diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5621 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.8313847752663029,
+  "eval_steps": 500,
+  "global_step": 4000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 18.487945556640625,
+      "learning_rate": 5e-06,
+      "loss": 13.0056,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 14.99580192565918,
+      "learning_rate": 1e-05,
+      "loss": 12.7549,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 12.252545356750488,
+      "learning_rate": 1.5e-05,
+      "loss": 12.1667,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 11.030613899230957,
+      "learning_rate": 2e-05,
+      "loss": 11.3488,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.9062275886535645,
+      "learning_rate": 2.5e-05,
+      "loss": 10.3979,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.439232349395752,
+      "learning_rate": 3e-05,
+      "loss": 9.5814,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.189538955688477,
+      "learning_rate": 3.5e-05,
+      "loss": 8.9368,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.1056129932403564,
+      "learning_rate": 4e-05,
+      "loss": 8.4454,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.7891372442245483,
+      "learning_rate": 4.5e-05,
+      "loss": 8.0365,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1064869165420532,
+      "learning_rate": 5e-05,
+      "loss": 7.8555,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9185166954994202,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 7.7611,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8066293001174927,
+      "learning_rate": 6e-05,
+      "loss": 7.6786,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7840200662612915,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 7.5377,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6963502168655396,
+      "learning_rate": 7e-05,
+      "loss": 7.4308,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6896659731864929,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 7.2941,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9641470909118652,
+      "learning_rate": 8e-05,
+      "loss": 7.113,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6882058382034302,
+      "learning_rate": 8.5e-05,
+      "loss": 6.9274,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9275090098381042,
+      "learning_rate": 9e-05,
+      "loss": 6.7585,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9500148892402649,
+      "learning_rate": 9.5e-05,
+      "loss": 6.6093,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4086371660232544,
+      "learning_rate": 0.0001,
+      "loss": 6.4294,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.023192286491394,
+      "learning_rate": 9.999972205865686e-05,
+      "loss": 6.3935,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.651755928993225,
+      "learning_rate": 9.999888823771751e-05,
+      "loss": 6.1994,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.9642592668533325,
+      "learning_rate": 9.999749854645204e-05,
+      "loss": 6.1148,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4170254468917847,
+      "learning_rate": 9.99955530003106e-05,
+      "loss": 6.017,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.4251655340194702,
+      "learning_rate": 9.99930516209231e-05,
+      "loss": 5.8718,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2349293231964111,
+      "learning_rate": 9.998999443609897e-05,
+      "loss": 5.8388,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.8177589178085327,
+      "learning_rate": 9.998638147982696e-05,
+      "loss": 5.8051,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.7111365795135498,
+      "learning_rate": 9.998221279227467e-05,
+      "loss": 5.7493,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.7238991260528564,
+      "learning_rate": 9.997748841978812e-05,
+      "loss": 5.677,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.7837049961090088,
+      "learning_rate": 9.997220841489122e-05,
+      "loss": 5.5433,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.4730719327926636,
+      "learning_rate": 9.996637283628528e-05,
+      "loss": 5.6268,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.5730408430099487,
+      "learning_rate": 9.995998174884821e-05,
+      "loss": 5.412,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.4034667015075684,
+      "learning_rate": 9.995303522363394e-05,
+      "loss": 5.4502,
+      "step": 165
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.4483870267868042,
+      "learning_rate": 9.99455333378715e-05,
+      "loss": 5.4505,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.6795587539672852,
+      "learning_rate": 9.993747617496428e-05,
+      "loss": 5.3595,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2084821462631226,
+      "learning_rate": 9.9928863824489e-05,
+      "loss": 5.3636,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.4595478773117065,
+      "learning_rate": 9.99196963821948e-05,
+      "loss": 5.2336,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.4178177118301392,
+      "learning_rate": 9.990997395000217e-05,
+      "loss": 5.2247,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.4685578346252441,
+      "learning_rate": 9.989969663600169e-05,
+      "loss": 5.1575,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.5389108657836914,
+      "learning_rate": 9.9888864554453e-05,
+      "loss": 5.1734,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.5792580842971802,
+      "learning_rate": 9.987747782578342e-05,
+      "loss": 5.1492,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.6819140911102295,
+      "learning_rate": 9.986553657658668e-05,
+      "loss": 5.1638,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.5122543573379517,
+      "learning_rate": 9.985304093962145e-05,
+      "loss": 5.1114,
+      "step": 215
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.5873609781265259,
+      "learning_rate": 9.983999105380988e-05,
+      "loss": 5.0102,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.7786037921905518,
+      "learning_rate": 9.982638706423608e-05,
+      "loss": 4.9825,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.9327354431152344,
+      "learning_rate": 9.98122291221445e-05,
+      "loss": 5.0193,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.6568500995635986,
+      "learning_rate": 9.979751738493826e-05,
+      "loss": 4.9165,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.7649105787277222,
+      "learning_rate": 9.978225201617732e-05,
+      "loss": 4.9153,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2607721090316772,
