363,
+      "learning_rate": 1.998833384824874e-05,
+      "loss": 0.4613,
+      "step": 51
+    },
+    {
+      "epoch": 0.10206084396467124,
+      "grad_norm": 0.1645777226690783,
+      "learning_rate": 1.9987569212189224e-05,
+      "loss": 0.477,
+      "step": 52
+    },
+    {
+      "epoch": 0.10402355250245339,
+      "grad_norm": 0.17954527682275045,
+      "learning_rate": 1.998678032240723e-05,
+      "loss": 0.4718,
+      "step": 53
+    },
+    {
+      "epoch": 0.10598626104023552,
+      "grad_norm": 0.16162280011064276,
+      "learning_rate": 1.9985967180818493e-05,
+      "loss": 0.4927,
+      "step": 54
+    },
+    {
+      "epoch": 0.10794896957801767,
+      "grad_norm": 0.16773939638216998,
+      "learning_rate": 1.9985129789397633e-05,
+      "loss": 0.4797,
+      "step": 55
+    },
+    {
+      "epoch": 0.1099116781157998,
+      "grad_norm": 0.18424558774001681,
+      "learning_rate": 1.998426815017817e-05,
+      "loss": 0.4747,
+      "step": 56
+    },
+    {
+      "epoch": 0.11187438665358194,
+      "grad_norm": 0.1388232795248698,
+      "learning_rate": 1.9983382265252494e-05,
+      "loss": 0.4633,
+      "step": 57
+    },
+    {
+      "epoch": 0.11383709519136408,
+      "grad_norm": 0.17752346108818134,
+      "learning_rate": 1.998247213677188e-05,
+      "loss": 0.4613,
+      "step": 58
+    },
+    {
+      "epoch": 0.11579980372914622,
+      "grad_norm": 0.14030828634333592,
+      "learning_rate": 1.9981537766946486e-05,
+      "loss": 0.4726,
+      "step": 59
+    },
+    {
+      "epoch": 0.11776251226692837,
+      "grad_norm": 0.17867869880789194,
+      "learning_rate": 1.9980579158045322e-05,
+      "loss": 0.4659,
+      "step": 60
+    },
+    {
+      "epoch": 0.1197252208047105,
+      "grad_norm": 0.17555178239215463,
+      "learning_rate": 1.9979596312396258e-05,
+      "loss": 0.4882,
+      "step": 61
+    },
+    {
+      "epoch": 0.12168792934249265,
+      "grad_norm": 0.1525166881836132,
+      "learning_rate": 1.9978589232386036e-05,
+      "loss": 0.4668,
+      "step": 62
+    },
+    {
+      "epoch": 0.12365063788027478,
+      "grad_norm": 0.16104011998686385,
+      "learning_rate": 1.9977557920460237e-05,
+      "loss": 0.5001,
+      "step": 63
+    },
+    {
+      "epoch": 0.1256133464180569,
+      "grad_norm": 0.16256381990686963,
+      "learning_rate": 1.997650237912329e-05,
+      "loss": 0.4615,
+      "step": 64
+    },
+    {
+      "epoch": 0.12757605495583907,
+      "grad_norm": 0.14954423248271237,
+      "learning_rate": 1.9975422610938463e-05,
+      "loss": 0.4686,
+      "step": 65
+    },
+    {
+      "epoch": 0.1295387634936212,
+      "grad_norm": 0.19293534758098466,
+      "learning_rate": 1.997431861852785e-05,
+      "loss": 0.4607,
+      "step": 66
+    },
+    {
+      "epoch": 0.13150147203140333,
+      "grad_norm": 0.13976174634255645,
+      "learning_rate": 1.997319040457238e-05,
+      "loss": 0.4727,
+      "step": 67
+    },
+    {
+      "epoch": 0.13346418056918546,
+      "grad_norm": 0.18195008776452798,
+      "learning_rate": 1.9972037971811802e-05,
+      "loss": 0.4681,
+      "step": 68
+    },
+    {
+      "epoch": 0.13542688910696762,
+      "grad_norm": 0.13607271529544315,
+      "learning_rate": 1.9970861323044667e-05,
+      "loss": 0.4747,
+      "step": 69
+    },
+    {
+      "epoch": 0.13738959764474976,
+      "grad_norm": 0.14246714469054536,
+      "learning_rate": 1.996966046112834e-05,
+      "loss": 0.457,
+      "step": 70
+    },
+    {
+      "epoch": 0.1393523061825319,
+      "grad_norm": 0.15545571681772094,
+      "learning_rate": 1.9968435388978986e-05,
+      "loss": 0.4654,
+      "step": 71
+    },
+    {
+      "epoch": 0.14131501472031405,
+      "grad_norm": 0.1631020324563178,
+      "learning_rate": 1.996718610957155e-05,
+      "loss": 0.4663,
+      "step": 72
+    },
+    {
+      "epoch": 0.14327772325809618,
+      "grad_norm": 0.14479401685365606,
+      "learning_rate": 1.996591262593978e-05,
+      "loss": 0.4522,
+      "step": 73
+    },
+    {
+      "epoch": 0.1452404317958783,
+      "grad_norm": 0.15470054571534417,
+      "learning_rate": 1.9964614941176194e-05,
+      "loss": 0.4596,
+      "step": 74
+    },
+    {
+      "epoch": 0.14720314033366044,
+      "grad_norm": 0.13412246717171877,
+      "learning_rate": 1.9963293058432066e-05,
+      "loss": 0.4465,
+      "step": 75
+    },
+    {
+      "epoch": 0.1491658488714426,
+      "grad_norm": 0.14095810281768514,
+      "learning_rate": 1.9961946980917457e-05,
+      "loss": 0.4537,
+      "step": 76
+    },
+    {
+      "epoch": 0.15112855740922473,
+      "grad_norm": 0.16396930622944708,
+      "learning_rate": 1.996057671190116e-05,
+      "loss": 0.453,
+      "step": 77
+    },
+    {
+      "epoch": 0.15309126594700687,
+      "grad_norm": 0.5108895960415857,
+      "learning_rate": 1.995918225471073e-05,
+      "loss": 0.4819,
+      "step": 78
+    },
+    {
+      "epoch": 0.155053974484789,
+      "grad_norm": 0.16436026012077695,
+      "learning_rate": 1.995776361273245e-05,
+      "loss": 0.4481,
+      "step": 79
+    },
+    {
+      "epoch": 0.15701668302257116,
+      "grad_norm": 0.18055053584795847,
+      "learning_rate": 1.9956320789411338e-05,
+      "loss": 0.4561,
+      "step": 80
+    },
+    {
+      "epoch": 0.1589793915603533,
+      "grad_norm": 0.17300584614095313,
+      "learning_rate": 1.9954853788251137e-05,
+      "loss": 0.4626,
+      "step": 81
+    },
+    {
+      "epoch": 0.16094210009813542,
+      "grad_norm": 0.18537091420945745,
+      "learning_rate": 1.9953362612814294e-05,
+      "loss": 0.467,
+      "step": 82
+    },
+    {
+      "epoch": 0.16290480863591755,
+      "grad_norm": 0.1831953893733552,
+      "learning_rate": 1.995184726672197e-05,
+      "loss": 0.4533,
+      "step": 83
+    },
+    {
+      "epoch": 0.1648675171736997,
+      "grad_norm": 0.1785397387970331,
+      "learning_rate": 1.9950307753654016e-05,
+      "loss": 0.4714,
+      "step": 84
+    },
+    {
+      "epoch": 0.16683022571148184,
+      "grad_norm": 0.1696225474699336,
+      "learning_rate": 1.994874407734897e-05,
+      "loss": 0.4695,
+      "step": 85
+    },
+    {
+      "epoch": 0.16879293424926398,
+      "grad_norm": 0.20269683913377595,
+      "learning_rate": 1.994715624160405e-05,
+      "loss": 0.4648,
+      "step": 86
+    },
+    {
+      "epoch": 0.17075564278704614,
+      "grad_norm": 0.18739457130939605,
+      "learning_rate": 1.9945544250275147e-05,
+      "loss": 0.4608,
+      "step": 87
+    },
+    {
+      "epoch": 0.17271835132482827,
+      "grad_norm": 0.21960029028555894,
+      "learning_rate": 1.99439081072768e-05,
+      "loss": 0.4576,
+      "step": 88
+    },
+    {
+      "epoch": 0.1746810598626104,
+      "grad_norm": 0.1959495489832016,
+      "learning_rate": 1.9942247816582206e-05,
+      "loss": 0.4596,
+      "step": 89
+    },
+    {
+      "epoch": 0.17664376840039253,
+      "grad_norm": 0.17212969050820215,
+      "learning_rate": 1.9940563382223196e-05,
+      "loss": 0.4522,
+      "step": 90
+    },
+    {
+      "epoch": 0.1786064769381747,
+      "grad_norm": 0.1663706916020894,
+      "learning_rate": 1.9938854808290244e-05,
+      "loss": 0.4504,
+      "step": 91
+    },
+    {
+      "epoch": 0.18056918547595682,
+      "grad_norm": 0.31290735737771297,
+      "learning_rate": 1.9937122098932428e-05,
+      "loss": 0.4608,
+      "step": 92
+    },
+    {
+      "epoch": 0.18253189401373895,
+      "grad_norm": 0.23409310633529254,
+      "learning_rate": 1.9935365258357446e-05,
+      "loss": 0.4559,
+      "step": 93
+    },
+    {
+      "epoch": 0.1844946025515211,
+      "grad_norm": 0.2596933197806358,
+      "learning_rate": 1.9933584290831593e-05,
+      "loss": 0.4698,
+      "step": 94
+    },
+    {
+      "epoch": 0.18645731108930325,
+      "grad_norm": 0.19766333203388936,
+      "learning_rate": 1.9931779200679754e-05,
+      "loss": 0.443,
+      "step": 95
+    },
+    {
+      "epoch": 0.18842001962708538,
+      "grad_norm": 0.18968390210321073,
+      "learning_rate": 1.9929949992285397e-05,
+      "loss": 0.4623,
+      "step": 96
+    },
+    {
+      "epoch": 0.1903827281648675,
+      "grad_norm": 0.16223204516832543,
+      "learning_rate": 1.9928096670090552e-05,
+      "loss": 0.4772,
+      "step": 97
+    },
+    {
+      "epoch": 0.19234543670264967,
+      "grad_norm": 0.18715665258083516,
+      "learning_rate": 1.992621923859581e-05,
+      "loss": 0.4771,
+      "step": 98
+    },
+    {
+      "epoch": 0.1943081452404318,
+      "grad_norm": 0.2670823974648894,
+      "learning_rate": 1.992431770236031e-05,
+      "loss": 0.4625,
+      "step": 99
+    },
+    {
+      "epoch": 0.19627085377821393,
+      "grad_norm": 0.2389458544722749,
+      "learning_rate": 1.9922392066001724e-05,
+      "loss": 0.4496,
+      "step": 100
+    },
+    {
+      "epoch": 0.19823356231599606,
+      "grad_norm": 0.18959477739852718,
+      "learning_rate": 1.9920442334196248e-05,
+      "loss": 0.4685,
+      "step": 101
+    },
+    {
+      "epoch": 0.20019627085377822,
+      "grad_norm": 0.1872065955224367,
+      "learning_rate": 1.99184685116786e-05,
+      "loss": 0.4724,
+      "step": 102
+    },
+    {
+      "epoch": 0.20215897939156036,
+      "grad_norm": 0.2002235120172999,
+      "learning_rate": 1.991647060324198e-05,
+      "loss": 0.456,
+      "step": 103
+    },
+    {
+      "epoch": 0.2041216879293425,
+      "grad_norm": 0.2764357846718622,
+      "learning_rate": 1.9914448613738107e-05,
+      "loss": 0.4496,
+      "step": 104
+    },
+    {
+      "epoch": 0.20608439646712462,
+      "grad_norm": 0.1916898969998294,
+      "learning_rate": 1.991240254807715e-05,
+      "loss": 0.4584,
+      "step": 105
+    },
+    {
+      "epoch": 0.20804710500490678,
+      "grad_norm": 0.3108126102936448,
+      "learning_rate": 1.991033241122776e-05,
+      "loss": 0.4595,
+      "step": 106
+    },
+    {
+      "epoch": 0.2100098135426889,
+      "grad_norm": 0.21101958681311,
+      "learning_rate": 1.990823820821704e-05,
+      "loss": 0.4911,
+      "step": 107
+    },
+    {
+      "epoch": 0.21197252208047104,
+      "grad_norm": 0.42110628942523765,
+      "learning_rate": 1.9906119944130527e-05,
+      "loss": 0.4605,
+      "step": 108
+    },
+    {
+      "epoch": 0.2139352306182532,
+      "grad_norm": 0.17728266885723962,
+      "learning_rate": 1.9903977624112204e-05,
+      "loss": 0.461,
+      "step": 109
+    },
+    {
+      "epoch": 0.21589793915603533,
+      "grad_norm": 0.26344485864001554,
+      "learning_rate": 1.9901811253364458e-05,
+      "loss": 0.456,
+      "step": 110
+    },
+    {
+      "epoch": 0.21786064769381747,
+      "grad_norm": 0.24928437921890148,
+      "learning_rate": 1.989962083714808e-05,
+      "loss": 0.4565,
+      "step": 111
+    },
+    {
+      "epoch": 0.2198233562315996,
+      "grad_norm": 0.22760146691028113,
+      "learning_rate": 1.9897406380782262e-05,
+      "loss": 0.4617,
+      "step": 112
+    },
+    {
+      "epoch": 0.22178606476938176,
+      "grad_norm": 0.33505042507064053,
+      "learning_rate": 1.9895167889644568e-05,
+      "loss": 0.4619,
+      "step": 113
+    },
+    {
+      "epoch": 0.2237487733071639,
+      "grad_norm": 0.23380989191893622,
+      "learning_rate": 1.989290536917093e-05,
+      "loss": 0.4682,
+      "step": 114
+    },
+    {
+      "epoch": 0.22571148184494602,
+      "grad_norm": 0.16292574235785917,
+      "learning_rate": 1.9890618824855624e-05,
+      "loss": 0.4609,
+      "step": 115
+    },
+    {
+      "epoch": 0.22767419038272815,
+      "grad_norm": 0.21123922747241927,
+      "learning_rate": 1.9888308262251286e-05,
+      "loss": 0.4535,
+      "step": 116
+    },
+    {
+      "epoch": 0.2296368989205103,
+      "grad_norm": 0.19216925224306786,
+      "learning_rate": 1.988597368696886e-05,
+      "loss": 0.4666,
+      "step": 117
+    },
+    {
+      "epoch": 0.23159960745829244,
+      "grad_norm": 0.1978727333868667,
+      "learning_rate": 1.988361510467761e-05,
+      "loss": 0.4483,
+      "step": 118
+    },
+    {
+      "epoch": 0.23356231599607458,
+      "grad_norm": 0.2024852375366097,
+      "learning_rate": 1.988123252110509e-05,
+      "loss": 0.4575,
+      "step": 119
+    },
+    {
+      "epoch": 0.23552502453385674,
+      "grad_norm": 0.2275812980861413,
+      "learning_rate": 1.9878825942037147e-05,
+      "loss": 0.4733,
+      "step": 120
+    },
+    {
+      "epoch": 0.23748773307163887,
+      "grad_norm": 0.16622185526393202,
+      "learning_rate": 1.98763953733179e-05,
+      "loss": 0.4638,
+      "step": 121
+    },
+    {
+      "epoch": 0.239450441609421,
+      "grad_norm": 0.1599896679961423,
+      "learning_rate": 1.9873940820849714e-05,
+      "loss": 0.4676,
+      "step": 122
+    },
+    {
+      "epoch": 0.24141315014720313,
+      "grad_norm": 0.17524027525754993,
+      "learning_rate": 1.9871462290593206e-05,
+      "loss": 0.4671,
+      "step": 123
+    },
+    {
+      "epoch": 0.2433758586849853,
+      "grad_norm": 0.18451386556539923,
+      "learning_rate": 1.9868959788567213e-05,
+      "loss": 0.4714,
+      "step": 124
+    },
+    {
+      "epoch": 0.24533856722276742,
+      "grad_norm": 0.20332604727761022,
+      "learning_rate": 1.9866433320848793e-05,
+      "loss": 0.4395,
+      "step": 125
+    },
+    {
+      "epoch": 0.24730127576054955,
+      "grad_norm": 0.13917105010229833,
+      "learning_rate": 1.9863882893573188e-05,
+      "loss": 0.4568,
+      "step": 126
+    },
+    {
+      "epoch": 0.2492639842983317,
+      "grad_norm": 0.17751758499748868,
+      "learning_rate": 1.9861308512933846e-05,
+      "loss": 0.4746,
+      "step": 127
+    },
+    {
+      "epoch": 0.2512266928361138,
+      "grad_norm": 0.15877578177692772,
+      "learning_rate": 1.985871018518236e-05,
+      "loss": 0.4515,
+      "step": 128
+    },
+    {
+      "epoch": 0.2512266928361138,
+      "eval_loss": 0.4546394944190979,
+      "eval_runtime": 245.4163,
+      "eval_samples_per_second": 123.68,
+      "eval_steps_per_second": 3.867,
+      "step": 128
+    },
+    {
+      "epoch": 0.25318940137389595,
+      "grad_norm": 0.16746870749832415,
+      "learning_rate": 1.9856087916628487e-05,
+      "loss": 0.4564,
+      "step": 129
+    },
+    {
+      "epoch": 0.25515210991167814,
+      "grad_norm": 0.32948890713463225,
+      "learning_rate": 1.9853441713640123e-05,
+      "loss": 0.4604,
+      "step": 130
+    },
+    {
+      "epoch": 0.25711481844946027,
+      "grad_norm": 0.19647029953806538,
+      "learning_rate": 1.985077158264328e-05,
+      "loss": 0.4439,
+      "step": 131
+    },
+    {
+      "epoch": 0.2590775269872424,
+      "grad_norm": 0.251393507519347,
+      "learning_rate": 1.9848077530122083e-05,
+      "loss": 0.4574,
+      "step": 132
+    },
+    {
+      "epoch": 0.26104023552502453,
+      "grad_norm": 0.24198770425177948,
+      "learning_rate": 1.984535956261874e-05,
+      "loss": 0.4725,
+      "step": 133
+    },
+    {
+      "epoch": 0.26300294406280667,
+      "grad_norm": 0.19883583006200609,
+      "learning_rate": 1.9842617686733546e-05,
+      "loss": 0.4509,
+      "step": 134
+    },
+    {
+      "epoch": 0.2649656526005888,
+      "grad_norm": 0.19609522995459108,
+      "learning_rate": 1.983985190912484e-05,
+      "loss": 0.4513,
+      "step": 135
+    },
+    {
+      "epoch": 0.26692836113837093,
+      "grad_norm": 0.18132847552230033,
+      "learning_rate": 1.9837062236509013e-05,
+      "loss": 0.4614,
+      "step": 136
+    },
+    {
+      "epoch": 0.2688910696761531,
+      "grad_norm": 0.2617064903935584,
+      "learning_rate": 1.9834248675660484e-05,
+      "loss": 0.4729,
+      "step": 137
+    },
+    {
+      "epoch": 0.27085377821393525,
+      "grad_norm": 0.15567291724986626,
+      "learning_rate": 1.983141123341168e-05,
+      "loss": 0.4496,
+      "step": 138
+    },
+    {
+      "epoch": 0.2728164867517174,
+      "grad_norm": 0.8337178932954963,
+      "learning_rate": 1.9828549916653013e-05,
+      "loss": 0.4464,
+      "step": 139
+    },
+    {
+      "epoch": 0.2747791952894995,
+      "grad_norm": 0.13791288977874855,
+      "learning_rate": 1.9825664732332886e-05,
+      "loss": 0.4484,
+      "step": 140
+    },
+    {
+      "epoch": 0.27674190382728164,
+      "grad_norm": 0.15828576470912684,
+      "learning_rate": 1.9822755687457645e-05,
+      "loss": 0.4568,
+      "step": 141
+    },
+    {
+      "epoch": 0.2787046123650638,
+      "grad_norm": 0.18001770804302206,
+      "learning_rate": 1.9819822789091597e-05,
+      "loss": 0.4651,
+      "step": 142
+    },
+    {
+      "epoch": 0.2806673209028459,
+      "grad_norm": 0.39836155933646317,
+      "learning_rate": 1.9816866044356968e-05,
+      "loss": 0.4481,
+      "step": 143
+    },
+    {
+      "epoch": 0.2826300294406281,
+      "grad_norm": 0.15389096751590023,
+      "learning_rate": 1.981388546043388e-05,
+      "loss": 0.4604,
+      "step": 144
+    },
+    {
+      "epoch": 0.2845927379784102,
+      "grad_norm": 0.23663992095817404,
+      "learning_rate": 1.981088104456036e-05,
+      "loss": 0.4587,
+      "step": 145
+    },
+    {
+      "epoch": 0.28655544651619236,
+      "grad_norm": 0.18263790394872537,
+      "learning_rate": 1.9807852804032306e-05,
+      "loss": 0.4606,
+      "step": 146
+    },
+    {
+      "epoch": 0.2885181550539745,
+      "grad_norm": 0.1790389596426215,
+      "learning_rate": 1.980480074620347e-05,
+      "loss": 0.4492,
+      "step": 147
+    },
+    {
+      "epoch": 0.2904808635917566,
+      "grad_norm": 0.27451162166681453,
+      "learning_rate": 1.9801724878485438e-05,
+      "loss": 0.458,
+      "step": 148
+    },
+    {
+      "epoch": 0.29244357212953875,
+      "grad_norm": 0.17498754542123285,
+      "learning_rate": 1.9798625208347627e-05,
+      "loss": 0.4583,
+      "step": 149
+    },
+    {
+      "epoch": 0.2944062806673209,
+      "grad_norm": 0.9166003434108919,
+      "learning_rate": 1.979550174331724e-05,
+      "loss": 0.4599,
+      "step": 150
+    },
+    {
+      "epoch": 0.296368989205103,
+      "grad_norm": 1.7151984325661747,
+      "learning_rate": 1.9792354490979275e-05,
+      "loss": 0.4652,
+      "step": 151
+    },
+    {
+      "epoch": 0.2983316977428852,
+      "grad_norm": 3.7084263741544903,
+      "learning_rate": 1.9789183458976485e-05,
+      "loss": 0.5222,
+      "step": 152
+    },
+    {
+      "epoch": 0.30029440628066734,
+      "grad_norm": 0.23854084460339886,
+      "learning_rate": 1.9785988655009386e-05,
+      "loss": 0.4612,
+      "step": 153
+    },
+    {
+      "epoch": 0.30225711481844947,
+      "grad_norm": 0.4504871270083281,
+      "learning_rate": 1.97827700868362e-05,
+      "loss": 0.4812,
+      "step": 154
+    },
+    {
+      "epoch": 0.3042198233562316,
+      "grad_norm": 0.4201391978088661,
+      "learning_rate": 1.9779527762272877e-05,
+      "loss": 0.4855,
+      "step": 155
+    },
+    {
+      "epoch": 0.30618253189401373,
+      "grad_norm": 0.44160943541930997,
+      "learning_rate": 1.977626168919305e-05,
+      "loss": 0.4723,
+      "step": 156
+    },
+    {
+      "epoch": 0.30814524043179586,
+      "grad_norm": 0.4231731091405931,
+      "learning_rate": 1.977297187552801e-05,
+      "loss": 0.4648,
+      "step": 157
+    },
+    {
+      "epoch": 0.310107948969578,
+      "grad_norm": 0.3638067041093047,
+      "learning_rate": 1.9769658329266718e-05,
+      "loss": 0.4545,
+      "step": 158
+    },
+    {
+      "epoch": 0.3120706575073602,
+      "grad_norm": 0.4239566722192982,
+      "learning_rate": 1.976632105845576e-05,
+      "loss": 0.4849,
+      "step": 159
+    },
+    {
+      "epoch": 0.3140333660451423,
+      "grad_norm": 0.3461796288026477,
+      "learning_rate": 1.9762960071199334e-05,
+      "loss": 0.4794,
+      "step": 160
+    },
+    {
+      "epoch": 0.31599607458292445,
+      "grad_norm": 0.24887697538008421,
+      "learning_rate": 1.9759575375659232e-05,
+      "loss": 0.4814,
+      "step": 161
+    },
+    {
+      "epoch": 0.3179587831207066,
+      "grad_norm": 0.24050744980645108,
+      "learning_rate": 1.9756166980054812e-05,
+      "loss": 0.4516,
+      "step": 162
+    },
+    {
+      "epoch": 0.3199214916584887,
+      "grad_norm": 0.2345271899376135,
+      "learning_rate": 1.9752734892663e-05,
+      "loss": 0.4779,
+      "step": 163
+    },
+    {
+      "epoch": 0.32188420019627084,
+      "grad_norm": 0.27248178409002094,
+      "learning_rate": 1.9749279121818235e-05,
+      "loss": 0.4677,
+      "step": 164
+    },
+    {
+      "epoch": 0.323846908734053,
+      "grad_norm": 0.19021861355008923,
+      "learning_rate": 1.9745799675912492e-05,
+      "loss": 0.4491,
+      "step": 165
+    },
+    {
+      "epoch": 0.3258096172718351,
+      "grad_norm": 0.20109086134349027,
+      "learning_rate": 1.9742296563395218e-05,
+      "loss": 0.4485,
+      "step": 166
+    },
+    {
+      "epoch": 0.3277723258096173,
+      "grad_norm": 0.19466230218717548,
+      "learning_rate": 1.9738769792773338e-05,
+      "loss": 0.4429,
+      "step": 167
+    },
+    {
+      "epoch": 0.3297350343473994,
+      "grad_norm": 0.22618689389367447,
+      "learning_rate": 1.9735219372611232e-05,
+      "loss": 0.4776,
+      "step": 168
+    },
+    {
+      "epoch": 0.33169774288518156,
+      "grad_norm": 0.2548001164602319,
+      "learning_rate": 1.9731645311530718e-05,
+      "loss": 0.4544,
+      "step": 169
+    },
+    {
+      "epoch": 0.3336604514229637,
+      "grad_norm": 0.1791237461163614,
+      "learning_rate": 1.9728047618210995e-05,
+      "loss": 0.4554,
+      "step": 170
+    },
+    {
+      "epoch": 0.3356231599607458,
+      "grad_norm": 0.3948418555429021,
+      "learning_rate": 1.9724426301388683e-05,
+      "loss": 0.4638,
+      "step": 171
+    },
+    {
+      "epoch": 0.33758586849852795,
+      "grad_norm": 0.17378090038653812,
+      "learning_rate": 1.9720781369857747e-05,
+      "loss": 0.4454,
+      "step": 172
+    },
+    {
+      "epoch": 0.3395485770363101,
+      "grad_norm": 0.3222426037289685,
+      "learning_rate": 1.971711283246951e-05,
+      "loss": 0.4535,
+      "step": 173
+    },
+    {
+      "epoch": 0.34151128557409227,
+      "grad_norm": 0.18445752660643416,
+      "learning_rate": 1.9713420698132614e-05,
+      "loss": 0.4529,
+      "step": 174
+    },
+    {
+      "epoch": 0.3434739941118744,
+      "grad_norm": 0.16185923929968957,
+      "learning_rate": 1.9709704975813007e-05,
+      "loss": 0.4368,
+      "step": 175
+    },
+    {
+      "epoch": 0.34543670264965654,
+      "grad_norm": 1.1894973875213863,
+      "learning_rate": 1.970596567453391e-05,
+      "loss": 0.4544,
+      "step": 176
+    },
+    {
+      "epoch": 0.34739941118743867,
+      "grad_norm": 0.16134561772472725,
+      "learning_rate": 1.9702202803375813e-05,
+      "loss": 0.4622,
+      "step": 177
+    },
+    {
+      "epoch": 0.3493621197252208,
+      "grad_norm": 0.1970049090626823,
+      "learning_rate": 1.9698416371476434e-05,
+      "loss": 0.4671,
+      "step": 178
+    },
+    {
+      "epoch": 0.35132482826300293,
+      "grad_norm": 0.36331006012853106,
+      "learning_rate": 1.9694606388030715e-05,
+      "loss": 0.457,
+      "step": 179
+    },
+    {
+      "epoch": 0.35328753680078506,
+      "grad_norm": 0.17936278528599436,
+      "learning_rate": 1.969077286229078e-05,
+      "loss": 0.4548,
+      "step": 180
+    },
+    {
+      "epoch": 0.35525024533856725,
+      "grad_norm": 1.9159888836797607,
+      "learning_rate": 1.9686915803565934e-05,
+      "loss": 0.4645,
+      "step": 181
+    },
+    {
+      "epoch": 0.3572129538763494,
+      "grad_norm": 10.74627429291283,
+      "learning_rate": 1.9683035221222617e-05,
+      "loss": 0.6072,
+      "step": 182
+    },
+    {
+      "epoch": 0.3591756624141315,
+      "grad_norm": 52.13270281331323,
+      "learning_rate": 1.9679131124684403e-05,
+      "loss": 1.8264,
+      "step": 183
+    },
+    {
+      "epoch": 0.36113837095191365,
+      "grad_norm": 4.418687411757653,
+      "learning_rate": 1.9675203523431964e-05,
+      "loss": 0.5769,
+      "step": 184
+    },
+    {
+      "epoch": 0.3631010794896958,
+      "grad_norm": 0.5887657411717333,
+      "learning_rate": 1.9671252427003052e-05,
+      "loss": 0.5168,
+      "step": 185
+    },
+    {
+      "epoch": 0.3650637880274779,
+      "grad_norm": 0.674215963992249,
+      "learning_rate": 1.9667277844992476e-05,
+      "loss": 0.5026,
+      "step": 186
+    },
+    {
+      "epoch": 0.36702649656526004,
+      "grad_norm": 1.0528021833617283,
+      "learning_rate": 1.966327978705207e-05,
+      "loss": 0.5349,
+      "step": 187
+    },
+    {
+      "epoch": 0.3689892051030422,
+      "grad_norm": 0.9422062018416699,
+      "learning_rate": 1.9659258262890683e-05,
+      "loss": 0.5042,
+      "step": 188
+    },
+    {
+      "epoch": 0.37095191364082436,
+      "grad_norm": 0.38055499460304,
+      "learning_rate": 1.9655213282274153e-05,
+      "loss": 0.494,
+      "step": 189
+    },
+    {
+      "epoch": 0.3729146221786065,
+      "grad_norm": 0.7565736806021138,
+      "learning_rate": 1.9651144855025265e-05,
+      "loss": 0.4757,
+      "step": 190
+    },
+    {
+      "epoch": 0.3748773307163886,
+      "grad_norm": 0.6604416755715676,
+      "learning_rate": 1.964705299102376e-05,
+      "loss": 0.4694,
+      "step": 191
+    },
+    {
+      "epoch": 0.37684003925417076,
+      "grad_norm": 0.5198922569733609,
+      "learning_rate": 1.964293770020628e-05,
+      "loss": 0.4896,
+      "step": 192
+    },
+    {
+      "epoch": 0.3788027477919529,
+      "grad_norm": 0.3341751217641192,
+      "learning_rate": 1.9638798992566354e-05,
+      "loss": 0.482,
+      "step": 193
+    },
+    {
+      "epoch": 0.380765456329735,
+      "grad_norm": 0.34613355879167534,
+      "learning_rate": 1.9634636878154393e-05,
+      "loss": 0.4903,
+      "step": 194
+    },
+    {
+      "epoch": 0.38272816486751715,
+      "grad_norm": 0.29136481910346357,
+      "learning_rate": 1.963045136707763e-05,
+      "loss": 0.4795,
+      "step": 195
+    },
+    {
+      "epoch": 0.38469087340529934,
+      "grad_norm": 0.34118862521510146,
+      "learning_rate": 1.962624246950012e-05,
+      "loss": 0.4826,
+      "step": 196
+    },
+    {
+      "epoch": 0.38665358194308147,
+      "grad_norm": 0.45727416620673994,
+      "learning_rate": 1.9622010195642724e-05,
+      "loss": 0.4653,
+      "step": 197
+    },
+    {
+      "epoch": 0.3886162904808636,
+      "grad_norm": 0.312131527349307,
+      "learning_rate": 1.9617754555783045e-05,
+      "loss": 0.4761,
+      "step": 198
+    },
+    {
+      "epoch": 0.39057899901864573,
+      "grad_norm": 0.26137988821611824,
+      "learning_rate": 1.9613475560255445e-05,
+      "loss": 0.4792,
+      "step": 199
+    },
+    {
+      "epoch": 0.39254170755642787,
+      "grad_norm": 0.2571733191752843,
+      "learning_rate": 1.9609173219450998e-05,
+      "loss": 0.4636,
+      "step": 200
+    },
+    {
+      "epoch": 0.39450441609421,
+      "grad_norm": 0.286975647231177,
+      "learning_rate": 1.9604847543817467e-05,
+      "loss": 0.4733,
+      "step": 201
+    },
+    {
+      "epoch": 0.39646712463199213,
+      "grad_norm": 0.18272338104532077,
+      "learning_rate": 1.960049854385929e-05,
+      "loss": 0.437,
+      "step": 202
+    },
+    {
+      "epoch": 0.39842983316977426,
+      "grad_norm": 0.22862620377890577,
+      "learning_rate": 1.9596126230137532e-05,
+      "loss": 0.4634,
+      "step": 203
+    },
+    {
+      "epoch": 0.40039254170755645,
+      "grad_norm": 0.3492075746901584,
+      "learning_rate": 1.9591730613269878e-05,
+      "loss": 0.4776,
+      "step": 204
+    },
+    {
+      "epoch": 0.4023552502453386,
+      "grad_norm": 0.21584801782728166,
+      "learning_rate": 1.9587311703930615e-05,
+      "loss": 0.45,
+      "step": 205
+    },
+    {
+      "epoch": 0.4043179587831207,
+      "grad_norm": 0.1775551780611007,
+      "learning_rate": 1.9582869512850576e-05,
+      "loss": 0.4588,
+      "step": 206
+    },
+    {
+      "epoch": 0.40628066732090284,
+      "grad_norm": 0.19395395115771327,
+      "learning_rate": 1.9578404050817135e-05,
+      "loss": 0.4476,
+      "step": 207
+    },
+    {
+      "epoch": 0.408243375858685,
+      "grad_norm": 0.1756543157586606,
+      "learning_rate": 1.957391532867418e-05,
+      "loss": 0.4541,
+      "step": 208
+    },
+    {
+      "epoch": 0.4102060843964671,
+      "grad_norm": 0.2033504241274638,
+      "learning_rate": 1.956940335732209e-05,
+      "loss": 0.4517,
+      "step": 209
+    },
+    {
+      "epoch": 0.41216879293424924,
+      "grad_norm": 0.17283372188002102,
+      "learning_rate": 1.956486814771769e-05,
+      "loss": 0.4388,
+      "step": 210
+    },
+    {
+      "epoch": 0.4141315014720314,
+      "grad_norm": 0.3603837587378834,
+      "learning_rate": 1.956030971087424e-05,
+      "loss": 0.4589,
+      "step": 211
+    },
+    {
+      "epoch": 0.41609421000981356,
+      "grad_norm": 0.14136642540755467,
+      "learning_rate": 1.955572805786141e-05,
+      "loss": 0.4458,
+      "step": 212
+    },
+    {
+      "epoch": 0.4180569185475957,
+      "grad_norm": 0.5877313778359132,
+      "learning_rate": 1.9551123199805243e-05,
+      "loss": 0.4697,
+      "step": 213
+    },
+    {
+      "epoch": 0.4200196270853778,
+      "grad_norm": 0.14159168090968688,
+      "learning_rate": 1.9546495147888134e-05,
+      "loss": 0.4615,
+      "step": 214
+    },
+    {
+      "epoch": 0.42198233562315995,
+      "grad_norm": 0.14811573102512038,
+      "learning_rate": 1.9541843913348804e-05,
+      "loss": 0.4434,
+      "step": 215
+    },
+    {
+      "epoch": 0.4239450441609421,
+      "grad_norm": 0.1892192591601388,
+      "learning_rate": 1.953716950748227e-05,
+      "loss": 0.4502,
+      "step": 216
+    },
+    {
+      "epoch": 0.4259077526987242,
+      "grad_norm": 0.18436679871777018,
+      "learning_rate": 1.9532471941639816e-05,
+      "loss": 0.4583,
+      "step": 217
+    },
+    {
+      "epoch": 0.4278704612365064,
+      "grad_norm": 0.16076460592232203,
+      "learning_rate": 1.9527751227228964e-05,
+      "loss": 0.4782,
+      "step": 218
+    },
+    {
+      "epoch": 0.42983316977428854,
+      "grad_norm": 0.16974596439970596,
+      "learning_rate": 1.952300737571346e-05,
+      "loss": 0.4417,
+      "step": 219
+    },
+    {
+      "epoch": 0.43179587831207067,
+      "grad_norm": 0.1489166172695092,
+      "learning_rate": 1.9518240398613226e-05,
+      "loss": 0.45,
+      "step": 220
+    },
+    {
+      "epoch": 0.4337585868498528,
+      "grad_norm": 0.19974086406341826,
+      "learning_rate": 1.9513450307504346e-05,
+      "loss": 0.4457,
+      "step": 221
+    },
+    {
+      "epoch": 0.43572129538763493,
+      "grad_norm": 0.14988334374863602,
+      "learning_rate": 1.9508637114019037e-05,
+      "loss": 0.4416,
+      "step": 222
+    },
+    {
+      "epoch": 0.43768400392541706,
+      "grad_norm": 0.15223238304202846,
+      "learning_rate": 1.9503800829845613e-05,
+      "loss": 0.4463,
+      "step": 223
+    },
+    {
+      "epoch": 0.4396467124631992,
+      "grad_norm": 0.14747133903175189,
+      "learning_rate": 1.9498941466728462e-05,
+      "loss": 0.4408,
+      "step": 224
+    },
+    {
+      "epoch": 0.44160942100098133,
+      "grad_norm": 0.1526636764634093,
+      "learning_rate": 1.9494059036468016e-05,
+      "loss": 0.4499,
+      "step": 225
+    },
+    {
+      "epoch": 0.4435721295387635,
+      "grad_norm": 0.1628546826649206,
+      "learning_rate": 1.9489153550920726e-05,
+      "loss": 0.4439,
+      "step": 226
+    },
+    {
+      "epoch": 0.44553483807654565,
+      "grad_norm": 0.15901123933171535,
+      "learning_rate": 1.9484225021999032e-05,
+      "loss": 0.4412,
+      "step": 227
+    },
+    {
+      "epoch": 0.4474975466143278,
+      "grad_norm": 0.1490150024919776,
+      "learning_rate": 1.947927346167132e-05,
+      "loss": 0.455,
+      "step": 228
+    },
+    {
+      "epoch": 0.4494602551521099,
+      "grad_norm": 0.5516033577713996,
+      "learning_rate": 1.9474298881961918e-05,
+      "loss": 0.4566,
+      "step": 229
+    },
+    {
+      "epoch": 0.45142296368989204,
+      "grad_norm": 0.13967143480519018,
+      "learning_rate": 1.946930129495106e-05,
+      "loss": 0.4628,
+      "step": 230
+    },
+    {
+      "epoch": 0.4533856722276742,
+      "grad_norm": 0.1305332529656271,
+      "learning_rate": 1.9464280712774828e-05,
+      "loss": 0.4393,
+      "step": 231
+    },
+    {
+      "epoch": 0.4553483807654563,
+      "grad_norm": 0.1345554686370529,
+      "learning_rate": 1.945923714762516e-05,
+      "loss": 0.4545,
+      "step": 232
+    },
+    {
+      "epoch": 0.4573110893032385,
+      "grad_norm": 0.136671297145043,
+      "learning_rate": 1.9454170611749812e-05,
+      "loss": 0.4538,
+      "step": 233
+    },
+    {
+      "epoch": 0.4592737978410206,
+      "grad_norm": 0.18983035932318434,
+      "learning_rate": 1.9449081117452304e-05,
+      "loss": 0.4551,
+      "step": 234
+    },
+    {
+      "epoch": 0.46123650637880276,
+      "grad_norm": 0.13254449859971484,
+      "learning_rate": 1.9443968677091925e-05,
+      "loss": 0.4364,
+      "step": 235
+    },
+    {
+      "epoch": 0.4631992149165849,
+      "grad_norm": 0.12020638993133913,
+      "learning_rate": 1.9438833303083677e-05,
+      "loss": 0.4442,
+      "step": 236
+    },
+    {
+      "epoch": 0.465161923454367,
+      "grad_norm": 0.13012001270770882,
+      "learning_rate": 1.9433675007898255e-05,
+      "loss": 0.4397,
+      "step": 237
+    },
+    {
+      "epoch": 0.46712463199214915,
+      "grad_norm": 0.260116572268421,
+      "learning_rate": 1.9428493804062013e-05,
+      "loss": 0.4446,
+      "step": 238
+    },
+    {
+      "epoch": 0.4690873405299313,
+      "grad_norm": 0.1231982496143958,
+      "learning_rate": 1.9423289704156945e-05,
+      "loss": 0.4616,
+      "step": 239
+    },
+    {
+      "epoch": 0.47105004906771347,
+      "grad_norm": 0.11787031140608273,
+      "learning_rate": 1.9418062720820636e-05,
+      "loss": 0.4438,
+      "step": 240
+    },
+    {
+      "epoch": 0.4730127576054956,
+      "grad_norm": 0.13848283459243596,
+      "learning_rate": 1.9412812866746248e-05,
+      "loss": 0.4479,
+      "step": 241
+    },
+    {
+      "epoch": 0.47497546614327774,
+      "grad_norm": 0.12036969167700294,
+      "learning_rate": 1.9407540154682473e-05,
+      "loss": 0.4508,
+      "step": 242
+    },
+    {
+      "epoch": 0.47693817468105987,
+      "grad_norm": 0.17542281239029114,
+      "learning_rate": 1.9402244597433526e-05,
+      "loss": 0.4393,
+      "step": 243
+    },
+    {
+      "epoch": 0.478900883218842,
+      "grad_norm": 0.1137790859381083,
+      "learning_rate": 1.9396926207859085e-05,
+      "loss": 0.4444,
+      "step": 244
+    },
+    {
+      "epoch": 0.48086359175662413,
+      "grad_norm": 0.18308427216283138,
+      "learning_rate": 1.939158499887428e-05,
+      "loss": 0.4272,
+      "step": 245
+    },
+    {
+      "epoch": 0.48282630029440626,
+      "grad_norm": 0.13610805029149947,
+      "learning_rate": 1.9386220983449652e-05,
+      "loss": 0.4453,
+      "step": 246
+    },
+    {
+      "epoch": 0.4847890088321884,
+      "grad_norm": 0.12057983580084324,
+      "learning_rate": 1.938083417461113e-05,
+      "loss": 0.4298,
+      "step": 247
+    },
+    {
+      "epoch": 0.4867517173699706,
+      "grad_norm": 0.11085757126201562,
+      "learning_rate": 1.9375424585439994e-05,
+      "loss": 0.4389,
+      "step": 248
+    },
+    {
+      "epoch": 0.4887144259077527,
+      "grad_norm": 0.11914366344102803,
+      "learning_rate": 1.9369992229072834e-05,
+      "loss": 0.4427,
+      "step": 249
+    },
+    {
+      "epoch": 0.49067713444553485,
+      "grad_norm": 0.13244864999661263,
+      "learning_rate": 1.9364537118701542e-05,
+      "loss": 0.4347,
+      "step": 250
+    },
+    {
+      "epoch": 0.492639842983317,
+      "grad_norm": 0.19302953420237423,
+      "learning_rate": 1.935905926757326e-05,
+      "loss": 0.4499,
+      "step": 251
+    },
+    {
+      "epoch": 0.4946025515210991,
+      "grad_norm": 0.11571291217554695,
+      "learning_rate": 1.935355868899034e-05,
+      "loss": 0.4375,
+      "step": 252
+    },
+    {
+      "epoch": 0.49656526005888124,
+      "grad_norm": 0.11702904926416031,
+      "learning_rate": 1.9348035396310352e-05,
+      "loss": 0.4354,
+      "step": 253
+    },
+    {
+      "epoch": 0.4985279685966634,
+      "grad_norm": 0.12006868637499236,
+      "learning_rate": 1.9342489402945997e-05,
+      "loss": 0.4338,
+      "step": 254
+    },
+    {
+      "epoch": 0.5004906771344455,
+      "grad_norm": 0.12190055167354476,
+      "learning_rate": 1.9336920722365126e-05,
+      "loss": 0.4434,
+      "step": 255
+    },
+    {
+      "epoch": 0.5024533856722276,
+      "grad_norm": 0.12867582275673506,
+      "learning_rate": 1.9331329368090664e-05,
+      "loss": 0.435,
+      "step": 256
+    },
+    {
+      "epoch": 0.5024533856722276,
+      "eval_loss": 0.44365355372428894,
+      "eval_runtime": 245.6487,
+      "eval_samples_per_second": 123.563,
+      "eval_steps_per_second": 3.863,
+      "step": 256
+    },
+    {
+      "epoch": 0.5044160942100098,
+      "grad_norm": 0.1233612138867142,
+      "learning_rate": 1.932571535370061e-05,
+      "loss": 0.4417,
+      "step": 257
+    },
+    {
+      "epoch": 0.5063788027477919,
+      "grad_norm": 2.5334855914455368,
+      "learning_rate": 1.932007869282799e-05,
+      "loss": 0.4645,
+      "step": 258
+    },
+    {
+      "epoch": 0.5083415112855741,
+      "grad_norm": 0.15972840390446322,
+      "learning_rate": 1.9314419399160805e-05,
+      "loss": 0.4328,
+      "step": 259
+    },
+    {
+      "epoch": 0.5103042198233563,
+      "grad_norm": 0.21045316053724383,
+      "learning_rate": 1.9308737486442045e-05,
+      "loss": 0.437,
+      "step": 260
+    },
+    {
+      "epoch": 0.5122669283611384,
+      "grad_norm": 0.1353556883894671,
+      "learning_rate": 1.930303296846961e-05,
+      "loss": 0.4532,
+      "step": 261
+    },
+    {
+      "epoch": 0.5142296368989205,
+      "grad_norm": 0.13254413062805587,
+      "learning_rate": 1.9297305859096305e-05,
+      "loss": 0.452,
+      "step": 262
+    },
+    {
+      "epoch": 0.5161923454367027,
+      "grad_norm": 0.19171081658979544,
+      "learning_rate": 1.9291556172229784e-05,
+      "loss": 0.4302,
+      "step": 263
+    },
+    {
+      "epoch": 0.5181550539744848,
+      "grad_norm": 0.13134834595730802,
+      "learning_rate": 1.9285783921832537e-05,
+      "loss": 0.4405,
+      "step": 264
+    },
+    {
+      "epoch": 0.5201177625122669,
+      "grad_norm": 0.1301982247003894,
+      "learning_rate": 1.9279989121921846e-05,
+      "loss": 0.4462,
+      "step": 265
+    },
+    {
+      "epoch": 0.5220804710500491,
+      "grad_norm": 0.1321466010763688,
+      "learning_rate": 1.927417178656975e-05,
+      "loss": 0.4336,
+      "step": 266
+    },
+    {
+      "epoch": 0.5240431795878312,
+      "grad_norm": 0.1495881568007904,
+      "learning_rate": 1.9268331929903013e-05,
+      "loss": 0.4365,
+      "step": 267
+    },
+    {
+      "epoch": 0.5260058881256133,
+      "grad_norm": 0.2612207807396482,
+      "learning_rate": 1.926246956610309e-05,
+      "loss": 0.4457,
+      "step": 268
+    },
+    {
+      "epoch": 0.5279685966633955,
+      "grad_norm": 0.9154910569331776,
+      "learning_rate": 1.9256584709406093e-05,
+      "loss": 0.4553,
+      "step": 269
+    },
+    {
+      "epoch": 0.5299313052011776,
+      "grad_norm": 0.1534831664232736,
+      "learning_rate": 1.9250677374102752e-05,
+      "loss": 0.4441,
+      "step": 270
+    },
+    {
+      "epoch": 0.5318940137389597,
+      "grad_norm": 0.14984085964867322,
+      "learning_rate": 1.9244747574538387e-05,
+      "loss": 0.4306,
+      "step": 271
+    },
+    {
+      "epoch": 0.5338567222767419,
+      "grad_norm": 0.18760798027966705,
+      "learning_rate": 1.9238795325112867e-05,
+      "loss": 0.4538,
+      "step": 272
+    },
+    {
+      "epoch": 0.535819430814524,
+      "grad_norm": 0.1580188625628718,
+      "learning_rate": 1.923282064028059e-05,
+      "loss": 0.4443,
+      "step": 273
+    },
+    {
+      "epoch": 0.5377821393523062,
+      "grad_norm": 0.14444473834290128,
+      "learning_rate": 1.9226823534550418e-05,
+      "loss": 0.4443,
+      "step": 274
+    },
+    {
+      "epoch": 0.5397448478900884,
+      "grad_norm": 0.7330314247606119,
+      "learning_rate": 1.9220804022485674e-05,
+      "loss": 0.4638,
+      "step": 275
+    },
+    {
+      "epoch": 0.5417075564278705,
+      "grad_norm": 0.17268497255370896,
+      "learning_rate": 1.921476211870408e-05,
+      "loss": 0.4503,
+      "step": 276
+    },
+    {
+      "epoch": 0.5436702649656526,
+      "grad_norm": 0.18226707950341117,
+      "learning_rate": 1.920869783787774e-05,
+      "loss": 0.4383,
+      "step": 277
+    },
+    {
+      "epoch": 0.5456329735034348,
+      "grad_norm": 0.5632696487176844,
+      "learning_rate": 1.9202611194733107e-05,
+      "loss": 0.4452,
+      "step": 278
+    },
+    {
+      "epoch": 0.5475956820412169,
+      "grad_norm": 0.14836959675348457,
+      "learning_rate": 1.9196502204050925e-05,
+      "loss": 0.4314,
+      "step": 279
+    },
+    {
+      "epoch": 0.549558390578999,
+      "grad_norm": 0.34506199709023483,
+      "learning_rate": 1.9190370880666206e-05,
+      "loss": 0.4544,
+      "step": 280
+    },
+    {
+      "epoch": 0.5515210991167812,
+      "grad_norm": 0.16441747745890337,
+      "learning_rate": 1.9184217239468213e-05,
+      "loss": 0.4429,
+      "step": 281
+    },
+    {
+      "epoch": 0.5534838076545633,
+      "grad_norm": 0.16385435349212213,
+      "learning_rate": 1.9178041295400383e-05,
+      "loss": 0.46,
+      "step": 282
+    },
+    {
+      "epoch": 0.5554465161923454,
+      "grad_norm": 0.16456135455973658,
+      "learning_rate": 1.917184306346032e-05,
+      "loss": 0.4399,
+      "step": 283
+    },
+    {
+      "epoch": 0.5574092247301276,
+      "grad_norm": 0.14576272645037303,
+      "learning_rate": 1.9165622558699763e-05,
+      "loss": 0.4419,
+      "step": 284
+    },
+    {
+      "epoch": 0.5593719332679097,
+      "grad_norm": 0.15456597025956298,
+      "learning_rate": 1.9159379796224524e-05,
+      "loss": 0.449,
+      "step": 285
+    },
+    {
+      "epoch": 0.5613346418056918,
+      "grad_norm": 0.14297456086426616,
+      "learning_rate": 1.9153114791194475e-05,
+      "loss": 0.4371,
+      "step": 286
+    },
+    {
+      "epoch": 0.563297350343474,
+      "grad_norm": 0.14012901187447932,
+      "learning_rate": 1.914682755882349e-05,
+      "loss": 0.4408,
+      "step": 287
+    },
+    {
+      "epoch": 0.5652600588812562,
+      "grad_norm": 0.13060314277737214,
+      "learning_rate": 1.9140518114379433e-05,
+      "loss": 0.4437,
+      "step": 288
+    },
+    {
+      "epoch": 0.5672227674190383,
+      "grad_norm": 0.3090626051049369,
+      "learning_rate": 1.91341864731841e-05,
+      "loss": 0.4542,
+      "step": 289
+    },
+    {
+      "epoch": 0.5691854759568205,
+      "grad_norm": 0.2179414551780141,
+      "learning_rate": 1.912783265061319e-05,
+      "loss": 0.4408,
+      "step": 290
+    },
+    {
+      "epoch": 0.5711481844946026,
+      "grad_norm": 0.620377656428685,
+      "learning_rate": 1.912145666209627e-05,
+      "loss": 0.4879,
+      "step": 291
+    },
+    {
+      "epoch": 0.5731108930323847,
+      "grad_norm": 0.1689086534993839,
+      "learning_rate": 1.9115058523116734e-05,
+      "loss": 0.4462,
+      "step": 292
+    },
+    {