+      "learning_rate": 9.976643318557678e-05,
+      "loss": 4.9862,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.7145535945892334,
+      "learning_rate": 9.975006106900495e-05,
+      "loss": 4.8554,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.1020963191986084,
+      "learning_rate": 9.973313584848132e-05,
+      "loss": 4.8396,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.910445213317871,
+      "learning_rate": 9.971565771217464e-05,
+      "loss": 4.8259,
+      "step": 260
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.9731266498565674,
+      "learning_rate": 9.969762685440076e-05,
+      "loss": 4.7914,
+      "step": 265
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.589390754699707,
+      "learning_rate": 9.967904347562054e-05,
+      "loss": 4.8357,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2861696481704712,
+      "learning_rate": 9.965990778243755e-05,
+      "loss": 4.7907,
+      "step": 275
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.6012290716171265,
+      "learning_rate": 9.964021998759577e-05,
+      "loss": 4.7597,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.6387654542922974,
+      "learning_rate": 9.961998030997733e-05,
+      "loss": 4.6971,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.5079236030578613,
+      "learning_rate": 9.95991889745999e-05,
+      "loss": 4.7803,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.6518876552581787,
+      "learning_rate": 9.957784621261441e-05,
+      "loss": 4.7649,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.4536391496658325,
+      "learning_rate": 9.955595226130226e-05,
+      "loss": 4.7011,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.4148342609405518,
+      "learning_rate": 9.953350736407282e-05,
+      "loss": 4.7163,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.705102562904358,
+      "learning_rate": 9.951051177046069e-05,
+      "loss": 4.6227,
+      "step": 310
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.2245330810546875,
+      "learning_rate": 9.948696573612292e-05,
+      "loss": 4.6069,
+      "step": 315
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.7924284934997559,
+      "learning_rate": 9.946286952283618e-05,
+      "loss": 4.6406,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.903428554534912,
+      "learning_rate": 9.943822339849381e-05,
+      "loss": 4.6441,
+      "step": 325
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.7252001762390137,
+      "learning_rate": 9.941302763710288e-05,
+      "loss": 4.5764,
+      "step": 330
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.652077078819275,
+      "learning_rate": 9.938728251878116e-05,
+      "loss": 4.5748,
+      "step": 335
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.6563273668289185,
+      "learning_rate": 9.936098832975393e-05,
+      "loss": 4.6605,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.196190357208252,
+      "learning_rate": 9.933414536235091e-05,
+      "loss": 4.5116,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.7059569358825684,
+      "learning_rate": 9.93067539150029e-05,
+      "loss": 4.545,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.919015645980835,
+      "learning_rate": 9.927881429223853e-05,
+      "loss": 4.4465,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.4034647941589355,
+      "learning_rate": 9.925032680468085e-05,
+      "loss": 4.5262,
+      "step": 360
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.4941623210906982,
+      "learning_rate": 9.922129176904388e-05,
+      "loss": 4.5184,
+      "step": 365
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.7192869186401367,
+      "learning_rate": 9.919170950812911e-05,
+      "loss": 4.5419,
+      "step": 370
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.8791921138763428,
+      "learning_rate": 9.916158035082184e-05,
+      "loss": 4.4524,
+      "step": 375
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.7824347019195557,
+      "learning_rate": 9.913090463208763e-05,
+      "loss": 4.5193,
+      "step": 380
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.8421388864517212,
+      "learning_rate": 9.90996826929685e-05,
+      "loss": 4.4552,
+      "step": 385
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.1372251510620117,
+      "learning_rate": 9.906791488057916e-05,
+      "loss": 4.5304,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.8615418672561646,
+      "learning_rate": 9.903560154810313e-05,
+      "loss": 4.4751,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.6029841899871826,
+      "learning_rate": 9.900274305478887e-05,
+      "loss": 4.3688,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.8817858695983887,
+      "learning_rate": 9.896933976594572e-05,
+      "loss": 4.4703,
+      "step": 405
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.6392571926116943,
+      "learning_rate": 9.893539205293989e-05,
+      "loss": 4.4149,
+      "step": 410
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.00087308883667,
+      "learning_rate": 9.890090029319028e-05,
+      "loss": 4.4234,
+      "step": 415
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.6066954135894775,
+      "learning_rate": 9.886586487016433e-05,
+      "loss": 4.4217,
+      "step": 420
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.6025770902633667,
+      "learning_rate": 9.883028617337378e-05,
+      "loss": 4.4763,
+      "step": 425
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.266241431236267,
+      "learning_rate": 9.879416459837022e-05,
+      "loss": 4.4888,
+      "step": 430
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.8539241552352905,
+      "learning_rate": 9.875750054674082e-05,
+      "loss": 4.3752,
+      "step": 435
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.6009455919265747,
+      "learning_rate": 9.872029442610382e-05,
+      "loss": 4.4046,
+      "step": 440
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.015383243560791,
+      "learning_rate": 9.8682546650104e-05,
+      "loss": 4.261,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.6920061111450195,
+      "learning_rate": 9.864425763840802e-05,
+      "loss": 4.2679,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.5430705547332764,
+      "learning_rate": 9.860542781669988e-05,
+      "loss": 4.2743,
+      "step": 455
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.410170555114746,
+      "learning_rate": 9.85660576166761e-05,
+      "loss": 4.2691,
+      "step": 460
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.5226540565490723,
+      "learning_rate": 9.852614747604093e-05,