+      "epoch": 0.5750736015701668,
+      "grad_norm": 0.14611412522744563,
+      "learning_rate": 1.910863824921176e-05,
+      "loss": 0.4255,
+      "step": 293
+    },
+    {
+      "epoch": 0.577036310107949,
+      "grad_norm": 0.1882222162638847,
+      "learning_rate": 1.9102195855972287e-05,
+      "loss": 0.4425,
+      "step": 294
+    },
+    {
+      "epoch": 0.5789990186457311,
+      "grad_norm": 0.1656815361127913,
+      "learning_rate": 1.909573135904296e-05,
+      "loss": 0.4283,
+      "step": 295
+    },
+    {
+      "epoch": 0.5809617271835132,
+      "grad_norm": 0.13073397914317533,
+      "learning_rate": 1.908924477412211e-05,
+      "loss": 0.4558,
+      "step": 296
+    },
+    {
+      "epoch": 0.5829244357212954,
+      "grad_norm": 0.1441943832714689,
+      "learning_rate": 1.90827361169617e-05,
+      "loss": 0.4427,
+      "step": 297
+    },
+    {
+      "epoch": 0.5848871442590775,
+      "grad_norm": 0.12583976569411942,
+      "learning_rate": 1.9076205403367287e-05,
+      "loss": 0.4223,
+      "step": 298
+    },
+    {
+      "epoch": 0.5868498527968596,
+      "grad_norm": 0.1755315808099223,
+      "learning_rate": 1.9069652649198004e-05,
+      "loss": 0.4421,
+      "step": 299
+    },
+    {
+      "epoch": 0.5888125613346418,
+      "grad_norm": 0.12236948996858302,
+      "learning_rate": 1.9063077870366504e-05,
+      "loss": 0.4403,
+      "step": 300
+    },
+    {
+      "epoch": 0.5907752698724239,
+      "grad_norm": 0.13958618892557004,
+      "learning_rate": 1.905648108283891e-05,
+      "loss": 0.4304,
+      "step": 301
+    },
+    {
+      "epoch": 0.592737978410206,
+      "grad_norm": 0.14279319639440627,
+      "learning_rate": 1.90498623026348e-05,
+      "loss": 0.4461,
+      "step": 302
+    },
+    {
+      "epoch": 0.5947006869479883,
+      "grad_norm": 0.3909906198333688,
+      "learning_rate": 1.9043221545827172e-05,
+      "loss": 0.4845,
+      "step": 303
+    },
+    {
+      "epoch": 0.5966633954857704,
+      "grad_norm": 0.37297343958892704,
+      "learning_rate": 1.903655882854237e-05,
+      "loss": 0.4494,
+      "step": 304
+    },
+    {
+      "epoch": 0.5986261040235525,
+      "grad_norm": 0.8045487515732387,
+      "learning_rate": 1.9029874166960075e-05,
+      "loss": 0.4651,
+      "step": 305
+    },
+    {
+      "epoch": 0.6005888125613347,
+      "grad_norm": 0.17675678913043202,
+      "learning_rate": 1.9023167577313267e-05,
+      "loss": 0.4342,
+      "step": 306
+    },
+    {
+      "epoch": 0.6025515210991168,
+      "grad_norm": 0.4614395798632603,
+      "learning_rate": 1.901643907588816e-05,
+      "loss": 0.4666,
+      "step": 307
+    },
+    {
+      "epoch": 0.6045142296368989,
+      "grad_norm": 0.24112777104201236,
+      "learning_rate": 1.900968867902419e-05,
+      "loss": 0.4608,
+      "step": 308
+    },
+    {
+      "epoch": 0.6064769381746811,
+      "grad_norm": 0.16289640981921516,
+      "learning_rate": 1.900291640311396e-05,
+      "loss": 0.4347,
+      "step": 309
+    },
+    {
+      "epoch": 0.6084396467124632,
+      "grad_norm": 0.16490807898196183,
+      "learning_rate": 1.8996122264603202e-05,
+      "loss": 0.4286,
+      "step": 310
+    },
+    {
+      "epoch": 0.6104023552502453,
+      "grad_norm": 0.31752463155917027,
+      "learning_rate": 1.8989306279990736e-05,
+      "loss": 0.4721,
+      "step": 311
+    },
+    {
+      "epoch": 0.6123650637880275,
+      "grad_norm": 0.16490476925454267,
+      "learning_rate": 1.898246846582844e-05,
+      "loss": 0.4482,
+      "step": 312
+    },
+    {
+      "epoch": 0.6143277723258096,
+      "grad_norm": 0.15956184991382258,
+      "learning_rate": 1.897560883872121e-05,
+      "loss": 0.4417,
+      "step": 313
+    },
+    {
+      "epoch": 0.6162904808635917,
+      "grad_norm": 0.1831073214143379,
+      "learning_rate": 1.8968727415326885e-05,
+      "loss": 0.4441,
+      "step": 314
+    },
+    {
+      "epoch": 0.6182531894013739,
+      "grad_norm": 0.3402383015575775,
+      "learning_rate": 1.896182421235626e-05,
+      "loss": 0.4643,
+      "step": 315
+    },
+    {
+      "epoch": 0.620215897939156,
+      "grad_norm": 0.15056250902532822,
+      "learning_rate": 1.895489924657301e-05,
+      "loss": 0.4472,
+      "step": 316
+    },
+    {
+      "epoch": 0.6221786064769381,
+      "grad_norm": 0.17384419824788128,
+      "learning_rate": 1.8947952534793663e-05,
+      "loss": 0.4455,
+      "step": 317
+    },
+    {
+      "epoch": 0.6241413150147204,
+      "grad_norm": 0.18283666562150924,
+      "learning_rate": 1.894098409388754e-05,
+      "loss": 0.4425,
+      "step": 318
+    },
+    {
+      "epoch": 0.6261040235525025,
+      "grad_norm": 0.18363348573791377,
+      "learning_rate": 1.8933993940776753e-05,
+      "loss": 0.4347,
+      "step": 319
+    },
+    {
+      "epoch": 0.6280667320902846,
+      "grad_norm": 0.2173940703554356,
+      "learning_rate": 1.8926982092436117e-05,
+      "loss": 0.4426,
+      "step": 320
+    },
+    {
+      "epoch": 0.6300294406280668,
+      "grad_norm": 0.13338358899432057,
+      "learning_rate": 1.8919948565893144e-05,
+      "loss": 0.4348,
+      "step": 321
+    },
+    {
+      "epoch": 0.6319921491658489,
+      "grad_norm": 0.15291106806170288,
+      "learning_rate": 1.8912893378227984e-05,
+      "loss": 0.4441,
+      "step": 322
+    },
+    {
+      "epoch": 0.633954857703631,
+      "grad_norm": 0.19828098406270409,
+      "learning_rate": 1.8905816546573398e-05,
+      "loss": 0.4498,
+      "step": 323
+    },
+    {
+      "epoch": 0.6359175662414132,
+      "grad_norm": 0.12561855756704574,
+      "learning_rate": 1.8898718088114688e-05,
+      "loss": 0.4501,
+      "step": 324
+    },
+    {
+      "epoch": 0.6378802747791953,
+      "grad_norm": 0.413637068467992,
+      "learning_rate": 1.8891598020089695e-05,
+      "loss": 0.4522,
+      "step": 325
+    },
+    {
+      "epoch": 0.6398429833169774,
+      "grad_norm": 0.13065710530489547,
+      "learning_rate": 1.8884456359788725e-05,
+      "loss": 0.4517,
+      "step": 326
+    },
+    {
+      "epoch": 0.6418056918547596,
+      "grad_norm": 0.1255320805445076,
+      "learning_rate": 1.887729312455452e-05,
+      "loss": 0.4328,
+      "step": 327
+    },
+    {
+      "epoch": 0.6437684003925417,
+      "grad_norm": 0.17052107382561416,
+      "learning_rate": 1.887010833178222e-05,
+      "loss": 0.4335,
+      "step": 328
+    },
+    {
+      "epoch": 0.6457311089303238,
+      "grad_norm": 0.1463632116200419,
+      "learning_rate": 1.8862901998919305e-05,
+      "loss": 0.4489,
+      "step": 329
+    },
+    {
+      "epoch": 0.647693817468106,
+      "grad_norm": 0.20054042393653457,
+      "learning_rate": 1.8855674143465567e-05,
+      "loss": 0.4415,
+      "step": 330
+    },
+    {
+      "epoch": 0.6496565260058881,
+      "grad_norm": 0.12528395471108425,
+      "learning_rate": 1.8848424782973075e-05,
+      "loss": 0.4251,
+      "step": 331
+    },
+    {
+      "epoch": 0.6516192345436702,
+      "grad_norm": 0.12724271501474496,
+      "learning_rate": 1.8841153935046098e-05,
+      "loss": 0.4226,
+      "step": 332
+    },
+    {
+      "epoch": 0.6535819430814525,
+      "grad_norm": 0.12815480846276966,
+      "learning_rate": 1.8833861617341108e-05,
+      "loss": 0.4389,
+      "step": 333
+    },
+    {
+      "epoch": 0.6555446516192346,
+      "grad_norm": 0.11881092687539777,
+      "learning_rate": 1.8826547847566692e-05,
+      "loss": 0.4301,
+      "step": 334
+    },
+    {
+      "epoch": 0.6575073601570167,
+      "grad_norm": 0.11279466332160643,
+      "learning_rate": 1.881921264348355e-05,
+      "loss": 0.4441,
+      "step": 335
+    },
+    {
+      "epoch": 0.6594700686947988,
+      "grad_norm": 0.12334758226552325,
+      "learning_rate": 1.8811856022904423e-05,
+      "loss": 0.4333,
+      "step": 336
+    },
+    {
+      "epoch": 0.661432777232581,
+      "grad_norm": 0.1268483867046061,
+      "learning_rate": 1.8804478003694066e-05,
+      "loss": 0.4317,
+      "step": 337
+    },
+    {
+      "epoch": 0.6633954857703631,
+      "grad_norm": 0.12443337272902948,
+      "learning_rate": 1.8797078603769184e-05,
+      "loss": 0.4282,
+      "step": 338
+    },
+    {
+      "epoch": 0.6653581943081452,
+      "grad_norm": 0.12713474397441266,
+      "learning_rate": 1.878965784109842e-05,
+      "loss": 0.4482,
+      "step": 339
+    },
+    {
+      "epoch": 0.6673209028459274,
+      "grad_norm": 0.12143777103722954,
+      "learning_rate": 1.8782215733702286e-05,
+      "loss": 0.4342,
+      "step": 340
+    },
+    {
+      "epoch": 0.6692836113837095,
+      "grad_norm": 0.23691651644437595,
+      "learning_rate": 1.877475229965313e-05,
+      "loss": 0.4467,
+      "step": 341
+    },
+    {
+      "epoch": 0.6712463199214916,
+      "grad_norm": 0.13966524868065613,
+      "learning_rate": 1.876726755707508e-05,
+      "loss": 0.4119,
+      "step": 342
+    },
+    {
+      "epoch": 0.6732090284592738,
+      "grad_norm": 0.1324720435854753,
+      "learning_rate": 1.875976152414402e-05,
+      "loss": 0.4327,
+      "step": 343
+    },
+    {
+      "epoch": 0.6751717369970559,
+      "grad_norm": 0.1305963044017511,
+      "learning_rate": 1.8752234219087538e-05,
+      "loss": 0.4345,
+      "step": 344
+    },
+    {
+      "epoch": 0.677134445534838,
+      "grad_norm": 0.1256369439378997,
+      "learning_rate": 1.8744685660184868e-05,
+      "loss": 0.4373,
+      "step": 345
+    },
+    {
+      "epoch": 0.6790971540726202,
+      "grad_norm": 0.1306844371007657,
+      "learning_rate": 1.8737115865766865e-05,
+      "loss": 0.4326,
+      "step": 346
+    },
+    {
+      "epoch": 0.6810598626104023,
+      "grad_norm": 0.122862696340445,
+      "learning_rate": 1.8729524854215942e-05,
+      "loss": 0.4166,
+      "step": 347
+    },
+    {
+      "epoch": 0.6830225711481845,
+      "grad_norm": 0.13535420265000028,
+      "learning_rate": 1.8721912643966055e-05,
+      "loss": 0.4237,
+      "step": 348
+    },
+    {
+      "epoch": 0.6849852796859667,
+      "grad_norm": 0.11284294465947456,
+      "learning_rate": 1.8714279253502616e-05,
+      "loss": 0.4259,
+      "step": 349
+    },
+    {
+      "epoch": 0.6869479882237488,
+      "grad_norm": 0.1193357989589537,
+      "learning_rate": 1.8706624701362485e-05,
+      "loss": 0.4467,
+      "step": 350
+    },
+    {
+      "epoch": 0.6889106967615309,
+      "grad_norm": 0.1131129128247003,
+      "learning_rate": 1.8698949006133903e-05,
+      "loss": 0.4366,
+      "step": 351
+    },
+    {
+      "epoch": 0.6908734052993131,
+      "grad_norm": 0.111747855862703,
+      "learning_rate": 1.8691252186456465e-05,
+      "loss": 0.4433,
+      "step": 352
+    },
+    {
+      "epoch": 0.6928361138370952,
+      "grad_norm": 0.10862949730132815,
+      "learning_rate": 1.8683534261021058e-05,
+      "loss": 0.4299,
+      "step": 353
+    },
+    {
+      "epoch": 0.6947988223748773,
+      "grad_norm": 0.12299814873161867,
+      "learning_rate": 1.8675795248569816e-05,
+      "loss": 0.4366,
+      "step": 354
+    },
+    {
+      "epoch": 0.6967615309126595,
+      "grad_norm": 0.11664167882155978,
+      "learning_rate": 1.866803516789609e-05,
+      "loss": 0.4379,
+      "step": 355
+    },
+    {
+      "epoch": 0.6987242394504416,
+      "grad_norm": 0.12925426587807418,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.4333,
+      "step": 356
+    },
+    {
+      "epoch": 0.7006869479882237,
+      "grad_norm": 0.10893935119976447,
+      "learning_rate": 1.8652451877310337e-05,
+      "loss": 0.4293,
+      "step": 357
+    },
+    {
+      "epoch": 0.7026496565260059,
+      "grad_norm": 0.1227760342543677,
+      "learning_rate": 1.8644628705240636e-05,
+      "loss": 0.4255,
+      "step": 358
+    },
+    {
+      "epoch": 0.704612365063788,
+      "grad_norm": 0.11162933142481653,
+      "learning_rate": 1.8636784540633002e-05,
+      "loss": 0.4277,
+      "step": 359
+    },
+    {
+      "epoch": 0.7065750736015701,
+      "grad_norm": 0.11784511277232618,
+      "learning_rate": 1.862891940253613e-05,
+      "loss": 0.43,
+      "step": 360
+    },
+    {
+      "epoch": 0.7085377821393523,
+      "grad_norm": 0.19349027686476353,
+      "learning_rate": 1.8621033310049658e-05,
+      "loss": 0.4388,
+      "step": 361
+    },
+    {
+      "epoch": 0.7105004906771345,
+      "grad_norm": 0.11461754333230102,
+      "learning_rate": 1.8613126282324092e-05,
+      "loss": 0.4459,
+      "step": 362
+    },
+    {
+      "epoch": 0.7124631992149166,
+      "grad_norm": 0.10611440044540593,
+      "learning_rate": 1.860519833856079e-05,
+      "loss": 0.4286,
+      "step": 363
+    },
+    {
+      "epoch": 0.7144259077526988,
+      "grad_norm": 1.3919982408500475,
+      "learning_rate": 1.8597249498011906e-05,
+      "loss": 0.4271,
+      "step": 364
+    },
+    {
+      "epoch": 0.7163886162904809,
+      "grad_norm": 0.23104503037861346,
+      "learning_rate": 1.858927977998032e-05,
+      "loss": 0.4489,
+      "step": 365
+    },
+    {
+      "epoch": 0.718351324828263,
+      "grad_norm": 0.13423371418728916,
+      "learning_rate": 1.858128920381963e-05,
+      "loss": 0.4432,
+      "step": 366
+    },
+    {
+      "epoch": 0.7203140333660452,
+      "grad_norm": 0.14035310994358413,
+      "learning_rate": 1.8573277788934084e-05,
+      "loss": 0.4436,
+      "step": 367
+    },
+    {
+      "epoch": 0.7222767419038273,
+      "grad_norm": 0.12538177756048358,
+      "learning_rate": 1.8565245554778516e-05,
+      "loss": 0.4152,
+      "step": 368
+    },
+    {
+      "epoch": 0.7242394504416094,
+      "grad_norm": 0.1393704340215518,
+      "learning_rate": 1.855719252085834e-05,
+      "loss": 0.4599,
+      "step": 369
+    },
+    {
+      "epoch": 0.7262021589793916,
+      "grad_norm": 0.15221198393758803,
+      "learning_rate": 1.854911870672947e-05,
+      "loss": 0.4346,
+      "step": 370
+    },
+    {
+      "epoch": 0.7281648675171737,
+      "grad_norm": 0.13205252608835663,
+      "learning_rate": 1.8541024131998277e-05,
+      "loss": 0.433,
+      "step": 371
+    },
+    {
+      "epoch": 0.7301275760549558,
+      "grad_norm": 0.1824152481964556,
+      "learning_rate": 1.8532908816321557e-05,
+      "loss": 0.4258,
+      "step": 372
+    },
+    {
+      "epoch": 0.732090284592738,
+      "grad_norm": 0.11473500921109103,
+      "learning_rate": 1.852477277940647e-05,
+      "loss": 0.4364,
+      "step": 373
+    },
+    {
+      "epoch": 0.7340529931305201,
+      "grad_norm": 0.17550775961163126,
+      "learning_rate": 1.8516616041010495e-05,
+      "loss": 0.4311,
+      "step": 374
+    },
+    {
+      "epoch": 0.7360157016683022,
+      "grad_norm": 0.11491081744632524,
+      "learning_rate": 1.850843862094138e-05,
+      "loss": 0.4299,
+      "step": 375
+    },
+    {
+      "epoch": 0.7379784102060843,
+      "grad_norm": 0.12893078747082584,
+      "learning_rate": 1.8500240539057093e-05,
+      "loss": 0.4375,
+      "step": 376
+    },
+    {
+      "epoch": 0.7399411187438666,
+      "grad_norm": 0.1386244465195304,
+      "learning_rate": 1.849202181526579e-05,
+      "loss": 0.4411,
+      "step": 377
+    },
+    {
+      "epoch": 0.7419038272816487,
+      "grad_norm": 0.1363795250121767,
+      "learning_rate": 1.848378246952574e-05,
+      "loss": 0.4294,
+      "step": 378
+    },
+    {
+      "epoch": 0.7438665358194309,
+      "grad_norm": 0.11138728492984143,
+      "learning_rate": 1.8475522521845296e-05,
+      "loss": 0.4403,
+      "step": 379
+    },
+    {
+      "epoch": 0.745829244357213,
+      "grad_norm": 0.1238513979749556,
+      "learning_rate": 1.8467241992282842e-05,
+      "loss": 0.4267,
+      "step": 380
+    },
+    {
+      "epoch": 0.7477919528949951,
+      "grad_norm": 0.12373568951195847,
+      "learning_rate": 1.845894090094674e-05,
+      "loss": 0.4312,
+      "step": 381
+    },
+    {
+      "epoch": 0.7497546614327772,
+      "grad_norm": 0.19653148756122196,
+      "learning_rate": 1.8450619267995283e-05,
+      "loss": 0.4343,
+      "step": 382
+    },
+    {
+      "epoch": 0.7517173699705594,
+      "grad_norm": 0.1394570505619246,
+      "learning_rate": 1.8442277113636654e-05,
+      "loss": 0.4373,
+      "step": 383
+    },
+    {
+      "epoch": 0.7536800785083415,
+      "grad_norm": 0.12251934848515013,
+      "learning_rate": 1.843391445812886e-05,
+      "loss": 0.4267,
+      "step": 384
+    },
+    {
+      "epoch": 0.7536800785083415,
+      "eval_loss": 0.43717867136001587,
+      "eval_runtime": 245.9442,
+      "eval_samples_per_second": 123.414,
+      "eval_steps_per_second": 3.859,
+      "step": 384
+    },
+    {
+      "epoch": 0.7556427870461236,
+      "grad_norm": 0.11928267779998,
+      "learning_rate": 1.84255313217797e-05,
+      "loss": 0.4411,
+      "step": 385
+    },
+    {
+      "epoch": 0.7576054955839058,
+      "grad_norm": 0.10973043663025576,
+      "learning_rate": 1.84171277249467e-05,
+      "loss": 0.4266,
+      "step": 386
+    },
+    {
+      "epoch": 0.7595682041216879,
+      "grad_norm": 0.6751166131440709,
+      "learning_rate": 1.8408703688037088e-05,
+      "loss": 0.4442,
+      "step": 387
+    },
+    {
+      "epoch": 0.76153091265947,
+      "grad_norm": 0.11525859230534183,
+      "learning_rate": 1.8400259231507716e-05,
+      "loss": 0.4257,
+      "step": 388
+    },
+    {
+      "epoch": 0.7634936211972522,
+      "grad_norm": 0.14081912497481,
+      "learning_rate": 1.8391794375865025e-05,
+      "loss": 0.4559,
+      "step": 389
+    },
+    {
+      "epoch": 0.7654563297350343,
+      "grad_norm": 0.1430446337142817,
+      "learning_rate": 1.8383309141664992e-05,
+      "loss": 0.4348,
+      "step": 390
+    },
+    {
+      "epoch": 0.7674190382728164,
+      "grad_norm": 0.13701747456448043,
+      "learning_rate": 1.837480354951308e-05,
+      "loss": 0.4355,
+      "step": 391
+    },
+    {
+      "epoch": 0.7693817468105987,
+      "grad_norm": 0.15468002573828674,
+      "learning_rate": 1.83662776200642e-05,
+      "loss": 0.4247,
+      "step": 392
+    },
+    {
+      "epoch": 0.7713444553483808,
+      "grad_norm": 0.1359423923850112,
+      "learning_rate": 1.8357731374022635e-05,
+      "loss": 0.4378,
+      "step": 393
+    },
+    {
+      "epoch": 0.7733071638861629,
+      "grad_norm": 0.12156559871467941,
+      "learning_rate": 1.8349164832142015e-05,
+      "loss": 0.433,
+      "step": 394
+    },
+    {
+      "epoch": 0.7752698724239451,
+      "grad_norm": 1.1029116856223753,
+      "learning_rate": 1.834057801522525e-05,
+      "loss": 0.456,
+      "step": 395
+    },
+    {
+      "epoch": 0.7772325809617272,
+      "grad_norm": 0.1498532799383752,
+      "learning_rate": 1.833197094412449e-05,
+      "loss": 0.4416,
+      "step": 396
+    },
+    {
+      "epoch": 0.7791952894995093,
+      "grad_norm": 0.4425396937410584,
+      "learning_rate": 1.832334363974107e-05,
+      "loss": 0.4358,
+      "step": 397
+    },
+    {
+      "epoch": 0.7811579980372915,
+      "grad_norm": 0.14741438537186383,
+      "learning_rate": 1.8314696123025456e-05,
+      "loss": 0.4432,
+      "step": 398
+    },
+    {
+      "epoch": 0.7831207065750736,
+      "grad_norm": 1.3635940422702577,
+      "learning_rate": 1.8306028414977196e-05,
+      "loss": 0.4473,
+      "step": 399
+    },
+    {
+      "epoch": 0.7850834151128557,
+      "grad_norm": 0.31067568125771744,
+      "learning_rate": 1.8297340536644877e-05,
+      "loss": 0.4341,
+      "step": 400
+    },
+    {
+      "epoch": 0.7870461236506379,
+      "grad_norm": 0.23052926471546017,
+      "learning_rate": 1.8288632509126064e-05,
+      "loss": 0.4469,
+      "step": 401
+    },
+    {
+      "epoch": 0.78900883218842,
+      "grad_norm": 1.4150298223131084,
+      "learning_rate": 1.827990435356725e-05,
+      "loss": 0.4573,
+      "step": 402
+    },
+    {
+      "epoch": 0.7909715407262021,
+      "grad_norm": 1.2166261643274474,
+      "learning_rate": 1.8271156091163813e-05,
+      "loss": 0.4623,
+      "step": 403
+    },
+    {
+      "epoch": 0.7929342492639843,
+      "grad_norm": 8.194510529799096,
+      "learning_rate": 1.826238774315995e-05,
+      "loss": 0.9405,
+      "step": 404
+    },
+    {
+      "epoch": 0.7948969578017664,
+      "grad_norm": 1.780892469368783,
+      "learning_rate": 1.8253599330848638e-05,
+      "loss": 0.6091,
+      "step": 405
+    },
+    {
+      "epoch": 0.7968596663395485,
+      "grad_norm": 7.535603719678218,
+      "learning_rate": 1.8244790875571582e-05,
+      "loss": 0.9524,
+      "step": 406
+    },
+    {
+      "epoch": 0.7988223748773308,
+      "grad_norm": 1.7773136251051938,
+      "learning_rate": 1.823596239871915e-05,
+      "loss": 0.6362,
+      "step": 407
+    },
+    {
+      "epoch": 0.8007850834151129,
+      "grad_norm": 0.9044661668135322,
+      "learning_rate": 1.8227113921730336e-05,
+      "loss": 0.5439,
+      "step": 408
+    },
+    {
+      "epoch": 0.802747791952895,
+      "grad_norm": 2.7139783131315376,
+      "learning_rate": 1.8218245466092704e-05,
+      "loss": 0.5663,
+      "step": 409
+    },
+    {
+      "epoch": 0.8047105004906772,
+      "grad_norm": 0.8834413599502067,
+      "learning_rate": 1.8209357053342325e-05,
+      "loss": 0.5256,
+      "step": 410
+    },
+    {
+      "epoch": 0.8066732090284593,
+      "grad_norm": 1.1467415784558643,
+      "learning_rate": 1.8200448705063748e-05,
+      "loss": 0.5348,
+      "step": 411
+    },
+    {
+      "epoch": 0.8086359175662414,
+      "grad_norm": 0.515184372706671,
+      "learning_rate": 1.819152044288992e-05,
+      "loss": 0.5062,
+      "step": 412
+    },
+    {
+      "epoch": 0.8105986261040236,
+      "grad_norm": 0.6461620091644302,
+      "learning_rate": 1.8182572288502154e-05,
+      "loss": 0.5077,
+      "step": 413
+    },
+    {
+      "epoch": 0.8125613346418057,