+      "loss": 4.2518,
+      "step": 465
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.3666799068450928,
+      "learning_rate": 9.848569783850145e-05,
+      "loss": 4.2504,
+      "step": 470
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.4098422527313232,
+      "learning_rate": 9.844470915376278e-05,
+      "loss": 4.274,
+      "step": 475
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.5550212860107422,
+      "learning_rate": 9.840318187752292e-05,
+      "loss": 4.2033,
+      "step": 480
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.451125144958496,
+      "learning_rate": 9.836111647146771e-05,
+      "loss": 4.2993,
+      "step": 485
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.3934297561645508,
+      "learning_rate": 9.831851340326577e-05,
+      "loss": 4.2035,
+      "step": 490
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.2739324569702148,
+      "learning_rate": 9.82753731465633e-05,
+      "loss": 4.2577,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.430321455001831,
+      "learning_rate": 9.823169618097871e-05,
+      "loss": 4.2677,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.8821338415145874,
+      "learning_rate": 9.81874829920974e-05,
+      "loss": 4.2304,
+      "step": 505
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.5086530447006226,
+      "learning_rate": 9.814273407146623e-05,
+      "loss": 4.2587,
+      "step": 510
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.5528761148452759,
+      "learning_rate": 9.809744991658829e-05,
+      "loss": 4.1734,
+      "step": 515
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.376980185508728,
+      "learning_rate": 9.805163103091708e-05,
+      "loss": 4.2477,
+      "step": 520
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.3953189849853516,
+      "learning_rate": 9.800527792385112e-05,
+      "loss": 4.2327,
+      "step": 525
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.5399726629257202,
+      "learning_rate": 9.79583911107282e-05,
+      "loss": 4.313,
+      "step": 530
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.4923655986785889,
+      "learning_rate": 9.791097111281968e-05,
+      "loss": 4.2384,
+      "step": 535
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.3612940311431885,
+      "learning_rate": 9.786301845732467e-05,
+      "loss": 4.1915,
+      "step": 540
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.4748578071594238,
+      "learning_rate": 9.781453367736418e-05,
+      "loss": 4.2318,
+      "step": 545
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.0609843730926514,
+      "learning_rate": 9.776551731197524e-05,
+      "loss": 4.2361,
+      "step": 550
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.8525745868682861,
+      "learning_rate": 9.771596990610478e-05,
+      "loss": 4.1553,
+      "step": 555
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.5976070165634155,
+      "learning_rate": 9.766589201060372e-05,
+      "loss": 4.1892,
+      "step": 560
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.5703035593032837,
+      "learning_rate": 9.761528418222077e-05,
+      "loss": 4.1657,
+      "step": 565
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.308478593826294,
+      "learning_rate": 9.756414698359624e-05,
+      "loss": 4.1322,
+      "step": 570
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.4874699115753174,
+      "learning_rate": 9.75124809832558e-05,
+      "loss": 4.0567,
+      "step": 575
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.1793978214263916,
+      "learning_rate": 9.746028675560413e-05,
+      "loss": 4.171,
+      "step": 580
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.6352083683013916,
+      "learning_rate": 9.740756488091861e-05,
+      "loss": 4.0955,
+      "step": 585
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.008838176727295,
+      "learning_rate": 9.735431594534277e-05,
+      "loss": 4.1521,
+      "step": 590
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.4547291994094849,
+      "learning_rate": 9.730054054087983e-05,
+      "loss": 4.098,
+      "step": 595
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.298056960105896,
+      "learning_rate": 9.724623926538612e-05,
+      "loss": 4.161,
+      "step": 600
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.6623826026916504,
+      "learning_rate": 9.719141272256443e-05,
+      "loss": 4.1901,
+      "step": 605
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.4018698930740356,
+      "learning_rate": 9.713606152195726e-05,
+      "loss": 4.1762,
+      "step": 610
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.5026352405548096,
+      "learning_rate": 9.708018627894011e-05,
+      "loss": 4.182,
+      "step": 615
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8811168670654297,
+      "learning_rate": 9.702378761471456e-05,
+      "loss": 4.0757,
+      "step": 620
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.5124515295028687,
+      "learning_rate": 9.696686615630146e-05,
+      "loss": 4.0314,
+      "step": 625
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.6260911226272583,
+      "learning_rate": 9.690942253653385e-05,
+      "loss": 4.0444,
+      "step": 630
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.3121532201766968,
+      "learning_rate": 9.685145739405002e-05,
+      "loss": 4.0608,
+      "step": 635
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.7652568817138672,
+      "learning_rate": 9.679297137328634e-05,
+      "loss": 4.0556,
+      "step": 640
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.2742689847946167,
+      "learning_rate": 9.673396512447013e-05,
+      "loss": 4.0923,
+      "step": 645
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.4735512733459473,
+      "learning_rate": 9.667443930361247e-05,
+      "loss": 4.0726,
+      "step": 650
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.2956384420394897,
+      "learning_rate": 9.661439457250076e-05,
+      "loss": 4.076,
+      "step": 655
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.651337742805481,
+      "learning_rate": 9.655383159869158e-05,
+      "loss": 4.0232,
+      "step": 660
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.560766577720642,
+      "learning_rate": 9.649275105550309e-05,
+      "loss": 4.1068,
+      "step": 665
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.3690272569656372,
+      "learning_rate": 9.643115362200762e-05,
+      "loss": 4.1346,
+      "step": 670
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.3641835451126099,
+      "learning_rate": 9.636903998302409e-05,
+      "loss": 4.0315,
+      "step": 675
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.258906364440918,
+      "learning_rate": 9.630641082911045e-05,
+      "loss": 3.9937,
+      "step": 680
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.234425663948059,
+      "learning_rate": 9.624326685655593e-05,
+      "loss": 4.0275,
+      "step": 685
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.208127737045288,
+      "learning_rate": 9.617960876737337e-05,
+      "loss": 4.0103,
+      "step": 690
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.3401658535003662,
+      "learning_rate": 9.611543726929134e-05,
+      "loss": 4.1086,
+      "step": 695
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.2559030055999756,
+      "learning_rate": 9.605075307574635e-05,
+      "loss": 4.0595,
+      "step": 700
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.3014733791351318,
+      "learning_rate": 9.598555690587487e-05,
+      "loss": 3.9846,
+      "step": 705
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.2272580862045288,
+      "learning_rate": 9.591984948450532e-05,
+      "loss": 3.9744,
+      "step": 710
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.804677963256836,
+      "learning_rate": 9.585363154215008e-05,
+      "loss": 4.0576,
+      "step": 715
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.5206153392791748,
+      "learning_rate": 9.578690381499728e-05,
+      "loss": 4.0031,
+      "step": 720
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.6624362468719482,
+      "learning_rate": 9.571966704490271e-05,
+      "loss": 3.9699,
+      "step": 725
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.380760908126831,
+      "learning_rate": 9.565192197938148e-05,
+      "loss": 4.0038,
+      "step": 730
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.5826762914657593,
+      "learning_rate": 9.558366937159977e-05,
+      "loss": 4.0484,
+      "step": 735
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.5320595502853394,
+      "learning_rate": 9.551490998036646e-05,
+      "loss": 3.9104,
+      "step": 740
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.4461925029754639,
+      "learning_rate": 9.544564457012463e-05,
+      "loss": 3.9483,
+      "step": 745
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.2367421388626099,
+      "learning_rate": 9.537587391094314e-05,
+      "loss": 4.0613,
+      "step": 750
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.4271812438964844,
+      "learning_rate": 9.5305598778508e-05,
+      "loss": 3.9746,
+      "step": 755
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.7837293148040771,
+      "learning_rate": 9.52348199541138e-05,
+      "loss": 3.9011,
+      "step": 760
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.224382996559143,
+      "learning_rate": 9.516353822465504e-05,
+      "loss": 3.9156,
+      "step": 765
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.3299424648284912,
+      "learning_rate": 9.509175438261726e-05,
+      "loss": 3.9682,
+      "step": 770
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.3308087587356567,
+      "learning_rate": 9.501946922606838e-05,
+      "loss": 3.9638,
+      "step": 775
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.463572382926941,
+      "learning_rate": 9.494668355864973e-05,
+      "loss": 3.9593,
+      "step": 780
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.3762937784194946,
+      "learning_rate": 9.487339818956716e-05,
+      "loss": 3.8413,
+      "step": 785
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.4410449266433716,
+      "learning_rate": 9.479961393358203e-05,
+      "loss": 3.9589,
+      "step": 790
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.4337133169174194,
+      "learning_rate": 9.472533161100215e-05,
+      "loss": 3.9466,
+      "step": 795
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.4152320623397827,
+      "learning_rate": 9.465055204767265e-05,
+      "loss": 4.0096,
+      "step": 800
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.2973538637161255,
+      "learning_rate": 9.457527607496685e-05,
+      "loss": 3.9461,
+      "step": 805
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.2745497226715088,
+      "learning_rate": 9.44995045297769e-05,
+      "loss": 3.91,
+      "step": 810
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.2977548837661743,
+      "learning_rate": 9.442323825450464e-05,
+      "loss": 4.0103,
+      "step": 815
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.2384387254714966,
+      "learning_rate": 9.43464780970521e-05,
+      "loss": 3.9053,
+      "step": 820
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.2694612741470337,
+      "learning_rate": 9.426922491081212e-05,
+      "loss": 3.9256,
+      "step": 825
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.1504689455032349,
+      "learning_rate": 9.419147955465888e-05,
+      "loss": 3.9034,
+      "step": 830
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.1995341777801514,
+      "learning_rate": 9.411324289293832e-05,
+      "loss": 3.9751,
+      "step": 835
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.3309929370880127,
+      "learning_rate": 9.403451579545859e-05,
+      "loss": 3.8875,
+      "step": 840
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.1273236274719238,
+      "learning_rate": 9.395529913748025e-05,
+      "loss": 3.8719,
+      "step": 845
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.4124709367752075,
+      "learning_rate": 9.387559379970672e-05,
+      "loss": 3.9124,
+      "step": 850
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.2471351623535156,
+      "learning_rate": 9.379540066827431e-05,
+      "loss": 3.8352,
+      "step": 855
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.2889108657836914,
+      "learning_rate": 9.371472063474248e-05,
+      "loss": 3.878,
+      "step": 860
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.1750242710113525,
+      "learning_rate": 9.363355459608394e-05,
+      "loss": 3.9879,
+      "step": 865
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.1907912492752075,
+      "learning_rate": 9.355190345467457e-05,
+      "loss": 3.9429,
+      "step": 870
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.1586787700653076,
+      "learning_rate": 9.346976811828352e-05,
+      "loss": 3.9224,
+      "step": 875
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.3565126657485962,
+      "learning_rate": 9.338714950006297e-05,
+      "loss": 3.836,
+      "step": 880
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.2921334505081177,
+      "learning_rate": 9.330404851853817e-05,