+      "grad_norm": 0.4908151447534473,
+      "learning_rate": 1.8173604263630066e-05,
+      "loss": 0.4806,
+      "step": 414
+    },
+    {
+      "epoch": 0.8145240431795878,
+      "grad_norm": 0.3802375716765609,
+      "learning_rate": 1.8164616390051523e-05,
+      "loss": 0.4802,
+      "step": 415
+    },
+    {
+      "epoch": 0.81648675171737,
+      "grad_norm": 0.3805715926219177,
+      "learning_rate": 1.8155608689592604e-05,
+      "loss": 0.4814,
+      "step": 416
+    },
+    {
+      "epoch": 0.8184494602551521,
+      "grad_norm": 0.3945834169579088,
+      "learning_rate": 1.814658118412752e-05,
+      "loss": 0.4724,
+      "step": 417
+    },
+    {
+      "epoch": 0.8204121687929342,
+      "grad_norm": 0.2883495416963042,
+      "learning_rate": 1.8137533895578585e-05,
+      "loss": 0.4645,
+      "step": 418
+    },
+    {
+      "epoch": 0.8223748773307163,
+      "grad_norm": 0.2790010413679452,
+      "learning_rate": 1.8128466845916156e-05,
+      "loss": 0.4617,
+      "step": 419
+    },
+    {
+      "epoch": 0.8243375858684985,
+      "grad_norm": 0.2786295258327648,
+      "learning_rate": 1.811938005715857e-05,
+      "loss": 0.4791,
+      "step": 420
+    },
+    {
+      "epoch": 0.8263002944062807,
+      "grad_norm": 0.23317317430153287,
+      "learning_rate": 1.81102735513721e-05,
+      "loss": 0.4434,
+      "step": 421
+    },
+    {
+      "epoch": 0.8282630029440629,
+      "grad_norm": 0.2451241340574092,
+      "learning_rate": 1.8101147350670905e-05,
+      "loss": 0.4528,
+      "step": 422
+    },
+    {
+      "epoch": 0.830225711481845,
+      "grad_norm": 0.22624335497909545,
+      "learning_rate": 1.8092001477216975e-05,
+      "loss": 0.4644,
+      "step": 423
+    },
+    {
+      "epoch": 0.8321884200196271,
+      "grad_norm": 0.23107219922308564,
+      "learning_rate": 1.8082835953220055e-05,
+      "loss": 0.4352,
+      "step": 424
+    },
+    {
+      "epoch": 0.8341511285574092,
+      "grad_norm": 0.21335476872866976,
+      "learning_rate": 1.8073650800937627e-05,
+      "loss": 0.4515,
+      "step": 425
+    },
+    {
+      "epoch": 0.8361138370951914,
+      "grad_norm": 0.2076816094768591,
+      "learning_rate": 1.806444604267483e-05,
+      "loss": 0.4544,
+      "step": 426
+    },
+    {
+      "epoch": 0.8380765456329735,
+      "grad_norm": 0.19718486185623377,
+      "learning_rate": 1.805522170078441e-05,
+      "loss": 0.4428,
+      "step": 427
+    },
+    {
+      "epoch": 0.8400392541707556,
+      "grad_norm": 0.227660993052998,
+      "learning_rate": 1.8045977797666685e-05,
+      "loss": 0.4443,
+      "step": 428
+    },
+    {
+      "epoch": 0.8420019627085378,
+      "grad_norm": 0.21176408168289124,
+      "learning_rate": 1.803671435576946e-05,
+      "loss": 0.4617,
+      "step": 429
+    },
+    {
+      "epoch": 0.8439646712463199,
+      "grad_norm": 0.16842294987099302,
+      "learning_rate": 1.8027431397587993e-05,
+      "loss": 0.462,
+      "step": 430
+    },
+    {
+      "epoch": 0.845927379784102,
+      "grad_norm": 0.2398849119706587,
+      "learning_rate": 1.8018128945664936e-05,
+      "loss": 0.4408,
+      "step": 431
+    },
+    {
+      "epoch": 0.8478900883218842,
+      "grad_norm": 0.493692443958903,
+      "learning_rate": 1.8008807022590283e-05,
+      "loss": 0.4592,
+      "step": 432
+    },
+    {
+      "epoch": 0.8498527968596663,
+      "grad_norm": 0.544249601841786,
+      "learning_rate": 1.7999465651001297e-05,
+      "loss": 0.4476,
+      "step": 433
+    },
+    {
+      "epoch": 0.8518155053974484,
+      "grad_norm": 0.13924680999659736,
+      "learning_rate": 1.7990104853582494e-05,
+      "loss": 0.4281,
+      "step": 434
+    },
+    {
+      "epoch": 0.8537782139352306,
+      "grad_norm": 0.173740825747816,
+      "learning_rate": 1.7980724653065538e-05,
+      "loss": 0.4269,
+      "step": 435
+    },
+    {
+      "epoch": 0.8557409224730128,
+      "grad_norm": 0.15075725057898212,
+      "learning_rate": 1.7971325072229227e-05,
+      "loss": 0.4276,
+      "step": 436
+    },
+    {
+      "epoch": 0.8577036310107949,
+      "grad_norm": 0.1448476326691961,
+      "learning_rate": 1.7961906133899417e-05,
+      "loss": 0.4259,
+      "step": 437
+    },
+    {
+      "epoch": 0.8596663395485771,
+      "grad_norm": 0.15099110760456436,
+      "learning_rate": 1.7952467860948975e-05,
+      "loss": 0.4385,
+      "step": 438
+    },
+    {
+      "epoch": 0.8616290480863592,
+      "grad_norm": 0.2908525767466193,
+      "learning_rate": 1.7943010276297717e-05,
+      "loss": 0.4532,
+      "step": 439
+    },
+    {
+      "epoch": 0.8635917566241413,
+      "grad_norm": 0.19591315110166763,
+      "learning_rate": 1.7933533402912354e-05,
+      "loss": 0.4443,
+      "step": 440
+    },
+    {
+      "epoch": 0.8655544651619235,
+      "grad_norm": 0.1427892817158057,
+      "learning_rate": 1.792403726380644e-05,
+      "loss": 0.4497,
+      "step": 441
+    },
+    {
+      "epoch": 0.8675171736997056,
+      "grad_norm": 0.1557386790206549,
+      "learning_rate": 1.791452188204031e-05,
+      "loss": 0.4455,
+      "step": 442
+    },
+    {
+      "epoch": 0.8694798822374877,
+      "grad_norm": 0.13076931277969595,
+      "learning_rate": 1.7904987280721037e-05,
+      "loss": 0.4273,
+      "step": 443
+    },
+    {
+      "epoch": 0.8714425907752699,
+      "grad_norm": 0.17697976434079088,
+      "learning_rate": 1.7895433483002356e-05,
+      "loss": 0.4411,
+      "step": 444
+    },
+    {
+      "epoch": 0.873405299313052,
+      "grad_norm": 0.12603127769001743,
+      "learning_rate": 1.7885860512084622e-05,
+      "loss": 0.4254,
+      "step": 445
+    },
+    {
+      "epoch": 0.8753680078508341,
+      "grad_norm": 0.7782445781966569,
+      "learning_rate": 1.7876268391214756e-05,
+      "loss": 0.4558,
+      "step": 446
+    },
+    {
+      "epoch": 0.8773307163886163,
+      "grad_norm": 0.12647292734359114,
+      "learning_rate": 1.786665714368617e-05,
+      "loss": 0.4407,
+      "step": 447
+    },
+    {
+      "epoch": 0.8792934249263984,
+      "grad_norm": 0.20575076444088897,
+      "learning_rate": 1.785702679283874e-05,
+      "loss": 0.438,
+      "step": 448
+    },
+    {
+      "epoch": 0.8812561334641805,
+      "grad_norm": 0.19417352083348613,
+      "learning_rate": 1.7847377362058712e-05,
+      "loss": 0.4484,
+      "step": 449
+    },
+    {
+      "epoch": 0.8832188420019627,
+      "grad_norm": 0.12756724963135646,
+      "learning_rate": 1.7837708874778683e-05,
+      "loss": 0.4348,
+      "step": 450
+    },
+    {
+      "epoch": 0.8851815505397449,
+      "grad_norm": 0.11799761970570792,
+      "learning_rate": 1.7828021354477515e-05,
+      "loss": 0.437,
+      "step": 451
+    },
+    {
+      "epoch": 0.887144259077527,
+      "grad_norm": 0.1251248189106141,
+      "learning_rate": 1.78183148246803e-05,
+      "loss": 0.4302,
+      "step": 452
+    },
+    {
+      "epoch": 0.8891069676153092,
+      "grad_norm": 0.11474241608076738,
+      "learning_rate": 1.7808589308958284e-05,
+      "loss": 0.4343,
+      "step": 453
+    },
+    {
+      "epoch": 0.8910696761530913,
+      "grad_norm": 0.12586354241305017,
+      "learning_rate": 1.7798844830928818e-05,
+      "loss": 0.4348,
+      "step": 454
+    },
+    {
+      "epoch": 0.8930323846908734,
+      "grad_norm": 0.19899892032690525,
+      "learning_rate": 1.778908141425531e-05,
+      "loss": 0.4416,
+      "step": 455
+    },
+    {
+      "epoch": 0.8949950932286556,
+      "grad_norm": 0.21889266119451378,
+      "learning_rate": 1.777929908264715e-05,
+      "loss": 0.431,
+      "step": 456
+    },
+    {
+      "epoch": 0.8969578017664377,
+      "grad_norm": 0.1247721576190928,
+      "learning_rate": 1.7769497859859664e-05,
+      "loss": 0.4268,
+      "step": 457
+    },
+    {
+      "epoch": 0.8989205103042198,
+      "grad_norm": 0.1848738794336329,
+      "learning_rate": 1.775967776969405e-05,
+      "loss": 0.4278,
+      "step": 458
+    },
+    {
+      "epoch": 0.900883218842002,
+      "grad_norm": 0.14033082433848945,
+      "learning_rate": 1.774983883599733e-05,
+      "loss": 0.4202,
+      "step": 459
+    },
+    {
+      "epoch": 0.9028459273797841,
+      "grad_norm": 0.1730390019850834,
+      "learning_rate": 1.7739981082662275e-05,
+      "loss": 0.4298,
+      "step": 460
+    },
+    {
+      "epoch": 0.9048086359175662,
+      "grad_norm": 0.11693891835416576,
+      "learning_rate": 1.773010453362737e-05,
+      "loss": 0.4448,
+      "step": 461
+    },
+    {
+      "epoch": 0.9067713444553483,
+      "grad_norm": 0.12196405283755252,
+      "learning_rate": 1.772020921287674e-05,
+      "loss": 0.4378,
+      "step": 462
+    },
+    {
+      "epoch": 0.9087340529931305,
+      "grad_norm": 0.11586861714009496,
+      "learning_rate": 1.771029514444008e-05,
+      "loss": 0.4471,
+      "step": 463
+    },
+    {
+      "epoch": 0.9106967615309126,
+      "grad_norm": 0.12574469171427458,
+      "learning_rate": 1.7700362352392632e-05,
+      "loss": 0.4385,
+      "step": 464
+    },
+    {
+      "epoch": 0.9126594700686947,
+      "grad_norm": 0.12237397310589682,
+      "learning_rate": 1.7690410860855095e-05,
+      "loss": 0.4309,
+      "step": 465
+    },
+    {
+      "epoch": 0.914622178606477,
+      "grad_norm": 0.11990527876395565,
+      "learning_rate": 1.7680440693993586e-05,
+      "loss": 0.4336,
+      "step": 466
+    },
+    {
+      "epoch": 0.9165848871442591,
+      "grad_norm": 0.11451672776029655,
+      "learning_rate": 1.7670451876019562e-05,
+      "loss": 0.4298,
+      "step": 467
+    },
+    {
+      "epoch": 0.9185475956820413,
+      "grad_norm": 0.10316051482791179,
+      "learning_rate": 1.766044443118978e-05,
+      "loss": 0.4331,
+      "step": 468
+    },
+    {
+      "epoch": 0.9205103042198234,
+      "grad_norm": 0.12496652874488692,
+      "learning_rate": 1.7650418383806233e-05,
+      "loss": 0.4354,
+      "step": 469
+    },
+    {
+      "epoch": 0.9224730127576055,
+      "grad_norm": 0.11863765454148498,
+      "learning_rate": 1.7640373758216075e-05,
+      "loss": 0.4352,
+      "step": 470
+    },
+    {
+      "epoch": 0.9244357212953876,
+      "grad_norm": 0.11413588741219476,
+      "learning_rate": 1.763031057881159e-05,
+      "loss": 0.4302,
+      "step": 471
+    },
+    {
+      "epoch": 0.9263984298331698,
+      "grad_norm": 0.11625749644832431,
+      "learning_rate": 1.762022887003011e-05,
+      "loss": 0.4336,
+      "step": 472
+    },
+    {
+      "epoch": 0.9283611383709519,
+      "grad_norm": 0.11492501049851209,
+      "learning_rate": 1.761012865635396e-05,
+      "loss": 0.422,
+      "step": 473
+    },
+    {
+      "epoch": 0.930323846908734,
+      "grad_norm": 0.12563889502937342,
+      "learning_rate": 1.7600009962310417e-05,
+      "loss": 0.4348,
+      "step": 474
+    },
+    {
+      "epoch": 0.9322865554465162,
+      "grad_norm": 0.12443828665681882,
+      "learning_rate": 1.758987281247162e-05,
+      "loss": 0.4242,
+      "step": 475
+    },
+    {
+      "epoch": 0.9342492639842983,
+      "grad_norm": 0.11644777576344677,
+      "learning_rate": 1.757971723145453e-05,
+      "loss": 0.4194,
+      "step": 476
+    },
+    {
+      "epoch": 0.9362119725220804,
+      "grad_norm": 0.10793194810684831,
+      "learning_rate": 1.7569543243920873e-05,
+      "loss": 0.4345,
+      "step": 477
+    },
+    {
+      "epoch": 0.9381746810598626,
+      "grad_norm": 0.11594061352654564,
+      "learning_rate": 1.7559350874577066e-05,
+      "loss": 0.4314,
+      "step": 478
+    },
+    {
+      "epoch": 0.9401373895976447,
+      "grad_norm": 0.10332500150437529,
+      "learning_rate": 1.754914014817416e-05,
+      "loss": 0.4296,
+      "step": 479
+    },
+    {
+      "epoch": 0.9421000981354269,
+      "grad_norm": 0.11644279502007197,
+      "learning_rate": 1.75389110895078e-05,
+      "loss": 0.429,
+      "step": 480
+    },
+    {
+      "epoch": 0.9440628066732091,
+      "grad_norm": 0.10948538558656842,
+      "learning_rate": 1.7528663723418137e-05,
+      "loss": 0.4173,
+      "step": 481
+    },
+    {
+      "epoch": 0.9460255152109912,
+      "grad_norm": 0.0994509640223274,
+      "learning_rate": 1.7518398074789776e-05,
+      "loss": 0.4205,
+      "step": 482
+    },
+    {
+      "epoch": 0.9479882237487733,
+      "grad_norm": 0.9276868697631023,
+      "learning_rate": 1.750811416855173e-05,
+      "loss": 0.4673,
+      "step": 483
+    },
+    {
+      "epoch": 0.9499509322865555,
+      "grad_norm": 0.1374275322216112,
+      "learning_rate": 1.7497812029677344e-05,
+      "loss": 0.4242,
+      "step": 484
+    },
+    {
+      "epoch": 0.9519136408243376,
+      "grad_norm": 0.38132564392520824,
+      "learning_rate": 1.7487491683184236e-05,
+      "loss": 0.4371,
+      "step": 485
+    },
+    {
+      "epoch": 0.9538763493621197,
+      "grad_norm": 1.7151988276134096,
+      "learning_rate": 1.7477153154134244e-05,
+      "loss": 0.4803,
+      "step": 486
+    },
+    {
+      "epoch": 0.9558390578999019,
+      "grad_norm": 0.32545933630412127,
+      "learning_rate": 1.7466796467633357e-05,
+      "loss": 0.4369,
+      "step": 487
+    },
+    {
+      "epoch": 0.957801766437684,
+      "grad_norm": 0.1866079133255275,
+      "learning_rate": 1.7456421648831658e-05,
+      "loss": 0.4334,
+      "step": 488
+    },
+    {
+      "epoch": 0.9597644749754661,
+      "grad_norm": 0.3272314043287591,
+      "learning_rate": 1.7446028722923266e-05,
+      "loss": 0.4443,
+      "step": 489
+    },
+    {
+      "epoch": 0.9617271835132483,
+      "grad_norm": 0.16996374422654656,
+      "learning_rate": 1.743561771514626e-05,
+      "loss": 0.4366,
+      "step": 490
+    },
+    {
+      "epoch": 0.9636898920510304,
+      "grad_norm": 0.15827075993181033,
+      "learning_rate": 1.7425188650782648e-05,
+      "loss": 0.436,
+      "step": 491
+    },
+    {
+      "epoch": 0.9656526005888125,
+      "grad_norm": 0.43165470913413256,
+      "learning_rate": 1.741474155515827e-05,
+      "loss": 0.4433,
+      "step": 492
+    },
+    {
+      "epoch": 0.9676153091265947,
+      "grad_norm": 1.745564670186826,
+      "learning_rate": 1.7404276453642755e-05,
+      "loss": 0.546,
+      "step": 493
+    },
+    {
+      "epoch": 0.9695780176643768,
+      "grad_norm": 0.8077780518520157,
+      "learning_rate": 1.739379337164946e-05,
+      "loss": 0.4497,
+      "step": 494
+    },
+    {
+      "epoch": 0.971540726202159,
+      "grad_norm": 0.474709975296865,
+      "learning_rate": 1.738329233463542e-05,
+      "loss": 0.4496,
+      "step": 495
+    },
+    {
+      "epoch": 0.9735034347399412,
+      "grad_norm": 0.26095766576170426,
+      "learning_rate": 1.737277336810124e-05,
+      "loss": 0.4544,
+      "step": 496
+    },
+    {
+      "epoch": 0.9754661432777233,
+      "grad_norm": 0.2912691308833845,
+      "learning_rate": 1.7362236497591097e-05,
+      "loss": 0.4423,
+      "step": 497
+    },
+    {
+      "epoch": 0.9774288518155054,
+      "grad_norm": 0.24976229543285455,
+      "learning_rate": 1.7351681748692622e-05,
+      "loss": 0.4398,
+      "step": 498
+    },
+    {
+      "epoch": 0.9793915603532876,
+      "grad_norm": 0.20231287847225168,
+      "learning_rate": 1.7341109147036873e-05,
+      "loss": 0.4467,
+      "step": 499
+    },
+    {
+      "epoch": 0.9813542688910697,
+      "grad_norm": 0.2412935338809819,
+      "learning_rate": 1.7330518718298263e-05,
+      "loss": 0.4357,
+      "step": 500
+    },
+    {
+      "epoch": 0.9833169774288518,
+      "grad_norm": 0.18003467387447356,
+      "learning_rate": 1.7319910488194494e-05,
+      "loss": 0.4318,
+      "step": 501
+    },
+    {
+      "epoch": 0.985279685966634,
+      "grad_norm": 0.20072691034722498,
+      "learning_rate": 1.7309284482486494e-05,
+      "loss": 0.4458,
+      "step": 502
+    },
+    {
+      "epoch": 0.9872423945044161,
+      "grad_norm": 0.19649513698293491,
+      "learning_rate": 1.7298640726978357e-05,
+      "loss": 0.4514,
+      "step": 503
+    },
+    {
+      "epoch": 0.9892051030421982,
+      "grad_norm": 0.2071441349062203,
+      "learning_rate": 1.7287979247517285e-05,
+      "loss": 0.439,
+      "step": 504
+    },
+    {
+      "epoch": 0.9911678115799804,
+      "grad_norm": 0.20373439560246617,
+      "learning_rate": 1.7277300069993515e-05,
+      "loss": 0.4205,
+      "step": 505
+    },
+    {
+      "epoch": 0.9931305201177625,
+      "grad_norm": 0.18699843473849642,
+      "learning_rate": 1.7266603220340273e-05,
+      "loss": 0.4264,
+      "step": 506
+    },
+    {
+      "epoch": 0.9950932286555446,
+      "grad_norm": 0.1600200894069467,
+      "learning_rate": 1.725588872453368e-05,
+      "loss": 0.4277,
+      "step": 507
+    },
+    {
+      "epoch": 0.9970559371933267,
+      "grad_norm": 0.18202529308383952,
+      "learning_rate": 1.7245156608592727e-05,
+      "loss": 0.4491,
+      "step": 508
+    },
+    {
+      "epoch": 0.9990186457311089,
+      "grad_norm": 0.15887455063463465,
+      "learning_rate": 1.7234406898579187e-05,
+      "loss": 0.4558,
+      "step": 509
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.15887455063463465,
+      "learning_rate": 1.7223639620597556e-05,
+      "loss": 0.209,
+      "step": 510
+    },
+    {
+      "epoch": 1.0019627085377822,
+      "grad_norm": 0.1825187909705365,
+      "learning_rate": 1.7212854800794998e-05,
+      "loss": 0.3468,
+      "step": 511
+    },
+    {
+      "epoch": 1.0039254170755643,
+      "grad_norm": 0.20957758969625895,
+      "learning_rate": 1.7202052465361268e-05,
+      "loss": 0.3492,
+      "step": 512
+    },
+    {
+      "epoch": 1.0039254170755643,
+      "eval_loss": 0.4533812999725342,
+      "eval_runtime": 245.7967,
+      "eval_samples_per_second": 123.488,
+      "eval_steps_per_second": 3.861,
+      "step": 512
+    },
+    {
+      "epoch": 1.0058881256133465,
+      "grad_norm": 0.28069305660927224,
+      "learning_rate": 1.719123264052866e-05,
+      "loss": 0.339,
+      "step": 513
+    },
+    {
+      "epoch": 1.0078508341511285,
+      "grad_norm": 0.19132375047173436,
+      "learning_rate": 1.718039535257194e-05,
+      "loss": 0.3261,
+      "step": 514
+    },
+    {
+      "epoch": 1.0098135426889108,
+      "grad_norm": 0.17431307028109386,
+      "learning_rate": 1.7169540627808276e-05,
+      "loss": 0.3514,
+      "step": 515
+    },
+    {
+      "epoch": 1.0117762512266928,
+      "grad_norm": 0.5111559402715384,
+      "learning_rate": 1.7158668492597186e-05,
+      "loss": 0.3241,
+      "step": 516
+    },
+    {
+      "epoch": 1.013738959764475,
+      "grad_norm": 0.16901933614697587,
+      "learning_rate": 1.7147778973340466e-05,
+      "loss": 0.3228,
+      "step": 517
+    },
+    {
+      "epoch": 1.015701668302257,
+      "grad_norm": 0.15876073778651265,
+      "learning_rate": 1.7136872096482123e-05,
+      "loss": 0.4057,
+      "step": 518
+    },
+    {
+      "epoch": 1.0176643768400393,
+      "grad_norm": 2.1272381042420556,
+      "learning_rate": 1.7125947888508322e-05,
+      "loss": 0.3404,
+      "step": 519
+    },
+    {
+      "epoch": 1.0196270853778213,
+      "grad_norm": 0.19402112582649061,
+      "learning_rate": 1.7115006375947304e-05,
+      "loss": 0.3396,
+      "step": 520
+    },
+    {
+      "epoch": 1.0215897939156036,
+      "grad_norm": 0.184906049478743,
+      "learning_rate": 1.7104047585369345e-05,
+      "loss": 0.3335,
+      "step": 521
+    },
+    {
+      "epoch": 1.0235525024533856,
+      "grad_norm": 0.2027640255413657,
+      "learning_rate": 1.7093071543386667e-05,
+      "loss": 0.3336,
+      "step": 522
+    },
+    {
+      "epoch": 1.0255152109911678,
+      "grad_norm": 0.1592888909141134,
+      "learning_rate": 1.7082078276653392e-05,
+      "loss": 0.3224,
+      "step": 523
+    },
+    {
+      "epoch": 1.0274779195289498,
+      "grad_norm": 0.2560733378030634,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.3356,
+      "step": 524
+    },
+    {
+      "epoch": 1.029440628066732,
+      "grad_norm": 0.15861327684708404,
+      "learning_rate": 1.7060040175760626e-05,
+      "loss": 0.3237,
+      "step": 525
+    },
+    {
+      "epoch": 1.0314033366045143,
+      "grad_norm": 0.135405942028162,
+      "learning_rate": 1.7048995395118253e-05,
+      "loss": 0.3417,
+      "step": 526
+    },
+    {
+      "epoch": 1.0333660451422964,
+      "grad_norm": 0.15133569201059896,
+      "learning_rate": 1.7037933496759404e-05,
+      "loss": 0.3343,
+      "step": 527
+    },
+    {
+      "epoch": 1.0353287536800786,
+      "grad_norm": 0.134069507215934,
+      "learning_rate": 1.7026854507546694e-05,
+      "loss": 0.3338,
+      "step": 528
+    },
+    {
+      "epoch": 1.0372914622178606,
+      "grad_norm": 18.90339450893893,
+      "learning_rate": 1.7015758454384234e-05,
+      "loss": 0.346,
+      "step": 529
+    },
+    {
+      "epoch": 1.0392541707556429,
+      "grad_norm": 0.19965137004126413,
+      "learning_rate": 1.7004645364217584e-05,
+      "loss": 0.3264,
+      "step": 530
+    },
+    {
+      "epoch": 1.0412168792934249,
+      "grad_norm": 0.16909173895106233,
+      "learning_rate": 1.699351526403367e-05,
+      "loss": 0.3292,
+      "step": 531
+    },
+    {
+      "epoch": 1.0431795878312071,
+      "grad_norm": 0.2342868214892783,
+      "learning_rate": 1.698236818086073e-05,
+      "loss": 0.3262,
+      "step": 532
+    },
+    {
+      "epoch": 1.0451422963689891,
+      "grad_norm": 0.18469564769004154,
+      "learning_rate": 1.6971204141768235e-05,
+      "loss": 0.3296,
+      "step": 533
+    },
+    {
+      "epoch": 1.0471050049067714,
+      "grad_norm": 0.157241496947566,
+      "learning_rate": 1.6960023173866834e-05,
+      "loss": 0.3347,
+      "step": 534
+    },
+    {
+      "epoch": 1.0490677134445534,
+      "grad_norm": 0.17900244071577498,