+      "loss": 3.9068,
+      "step": 885
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.2521257400512695,
+      "learning_rate": 9.3220466097597e-05,
+      "loss": 3.8314,
+      "step": 890
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.3298193216323853,
+      "learning_rate": 9.313640316647991e-05,
+      "loss": 3.7788,
+      "step": 895
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.3024024963378906,
+      "learning_rate": 9.305186065976945e-05,
+      "loss": 3.8647,
+      "step": 900
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.1826115846633911,
+      "learning_rate": 9.296683951737993e-05,
+      "loss": 3.7856,
+      "step": 905
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.1611673831939697,
+      "learning_rate": 9.288134068454697e-05,
+      "loss": 3.8901,
+      "step": 910
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.5218857526779175,
+      "learning_rate": 9.2795365111817e-05,
+      "loss": 3.8341,
+      "step": 915
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.2008612155914307,
+      "learning_rate": 9.270891375503665e-05,
+      "loss": 3.8005,
+      "step": 920
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.2103996276855469,
+      "learning_rate": 9.262198757534218e-05,
+      "loss": 3.8177,
+      "step": 925
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.1930011510849,
+      "learning_rate": 9.253458753914874e-05,
+      "loss": 3.7792,
+      "step": 930
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.2458661794662476,
+      "learning_rate": 9.244671461813969e-05,
+      "loss": 3.8579,
+      "step": 935
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.2035701274871826,
+      "learning_rate": 9.235836978925572e-05,
+      "loss": 3.7795,
+      "step": 940
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.2439464330673218,
+      "learning_rate": 9.226955403468406e-05,
+      "loss": 3.8381,
+      "step": 945
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.3288521766662598,
+      "learning_rate": 9.21802683418475e-05,
+      "loss": 3.8374,
+      "step": 950
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.4271641969680786,
+      "learning_rate": 9.209051370339347e-05,
+      "loss": 3.8206,
+      "step": 955
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.034192442893982,
+      "learning_rate": 9.200029111718295e-05,
+      "loss": 3.9133,
+      "step": 960
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.2292518615722656,
+      "learning_rate": 9.190960158627941e-05,
+      "loss": 3.8224,
+      "step": 965
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.3718514442443848,
+      "learning_rate": 9.181844611893766e-05,
+      "loss": 3.8073,
+      "step": 970
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.1448020935058594,
+      "learning_rate": 9.172682572859261e-05,
+      "loss": 3.7932,
+      "step": 975
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.2375144958496094,
+      "learning_rate": 9.163474143384806e-05,
+      "loss": 3.8355,
+      "step": 980
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.2168779373168945,
+      "learning_rate": 9.154219425846528e-05,
+      "loss": 3.7932,
+      "step": 985
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.285558819770813,
+      "learning_rate": 9.144918523135175e-05,
+      "loss": 3.7946,
+      "step": 990
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.4132548570632935,
+      "learning_rate": 9.13557153865496e-05,
+      "loss": 3.7393,
+      "step": 995
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.2185122966766357,
+      "learning_rate": 9.12617857632242e-05,
+      "loss": 3.8195,
+      "step": 1000
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.0606962442398071,
+      "learning_rate": 9.116739740565259e-05,
+      "loss": 3.6942,
+      "step": 1005
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.2439757585525513,
+      "learning_rate": 9.107255136321184e-05,
+      "loss": 3.7296,
+      "step": 1010
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.2272452116012573,
+      "learning_rate": 9.09772486903674e-05,
+      "loss": 3.7419,
+      "step": 1015
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.1247087717056274,
+      "learning_rate": 9.08814904466614e-05,
+      "loss": 3.7929,
+      "step": 1020
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.1768749952316284,
+      "learning_rate": 9.078527769670085e-05,
+      "loss": 3.7367,
+      "step": 1025
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.0592442750930786,
+      "learning_rate": 9.068861151014575e-05,
+      "loss": 3.8446,
+      "step": 1030
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.0224711894989014,
+      "learning_rate": 9.05914929616973e-05,
+      "loss": 3.7538,
+      "step": 1035
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.0650990009307861,
+      "learning_rate": 9.04939231310859e-05,
+      "loss": 3.747,
+      "step": 1040
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.1367437839508057,
+      "learning_rate": 9.039590310305914e-05,
+      "loss": 3.7478,
+      "step": 1045
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.111429214477539,
+      "learning_rate": 9.029743396736974e-05,
+      "loss": 3.6822,
+      "step": 1050
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.1658196449279785,
+      "learning_rate": 9.019851681876348e-05,
+      "loss": 3.7084,
+      "step": 1055
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.2753981351852417,
+      "learning_rate": 9.009915275696693e-05,
+      "loss": 3.678,
+      "step": 1060
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.1910932064056396,
+      "learning_rate": 8.999934288667534e-05,
+      "loss": 3.7276,
+      "step": 1065
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.3303883075714111,
+      "learning_rate": 8.989908831754028e-05,
+      "loss": 3.7377,
+      "step": 1070
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.1198793649673462,
+      "learning_rate": 8.979839016415735e-05,
+      "loss": 3.7212,
+      "step": 1075
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.1606683731079102,
+      "learning_rate": 8.969724954605373e-05,
+      "loss": 3.7697,
+      "step": 1080
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.1115267276763916,
+      "learning_rate": 8.959566758767581e-05,
+      "loss": 3.6644,
+      "step": 1085
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.1623845100402832,
+      "learning_rate": 8.949364541837661e-05,
+      "loss": 3.7393,
+      "step": 1090
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.153632640838623,