+      "learning_rate": 1.6948825304308293e-05,
+      "loss": 0.3309,
+      "step": 535
+    },
+    {
+      "epoch": 1.0510304219823356,
+      "grad_norm": 0.14808820544169662,
+      "learning_rate": 1.693761056028542e-05,
+      "loss": 0.3302,
+      "step": 536
+    },
+    {
+      "epoch": 1.0529931305201177,
+      "grad_norm": 0.13596938535164,
+      "learning_rate": 1.6926378969031988e-05,
+      "loss": 0.3311,
+      "step": 537
+    },
+    {
+      "epoch": 1.0549558390579,
+      "grad_norm": 0.14432634269560748,
+      "learning_rate": 1.6915130557822698e-05,
+      "loss": 0.3305,
+      "step": 538
+    },
+    {
+      "epoch": 1.056918547595682,
+      "grad_norm": 0.1499905448155453,
+      "learning_rate": 1.6903865353973087e-05,
+      "loss": 0.32,
+      "step": 539
+    },
+    {
+      "epoch": 1.0588812561334642,
+      "grad_norm": 0.15652727350645462,
+      "learning_rate": 1.689258338483947e-05,
+      "loss": 0.3289,
+      "step": 540
+    },
+    {
+      "epoch": 1.0608439646712464,
+      "grad_norm": 0.12161904132806402,
+      "learning_rate": 1.6881284677818892e-05,
+      "loss": 0.3167,
+      "step": 541
+    },
+    {
+      "epoch": 1.0628066732090284,
+      "grad_norm": 0.12695759397898215,
+      "learning_rate": 1.686996926034902e-05,
+      "loss": 0.3227,
+      "step": 542
+    },
+    {
+      "epoch": 1.0647693817468107,
+      "grad_norm": 0.12620759014174157,
+      "learning_rate": 1.685863715990811e-05,
+      "loss": 0.3273,
+      "step": 543
+    },
+    {
+      "epoch": 1.0667320902845927,
+      "grad_norm": 0.12300912173446868,
+      "learning_rate": 1.6847288404014937e-05,
+      "loss": 0.3234,
+      "step": 544
+    },
+    {
+      "epoch": 1.068694798822375,
+      "grad_norm": 0.13604294154259072,
+      "learning_rate": 1.6835923020228714e-05,
+      "loss": 0.3218,
+      "step": 545
+    },
+    {
+      "epoch": 1.070657507360157,
+      "grad_norm": 0.12911000723405103,
+      "learning_rate": 1.682454103614904e-05,
+      "loss": 0.3199,
+      "step": 546
+    },
+    {
+      "epoch": 1.0726202158979392,
+      "grad_norm": 0.12394146843939333,
+      "learning_rate": 1.6813142479415815e-05,
+      "loss": 0.318,
+      "step": 547
+    },
+    {
+      "epoch": 1.0745829244357212,
+      "grad_norm": 0.11914666168696621,
+      "learning_rate": 1.6801727377709195e-05,
+      "loss": 0.3225,
+      "step": 548
+    },
+    {
+      "epoch": 1.0765456329735035,
+      "grad_norm": 0.3745213254445056,
+      "learning_rate": 1.6790295758749512e-05,
+      "loss": 0.3244,
+      "step": 549
+    },
+    {
+      "epoch": 1.0785083415112855,
+      "grad_norm": 0.11387952522253666,
+      "learning_rate": 1.67788476502972e-05,
+      "loss": 0.327,
+      "step": 550
+    },
+    {
+      "epoch": 1.0804710500490677,
+      "grad_norm": 0.1996105910656074,
+      "learning_rate": 1.6767383080152744e-05,
+      "loss": 0.3461,
+      "step": 551
+    },
+    {
+      "epoch": 1.0824337585868498,
+      "grad_norm": 0.12175257973633413,
+      "learning_rate": 1.6755902076156606e-05,
+      "loss": 0.3153,
+      "step": 552
+    },
+    {
+      "epoch": 1.084396467124632,
+      "grad_norm": 0.14527489586682849,
+      "learning_rate": 1.6744404666189146e-05,
+      "loss": 0.3265,
+      "step": 553
+    },
+    {
+      "epoch": 1.0863591756624142,
+      "grad_norm": 0.15226241276567778,
+      "learning_rate": 1.6732890878170573e-05,
+      "loss": 0.3289,
+      "step": 554
+    },
+    {
+      "epoch": 1.0883218842001963,
+      "grad_norm": 0.16932705018976396,
+      "learning_rate": 1.6721360740060864e-05,
+      "loss": 0.3256,
+      "step": 555
+    },
+    {
+      "epoch": 1.0902845927379785,
+      "grad_norm": 0.3186308522830804,
+      "learning_rate": 1.67098142798597e-05,
+      "loss": 0.3293,
+      "step": 556
+    },
+    {
+      "epoch": 1.0922473012757605,
+      "grad_norm": 0.12034379856239824,
+      "learning_rate": 1.669825152560641e-05,
+      "loss": 0.3211,
+      "step": 557
+    },
+    {
+      "epoch": 1.0942100098135428,
+      "grad_norm": 0.2213258321448838,
+      "learning_rate": 1.668667250537987e-05,
+      "loss": 0.3157,
+      "step": 558
+    },
+    {
+      "epoch": 1.0961727183513248,
+      "grad_norm": 0.4014643343300227,
+      "learning_rate": 1.6675077247298475e-05,
+      "loss": 0.3307,
+      "step": 559
+    },
+    {
+      "epoch": 1.098135426889107,
+      "grad_norm": 0.14678019949020815,
+      "learning_rate": 1.6663465779520042e-05,
+      "loss": 0.3341,
+      "step": 560
+    },
+    {
+      "epoch": 1.100098135426889,
+      "grad_norm": 0.11691608425931684,
+      "learning_rate": 1.665183813024175e-05,
+      "loss": 0.3207,
+      "step": 561
+    },
+    {
+      "epoch": 1.1020608439646713,
+      "grad_norm": 0.13749331512464025,
+      "learning_rate": 1.6640194327700087e-05,
+      "loss": 0.3212,
+      "step": 562
+    },
+    {
+      "epoch": 1.1040235525024533,
+      "grad_norm": 0.11763407718323214,
+      "learning_rate": 1.6628534400170746e-05,
+      "loss": 0.3184,
+      "step": 563
+    },
+    {
+      "epoch": 1.1059862610402356,
+      "grad_norm": 0.10763049323804355,
+      "learning_rate": 1.6616858375968596e-05,
+      "loss": 0.3255,
+      "step": 564
+    },
+    {
+      "epoch": 1.1079489695780176,
+      "grad_norm": 0.19415854323907672,
+      "learning_rate": 1.6605166283447587e-05,
+      "loss": 0.3265,
+      "step": 565
+    },
+    {
+      "epoch": 1.1099116781157998,
+      "grad_norm": 0.11392759854867497,
+      "learning_rate": 1.659345815100069e-05,
+      "loss": 0.3207,
+      "step": 566
+    },
+    {
+      "epoch": 1.1118743866535818,
+      "grad_norm": 0.11596195257412814,
+      "learning_rate": 1.658173400705983e-05,
+      "loss": 0.3359,
+      "step": 567
+    },
+    {
+      "epoch": 1.113837095191364,
+      "grad_norm": 0.1447901138867746,
+      "learning_rate": 1.6569993880095807e-05,
+      "loss": 0.3104,
+      "step": 568
+    },
+    {
+      "epoch": 1.1157998037291463,
+      "grad_norm": 0.11274072609698226,
+      "learning_rate": 1.6558237798618243e-05,
+      "loss": 0.3207,
+      "step": 569
+    },
+    {
+      "epoch": 1.1177625122669284,
+      "grad_norm": 0.1066641260817179,
+      "learning_rate": 1.6546465791175498e-05,
+      "loss": 0.33,
+      "step": 570
+    },
+    {
+      "epoch": 1.1197252208047106,
+      "grad_norm": 0.10845080737415526,
+      "learning_rate": 1.6534677886354605e-05,
+      "loss": 0.3324,
+      "step": 571
+    },
+    {
+      "epoch": 1.1216879293424926,
+      "grad_norm": 0.109524857570649,
+      "learning_rate": 1.6522874112781213e-05,
+      "loss": 0.33,
+      "step": 572
+    },
+    {
+      "epoch": 1.1236506378802749,
+      "grad_norm": 0.1086194965309956,
+      "learning_rate": 1.6511054499119493e-05,
+      "loss": 0.3318,
+      "step": 573
+    },
+    {
+      "epoch": 1.1256133464180569,
+      "grad_norm": 0.1097485465672784,
+      "learning_rate": 1.6499219074072087e-05,
+      "loss": 0.3201,
+      "step": 574
+    },
+    {
+      "epoch": 1.1275760549558391,
+      "grad_norm": 0.12784106789186606,
+      "learning_rate": 1.6487367866380037e-05,
+      "loss": 0.3277,
+      "step": 575
+    },
+    {
+      "epoch": 1.1295387634936211,
+      "grad_norm": 0.11285311191327178,
+      "learning_rate": 1.6475500904822707e-05,
+      "loss": 0.3249,
+      "step": 576
+    },
+    {
+      "epoch": 1.1315014720314034,
+      "grad_norm": 0.1024337781066779,
+      "learning_rate": 1.646361821821772e-05,
+      "loss": 0.3366,
+      "step": 577
+    },
+    {
+      "epoch": 1.1334641805691854,
+      "grad_norm": 0.11906272118674736,
+      "learning_rate": 1.645171983542088e-05,
+      "loss": 0.3339,
+      "step": 578
+    },
+    {
+      "epoch": 1.1354268891069677,
+      "grad_norm": 0.11321533330226549,
+      "learning_rate": 1.6439805785326114e-05,
+      "loss": 0.3276,
+      "step": 579
+    },
+    {
+      "epoch": 1.1373895976447497,
+      "grad_norm": 0.10571636728133642,
+      "learning_rate": 1.6427876096865394e-05,
+      "loss": 0.3289,
+      "step": 580
+    },
+    {
+      "epoch": 1.139352306182532,
+      "grad_norm": 0.22333265835518828,
+      "learning_rate": 1.6415930799008668e-05,
+      "loss": 0.325,
+      "step": 581
+    },
+    {
+      "epoch": 1.141315014720314,
+      "grad_norm": 0.10810955681560351,
+      "learning_rate": 1.640396992076379e-05,
+      "loss": 0.4055,
+      "step": 582
+    },
+    {
+      "epoch": 1.1432777232580962,
+      "grad_norm": 0.26109788537078404,
+      "learning_rate": 1.6391993491176445e-05,
+      "loss": 0.3162,
+      "step": 583
+    },
+    {
+      "epoch": 1.1452404317958784,
+      "grad_norm": 0.11788156500671607,
+      "learning_rate": 1.6380001539330088e-05,
+      "loss": 0.3132,
+      "step": 584
+    },
+    {
+      "epoch": 1.1472031403336604,
+      "grad_norm": 0.111527885844509,
+      "learning_rate": 1.6367994094345864e-05,
+      "loss": 0.3213,
+      "step": 585
+    },
+    {
+      "epoch": 1.1491658488714427,
+      "grad_norm": 0.10923834950482783,
+      "learning_rate": 1.6355971185382547e-05,
+      "loss": 0.3235,
+      "step": 586
+    },
+    {
+      "epoch": 1.1511285574092247,
+      "grad_norm": 0.11352228058272218,
+      "learning_rate": 1.6343932841636455e-05,
+      "loss": 0.3242,
+      "step": 587
+    },
+    {
+      "epoch": 1.153091265947007,
+      "grad_norm": 2.442797565842492,
+      "learning_rate": 1.6331879092341402e-05,
+      "loss": 0.3605,
+      "step": 588
+    },
+    {
+      "epoch": 1.155053974484789,
+      "grad_norm": 0.1340497478004765,
+      "learning_rate": 1.631980996676859e-05,
+      "loss": 0.3231,
+      "step": 589
+    },
+    {
+      "epoch": 1.1570166830225712,
+      "grad_norm": 7.251138564125245,
+      "learning_rate": 1.6307725494226586e-05,
+      "loss": 0.464,
+      "step": 590
+    },
+    {
+      "epoch": 1.1589793915603532,
+      "grad_norm": 1.3172662093711907,
+      "learning_rate": 1.6295625704061204e-05,
+      "loss": 0.3337,
+      "step": 591
+    },
+    {
+      "epoch": 1.1609421000981355,
+      "grad_norm": 0.31327939692037343,
+      "learning_rate": 1.6283510625655474e-05,
+      "loss": 0.3278,
+      "step": 592
+    },
+    {
+      "epoch": 1.1629048086359175,
+      "grad_norm": 4.811713971885485,
+      "learning_rate": 1.6271380288429535e-05,
+      "loss": 0.3485,
+      "step": 593
+    },
+    {
+      "epoch": 1.1648675171736997,
+      "grad_norm": 1.2548300965536887,
+      "learning_rate": 1.6259234721840595e-05,
+      "loss": 0.333,
+      "step": 594
+    },
+    {
+      "epoch": 1.1668302257114818,
+      "grad_norm": 0.2330658914469722,
+      "learning_rate": 1.624707395538283e-05,
+      "loss": 0.3294,
+      "step": 595
+    },
+    {
+      "epoch": 1.168792934249264,
+      "grad_norm": 0.24251502191211774,
+      "learning_rate": 1.6234898018587336e-05,
+      "loss": 0.3244,
+      "step": 596
+    },
+    {
+      "epoch": 1.170755642787046,
+      "grad_norm": 0.235968588849149,
+      "learning_rate": 1.6222706941022054e-05,
+      "loss": 0.3284,
+      "step": 597
+    },
+    {
+      "epoch": 1.1727183513248283,
+      "grad_norm": 0.2248234241197579,
+      "learning_rate": 1.6210500752291682e-05,
+      "loss": 0.3391,
+      "step": 598
+    },
+    {
+      "epoch": 1.1746810598626105,
+      "grad_norm": 0.4542190820206638,
+      "learning_rate": 1.6198279482037617e-05,
+      "loss": 0.3306,
+      "step": 599
+    },
+    {
+      "epoch": 1.1766437684003925,
+      "grad_norm": 0.19360720341661192,
+      "learning_rate": 1.6186043159937884e-05,
+      "loss": 0.3293,
+      "step": 600
+    },
+    {
+      "epoch": 1.1786064769381748,
+      "grad_norm": 0.33521960291085745,
+      "learning_rate": 1.6173791815707053e-05,
+      "loss": 0.3296,
+      "step": 601
+    },
+    {
+      "epoch": 1.1805691854759568,
+      "grad_norm": 0.1676340468101461,
+      "learning_rate": 1.616152547909618e-05,
+      "loss": 0.3227,
+      "step": 602
+    },
+    {
+      "epoch": 1.182531894013739,
+      "grad_norm": 0.166589926871346,
+      "learning_rate": 1.614924417989272e-05,
+      "loss": 0.3124,
+      "step": 603
+    },
+    {
+      "epoch": 1.184494602551521,
+      "grad_norm": 0.6062179431946622,
+      "learning_rate": 1.6136947947920477e-05,
+      "loss": 0.3542,
+      "step": 604
+    },
+    {
+      "epoch": 1.1864573110893033,
+      "grad_norm": 0.3148233399749341,
+      "learning_rate": 1.6124636813039502e-05,
+      "loss": 0.3583,
+      "step": 605
+    },
+    {
+      "epoch": 1.1884200196270853,
+      "grad_norm": 0.16441596245061335,
+      "learning_rate": 1.611231080514605e-05,
+      "loss": 0.3295,
+      "step": 606
+    },
+    {
+      "epoch": 1.1903827281648676,
+      "grad_norm": 0.16022712151086638,
+      "learning_rate": 1.609996995417248e-05,
+      "loss": 0.3271,
+      "step": 607
+    },
+    {
+      "epoch": 1.1923454367026496,
+      "grad_norm": 0.29243754067572564,
+      "learning_rate": 1.608761429008721e-05,
+      "loss": 0.3337,
+      "step": 608
+    },
+    {
+      "epoch": 1.1943081452404318,
+      "grad_norm": 0.15303231575175122,
+      "learning_rate": 1.6075243842894614e-05,
+      "loss": 0.336,
+      "step": 609
+    },
+    {
+      "epoch": 1.1962708537782138,
+      "grad_norm": 0.14650148016034228,
+      "learning_rate": 1.606285864263498e-05,
+      "loss": 0.329,
+      "step": 610
+    },
+    {
+      "epoch": 1.198233562315996,
+      "grad_norm": 0.15261136727699032,
+      "learning_rate": 1.605045871938441e-05,
+      "loss": 0.3218,
+      "step": 611
+    },
+    {
+      "epoch": 1.2001962708537781,
+      "grad_norm": 0.21026340819790304,
+      "learning_rate": 1.6038044103254775e-05,
+      "loss": 0.3292,
+      "step": 612
+    },
+    {
+      "epoch": 1.2021589793915604,
+      "grad_norm": 0.13503564973923055,
+      "learning_rate": 1.6025614824393606e-05,
+      "loss": 0.326,
+      "step": 613
+    },
+    {
+      "epoch": 1.2041216879293426,
+      "grad_norm": 0.1312176723931598,
+      "learning_rate": 1.601317091298406e-05,
+      "loss": 0.3258,
+      "step": 614
+    },
+    {
+      "epoch": 1.2060843964671246,
+      "grad_norm": 0.14448732114233953,
+      "learning_rate": 1.6000712399244813e-05,
+      "loss": 0.3358,
+      "step": 615
+    },
+    {
+      "epoch": 1.2080471050049069,
+      "grad_norm": 0.12820019901824906,
+      "learning_rate": 1.5988239313430004e-05,
+      "loss": 0.3186,
+      "step": 616
+    },
+    {
+      "epoch": 1.2100098135426889,
+      "grad_norm": 0.2613273277747731,
+      "learning_rate": 1.5975751685829167e-05,
+      "loss": 0.337,
+      "step": 617
+    },
+    {
+      "epoch": 1.2119725220804711,
+      "grad_norm": 0.1134463713652465,
+      "learning_rate": 1.5963249546767144e-05,
+      "loss": 0.3217,
+      "step": 618
+    },
+    {
+      "epoch": 1.2139352306182531,
+      "grad_norm": 0.12611151264418188,
+      "learning_rate": 1.5950732926604012e-05,
+      "loss": 0.3182,
+      "step": 619
+    },
+    {
+      "epoch": 1.2158979391560354,
+      "grad_norm": 0.11276212188506385,
+      "learning_rate": 1.5938201855735017e-05,
+      "loss": 0.3227,
+      "step": 620
+    },
+    {
+      "epoch": 1.2178606476938174,
+      "grad_norm": 0.12586598314647854,
+      "learning_rate": 1.5925656364590504e-05,
+      "loss": 0.3228,
+      "step": 621
+    },
+    {
+      "epoch": 1.2198233562315997,
+      "grad_norm": 0.23240521031529912,
+      "learning_rate": 1.5913096483635827e-05,
+      "loss": 0.3173,
+      "step": 622
+    },
+    {
+      "epoch": 1.2217860647693817,
+      "grad_norm": 0.12306625560657121,
+      "learning_rate": 1.5900522243371283e-05,
+      "loss": 0.3176,
+      "step": 623
+    },
+    {
+      "epoch": 1.223748773307164,
+      "grad_norm": 0.11574270115288804,
+      "learning_rate": 1.5887933674332048e-05,
+      "loss": 0.3362,
+      "step": 624
+    },
+    {
+      "epoch": 1.225711481844946,
+      "grad_norm": 0.113799124679694,
+      "learning_rate": 1.587533080708809e-05,
+      "loss": 0.3279,
+      "step": 625
+    },
+    {
+      "epoch": 1.2276741903827282,
+      "grad_norm": 0.11649342842785392,
+      "learning_rate": 1.5862713672244092e-05,
+      "loss": 0.3276,
+      "step": 626
+    },
+    {
+      "epoch": 1.2296368989205102,
+      "grad_norm": 0.25352594498528047,
+      "learning_rate": 1.5850082300439395e-05,
+      "loss": 0.3241,
+      "step": 627
+    },
+    {
+      "epoch": 1.2315996074582924,
+      "grad_norm": 0.16227315088252345,
+      "learning_rate": 1.5837436722347902e-05,
+      "loss": 0.3303,
+      "step": 628
+    },
+    {
+      "epoch": 1.2335623159960747,
+      "grad_norm": 0.1678226345146868,
+      "learning_rate": 1.5824776968678024e-05,
+      "loss": 0.3315,
+      "step": 629
+    },
+    {
+      "epoch": 1.2355250245338567,
+      "grad_norm": 0.13075124341701522,
+      "learning_rate": 1.5812103070172592e-05,
+      "loss": 0.3108,
+      "step": 630
+    },
+    {
+      "epoch": 1.237487733071639,
+      "grad_norm": 0.11601688723318539,
+      "learning_rate": 1.5799415057608785e-05,
+      "loss": 0.3264,
+      "step": 631
+    },
+    {
+      "epoch": 1.239450441609421,
+      "grad_norm": 0.11324740404685361,
+      "learning_rate": 1.578671296179806e-05,
+      "loss": 0.3177,
+      "step": 632
+    },
+    {
+      "epoch": 1.2414131501472032,
+      "grad_norm": 0.11840350625128808,
+      "learning_rate": 1.5773996813586067e-05,
+      "loss": 0.3354,
+      "step": 633
+    },
+    {
+      "epoch": 1.2433758586849852,
+      "grad_norm": 0.13875834597920514,
+      "learning_rate": 1.5761266643852587e-05,
+      "loss": 0.317,
+      "step": 634
+    },
+    {
+      "epoch": 1.2453385672227675,
+      "grad_norm": 0.117375827430945,
+      "learning_rate": 1.574852248351145e-05,
+      "loss": 0.3234,
+      "step": 635
+    },
+    {
+      "epoch": 1.2473012757605495,
+      "grad_norm": 0.11942850208082942,
+      "learning_rate": 1.573576436351046e-05,
+      "loss": 0.3087,
+      "step": 636
+    },
+    {
+      "epoch": 1.2492639842983317,
+      "grad_norm": 0.1335866448171688,
+      "learning_rate": 1.572299231483132e-05,
+      "loss": 0.3296,
+      "step": 637
+    },
+    {
+      "epoch": 1.2512266928361138,
+      "grad_norm": 0.12470555907017847,
+      "learning_rate": 1.5710206368489555e-05,
+      "loss": 0.3273,
+      "step": 638
+    },
+    {
+      "epoch": 1.253189401373896,
+      "grad_norm": 0.1757334780641794,
+      "learning_rate": 1.569740655553444e-05,
+      "loss": 0.3206,
+      "step": 639
+    },
+    {
+      "epoch": 1.2551521099116782,
+      "grad_norm": 0.11262712029964689,
+      "learning_rate": 1.5684592907048925e-05,
+      "loss": 0.3265,
+      "step": 640
+    },
+    {
+      "epoch": 1.2551521099116782,
+      "eval_loss": 0.4366247355937958,
+      "eval_runtime": 245.4297,
+      "eval_samples_per_second": 123.673,
+      "eval_steps_per_second": 3.867,
+      "step": 640
+    },
+    {
+      "epoch": 1.2571148184494603,
+      "grad_norm": 0.13469264864574937,
+      "learning_rate": 1.5671765454149558e-05,
+      "loss": 0.3286,
+      "step": 641
+    },
+    {
+      "epoch": 1.2590775269872423,
+      "grad_norm": 0.12053610787587447,
+      "learning_rate": 1.5658924227986415e-05,
+      "loss": 0.3275,
+      "step": 642
+    },
+    {
+      "epoch": 1.2610402355250245,
+      "grad_norm": 0.14172736929838212,
+      "learning_rate": 1.5646069259743007e-05,
+      "loss": 0.3323,
+      "step": 643
+    },
+    {
+      "epoch": 1.2630029440628068,
+      "grad_norm": 0.13230460831867666,
+      "learning_rate": 1.563320058063622e-05,
+      "loss": 0.3406,
+      "step": 644
+    },
+    {
+      "epoch": 1.2649656526005888,
+      "grad_norm": 0.11391731055963321,
+      "learning_rate": 1.5620318221916245e-05,
+      "loss": 0.3321,
+      "step": 645
+    },
+    {
+      "epoch": 1.2669283611383708,
+      "grad_norm": 0.20710778775533598,
+      "learning_rate": 1.560742221486648e-05,
+      "loss": 0.3233,
+      "step": 646
+    },
+    {
+      "epoch": 1.268891069676153,
+      "grad_norm": 0.12086849482365936,
+      "learning_rate": 1.5594512590803476e-05,
+      "loss": 0.3227,
+      "step": 647
+    },
+    {
+      "epoch": 1.2708537782139353,
+      "grad_norm": 0.12570156067267854,
+      "learning_rate": 1.5581589381076843e-05,
+      "loss": 0.3219,
+      "step": 648
+    },
+    {
+      "epoch": 1.2728164867517173,
+      "grad_norm": 0.12011193178735968,
+      "learning_rate": 1.556865261706918e-05,
+      "loss": 0.3182,
+      "step": 649
+    },
+    {
+      "epoch": 1.2747791952894996,
+      "grad_norm": 0.1648783672119669,
+      "learning_rate": 1.5555702330196024e-05,
+      "loss": 0.3297,
+      "step": 650
+    },
+    {
+      "epoch": 1.2767419038272816,
+      "grad_norm": 0.12110445033914359,
+      "learning_rate": 1.554273855190572e-05,
+      "loss": 0.3254,
+      "step": 651
+    },
+    {
+      "epoch": 1.2787046123650638,
+      "grad_norm": 0.11020981189703709,
+      "learning_rate": 1.5529761313679396e-05,
+      "loss": 0.3235,
+      "step": 652
+    },
+    {
+      "epoch": 1.2806673209028459,
+      "grad_norm": 0.15853842771502125,
+      "learning_rate": 1.551677064703086e-05,
+      "loss": 0.3264,
+      "step": 653
+    },
+    {
+      "epoch": 1.282630029440628,
+      "grad_norm": 0.11133610328097518,
+      "learning_rate": 1.5503766583506522e-05,
+      "loss": 0.329,
+      "step": 654
+    },
+    {
+      "epoch": 1.2845927379784103,
+      "grad_norm": 0.1193501946144463,
+      "learning_rate": 1.549074915468534e-05,
+      "loss": 0.332,
+      "step": 655
+    },
+    {
+      "epoch": 1.2865554465161924,
+      "grad_norm": 0.12269380787533944,