+      "learning_rate": 8.939118417240329e-05,
+      "loss": 3.7721,
+      "step": 1095
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.0939881801605225,
+      "learning_rate": 8.92882849888845e-05,
+      "loss": 3.7285,
+      "step": 1100
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.1292450428009033,
+      "learning_rate": 8.918494901181773e-05,
+      "loss": 3.6388,
+      "step": 1105
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.1940792798995972,
+      "learning_rate": 8.908117739005659e-05,
+      "loss": 3.7399,
+      "step": 1110
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.0818376541137695,
+      "learning_rate": 8.897697127729805e-05,
+      "loss": 3.7775,
+      "step": 1115
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.0121492147445679,
+      "learning_rate": 8.887233183206957e-05,
+      "loss": 3.727,
+      "step": 1120
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.1409603357315063,
+      "learning_rate": 8.876726021771627e-05,
+      "loss": 3.6582,
+      "step": 1125
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.3128355741500854,
+      "learning_rate": 8.866175760238798e-05,
+      "loss": 3.7046,
+      "step": 1130
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.2662190198898315,
+      "learning_rate": 8.855582515902625e-05,
+      "loss": 3.7284,
+      "step": 1135
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.3230217695236206,
+      "learning_rate": 8.844946406535131e-05,
+      "loss": 3.6778,
+      "step": 1140
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.2414748668670654,
+      "learning_rate": 8.834267550384893e-05,
+      "loss": 3.7084,
+      "step": 1145
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.0208975076675415,
+      "learning_rate": 8.823546066175741e-05,
+      "loss": 3.6613,
+      "step": 1150
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.1987378597259521,
+      "learning_rate": 8.81278207310542e-05,
+      "loss": 3.7607,
+      "step": 1155
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.0733369588851929,
+      "learning_rate": 8.801975690844278e-05,
+      "loss": 3.6559,
+      "step": 1160
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.1883203983306885,
+      "learning_rate": 8.791127039533934e-05,
+      "loss": 3.7051,
+      "step": 1165
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.13148832321167,
+      "learning_rate": 8.780236239785935e-05,
+      "loss": 3.6763,
+      "step": 1170
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.9961485266685486,
+      "learning_rate": 8.76930341268042e-05,
+      "loss": 3.6938,
+      "step": 1175
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.0561619997024536,
+      "learning_rate": 8.758328679764776e-05,
+      "loss": 3.5663,
+      "step": 1180
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.127403736114502,
+      "learning_rate": 8.747312163052284e-05,
+      "loss": 3.679,
+      "step": 1185
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.0956592559814453,
+      "learning_rate": 8.736253985020761e-05,
+      "loss": 3.6786,
+      "step": 1190
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.08185613155365,
+      "learning_rate": 8.725154268611203e-05,
+      "loss": 3.757,
+      "step": 1195
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.0391801595687866,
+      "learning_rate": 8.714013137226411e-05,
+      "loss": 3.7139,
+      "step": 1200
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.1036536693572998,
+      "learning_rate": 8.702830714729628e-05,
+      "loss": 3.7101,
+      "step": 1205
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.2152525186538696,
+      "learning_rate": 8.691607125443153e-05,
+      "loss": 3.6212,
+      "step": 1210
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.1887563467025757,
+      "learning_rate": 8.680342494146967e-05,
+      "loss": 3.6976,
+      "step": 1215
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.0755887031555176,
+      "learning_rate": 8.66903694607734e-05,
+      "loss": 3.678,
+      "step": 1220
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.9953941106796265,
+      "learning_rate": 8.65769060692544e-05,
+      "loss": 3.6319,
+      "step": 1225
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.09784996509552,
+      "learning_rate": 8.646303602835936e-05,
+      "loss": 3.7209,
+      "step": 1230
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.162729263305664,
+      "learning_rate": 8.634876060405597e-05,
+      "loss": 3.6444,
+      "step": 1235
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.3574235439300537,
+      "learning_rate": 8.623408106681884e-05,
+      "loss": 3.5982,
+      "step": 1240
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.187386393547058,
+      "learning_rate": 8.611899869161535e-05,
+      "loss": 3.6177,
+      "step": 1245
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.1473116874694824,
+      "learning_rate": 8.600351475789147e-05,
+      "loss": 3.6681,
+      "step": 1250
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.0066689252853394,
+      "learning_rate": 8.588763054955764e-05,
+      "loss": 3.6256,
+      "step": 1255
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.1282976865768433,
+      "learning_rate": 8.57713473549743e-05,
+      "loss": 3.5512,
+      "step": 1260
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.0864704847335815,
+      "learning_rate": 8.565466646693778e-05,
+      "loss": 3.6497,
+      "step": 1265
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.0346888303756714,
+      "learning_rate": 8.553758918266578e-05,
+      "loss": 3.6071,
+      "step": 1270
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.0757558345794678,
+      "learning_rate": 8.5420116803783e-05,
+      "loss": 3.61,
+      "step": 1275
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.1552907228469849,
+      "learning_rate": 8.530225063630668e-05,
+      "loss": 3.6771,
+      "step": 1280
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.9322082996368408,
+      "learning_rate": 8.518399199063205e-05,
+      "loss": 3.6004,
+      "step": 1285
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.185263752937317,
+      "learning_rate": 8.50653421815178e-05,
+      "loss": 3.6133,
+      "step": 1290
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.0670915842056274,
+      "learning_rate": 8.494630252807138e-05,
+      "loss": 3.6131,
+      "step": 1295
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.0210850238800049,
+      "learning_rate": 8.482687435373449e-05,
+      "loss": 3.6778,
+      "step": 1300