+      "learning_rate": 1.5477718392178716e-05,
+      "loss": 0.3314,
+      "step": 656
+    },
+    {
+      "epoch": 1.2885181550539744,
+      "grad_norm": 0.11313715963097541,
+      "learning_rate": 1.5464674327630437e-05,
+      "loss": 0.3222,
+      "step": 657
+    },
+    {
+      "epoch": 1.2904808635917566,
+      "grad_norm": 0.12048024103575065,
+      "learning_rate": 1.545161699271659e-05,
+      "loss": 0.3294,
+      "step": 658
+    },
+    {
+      "epoch": 1.2924435721295389,
+      "grad_norm": 0.11947900701309468,
+      "learning_rate": 1.543854641914549e-05,
+      "loss": 0.3235,
+      "step": 659
+    },
+    {
+      "epoch": 1.2944062806673209,
+      "grad_norm": 0.11207590380323458,
+      "learning_rate": 1.5425462638657597e-05,
+      "loss": 0.3266,
+      "step": 660
+    },
+    {
+      "epoch": 1.296368989205103,
+      "grad_norm": 0.12528223605914246,
+      "learning_rate": 1.5412365683025447e-05,
+      "loss": 0.3321,
+      "step": 661
+    },
+    {
+      "epoch": 1.2983316977428851,
+      "grad_norm": 0.10633033440100298,
+      "learning_rate": 1.5399255584053568e-05,
+      "loss": 0.3165,
+      "step": 662
+    },
+    {
+      "epoch": 1.3002944062806674,
+      "grad_norm": 0.11406876546510498,
+      "learning_rate": 1.5386132373578405e-05,
+      "loss": 0.3247,
+      "step": 663
+    },
+    {
+      "epoch": 1.3022571148184494,
+      "grad_norm": 0.11139990576253381,
+      "learning_rate": 1.5372996083468242e-05,
+      "loss": 0.3261,
+      "step": 664
+    },
+    {
+      "epoch": 1.3042198233562317,
+      "grad_norm": 0.10272448533365049,
+      "learning_rate": 1.5359846745623128e-05,
+      "loss": 0.316,
+      "step": 665
+    },
+    {
+      "epoch": 1.3061825318940137,
+      "grad_norm": 0.10645469773905245,
+      "learning_rate": 1.5346684391974792e-05,
+      "loss": 0.3311,
+      "step": 666
+    },
+    {
+      "epoch": 1.308145240431796,
+      "grad_norm": 0.10518169859307575,
+      "learning_rate": 1.5333509054486583e-05,
+      "loss": 0.3245,
+      "step": 667
+    },
+    {
+      "epoch": 1.310107948969578,
+      "grad_norm": 0.2242263732415979,
+      "learning_rate": 1.5320320765153367e-05,
+      "loss": 0.3109,
+      "step": 668
+    },
+    {
+      "epoch": 1.3120706575073602,
+      "grad_norm": 0.12309952064417946,
+      "learning_rate": 1.5307119556001463e-05,
+      "loss": 0.3196,
+      "step": 669
+    },
+    {
+      "epoch": 1.3140333660451424,
+      "grad_norm": 0.1129490904179756,
+      "learning_rate": 1.529390545908857e-05,
+      "loss": 0.3222,
+      "step": 670
+    },
+    {
+      "epoch": 1.3159960745829244,
+      "grad_norm": 0.1196632767822302,
+      "learning_rate": 1.528067850650368e-05,
+      "loss": 0.3182,
+      "step": 671
+    },
+    {
+      "epoch": 1.3179587831207065,
+      "grad_norm": 0.11338552151174172,
+      "learning_rate": 1.526743873036701e-05,
+      "loss": 0.3419,
+      "step": 672
+    },
+    {
+      "epoch": 1.3199214916584887,
+      "grad_norm": 0.17835671344319715,
+      "learning_rate": 1.5254186162829903e-05,
+      "loss": 0.3169,
+      "step": 673
+    },
+    {
+      "epoch": 1.321884200196271,
+      "grad_norm": 0.10783382011254725,
+      "learning_rate": 1.5240920836074777e-05,
+      "loss": 0.3196,
+      "step": 674
+    },
+    {
+      "epoch": 1.323846908734053,
+      "grad_norm": 0.10869968464668042,
+      "learning_rate": 1.5227642782315037e-05,
+      "loss": 0.3208,
+      "step": 675
+    },
+    {
+      "epoch": 1.325809617271835,
+      "grad_norm": 0.10637526513057953,
+      "learning_rate": 1.5214352033794981e-05,
+      "loss": 0.3263,
+      "step": 676
+    },
+    {
+      "epoch": 1.3277723258096172,
+      "grad_norm": 0.1239954358141019,
+      "learning_rate": 1.5201048622789747e-05,
+      "loss": 0.3262,
+      "step": 677
+    },
+    {
+      "epoch": 1.3297350343473995,
+      "grad_norm": 0.10993374676490457,
+      "learning_rate": 1.5187732581605217e-05,
+      "loss": 0.3367,
+      "step": 678
+    },
+    {
+      "epoch": 1.3316977428851815,
+      "grad_norm": 0.12433721542893529,
+      "learning_rate": 1.5174403942577942e-05,
+      "loss": 0.3329,
+      "step": 679
+    },
+    {
+      "epoch": 1.3336604514229637,
+      "grad_norm": 0.10189759821643117,
+      "learning_rate": 1.5161062738075068e-05,
+      "loss": 0.3191,
+      "step": 680
+    },
+    {
+      "epoch": 1.3356231599607458,
+      "grad_norm": 0.13313656343105662,
+      "learning_rate": 1.5147709000494258e-05,
+      "loss": 0.32,
+      "step": 681
+    },
+    {
+      "epoch": 1.337585868498528,
+      "grad_norm": 0.13908249732191438,
+      "learning_rate": 1.5134342762263606e-05,
+      "loss": 0.3295,
+      "step": 682
+    },
+    {
+      "epoch": 1.33954857703631,
+      "grad_norm": 0.11489030169841424,
+      "learning_rate": 1.5120964055841563e-05,
+      "loss": 0.3316,
+      "step": 683
+    },
+    {
+      "epoch": 1.3415112855740923,
+      "grad_norm": 0.3799510436085688,
+      "learning_rate": 1.5107572913716859e-05,
+      "loss": 0.3162,
+      "step": 684
+    },
+    {
+      "epoch": 1.3434739941118745,
+      "grad_norm": 0.11349571041844954,
+      "learning_rate": 1.509416936840842e-05,
+      "loss": 0.333,
+      "step": 685
+    },
+    {
+      "epoch": 1.3454367026496565,
+      "grad_norm": 0.11005963527146614,
+      "learning_rate": 1.5080753452465296e-05,
+      "loss": 0.3272,
+      "step": 686
+    },
+    {
+      "epoch": 1.3473994111874386,
+      "grad_norm": 0.11959100787233001,
+      "learning_rate": 1.5067325198466576e-05,
+      "loss": 0.3311,
+      "step": 687
+    },
+    {
+      "epoch": 1.3493621197252208,
+      "grad_norm": 0.10886477394608883,
+      "learning_rate": 1.505388463902131e-05,
+      "loss": 0.3335,
+      "step": 688
+    },
+    {
+      "epoch": 1.351324828263003,
+      "grad_norm": 0.11344016969622332,
+      "learning_rate": 1.504043180676843e-05,
+      "loss": 0.3207,
+      "step": 689
+    },
+    {
+      "epoch": 1.353287536800785,
+      "grad_norm": 0.1102062785109799,
+      "learning_rate": 1.502696673437667e-05,
+      "loss": 0.3338,
+      "step": 690
+    },
+    {
+      "epoch": 1.3552502453385673,
+      "grad_norm": 0.1134536009046737,
+      "learning_rate": 1.5013489454544494e-05,
+      "loss": 0.3212,
+      "step": 691
+    },
+    {
+      "epoch": 1.3572129538763493,
+      "grad_norm": 0.10812804193252123,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.3244,
+      "step": 692
+    },
+    {
+      "epoch": 1.3591756624141316,
+      "grad_norm": 0.10930042393971806,
+      "learning_rate": 1.4986498403500864e-05,
+      "loss": 0.3196,
+      "step": 693
+    },
+    {
+      "epoch": 1.3611383709519136,
+      "grad_norm": 0.11325005359482505,
+      "learning_rate": 1.4972984697834238e-05,
+      "loss": 0.3283,
+      "step": 694
+    },
+    {
+      "epoch": 1.3631010794896958,
+      "grad_norm": 0.11944962015232705,
+      "learning_rate": 1.4959458915816681e-05,
+      "loss": 0.3071,
+      "step": 695
+    },
+    {
+      "epoch": 1.3650637880274779,
+      "grad_norm": 0.11035654504778693,
+      "learning_rate": 1.4945921090294076e-05,
+      "loss": 0.3258,
+      "step": 696
+    },
+    {
+      "epoch": 1.36702649656526,
+      "grad_norm": 0.2791207437797803,
+      "learning_rate": 1.4932371254141562e-05,
+      "loss": 0.3444,
+      "step": 697
+    },
+    {
+      "epoch": 1.3689892051030421,
+      "grad_norm": 0.17253680739822636,
+      "learning_rate": 1.4918809440263435e-05,
+      "loss": 0.3263,
+      "step": 698
+    },
+    {
+      "epoch": 1.3709519136408244,
+      "grad_norm": 0.11509592444315454,
+      "learning_rate": 1.4905235681593079e-05,
+      "loss": 0.3159,
+      "step": 699
+    },
+    {
+      "epoch": 1.3729146221786066,
+      "grad_norm": 0.11893850145942958,
+      "learning_rate": 1.4891650011092896e-05,
+      "loss": 0.3361,
+      "step": 700
+    },
+    {
+      "epoch": 1.3748773307163886,
+      "grad_norm": 0.11595816082844375,
+      "learning_rate": 1.4878052461754192e-05,
+      "loss": 0.3172,
+      "step": 701
+    },
+    {
+      "epoch": 1.3768400392541706,
+      "grad_norm": 0.18006969131556863,
+      "learning_rate": 1.486444306659714e-05,
+      "loss": 0.3262,
+      "step": 702
+    },
+    {
+      "epoch": 1.3788027477919529,
+      "grad_norm": 0.11978594076007715,
+      "learning_rate": 1.4850821858670668e-05,
+      "loss": 0.3223,
+      "step": 703
+    },
+    {
+      "epoch": 1.3807654563297351,
+      "grad_norm": 0.1214666563188013,
+      "learning_rate": 1.4837188871052399e-05,
+      "loss": 0.3287,
+      "step": 704
+    },
+    {
+      "epoch": 1.3827281648675172,
+      "grad_norm": 0.1087134010196951,
+      "learning_rate": 1.4823544136848554e-05,
+      "loss": 0.3211,
+      "step": 705
+    },
+    {
+      "epoch": 1.3846908734052994,
+      "grad_norm": 0.11263273643445283,
+      "learning_rate": 1.4809887689193878e-05,
+      "loss": 0.321,
+      "step": 706
+    },
+    {
+      "epoch": 1.3866535819430814,
+      "grad_norm": 0.10441056093366385,
+      "learning_rate": 1.4796219561251569e-05,
+      "loss": 0.3205,
+      "step": 707
+    },
+    {
+      "epoch": 1.3886162904808637,
+      "grad_norm": 0.11847864217596303,
+      "learning_rate": 1.4782539786213184e-05,
+      "loss": 0.3358,
+      "step": 708
+    },
+    {
+      "epoch": 1.3905789990186457,
+      "grad_norm": 0.11231538216944834,
+      "learning_rate": 1.4768848397298562e-05,
+      "loss": 0.3265,
+      "step": 709
+    },
+    {
+      "epoch": 1.392541707556428,
+      "grad_norm": 0.1033742948772916,
+      "learning_rate": 1.4755145427755755e-05,
+      "loss": 0.3295,
+      "step": 710
+    },
+    {
+      "epoch": 1.39450441609421,
+      "grad_norm": 0.11293962071828316,
+      "learning_rate": 1.4741430910860918e-05,
+      "loss": 0.3284,
+      "step": 711
+    },
+    {
+      "epoch": 1.3964671246319922,
+      "grad_norm": 0.10568902630893928,
+      "learning_rate": 1.4727704879918272e-05,
+      "loss": 0.324,
+      "step": 712
+    },
+    {
+      "epoch": 1.3984298331697742,
+      "grad_norm": 0.10454582574341005,
+      "learning_rate": 1.4713967368259981e-05,
+      "loss": 0.3294,
+      "step": 713
+    },
+    {
+      "epoch": 1.4003925417075564,
+      "grad_norm": 0.11151842422641142,
+      "learning_rate": 1.4700218409246087e-05,
+      "loss": 0.3272,
+      "step": 714
+    },
+    {
+      "epoch": 1.4023552502453387,
+      "grad_norm": 0.1918179917026182,
+      "learning_rate": 1.4686458036264446e-05,
+      "loss": 0.3157,
+      "step": 715
+    },
+    {
+      "epoch": 1.4043179587831207,
+      "grad_norm": 0.11036284447673995,
+      "learning_rate": 1.4672686282730622e-05,
+      "loss": 0.312,
+      "step": 716
+    },
+    {
+      "epoch": 1.4062806673209027,
+      "grad_norm": 0.13510751112765382,
+      "learning_rate": 1.4658903182087814e-05,
+      "loss": 0.3382,
+      "step": 717
+    },
+    {
+      "epoch": 1.408243375858685,
+      "grad_norm": 0.10984538370082228,
+      "learning_rate": 1.4645108767806778e-05,
+      "loss": 0.3186,
+      "step": 718
+    },
+    {
+      "epoch": 1.4102060843964672,
+      "grad_norm": 0.11386113381529725,
+      "learning_rate": 1.4631303073385745e-05,
+      "loss": 0.333,
+      "step": 719
+    },
+    {
+      "epoch": 1.4121687929342492,
+      "grad_norm": 0.12922683158535153,
+      "learning_rate": 1.4617486132350343e-05,
+      "loss": 0.3131,
+      "step": 720
+    },
+    {
+      "epoch": 1.4141315014720315,
+      "grad_norm": 0.11429762711313823,
+      "learning_rate": 1.4603657978253499e-05,
+      "loss": 0.3356,
+      "step": 721
+    },
+    {
+      "epoch": 1.4160942100098135,
+      "grad_norm": 0.11032902828120253,
+      "learning_rate": 1.4589818644675378e-05,
+      "loss": 0.324,
+      "step": 722
+    },
+    {
+      "epoch": 1.4180569185475957,
+      "grad_norm": 0.10896038916378659,
+      "learning_rate": 1.4575968165223297e-05,
+      "loss": 0.3273,
+      "step": 723
+    },
+    {
+      "epoch": 1.4200196270853778,
+      "grad_norm": 0.11197390514652242,
+      "learning_rate": 1.4562106573531632e-05,
+      "loss": 0.3212,
+      "step": 724
+    },
+    {
+      "epoch": 1.42198233562316,
+      "grad_norm": 0.11854344219452063,
+      "learning_rate": 1.4548233903261746e-05,
+      "loss": 0.3266,
+      "step": 725
+    },
+    {
+      "epoch": 1.423945044160942,
+      "grad_norm": 0.10788963190927026,
+      "learning_rate": 1.4534350188101905e-05,
+      "loss": 0.3261,
+      "step": 726
+    },
+    {
+      "epoch": 1.4259077526987243,
+      "grad_norm": 0.10409181843438578,
+      "learning_rate": 1.45204554617672e-05,
+      "loss": 0.3267,
+      "step": 727
+    },
+    {
+      "epoch": 1.4278704612365063,
+      "grad_norm": 0.11457863412214213,
+      "learning_rate": 1.4506549757999456e-05,
+      "loss": 0.3322,
+      "step": 728
+    },
+    {
+      "epoch": 1.4298331697742885,
+      "grad_norm": 1.3410888637080636,
+      "learning_rate": 1.4492633110567155e-05,
+      "loss": 0.3836,
+      "step": 729
+    },
+    {
+      "epoch": 1.4317958783120708,
+      "grad_norm": 0.12437392309356543,
+      "learning_rate": 1.4478705553265363e-05,
+      "loss": 0.3244,
+      "step": 730
+    },
+    {
+      "epoch": 1.4337585868498528,
+      "grad_norm": 0.11654470773611598,
+      "learning_rate": 1.446476711991563e-05,
+      "loss": 0.334,
+      "step": 731
+    },
+    {
+      "epoch": 1.4357212953876348,
+      "grad_norm": 0.295599491807141,
+      "learning_rate": 1.4450817844365924e-05,
+      "loss": 0.3292,
+      "step": 732
+    },
+    {
+      "epoch": 1.437684003925417,
+      "grad_norm": 0.23594221943762184,
+      "learning_rate": 1.4436857760490539e-05,
+      "loss": 0.3244,
+      "step": 733
+    },
+    {
+      "epoch": 1.4396467124631993,
+      "grad_norm": 0.13208951730861257,
+      "learning_rate": 1.4422886902190014e-05,
+      "loss": 0.3218,
+      "step": 734
+    },
+    {
+      "epoch": 1.4416094210009813,
+      "grad_norm": 0.12203894918097015,
+      "learning_rate": 1.4408905303391054e-05,
+      "loss": 0.3184,
+      "step": 735
+    },
+    {
+      "epoch": 1.4435721295387636,
+      "grad_norm": 0.12073834849861981,
+      "learning_rate": 1.4394912998046451e-05,
+      "loss": 0.3297,
+      "step": 736
+    },
+    {
+      "epoch": 1.4455348380765456,
+      "grad_norm": 0.42701652972686566,
+      "learning_rate": 1.4380910020134988e-05,
+      "loss": 0.332,
+      "step": 737
+    },
+    {
+      "epoch": 1.4474975466143278,
+      "grad_norm": 0.12608912444867507,
+      "learning_rate": 1.436689640366137e-05,
+      "loss": 0.3267,
+      "step": 738
+    },
+    {
+      "epoch": 1.4494602551521099,
+      "grad_norm": 0.16001847759687152,
+      "learning_rate": 1.435287218265614e-05,
+      "loss": 0.3315,
+      "step": 739
+    },
+    {
+      "epoch": 1.451422963689892,
+      "grad_norm": 0.12030660319385979,
+      "learning_rate": 1.4338837391175582e-05,
+      "loss": 0.3285,
+      "step": 740
+    },
+    {
+      "epoch": 1.4533856722276741,
+      "grad_norm": 0.1253610461061891,
+      "learning_rate": 1.4324792063301662e-05,
+      "loss": 0.3351,
+      "step": 741
+    },
+    {
+      "epoch": 1.4553483807654564,
+      "grad_norm": 0.11052209080423725,
+      "learning_rate": 1.4310736233141926e-05,
+      "loss": 0.3289,
+      "step": 742
+    },
+    {
+      "epoch": 1.4573110893032384,
+      "grad_norm": 0.128573183549157,
+      "learning_rate": 1.4296669934829425e-05,
+      "loss": 0.3281,
+      "step": 743
+    },
+    {
+      "epoch": 1.4592737978410206,
+      "grad_norm": 0.11507190335793269,
+      "learning_rate": 1.4282593202522627e-05,
+      "loss": 0.331,
+      "step": 744
+    },
+    {
+      "epoch": 1.4612365063788029,
+      "grad_norm": 0.11860253438573154,
+      "learning_rate": 1.4268506070405345e-05,
+      "loss": 0.3278,
+      "step": 745
+    },
+    {
+      "epoch": 1.463199214916585,
+      "grad_norm": 0.10619138540627954,
+      "learning_rate": 1.4254408572686642e-05,
+      "loss": 0.3211,
+      "step": 746
+    },
+    {
+      "epoch": 1.465161923454367,
+      "grad_norm": 0.2575777832730168,
+      "learning_rate": 1.424030074360075e-05,
+      "loss": 0.3263,
+      "step": 747
+    },
+    {
+      "epoch": 1.4671246319921492,
+      "grad_norm": 0.11876073393454559,
+      "learning_rate": 1.4226182617406996e-05,
+      "loss": 0.3416,
+      "step": 748
+    },
+    {
+      "epoch": 1.4690873405299314,
+      "grad_norm": 0.1093129912760756,
+      "learning_rate": 1.4212054228389712e-05,
+      "loss": 0.3233,
+      "step": 749
+    },
+    {
+      "epoch": 1.4710500490677134,
+      "grad_norm": 0.11230780423472099,
+      "learning_rate": 1.4197915610858143e-05,
+      "loss": 0.32,
+      "step": 750
+    },
+    {
+      "epoch": 1.4730127576054957,
+      "grad_norm": 0.12038548720872048,
+      "learning_rate": 1.4183766799146383e-05,
+      "loss": 0.3246,
+      "step": 751
+    },
+    {
+      "epoch": 1.4749754661432777,
+      "grad_norm": 0.15570969251778305,
+      "learning_rate": 1.4169607827613284e-05,
+      "loss": 0.3261,
+      "step": 752
+    },
+    {
+      "epoch": 1.47693817468106,
+      "grad_norm": 0.3408875294488595,
+      "learning_rate": 1.4155438730642354e-05,
+      "loss": 0.3372,
+      "step": 753
+    },
+    {
+      "epoch": 1.478900883218842,
+      "grad_norm": 0.10372129141329851,
+      "learning_rate": 1.4141259542641706e-05,
+      "loss": 0.3275,
+      "step": 754
+    },
+    {
+      "epoch": 1.4808635917566242,
+      "grad_norm": 0.10861393950490046,
+      "learning_rate": 1.4127070298043949e-05,
+      "loss": 0.3195,
+      "step": 755
+    },
+    {
+      "epoch": 1.4828263002944062,
+      "grad_norm": 0.2287314698384735,
+      "learning_rate": 1.4112871031306118e-05,
+      "loss": 0.3392,
+      "step": 756
+    },
+    {
+      "epoch": 1.4847890088321885,
+      "grad_norm": 0.14387105066919137,
+      "learning_rate": 1.4098661776909581e-05,
+      "loss": 0.3278,
+      "step": 757
+    },
+    {
+      "epoch": 1.4867517173699705,
+      "grad_norm": 0.11031577836845029,
+      "learning_rate": 1.4084442569359964e-05,
+      "loss": 0.3261,
+      "step": 758
+    },
+    {
+      "epoch": 1.4887144259077527,
+      "grad_norm": 0.10372345153909722,
+      "learning_rate": 1.4070213443187062e-05,
+      "loss": 0.304,
+      "step": 759
+    },
+    {
+      "epoch": 1.490677134445535,
+      "grad_norm": 1.1167059057492386,
+      "learning_rate": 1.4055974432944753e-05,
+      "loss": 0.3512,
+      "step": 760
+    },
+    {
+      "epoch": 1.492639842983317,
+      "grad_norm": 0.3507372435965062,
+      "learning_rate": 1.404172557321092e-05,
+      "loss": 0.3467,
+      "step": 761
+    },
+    {
+      "epoch": 1.494602551521099,
+      "grad_norm": 0.13192221407076998,
+      "learning_rate": 1.4027466898587375e-05,
+      "loss": 0.3255,
+      "step": 762
+    },
+    {
+      "epoch": 1.4965652600588812,
+      "grad_norm": 0.13116992214786896,
+      "learning_rate": 1.401319844369974e-05,
+      "loss": 0.3385,
+      "step": 763
+    },
+    {
+      "epoch": 1.4985279685966635,
+      "grad_norm": 0.14515008145396116,
+      "learning_rate": 1.3998920243197408e-05,
+      "loss": 0.331,
+      "step": 764
+    },
+    {
+      "epoch": 1.5004906771344455,
+      "grad_norm": 0.12599311264669502,
+      "learning_rate": 1.3984632331753436e-05,
+      "loss": 0.3184,
+      "step": 765
+    },
+    {
+      "epoch": 1.5024533856722275,
+      "grad_norm": 0.11722994192309723,
+      "learning_rate": 1.3970334744064451e-05,
+      "loss": 0.3186,
+      "step": 766
+    },
+    {
+      "epoch": 1.5044160942100098,
+      "grad_norm": 0.1297039712182582,
+      "learning_rate": 1.395602751485059e-05,
+      "loss": 0.3239,
+      "step": 767
+    },
+    {
+      "epoch": 1.506378802747792,
+      "grad_norm": 0.12096476484978132,
+      "learning_rate": 1.3941710678855396e-05,
+      "loss": 0.3299,
+      "step": 768
+    },
+    {
+      "epoch": 1.506378802747792,
+      "eval_loss": 0.4330715835094452,
+      "eval_runtime": 245.5392,
+      "eval_samples_per_second": 123.618,
+      "eval_steps_per_second": 3.865,
+      "step": 768
+    },
+    {
+      "epoch": 1.508341511285574,
+      "grad_norm": 0.12230386295101764,
+      "learning_rate": 1.3927384270845744e-05,
+      "loss": 0.3251,
+      "step": 769
+    },
+    {
+      "epoch": 1.5103042198233563,
+      "grad_norm": 0.5983535137475166,
+      "learning_rate": 1.391304832561175e-05,
+      "loss": 0.3648,
+      "step": 770
+    },
+    {
+      "epoch": 1.5122669283611385,
+      "grad_norm": 0.11674041507665114,
+      "learning_rate": 1.38987028779667e-05,
+      "loss": 0.3392,
+      "step": 771
+    },
+    {
+      "epoch": 1.5142296368989205,
+      "grad_norm": 0.1729456768990231,
+      "learning_rate": 1.3884347962746949e-05,
+      "loss": 0.3225,
+      "step": 772
+    },
+    {
+      "epoch": 1.5161923454367026,
+      "grad_norm": 0.1252493496614802,
+      "learning_rate": 1.3869983614811837e-05,
+      "loss": 0.3209,
+      "step": 773
+    },
+    {
+      "epoch": 1.5181550539744848,
+      "grad_norm": 0.11700364355947357,
+      "learning_rate": 1.3855609869043618e-05,
+      "loss": 0.3315,
+      "step": 774
+    },
+    {
+      "epoch": 1.520117762512267,
+      "grad_norm": 0.1462155960573189,
+      "learning_rate": 1.384122676034737e-05,
+      "loss": 0.3259,
+      "step": 775
+    },
+    {
+      "epoch": 1.522080471050049,
+      "grad_norm": 0.13319429573990046,
+      "learning_rate": 1.3826834323650899e-05,
+      "loss": 0.3334,
+      "step": 776
+    },
+    {
+      "epoch": 1.524043179587831,
+      "grad_norm": 0.12414522959921269,