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.0529567003250122,
+      "learning_rate": 8.470705898626817e-05,
+      "loss": 3.621,
+      "step": 1305
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.0043034553527832,
+      "learning_rate": 8.458685775773822e-05,
+      "loss": 3.6038,
+      "step": 1310
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.008782982826233,
+      "learning_rate": 8.446627200450025e-05,
+      "loss": 3.6061,
+      "step": 1315
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.0716972351074219,
+      "learning_rate": 8.434530306718493e-05,
+      "loss": 3.5565,
+      "step": 1320
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.9667292833328247,
+      "learning_rate": 8.4223952290683e-05,
+      "loss": 3.6784,
+      "step": 1325
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.0869324207305908,
+      "learning_rate": 8.41022210241304e-05,
+      "loss": 3.6062,
+      "step": 1330
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.1732205152511597,
+      "learning_rate": 8.398011062089316e-05,
+      "loss": 3.6649,
+      "step": 1335
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.9114173650741577,
+      "learning_rate": 8.385762243855249e-05,
+      "loss": 3.5658,
+      "step": 1340
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.0308265686035156,
+      "learning_rate": 8.373475783888958e-05,
+      "loss": 3.548,
+      "step": 1345
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.0726767778396606,
+      "learning_rate": 8.36115181878705e-05,
+      "loss": 3.6422,
+      "step": 1350
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.0571999549865723,
+      "learning_rate": 8.348790485563101e-05,
+      "loss": 3.5881,
+      "step": 1355
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.1550967693328857,
+      "learning_rate": 8.336391921646134e-05,
+      "loss": 3.5942,
+      "step": 1360
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.0572580099105835,
+      "learning_rate": 8.323956264879089e-05,
+      "loss": 3.641,
+      "step": 1365
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.9506217241287231,
+      "learning_rate": 8.311483653517294e-05,
+      "loss": 3.6348,
+      "step": 1370
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.0256168842315674,
+      "learning_rate": 8.298974226226919e-05,
+      "loss": 3.5617,
+      "step": 1375
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.9493563771247864,
+      "learning_rate": 8.28642812208345e-05,
+      "loss": 3.6105,
+      "step": 1380
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.9935876727104187,
+      "learning_rate": 8.273845480570123e-05,
+      "loss": 3.631,
+      "step": 1385
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.040040135383606,
+      "learning_rate": 8.26122644157639e-05,
+      "loss": 3.6058,
+      "step": 1390
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.1051440238952637,
+      "learning_rate": 8.248571145396362e-05,
+      "loss": 3.6573,
+      "step": 1395
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.9562785029411316,
+      "learning_rate": 8.235879732727236e-05,
+      "loss": 3.6642,
+      "step": 1400
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.9978268146514893,
+      "learning_rate": 8.223152344667745e-05,
+      "loss": 3.6837,
+      "step": 1405
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.0548397302627563,
+      "learning_rate": 8.21038912271658e-05,
+      "loss": 3.6198,
+      "step": 1410
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.9698254466056824,
+      "learning_rate": 8.197590208770824e-05,
+      "loss": 3.6531,
+      "step": 1415
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.8963156342506409,
+      "learning_rate": 8.184755745124371e-05,
+      "loss": 3.6352,
+      "step": 1420
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.9984579682350159,
+      "learning_rate": 8.171885874466342e-05,
+      "loss": 3.5373,
+      "step": 1425
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.9025471210479736,
+      "learning_rate": 8.158980739879507e-05,
+      "loss": 3.5601,
+      "step": 1430
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.0891411304473877,
+      "learning_rate": 8.146040484838677e-05,
+      "loss": 3.6682,
+      "step": 1435
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.0604552030563354,
+      "learning_rate": 8.133065253209132e-05,
+      "loss": 3.5918,
+      "step": 1440
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.0379425287246704,
+      "learning_rate": 8.120055189245e-05,
+      "loss": 3.5252,
+      "step": 1445
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.9820531010627747,
+      "learning_rate": 8.10701043758767e-05,
+      "loss": 3.5687,
+      "step": 1450
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.9933525919914246,
+      "learning_rate": 8.093931143264174e-05,
+      "loss": 3.6001,
+      "step": 1455
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.0662941932678223,
+      "learning_rate": 8.080817451685576e-05,
+      "loss": 3.5142,
+      "step": 1460
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.0238711833953857,
+      "learning_rate": 8.067669508645356e-05,
+      "loss": 3.6646,
+      "step": 1465
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.9672027826309204,
+      "learning_rate": 8.054487460317797e-05,
+      "loss": 3.6427,
+      "step": 1470
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.914375364780426,
+      "learning_rate": 8.041271453256345e-05,
+      "loss": 3.4607,
+      "step": 1475
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.0790928602218628,
+      "learning_rate": 8.02802163439199e-05,
+      "loss": 3.6118,
+      "step": 1480
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.070357322692871,
+      "learning_rate": 8.01473815103163e-05,
+      "loss": 3.5799,
+      "step": 1485
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.1024534702301025,
+      "learning_rate": 8.001421150856434e-05,
+      "loss": 3.4861,
+      "step": 1490
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.0195724964141846,
+      "learning_rate": 7.988070781920197e-05,
+      "loss": 3.6483,
+      "step": 1495
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.9426557421684265,
+      "learning_rate": 7.9746871926477e-05,
+      "loss": 3.547,
+      "step": 1500
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.0127331018447876,
+      "learning_rate": 7.961270531833052e-05,
+      "loss": 3.6059,
+      "step": 1505
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.9816798567771912,
+      "learning_rate": 7.947820948638045e-05,