+      "learning_rate": 1.381243259390467e-05,
+      "loss": 0.3316,
+      "step": 777
+    },
+    {
+      "epoch": 1.5260058881256133,
+      "grad_norm": 0.12126161126577431,
+      "learning_rate": 1.3798021606081713e-05,
+      "loss": 0.3178,
+      "step": 778
+    },
+    {
+      "epoch": 1.5279685966633956,
+      "grad_norm": 0.11507460737639888,
+      "learning_rate": 1.3783601395177537e-05,
+      "loss": 0.3294,
+      "step": 779
+    },
+    {
+      "epoch": 1.5299313052011776,
+      "grad_norm": 0.11575304901544674,
+      "learning_rate": 1.3769171996210053e-05,
+      "loss": 0.3223,
+      "step": 780
+    },
+    {
+      "epoch": 1.5318940137389596,
+      "grad_norm": 0.131991412651207,
+      "learning_rate": 1.3754733444219488e-05,
+      "loss": 0.3302,
+      "step": 781
+    },
+    {
+      "epoch": 1.5338567222767419,
+      "grad_norm": 0.190709934290855,
+      "learning_rate": 1.3740285774268282e-05,
+      "loss": 0.3371,
+      "step": 782
+    },
+    {
+      "epoch": 1.535819430814524,
+      "grad_norm": 0.47614707484422286,
+      "learning_rate": 1.372582902144103e-05,
+      "loss": 0.3193,
+      "step": 783
+    },
+    {
+      "epoch": 1.5377821393523061,
+      "grad_norm": 0.11311414215315257,
+      "learning_rate": 1.371136322084438e-05,
+      "loss": 0.3147,
+      "step": 784
+    },
+    {
+      "epoch": 1.5397448478900884,
+      "grad_norm": 0.12393935880179471,
+      "learning_rate": 1.3696888407606952e-05,
+      "loss": 0.3294,
+      "step": 785
+    },
+    {
+      "epoch": 1.5417075564278706,
+      "grad_norm": 0.12459873866982599,
+      "learning_rate": 1.3682404616879246e-05,
+      "loss": 0.3523,
+      "step": 786
+    },
+    {
+      "epoch": 1.5436702649656526,
+      "grad_norm": 1.5376937047472998,
+      "learning_rate": 1.3667911883833573e-05,
+      "loss": 0.3232,
+      "step": 787
+    },
+    {
+      "epoch": 1.5456329735034346,
+      "grad_norm": 0.17863323707586204,
+      "learning_rate": 1.3653410243663953e-05,
+      "loss": 0.3294,
+      "step": 788
+    },
+    {
+      "epoch": 1.547595682041217,
+      "grad_norm": 0.13291935976093486,
+      "learning_rate": 1.3638899731586036e-05,
+      "loss": 0.3343,
+      "step": 789
+    },
+    {
+      "epoch": 1.5495583905789991,
+      "grad_norm": 0.19803813677781645,
+      "learning_rate": 1.3624380382837017e-05,
+      "loss": 0.3359,
+      "step": 790
+    },
+    {
+      "epoch": 1.5515210991167812,
+      "grad_norm": 0.1595795029165099,
+      "learning_rate": 1.3609852232675558e-05,
+      "loss": 0.326,
+      "step": 791
+    },
+    {
+      "epoch": 1.5534838076545632,
+      "grad_norm": 0.14037586512565553,
+      "learning_rate": 1.3595315316381676e-05,
+      "loss": 0.3302,
+      "step": 792
+    },
+    {
+      "epoch": 1.5554465161923454,
+      "grad_norm": 0.16013145607494655,
+      "learning_rate": 1.3580769669256695e-05,
+      "loss": 0.3383,
+      "step": 793
+    },
+    {
+      "epoch": 1.5574092247301277,
+      "grad_norm": 0.13377340237611735,
+      "learning_rate": 1.3566215326623131e-05,
+      "loss": 0.3235,
+      "step": 794
+    },
+    {
+      "epoch": 1.5593719332679097,
+      "grad_norm": 0.1385245785766373,
+      "learning_rate": 1.3551652323824617e-05,
+      "loss": 0.3214,
+      "step": 795
+    },
+    {
+      "epoch": 1.5613346418056917,
+      "grad_norm": 0.15040533378760934,
+      "learning_rate": 1.3537080696225815e-05,
+      "loss": 0.3396,
+      "step": 796
+    },
+    {
+      "epoch": 1.563297350343474,
+      "grad_norm": 0.1329758232018135,
+      "learning_rate": 1.3522500479212337e-05,
+      "loss": 0.3358,
+      "step": 797
+    },
+    {
+      "epoch": 1.5652600588812562,
+      "grad_norm": 0.12257059717404521,
+      "learning_rate": 1.3507911708190646e-05,
+      "loss": 0.3176,
+      "step": 798
+    },
+    {
+      "epoch": 1.5672227674190382,
+      "grad_norm": 0.13209253543187308,
+      "learning_rate": 1.3493314418587982e-05,
+      "loss": 0.3314,
+      "step": 799
+    },
+    {
+      "epoch": 1.5691854759568205,
+      "grad_norm": 0.11457657070098855,
+      "learning_rate": 1.3478708645852272e-05,
+      "loss": 0.3306,
+      "step": 800
+    },
+    {
+      "epoch": 1.5711481844946027,
+      "grad_norm": 0.19371352224283817,
+      "learning_rate": 1.3464094425452046e-05,
+      "loss": 0.3145,
+      "step": 801
+    },
+    {
+      "epoch": 1.5731108930323847,
+      "grad_norm": 0.10702419786852241,
+      "learning_rate": 1.3449471792876333e-05,
+      "loss": 0.3227,
+      "step": 802
+    },
+    {
+      "epoch": 1.5750736015701667,
+      "grad_norm": 0.11647703035094253,
+      "learning_rate": 1.3434840783634611e-05,
+      "loss": 0.3246,
+      "step": 803
+    },
+    {
+      "epoch": 1.577036310107949,
+      "grad_norm": 0.14359657153339322,
+      "learning_rate": 1.342020143325669e-05,
+      "loss": 0.3198,
+      "step": 804
+    },
+    {
+      "epoch": 1.5789990186457312,
+      "grad_norm": 0.11430760797920907,
+      "learning_rate": 1.3405553777292627e-05,
+      "loss": 0.3174,
+      "step": 805
+    },
+    {
+      "epoch": 1.5809617271835132,
+      "grad_norm": 0.12045105701279527,
+      "learning_rate": 1.3390897851312667e-05,
+      "loss": 0.3281,
+      "step": 806
+    },
+    {
+      "epoch": 1.5829244357212953,
+      "grad_norm": 0.11152977267946707,
+      "learning_rate": 1.3376233690907126e-05,
+      "loss": 0.3343,
+      "step": 807
+    },
+    {
+      "epoch": 1.5848871442590775,
+      "grad_norm": 0.11046335035659553,
+      "learning_rate": 1.336156133168631e-05,
+      "loss": 0.3222,
+      "step": 808
+    },
+    {
+      "epoch": 1.5868498527968598,
+      "grad_norm": 0.2805415736645124,
+      "learning_rate": 1.3346880809280451e-05,
+      "loss": 0.3285,
+      "step": 809
+    },
+    {
+      "epoch": 1.5888125613346418,
+      "grad_norm": 0.1165026997728163,
+      "learning_rate": 1.3332192159339595e-05,
+      "loss": 0.3327,
+      "step": 810
+    },
+    {
+      "epoch": 1.5907752698724238,
+      "grad_norm": 0.10269266294028667,
+      "learning_rate": 1.3317495417533523e-05,
+      "loss": 0.3288,
+      "step": 811
+    },
+    {
+      "epoch": 1.592737978410206,
+      "grad_norm": 0.10882094664575565,
+      "learning_rate": 1.3302790619551673e-05,
+      "loss": 0.3226,
+      "step": 812
+    },
+    {
+      "epoch": 1.5947006869479883,
+      "grad_norm": 0.10402717390991092,
+      "learning_rate": 1.3288077801103041e-05,
+      "loss": 0.3242,
+      "step": 813
+    },
+    {
+      "epoch": 1.5966633954857703,
+      "grad_norm": 0.11773324479057384,
+      "learning_rate": 1.3273356997916106e-05,
+      "loss": 0.3312,
+      "step": 814
+    },
+    {
+      "epoch": 1.5986261040235525,
+      "grad_norm": 0.10710334700945738,
+      "learning_rate": 1.3258628245738726e-05,
+      "loss": 0.3249,
+      "step": 815
+    },
+    {
+      "epoch": 1.6005888125613348,
+      "grad_norm": 0.10399693106990227,
+      "learning_rate": 1.3243891580338074e-05,
+      "loss": 0.3272,
+      "step": 816
+    },
+    {
+      "epoch": 1.6025515210991168,
+      "grad_norm": 0.107596072668019,
+      "learning_rate": 1.3229147037500534e-05,
+      "loss": 0.3334,
+      "step": 817
+    },
+    {
+      "epoch": 1.6045142296368988,
+      "grad_norm": 0.1069444594286671,
+      "learning_rate": 1.3214394653031616e-05,
+      "loss": 0.318,
+      "step": 818
+    },
+    {
+      "epoch": 1.606476938174681,
+      "grad_norm": 0.2003747727263415,
+      "learning_rate": 1.3199634462755886e-05,
+      "loss": 0.321,
+      "step": 819
+    },
+    {
+      "epoch": 1.6084396467124633,
+      "grad_norm": 0.10473434193181139,
+      "learning_rate": 1.3184866502516846e-05,
+      "loss": 0.3289,
+      "step": 820
+    },
+    {
+      "epoch": 1.6104023552502453,
+      "grad_norm": 0.11035758045532282,
+      "learning_rate": 1.3170090808176883e-05,
+      "loss": 0.3302,
+      "step": 821
+    },
+    {
+      "epoch": 1.6123650637880274,
+      "grad_norm": 0.10519561826066223,
+      "learning_rate": 1.3155307415617156e-05,
+      "loss": 0.3264,
+      "step": 822
+    },
+    {
+      "epoch": 1.6143277723258096,
+      "grad_norm": 0.11015590527571209,
+      "learning_rate": 1.3140516360737523e-05,
+      "loss": 0.3173,
+      "step": 823
+    },
+    {
+      "epoch": 1.6162904808635918,
+      "grad_norm": 0.10092061607132297,
+      "learning_rate": 1.3125717679456447e-05,
+      "loss": 0.331,
+      "step": 824
+    },
+    {
+      "epoch": 1.6182531894013739,
+      "grad_norm": 0.1024320679905342,
+      "learning_rate": 1.3110911407710909e-05,
+      "loss": 0.3177,
+      "step": 825
+    },
+    {
+      "epoch": 1.6202158979391559,
+      "grad_norm": 0.10726261404380205,
+      "learning_rate": 1.309609758145633e-05,
+      "loss": 0.33,
+      "step": 826
+    },
+    {
+      "epoch": 1.6221786064769381,
+      "grad_norm": 0.30696269389558967,
+      "learning_rate": 1.308127623666647e-05,
+      "loss": 0.3217,
+      "step": 827
+    },
+    {
+      "epoch": 1.6241413150147204,
+      "grad_norm": 0.11038918441384825,
+      "learning_rate": 1.3066447409333345e-05,
+      "loss": 0.3428,
+      "step": 828
+    },
+    {
+      "epoch": 1.6261040235525024,
+      "grad_norm": 0.13114032892510316,
+      "learning_rate": 1.3051611135467145e-05,
+      "loss": 0.3217,
+      "step": 829
+    },
+    {
+      "epoch": 1.6280667320902846,
+      "grad_norm": 0.10209817496354183,
+      "learning_rate": 1.3036767451096148e-05,
+      "loss": 0.3169,
+      "step": 830
+    },
+    {
+      "epoch": 1.6300294406280669,
+      "grad_norm": 0.11163305235516735,
+      "learning_rate": 1.3021916392266618e-05,
+      "loss": 0.3215,
+      "step": 831
+    },
+    {
+      "epoch": 1.631992149165849,
+      "grad_norm": 0.10335841003276967,
+      "learning_rate": 1.300705799504273e-05,
+      "loss": 0.3179,
+      "step": 832
+    },
+    {
+      "epoch": 1.633954857703631,
+      "grad_norm": 0.10298399337006449,
+      "learning_rate": 1.2992192295506489e-05,
+      "loss": 0.3342,
+      "step": 833
+    },
+    {
+      "epoch": 1.6359175662414132,
+      "grad_norm": 0.10601682383580134,
+      "learning_rate": 1.2977319329757616e-05,
+      "loss": 0.3189,
+      "step": 834
+    },
+    {
+      "epoch": 1.6378802747791954,
+      "grad_norm": 0.1018229680915178,
+      "learning_rate": 1.296243913391349e-05,
+      "loss": 0.3231,
+      "step": 835
+    },
+    {
+      "epoch": 1.6398429833169774,
+      "grad_norm": 0.1085023326981021,
+      "learning_rate": 1.2947551744109044e-05,
+      "loss": 0.3184,
+      "step": 836
+    },
+    {
+      "epoch": 1.6418056918547594,
+      "grad_norm": 0.10661650845272141,
+      "learning_rate": 1.2932657196496678e-05,
+      "loss": 0.3167,
+      "step": 837
+    },
+    {
+      "epoch": 1.6437684003925417,
+      "grad_norm": 0.11175562549636477,
+      "learning_rate": 1.2917755527246179e-05,
+      "loss": 0.3316,
+      "step": 838
+    },
+    {
+      "epoch": 1.645731108930324,
+      "grad_norm": 0.10523122873271762,
+      "learning_rate": 1.2902846772544625e-05,
+      "loss": 0.3312,
+      "step": 839
+    },
+    {
+      "epoch": 1.647693817468106,
+      "grad_norm": 0.13408743582159743,
+      "learning_rate": 1.28879309685963e-05,
+      "loss": 0.3279,
+      "step": 840
+    },
+    {
+      "epoch": 1.649656526005888,
+      "grad_norm": 0.4305626635372785,
+      "learning_rate": 1.2873008151622606e-05,
+      "loss": 0.3143,
+      "step": 841
+    },
+    {
+      "epoch": 1.6516192345436702,
+      "grad_norm": 0.10214474856379448,
+      "learning_rate": 1.2858078357861979e-05,
+      "loss": 0.3398,
+      "step": 842
+    },
+    {
+      "epoch": 1.6535819430814525,
+      "grad_norm": 0.19415909087113795,
+      "learning_rate": 1.2843141623569792e-05,
+      "loss": 0.3205,
+      "step": 843
+    },
+    {
+      "epoch": 1.6555446516192345,
+      "grad_norm": 0.11011953169021917,
+      "learning_rate": 1.2828197985018276e-05,
+      "loss": 0.3336,
+      "step": 844
+    },
+    {
+      "epoch": 1.6575073601570167,
+      "grad_norm": 0.1503839751469065,
+      "learning_rate": 1.2813247478496428e-05,
+      "loss": 0.3279,
+      "step": 845
+    },
+    {
+      "epoch": 1.659470068694799,
+      "grad_norm": 0.11781800325747578,
+      "learning_rate": 1.2798290140309924e-05,
+      "loss": 0.3254,
+      "step": 846
+    },
+    {
+      "epoch": 1.661432777232581,
+      "grad_norm": 0.10932301796870458,
+      "learning_rate": 1.2783326006781023e-05,
+      "loss": 0.3131,
+      "step": 847
+    },
+    {
+      "epoch": 1.663395485770363,
+      "grad_norm": 0.11978600321445199,
+      "learning_rate": 1.2768355114248493e-05,
+      "loss": 0.3277,
+      "step": 848
+    },
+    {
+      "epoch": 1.6653581943081452,
+      "grad_norm": 0.11039786376284047,
+      "learning_rate": 1.2753377499067522e-05,
+      "loss": 0.3522,
+      "step": 849
+    },
+    {
+      "epoch": 1.6673209028459275,
+      "grad_norm": 0.2497098566113429,
+      "learning_rate": 1.2738393197609602e-05,
+      "loss": 0.3297,
+      "step": 850
+    },
+    {
+      "epoch": 1.6692836113837095,
+      "grad_norm": 0.11865444749257295,
+      "learning_rate": 1.2723402246262484e-05,
+      "loss": 0.3271,
+      "step": 851
+    },
+    {
+      "epoch": 1.6712463199214915,
+      "grad_norm": 0.11095090763334178,
+      "learning_rate": 1.2708404681430054e-05,
+      "loss": 0.3243,
+      "step": 852
+    },
+    {
+      "epoch": 1.6732090284592738,
+      "grad_norm": 0.11867828722172012,
+      "learning_rate": 1.2693400539532263e-05,
+      "loss": 0.3141,
+      "step": 853
+    },
+    {
+      "epoch": 1.675171736997056,
+      "grad_norm": 0.1131423717161425,
+      "learning_rate": 1.2678389857005033e-05,
+      "loss": 0.3256,
+      "step": 854
+    },
+    {
+      "epoch": 1.677134445534838,
+      "grad_norm": 0.8893669616224853,
+      "learning_rate": 1.266337267030017e-05,
+      "loss": 0.3338,
+      "step": 855
+    },
+    {
+      "epoch": 1.67909715407262,
+      "grad_norm": 0.1621276184736285,
+      "learning_rate": 1.2648349015885272e-05,
+      "loss": 0.331,
+      "step": 856
+    },
+    {
+      "epoch": 1.6810598626104023,
+      "grad_norm": 0.12724899204009446,
+      "learning_rate": 1.2633318930243647e-05,
+      "loss": 0.3168,
+      "step": 857
+    },
+    {
+      "epoch": 1.6830225711481845,
+      "grad_norm": 0.15162789739847926,
+      "learning_rate": 1.2618282449874221e-05,
+      "loss": 0.3147,
+      "step": 858
+    },
+    {
+      "epoch": 1.6849852796859666,
+      "grad_norm": 0.14035812455971758,
+      "learning_rate": 1.2603239611291445e-05,
+      "loss": 0.3215,
+      "step": 859
+    },
+    {
+      "epoch": 1.6869479882237488,
+      "grad_norm": 0.46008864002942473,
+      "learning_rate": 1.2588190451025209e-05,
+      "loss": 0.325,
+      "step": 860
+    },
+    {
+      "epoch": 1.688910696761531,
+      "grad_norm": 0.15188638903617155,
+      "learning_rate": 1.2573135005620757e-05,
+      "loss": 0.3091,
+      "step": 861
+    },
+    {
+      "epoch": 1.690873405299313,
+      "grad_norm": 0.1372788892477886,
+      "learning_rate": 1.2558073311638604e-05,
+      "loss": 0.3313,
+      "step": 862
+    },
+    {
+      "epoch": 1.692836113837095,
+      "grad_norm": 0.1317279830573809,
+      "learning_rate": 1.2543005405654418e-05,
+      "loss": 0.3169,
+      "step": 863
+    },
+    {
+      "epoch": 1.6947988223748773,
+      "grad_norm": 0.13349361942014895,
+      "learning_rate": 1.2527931324258975e-05,
+      "loss": 0.3203,
+      "step": 864
+    },
+    {
+      "epoch": 1.6967615309126596,
+      "grad_norm": 0.12909951263188774,
+      "learning_rate": 1.2512851104058038e-05,
+      "loss": 0.3202,
+      "step": 865
+    },
+    {
+      "epoch": 1.6987242394504416,
+      "grad_norm": 0.2566617860841747,
+      "learning_rate": 1.249776478167227e-05,
+      "loss": 0.3295,
+      "step": 866
+    },
+    {
+      "epoch": 1.7006869479882236,
+      "grad_norm": 0.11969281942134942,
+      "learning_rate": 1.2482672393737164e-05,
+      "loss": 0.3342,
+      "step": 867
+    },
+    {
+      "epoch": 1.7026496565260059,
+      "grad_norm": 0.12518924207179494,
+      "learning_rate": 1.2467573976902936e-05,
+      "loss": 0.3333,
+      "step": 868
+    },
+    {
+      "epoch": 1.704612365063788,
+      "grad_norm": 0.21335666453466184,
+      "learning_rate": 1.2452469567834449e-05,
+      "loss": 0.3198,
+      "step": 869
+    },
+    {
+      "epoch": 1.7065750736015701,
+      "grad_norm": 0.12157902223714591,
+      "learning_rate": 1.2437359203211109e-05,
+      "loss": 0.308,
+      "step": 870
+    },
+    {
+      "epoch": 1.7085377821393521,
+      "grad_norm": 0.13207634498097479,
+      "learning_rate": 1.2422242919726786e-05,
+      "loss": 0.3338,
+      "step": 871
+    },
+    {
+      "epoch": 1.7105004906771346,
+      "grad_norm": 0.10700849261671908,
+      "learning_rate": 1.2407120754089733e-05,
+      "loss": 0.3221,
+      "step": 872
+    },
+    {
+      "epoch": 1.7124631992149166,
+      "grad_norm": 0.12604995554733597,
+      "learning_rate": 1.2391992743022472e-05,
+      "loss": 0.3165,
+      "step": 873
+    },
+    {
+      "epoch": 1.7144259077526987,
+      "grad_norm": 0.1613701947168364,
+      "learning_rate": 1.2376858923261732e-05,
+      "loss": 0.3217,
+      "step": 874
+    },
+    {
+      "epoch": 1.716388616290481,
+      "grad_norm": 0.1328873058860125,
+      "learning_rate": 1.2361719331558346e-05,
+      "loss": 0.3262,
+      "step": 875
+    },
+    {
+      "epoch": 1.7183513248282631,
+      "grad_norm": 0.11726688731965726,
+      "learning_rate": 1.2346574004677154e-05,
+      "loss": 0.3298,
+      "step": 876
+    },
+    {
+      "epoch": 1.7203140333660452,
+      "grad_norm": 0.1214011987184142,
+      "learning_rate": 1.2331422979396936e-05,
+      "loss": 0.337,
+      "step": 877
+    },
+    {
+      "epoch": 1.7222767419038272,
+      "grad_norm": 0.12820530343064107,
+      "learning_rate": 1.2316266292510305e-05,
+      "loss": 0.327,
+      "step": 878
+    },
+    {
+      "epoch": 1.7242394504416094,
+      "grad_norm": 0.11451384636314305,
+      "learning_rate": 1.2301103980823619e-05,
+      "loss": 0.3235,
+      "step": 879
+    },
+    {
+      "epoch": 1.7262021589793917,
+      "grad_norm": 0.10873652825630871,
+      "learning_rate": 1.2285936081156897e-05,
+      "loss": 0.311,
+      "step": 880
+    },
+    {
+      "epoch": 1.7281648675171737,
+      "grad_norm": 0.10624002895012581,
+      "learning_rate": 1.2270762630343734e-05,
+      "loss": 0.3245,
+      "step": 881
+    },
+    {
+      "epoch": 1.7301275760549557,
+      "grad_norm": 0.10425222382859203,
+      "learning_rate": 1.2255583665231196e-05,
+      "loss": 0.3211,
+      "step": 882
+    },
+    {
+      "epoch": 1.732090284592738,
+      "grad_norm": 0.10943316735201042,
+      "learning_rate": 1.2240399222679747e-05,
+      "loss": 0.3166,
+      "step": 883
+    },
+    {
+      "epoch": 1.7340529931305202,
+      "grad_norm": 0.108007053530337,
+      "learning_rate": 1.2225209339563144e-05,
+      "loss": 0.3271,
+      "step": 884
+    },
+    {
+      "epoch": 1.7360157016683022,
+      "grad_norm": 0.1806278564221946,
+      "learning_rate": 1.221001405276837e-05,
+      "loss": 0.3342,
+      "step": 885
+    },
+    {
+      "epoch": 1.7379784102060842,
+      "grad_norm": 0.11217999203542753,
+      "learning_rate": 1.2194813399195518e-05,
+      "loss": 0.3174,
+      "step": 886
+    },
+    {
+      "epoch": 1.7399411187438667,
+      "grad_norm": 0.11446924843967385,
+      "learning_rate": 1.217960741575771e-05,
+      "loss": 0.3375,
+      "step": 887
+    },
+    {
+      "epoch": 1.7419038272816487,
+      "grad_norm": 0.11052403929142486,
+      "learning_rate": 1.2164396139381029e-05,
+      "loss": 0.3207,
+      "step": 888
+    },
+    {
+      "epoch": 1.7438665358194307,
+      "grad_norm": 0.12041833183157077,
+      "learning_rate": 1.2149179607004396e-05,
+      "loss": 0.3215,
+      "step": 889
+    },
+    {
+      "epoch": 1.745829244357213,
+      "grad_norm": 0.11164358188191847,
+      "learning_rate": 1.2133957855579501e-05,
+      "loss": 0.3124,
+      "step": 890
+    },
+    {
+      "epoch": 1.7477919528949952,
+      "grad_norm": 0.10293795446688632,
+      "learning_rate": 1.2118730922070707e-05,
+      "loss": 0.323,
+      "step": 891
+    },
+    {
+      "epoch": 1.7497546614327772,
+      "grad_norm": 0.10970003809084786,
+      "learning_rate": 1.210349884345496e-05,
+      "loss": 0.3267,
+      "step": 892
+    },
+    {
+      "epoch": 1.7517173699705593,
+      "grad_norm": 0.10828051591336894,
+      "learning_rate": 1.20882616567217e-05,
+      "loss": 0.329,
+      "step": 893
+    },
+    {
+      "epoch": 1.7536800785083415,
+      "grad_norm": 0.11148013332448067,
+      "learning_rate": 1.2073019398872778e-05,
+      "loss": 0.3201,
+      "step": 894
+    },
+    {
+      "epoch": 1.7556427870461238,
+      "grad_norm": 0.1154913650725476,
+      "learning_rate": 1.205777210692235e-05,
+      "loss": 0.3318,
+      "step": 895
+    },
+    {
+      "epoch": 1.7576054955839058,
+      "grad_norm": 0.10882465073502968,
+      "learning_rate": 1.2042519817896805e-05,
+      "loss": 0.3311,
+      "step": 896
+    },
+    {
+      "epoch": 1.7576054955839058,
+      "eval_loss": 0.4246857762336731,
+      "eval_runtime": 245.6728,
+      "eval_samples_per_second": 123.551,
+      "eval_steps_per_second": 3.863,
+      "step": 896
+    },
+    {
+      "epoch": 1.7595682041216878,
+      "grad_norm": 0.1164747664318698,
+      "learning_rate": 1.202726256883466e-05,
+      "loss": 0.317,
+      "step": 897
+    },
+    {
+      "epoch": 1.76153091265947,
+      "grad_norm": 0.11482257791759276,
+      "learning_rate": 1.2012000396786485e-05,
+      "loss": 0.3361,
+      "step": 898
+    },
+    {
+      "epoch": 1.7634936211972523,
+      "grad_norm": 0.10646048917975168,
+      "learning_rate": 1.1996733338814795e-05,
+      "loss": 0.3241,
+      "step": 899
+    },
+    {
+      "epoch": 1.7654563297350343,
+      "grad_norm": 0.10984226703740471,
+      "learning_rate": 1.1981461431993978e-05,
+      "loss": 0.3133,
+      "step": 900
+    },
+    {
+      "epoch": 1.7674190382728163,
+      "grad_norm": 0.1047308434934182,
+      "learning_rate": 1.1966184713410192e-05,
+      "loss": 0.3335,
+      "step": 901
+    },
+    {
+      "epoch": 1.7693817468105988,
+      "grad_norm": 0.10425069083553949,
+      "learning_rate": 1.1950903220161286e-05,
+      "loss": 0.3226,
+      "step": 902
+    },
+    {
+      "epoch": 1.7713444553483808,
+      "grad_norm": 0.10353476391750117,
+      "learning_rate": 1.1935616989356693e-05,
+      "loss": 0.3341,
+      "step": 903
+    },
+    {
+      "epoch": 1.7733071638861628,
+      "grad_norm": 0.13560691295515553,
+      "learning_rate": 1.1920326058117364e-05,
+      "loss": 0.3233,
+      "step": 904
+    },
+    {
+      "epoch": 1.775269872423945,
+      "grad_norm": 0.10178800675902581,
+      "learning_rate": 1.190503046357565e-05,
+      "loss": 0.3167,
+      "step": 905
+    },
+    {
+      "epoch": 1.7772325809617273,
+      "grad_norm": 0.21889157131254638,
+      "learning_rate": 1.1889730242875243e-05,
+      "loss": 0.3422,
+      "step": 906
+    },
+    {
+      "epoch": 1.7791952894995093,
+      "grad_norm": 0.10614581871424822,
+      "learning_rate": 1.1874425433171055e-05,
+      "loss": 0.309,
+      "step": 907
+    },
+    {
+      "epoch": 1.7811579980372914,
+      "grad_norm": 0.10365379424748211,
+      "learning_rate": 1.1859116071629148e-05,
+      "loss": 0.3137,
+      "step": 908
+    },
+    {
+      "epoch": 1.7831207065750736,
+      "grad_norm": 0.10348421216023944,
+      "learning_rate": 1.1843802195426634e-05,
+      "loss": 0.3225,
+      "step": 909
+    },
+    {
+      "epoch": 1.7850834151128558,
+      "grad_norm": 0.1060682052762049,
+      "learning_rate": 1.1828483841751597e-05,
+      "loss": 0.3139,
+      "step": 910
+    },
+    {
+      "epoch": 1.7870461236506379,
+      "grad_norm": 0.11042414275132055,
+      "learning_rate": 1.1813161047802986e-05,
+      "loss": 0.35,
+      "step": 911
+    },
+    {
+      "epoch": 1.7890088321884199,
+      "grad_norm": 0.11161988036910164,
+      "learning_rate": 1.1797833850790527e-05,
+      "loss": 0.3313,
+      "step": 912
+    },
+    {
+      "epoch": 1.7909715407262021,
+      "grad_norm": 0.11073431733692736,
+      "learning_rate": 1.1782502287934659e-05,
+      "loss": 0.3326,
+      "step": 913
+    },
+    {
+      "epoch": 1.7929342492639844,
+      "grad_norm": 0.10914796174284935,
+      "learning_rate": 1.1767166396466404e-05,
+      "loss": 0.3177,
+      "step": 914
+    },
+    {
+      "epoch": 1.7948969578017664,
+      "grad_norm": 0.10465459851626065,
+      "learning_rate": 1.1751826213627297e-05,
+      "loss": 0.3266,
+      "step": 915
+    },
+    {
+      "epoch": 1.7968596663395484,
+      "grad_norm": 0.10041864035051187,
+      "learning_rate": 1.1736481776669307e-05,
+      "loss": 0.3213,
+      "step": 916
+    },
+    {
+      "epoch": 1.7988223748773309,
+      "grad_norm": 0.11485845865027179,
+      "learning_rate": 1.172113312285472e-05,
+      "loss": 0.3265,
+      "step": 917
+    },
+    {
+      "epoch": 1.800785083415113,
+      "grad_norm": 0.12033881568016211,
+      "learning_rate": 1.1705780289456069e-05,
+      "loss": 0.3197,
+      "step": 918
+    },
+    {
+      "epoch": 1.802747791952895,
+      "grad_norm": 0.11332046772222625,
+      "learning_rate": 1.1690423313756037e-05,
+      "loss": 0.3139,
+      "step": 919
+    },
+    {
+      "epoch": 1.8047105004906772,
+      "grad_norm": 0.10473876200254037,
+      "learning_rate": 1.1675062233047365e-05,
+      "loss": 0.3281,
+      "step": 920
+    },
+    {
+      "epoch": 1.8066732090284594,
+      "grad_norm": 0.1184964198201036,
+      "learning_rate": 1.165969708463276e-05,
+      "loss": 0.3223,
+      "step": 921
+    },
+    {
+      "epoch": 1.8086359175662414,
+      "grad_norm": 0.12474305158961706,
+      "learning_rate": 1.1644327905824808e-05,
+      "loss": 0.3214,
+      "step": 922
+    },
+    {
+      "epoch": 1.8105986261040234,
+      "grad_norm": 0.09933741044605185,
+      "learning_rate": 1.162895473394589e-05,
+      "loss": 0.33,
+      "step": 923
+    },
+    {
+      "epoch": 1.8125613346418057,
+      "grad_norm": 0.25775919840623956,
+      "learning_rate": 1.1613577606328068e-05,
+      "loss": 0.3202,
+      "step": 924
+    },
+    {
+      "epoch": 1.814524043179588,
+      "grad_norm": 0.10094921766955992,
+      "learning_rate": 1.1598196560313024e-05,
+      "loss": 0.3092,
+      "step": 925
+    },
+    {
+      "epoch": 1.81648675171737,
+      "grad_norm": 0.11514280556001413,
+      "learning_rate": 1.1582811633251949e-05,
+      "loss": 0.3109,
+      "step": 926
+    },
+    {
+      "epoch": 1.818449460255152,
+      "grad_norm": 0.13680270431298738,
+      "learning_rate": 1.1567422862505465e-05,
+      "loss": 0.3275,
+      "step": 927
+    },
+    {
+      "epoch": 1.8204121687929342,
+      "grad_norm": 0.10821477293958764,
+      "learning_rate": 1.1552030285443516e-05,
+      "loss": 0.3185,
+      "step": 928
+    },
+    {
+      "epoch": 1.8223748773307165,
+      "grad_norm": 0.1020483099705299,
+      "learning_rate": 1.1536633939445302e-05,
+      "loss": 0.3216,
+      "step": 929
+    },
+    {
+      "epoch": 1.8243375858684985,
+      "grad_norm": 0.11310324846120763,
+      "learning_rate": 1.1521233861899168e-05,
+      "loss": 0.312,
+      "step": 930
+    },
+    {
+      "epoch": 1.8263002944062807,
+      "grad_norm": 0.10125352819285569,
+      "learning_rate": 1.1505830090202524e-05,
+      "loss": 0.3202,
+      "step": 931
+    },
+    {
+      "epoch": 1.828263002944063,
+      "grad_norm": 0.10518388367230969,
+      "learning_rate": 1.1490422661761744e-05,
+      "loss": 0.3161,
+      "step": 932
+    },
+    {
+      "epoch": 1.830225711481845,
+      "grad_norm": 0.10596564037766877,
+      "learning_rate": 1.1475011613992097e-05,
+      "loss": 0.3208,
+      "step": 933
+    },
+    {
+      "epoch": 1.832188420019627,
+      "grad_norm": 0.10832549491208261,
+      "learning_rate": 1.1459596984317622e-05,
+      "loss": 0.3198,
+      "step": 934
+    },
+    {
+      "epoch": 1.8341511285574092,
+      "grad_norm": 0.10340633509154877,
+      "learning_rate": 1.1444178810171074e-05,
+      "loss": 0.3347,
+      "step": 935
+    },
+    {
+      "epoch": 1.8361138370951915,
+      "grad_norm": 0.11187292516624683,
+      "learning_rate": 1.1428757128993801e-05,
+      "loss": 0.3159,
+      "step": 936
+    },
+    {
+      "epoch": 1.8380765456329735,
+      "grad_norm": 0.10201292596555332,
+      "learning_rate": 1.1413331978235677e-05,
+      "loss": 0.3166,
+      "step": 937
+    },
+    {
+      "epoch": 1.8400392541707555,
+      "grad_norm": 0.10804358781472774,
+      "learning_rate": 1.1397903395354996e-05,
+      "loss": 0.323,
+      "step": 938
+    },
+    {
+      "epoch": 1.8420019627085378,
+      "grad_norm": 0.10468656385490047,
+      "learning_rate": 1.138247141781839e-05,
+      "loss": 0.3185,
+      "step": 939
+    },
+    {
+      "epoch": 1.84396467124632,
+      "grad_norm": 0.21138099918352082,
+      "learning_rate": 1.1367036083100735e-05,
+      "loss": 0.3321,
+      "step": 940
+    },
+    {
+      "epoch": 1.845927379784102,
+      "grad_norm": 0.1007451545158691,
+      "learning_rate": 1.1351597428685055e-05,
+      "loss": 0.3201,
+      "step": 941
+    },
+    {
+      "epoch": 1.847890088321884,
+      "grad_norm": 0.10166102006184587,
+      "learning_rate": 1.1336155492062439e-05,
+      "loss": 0.3217,
+      "step": 942
+    },
+    {
+      "epoch": 1.8498527968596663,
+      "grad_norm": 0.10712467451663077,
+      "learning_rate": 1.132071031073195e-05,
+      "loss": 0.3207,
+      "step": 943
+    },
+    {
+      "epoch": 1.8518155053974485,
+      "grad_norm": 0.11530938712137909,
+      "learning_rate": 1.130526192220052e-05,
+      "loss": 0.3265,
+      "step": 944
+    },
+    {
+      "epoch": 1.8537782139352306,
+      "grad_norm": 0.102223465331792,
+      "learning_rate": 1.1289810363982875e-05,
+      "loss": 0.3109,
+      "step": 945
+    },
+    {
+      "epoch": 1.8557409224730128,
+      "grad_norm": 0.10435635169915954,
+      "learning_rate": 1.1274355673601446e-05,
+      "loss": 0.3215,
+      "step": 946
+    },
+    {
+      "epoch": 1.857703631010795,
+      "grad_norm": 0.10323000509883015,
+      "learning_rate": 1.1258897888586256e-05,
+      "loss": 0.3245,
+      "step": 947
+    },
+    {
+      "epoch": 1.859666339548577,
+      "grad_norm": 0.2356211384064762,
+      "learning_rate": 1.1243437046474854e-05,
+      "loss": 0.325,
+      "step": 948
+    },
+    {
+      "epoch": 1.861629048086359,
+      "grad_norm": 0.09975701738920462,
+      "learning_rate": 1.1227973184812207e-05,
+      "loss": 0.3125,
+      "step": 949
+    },
+    {
+      "epoch": 1.8635917566241413,
+      "grad_norm": 0.10475942757691821,
+      "learning_rate": 1.1212506341150615e-05,
+      "loss": 0.3168,
+      "step": 950
+    },
+    {
+      "epoch": 1.8655544651619236,
+      "grad_norm": 0.4497324410418576,
+      "learning_rate": 1.1197036553049626e-05,
+      "loss": 0.3325,
+      "step": 951
+    },
+    {
+      "epoch": 1.8675171736997056,
+      "grad_norm": 0.10885422432097863,
+      "learning_rate": 1.118156385807593e-05,
+      "loss": 0.3134,
+      "step": 952
+    },
+    {
+      "epoch": 1.8694798822374876,
+      "grad_norm": 0.9319115703562274,
+      "learning_rate": 1.1166088293803276e-05,
+      "loss": 0.3329,
+      "step": 953
+    },
+    {
+      "epoch": 1.8714425907752699,
+      "grad_norm": 0.1123785787687025,
+      "learning_rate": 1.1150609897812387e-05,
+      "loss": 0.311,
+      "step": 954
+    },
+    {
+      "epoch": 1.873405299313052,
+      "grad_norm": 0.10319256076853457,
+      "learning_rate": 1.1135128707690862e-05,
+      "loss": 0.3156,
+      "step": 955
+    },
+    {
+      "epoch": 1.8753680078508341,
+      "grad_norm": 0.7095695854915799,
+      "learning_rate": 1.1119644761033079e-05,
+      "loss": 0.3579,
+      "step": 956
+    },
+    {
+      "epoch": 1.8773307163886161,
+      "grad_norm": 0.167595727369702,
+      "learning_rate": 1.1104158095440115e-05,
+      "loss": 0.3277,
+      "step": 957
+    },
+    {
+      "epoch": 1.8792934249263984,
+      "grad_norm": 0.270060302607285,
+      "learning_rate": 1.1088668748519646e-05,
+      "loss": 0.3242,
+      "step": 958
+    },
+    {
+      "epoch": 1.8812561334641806,
+      "grad_norm": 0.15698959445501914,
+      "learning_rate": 1.1073176757885866e-05,
+      "loss": 0.361,
+      "step": 959
+    },
+    {
+      "epoch": 1.8832188420019627,
+      "grad_norm": 0.3221946038337756,
+      "learning_rate": 1.105768216115938e-05,
+      "loss": 0.3188,
+      "step": 960
+    },
+    {
+      "epoch": 1.885181550539745,
+      "grad_norm": 0.1406486088615483,
+      "learning_rate": 1.1042184995967127e-05,
+      "loss": 0.3312,
+      "step": 961
+    },
+    {
+      "epoch": 1.8871442590775271,
+      "grad_norm": 0.13397810433694357,
+      "learning_rate": 1.1026685299942286e-05,
+      "loss": 0.3261,
+      "step": 962
+    },
+    {
+      "epoch": 1.8891069676153092,
+      "grad_norm": 1.4841344998774781,
+      "learning_rate": 1.1011183110724173e-05,
+      "loss": 0.3272,
+      "step": 963
+    },
+    {
+      "epoch": 1.8910696761530912,
+      "grad_norm": 0.1659981257856718,
+      "learning_rate": 1.0995678465958168e-05,
+      "loss": 0.3288,
+      "step": 964
+    },
+    {
+      "epoch": 1.8930323846908734,
+      "grad_norm": 0.16778967117736607,
+      "learning_rate": 1.098017140329561e-05,
+      "loss": 0.3232,
+      "step": 965
+    },
+    {
+      "epoch": 1.8949950932286557,
+      "grad_norm": 0.21080374151960482,
+      "learning_rate": 1.0964661960393703e-05,
+      "loss": 0.3274,
+      "step": 966
+    },
+    {
+      "epoch": 1.8969578017664377,
+      "grad_norm": 0.16463715782689323,
+      "learning_rate": 1.0949150174915441e-05,
+      "loss": 0.3352,
+      "step": 967
+    },
+    {
+      "epoch": 1.8989205103042197,
+      "grad_norm": 0.2267681805465417,
+      "learning_rate": 1.0933636084529507e-05,
+      "loss": 0.3388,
+      "step": 968
+    },
+    {
+      "epoch": 1.900883218842002,
+      "grad_norm": 0.14488446998635424,
+      "learning_rate": 1.0918119726910175e-05,
+      "loss": 0.3202,
+      "step": 969
+    },
+    {
+      "epoch": 1.9028459273797842,
+      "grad_norm": 0.18140470030881337,
+      "learning_rate": 1.0902601139737225e-05,
+      "loss": 0.3329,
+      "step": 970
+    },
+    {
+      "epoch": 1.9048086359175662,
+      "grad_norm": 0.1546698941699075,
+      "learning_rate": 1.0887080360695855e-05,
+      "loss": 0.3306,
+      "step": 971
+    },
+    {
+      "epoch": 1.9067713444553482,
+      "grad_norm": 1.2221913272883775,
+      "learning_rate": 1.0871557427476585e-05,
+      "loss": 0.3399,
+      "step": 972
+    },
+    {
+      "epoch": 1.9087340529931305,
+      "grad_norm": 0.15426601384648553,
+      "learning_rate": 1.0856032377775161e-05,
+      "loss": 0.3358,
+      "step": 973
+    },
+    {
+      "epoch": 1.9106967615309127,
+      "grad_norm": 1.1114760751159305,
+      "learning_rate": 1.0840505249292477e-05,
+      "loss": 0.3356,
+      "step": 974
+    },
+    {
+      "epoch": 1.9126594700686947,
+      "grad_norm": 0.22900624172519007,
+      "learning_rate": 1.0824976079734472e-05,
+      "loss": 0.3134,
+      "step": 975
+    },
+    {
+      "epoch": 1.914622178606477,
+      "grad_norm": 0.9459775335785819,
+      "learning_rate": 1.0809444906812034e-05,
+      "loss": 0.3235,
+      "step": 976
+    },
+    {
+      "epoch": 1.9165848871442592,
+      "grad_norm": 0.17887149210801348,
+      "learning_rate": 1.079391176824093e-05,
+      "loss": 0.3304,
+      "step": 977
+    },
+    {
+      "epoch": 1.9185475956820413,
+      "grad_norm": 0.17484413383294636,
+      "learning_rate": 1.0778376701741688e-05,
+      "loss": 0.3298,
+      "step": 978
+    },
+    {
+      "epoch": 1.9205103042198233,
+      "grad_norm": 0.15868896830006096,
+      "learning_rate": 1.0762839745039526e-05,
+      "loss": 0.325,
+      "step": 979
+    },
+    {
+      "epoch": 1.9224730127576055,
+      "grad_norm": 0.13952635278363976,
+      "learning_rate": 1.0747300935864245e-05,
+      "loss": 0.3246,
+      "step": 980
+    },
+    {
+      "epoch": 1.9244357212953878,
+      "grad_norm": 0.14676838005524964,
+      "learning_rate": 1.073176031195015e-05,
+      "loss": 0.3367,
+      "step": 981
+    },
+    {
+      "epoch": 1.9263984298331698,
+      "grad_norm": 0.1393769904490578,
+      "learning_rate": 1.0716217911035952e-05,
+      "loss": 0.331,
+      "step": 982
+    },
+    {
+      "epoch": 1.9283611383709518,
+      "grad_norm": 0.12073516486793993,
+      "learning_rate": 1.0700673770864673e-05,
+      "loss": 0.3215,
+      "step": 983
+    },
+    {
+      "epoch": 1.930323846908734,
+      "grad_norm": 0.17261593509801376,
+      "learning_rate": 1.0685127929183567e-05,
+      "loss": 0.3377,
+      "step": 984
+    },
+    {
+      "epoch": 1.9322865554465163,
+      "grad_norm": 0.1288861939566597,
+      "learning_rate": 1.0669580423744014e-05,
+      "loss": 0.3268,
+      "step": 985
+    },
+    {
+      "epoch": 1.9342492639842983,
+      "grad_norm": 0.12242177876640145,
+      "learning_rate": 1.0654031292301432e-05,
+      "loss": 0.3331,
+      "step": 986
+    },
+    {
+      "epoch": 1.9362119725220803,
+      "grad_norm": 0.135438680122695,
+      "learning_rate": 1.063848057261519e-05,
+      "loss": 0.3276,
+      "step": 987
+    },
+    {
+      "epoch": 1.9381746810598626,
+      "grad_norm": 0.12113837642836334,
+      "learning_rate": 1.0622928302448523e-05,
+      "loss": 0.3196,
+      "step": 988
+    },
+    {
+      "epoch": 1.9401373895976448,
+      "grad_norm": 0.117144755644724,
+      "learning_rate": 1.0607374519568412e-05,
+      "loss": 0.3247,
+      "step": 989
+    },
+    {
+      "epoch": 1.9421000981354268,
+      "grad_norm": 0.11697508630127905,
+      "learning_rate": 1.0591819261745528e-05,
+      "loss": 0.3223,
+      "step": 990
+    },
+    {
+      "epoch": 1.944062806673209,
+      "grad_norm": 0.11238405496608947,
+      "learning_rate": 1.0576262566754121e-05,
+      "loss": 0.3212,
+      "step": 991
+    },
+    {
+      "epoch": 1.9460255152109913,
+      "grad_norm": 0.1139647888181511,
+      "learning_rate": 1.0560704472371919e-05,
+      "loss": 0.3253,
+      "step": 992
+    },
+    {
+      "epoch": 1.9479882237487733,
+      "grad_norm": 0.11466955150117668,
+      "learning_rate": 1.0545145016380065e-05,
+      "loss": 0.3273,
+      "step": 993
+    },
+    {
+      "epoch": 1.9499509322865554,
+      "grad_norm": 0.10970406400371188,
+      "learning_rate": 1.0529584236562995e-05,
+      "loss": 0.3254,
+      "step": 994
+    },
+    {
+      "epoch": 1.9519136408243376,
+      "grad_norm": 0.11114557334217089,
+      "learning_rate": 1.0514022170708374e-05,
+      "loss": 0.3246,
+      "step": 995
+    },
+    {
+      "epoch": 1.9538763493621198,
+      "grad_norm": 0.11801065397328833,
+      "learning_rate": 1.0498458856606972e-05,
+      "loss": 0.3294,
+      "step": 996
+    },
+    {
+      "epoch": 1.9558390578999019,
+      "grad_norm": 0.10405757613653188,
+      "learning_rate": 1.0482894332052607e-05,
+      "loss": 0.3261,
+      "step": 997
+    },
+    {
+      "epoch": 1.9578017664376839,
+      "grad_norm": 0.1103740005321899,
+      "learning_rate": 1.0467328634842024e-05,
+      "loss": 0.3289,
+      "step": 998
+    },
+    {
+      "epoch": 1.9597644749754661,
+      "grad_norm": 0.09987852489854114,
+      "learning_rate": 1.0451761802774824e-05,
+      "loss": 0.3172,
+      "step": 999
+    },
+    {
+      "epoch": 1.9617271835132484,
+      "grad_norm": 0.10893453828326174,
+      "learning_rate": 1.0436193873653362e-05,
+      "loss": 0.3228,
+      "step": 1000
+    },
+    {
+      "epoch": 1.9636898920510304,
+      "grad_norm": 0.10174600038711792,
+      "learning_rate": 1.0420624885282653e-05,
+      "loss": 0.3225,
+      "step": 1001
+    },
+    {
+      "epoch": 1.9656526005888124,
+      "grad_norm": 0.11640082237562227,
+      "learning_rate": 1.0405054875470287e-05,
+      "loss": 0.3193,
+      "step": 1002
+    },
+    {
+      "epoch": 1.9676153091265947,
+      "grad_norm": 0.15766357183117252,
+      "learning_rate": 1.0389483882026334e-05,
+      "loss": 0.3216,
+      "step": 1003
+    },
+    {
+      "epoch": 1.969578017664377,
+      "grad_norm": 0.10997358802993172,
+      "learning_rate": 1.037391194276326e-05,
+      "loss": 0.3142,
+      "step": 1004
+    },
+    {
+      "epoch": 1.971540726202159,
+      "grad_norm": 0.10152791127501247,
+      "learning_rate": 1.0358339095495811e-05,
+      "loss": 0.3169,
+      "step": 1005
+    },
+    {
+      "epoch": 1.9735034347399412,
+      "grad_norm": 0.20219624773435688,
+      "learning_rate": 1.0342765378040953e-05,
+      "loss": 0.324,
+      "step": 1006
+    },
+    {
+      "epoch": 1.9754661432777234,
+      "grad_norm": 0.10297719458762343,
+      "learning_rate": 1.0327190828217763e-05,
+      "loss": 0.3181,
+      "step": 1007
+    },
+    {
+      "epoch": 1.9774288518155054,
+      "grad_norm": 0.11052971601307274,
+      "learning_rate": 1.0311615483847333e-05,
+      "loss": 0.3245,
+      "step": 1008
+    },
+    {
+      "epoch": 1.9793915603532874,
+      "grad_norm": 0.09914734230876245,
+      "learning_rate": 1.0296039382752687e-05,
+      "loss": 0.324,
+      "step": 1009
+    },
+    {
+      "epoch": 1.9813542688910697,
+      "grad_norm": 0.10159497435434399,
+      "learning_rate": 1.028046256275869e-05,
+      "loss": 0.3119,
+      "step": 1010
+    },
+    {
+      "epoch": 1.983316977428852,
+      "grad_norm": 0.11056760688752829,
+      "learning_rate": 1.0264885061691954e-05,
+      "loss": 0.323,
+      "step": 1011
+    },
+    {
+      "epoch": 1.985279685966634,
+      "grad_norm": 0.10804474565070168,
+      "learning_rate": 1.0249306917380731e-05,
+      "loss": 0.3113,
+      "step": 1012
+    },
+    {
+      "epoch": 1.987242394504416,
+      "grad_norm": 0.13940608144653047,
+      "learning_rate": 1.023372816765485e-05,
+      "loss": 0.3123,
+      "step": 1013
+    },
+    {
+      "epoch": 1.9892051030421982,
+      "grad_norm": 0.10238052023308335,
+      "learning_rate": 1.0218148850345613e-05,
+      "loss": 0.3235,
+      "step": 1014
+    },
+    {
+      "epoch": 1.9911678115799805,
+      "grad_norm": 0.12527000090641732,
+      "learning_rate": 1.0202569003285683e-05,
+      "loss": 0.3337,
+      "step": 1015
+    },
+    {
+      "epoch": 1.9931305201177625,
+      "grad_norm": 0.11514202583090338,
+      "learning_rate": 1.0186988664309023e-05,
+      "loss": 0.333,
+      "step": 1016
+    },
+    {
+      "epoch": 1.9950932286555445,
+      "grad_norm": 0.11172112347247365,
+      "learning_rate": 1.017140787125079e-05,
+      "loss": 0.3227,
+      "step": 1017
+    },
+    {
+      "epoch": 1.9970559371933267,
+      "grad_norm": 0.15649181628225953,
+      "learning_rate": 1.0155826661947232e-05,
+      "loss": 0.3208,
+      "step": 1018
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 2036,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 509,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 6.450464370500567e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}