+      "loss": 3.5799,
+      "step": 1510
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.1183470487594604,
+      "learning_rate": 7.934338592590486e-05,
+      "loss": 3.6131,
+      "step": 1515
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.9896053671836853,
+      "learning_rate": 7.92082361358254e-05,
+      "loss": 3.5447,
+      "step": 1520
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.1003063917160034,
+      "learning_rate": 7.907276161869065e-05,
+      "loss": 3.5735,
+      "step": 1525
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.9578513503074646,
+      "learning_rate": 7.893696388065936e-05,
+      "loss": 3.5188,
+      "step": 1530
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.0876209735870361,
+      "learning_rate": 7.88008444314838e-05,
+      "loss": 3.505,
+      "step": 1535
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.95606529712677,
+      "learning_rate": 7.866440478449283e-05,
+      "loss": 3.5618,
+      "step": 1540
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.0113919973373413,
+      "learning_rate": 7.852764645657522e-05,
+      "loss": 3.5032,
+      "step": 1545
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.0219696760177612,
+      "learning_rate": 7.839057096816271e-05,
+      "loss": 3.4406,
+      "step": 1550
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.9677157998085022,
+      "learning_rate": 7.82531798432131e-05,
+      "loss": 3.4055,
+      "step": 1555
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.9238067269325256,
+      "learning_rate": 7.811547460919333e-05,
+      "loss": 3.5264,
+      "step": 1560
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.071569800376892,
+      "learning_rate": 7.797745679706254e-05,
+      "loss": 3.5282,
+      "step": 1565
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.0753663778305054,
+      "learning_rate": 7.783912794125496e-05,
+      "loss": 3.4538,
+      "step": 1570
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.0286469459533691,
+      "learning_rate": 7.770048957966291e-05,
+      "loss": 3.4562,
+      "step": 1575
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.0505075454711914,
+      "learning_rate": 7.756154325361967e-05,
+      "loss": 3.6064,
+      "step": 1580
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.976128876209259,
+      "learning_rate": 7.74222905078824e-05,
+      "loss": 3.4972,
+      "step": 1585
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.98134446144104,
+      "learning_rate": 7.728273289061489e-05,
+      "loss": 3.5518,
+      "step": 1590
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.9614392518997192,
+      "learning_rate": 7.714287195337044e-05,
+      "loss": 3.5749,
+      "step": 1595
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.9301189184188843,
+      "learning_rate": 7.700270925107448e-05,
+      "loss": 3.541,
+      "step": 1600
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.1225268840789795,
+      "learning_rate": 7.686224634200742e-05,
+      "loss": 3.5555,
+      "step": 1605
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.9103389978408813,
+      "learning_rate": 7.672148478778722e-05,
+      "loss": 3.5353,
+      "step": 1610
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.9482797384262085,
+      "learning_rate": 7.658042615335212e-05,
+      "loss": 3.4771,
+      "step": 1615
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.9419737458229065,
+      "learning_rate": 7.643907200694318e-05,
+      "loss": 3.5093,
+      "step": 1620
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.9081383347511292,
+      "learning_rate": 7.629742392008684e-05,
+      "loss": 3.5949,
+      "step": 1625
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.9159099459648132,
+      "learning_rate": 7.615548346757749e-05,
+      "loss": 3.4818,
+      "step": 1630
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.8918059468269348,
+      "learning_rate": 7.60132522274599e-05,
+      "loss": 3.6026,
+      "step": 1635
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.0188435316085815,
+      "learning_rate": 7.587073178101178e-05,
+      "loss": 3.5687,
+      "step": 1640
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.8577747941017151,
+      "learning_rate": 7.572792371272609e-05,
+      "loss": 3.5082,
+      "step": 1645
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.1805169582366943,
+      "learning_rate": 7.55848296102935e-05,
+      "loss": 3.5336,
+      "step": 1650
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.9859539270401001,
+      "learning_rate": 7.544145106458465e-05,
+      "loss": 3.5325,
+      "step": 1655
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.9058620929718018,
+      "learning_rate": 7.529778966963259e-05,
+      "loss": 3.5395,
+      "step": 1660
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.9886941909790039,
+      "learning_rate": 7.515384702261496e-05,
+      "loss": 3.5447,
+      "step": 1665
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.8978294134140015,
+      "learning_rate": 7.500962472383627e-05,
+      "loss": 3.4647,
+      "step": 1670
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.9597351551055908,
+      "learning_rate": 7.486512437671011e-05,
+      "loss": 3.4347,
+      "step": 1675
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.912813127040863,
+      "learning_rate": 7.472034758774128e-05,
+      "loss": 3.3937,
+      "step": 1680
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.9774367213249207,
+      "learning_rate": 7.457529596650797e-05,
+      "loss": 3.4253,
+      "step": 1685
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.9628680348396301,
+      "learning_rate": 7.442997112564392e-05,
+      "loss": 3.4534,
+      "step": 1690
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.0099488496780396,
+      "learning_rate": 7.428437468082037e-05,
+      "loss": 3.4865,
+      "step": 1695
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.9738776683807373,
+      "learning_rate": 7.413850825072817e-05,
+      "loss": 3.3938,
+      "step": 1700
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.8624957799911499,
+      "learning_rate": 7.39923734570598e-05,
+      "loss": 3.4803,
+      "step": 1705
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.8737362623214722,
+      "learning_rate": 7.384597192449126e-05,
+      "loss": 3.4579,
+      "step": 1710
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.9167929291725159,
+      "learning_rate": 7.369930528066412e-05,
+      "loss": 3.5313,
+      "step": 1715
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.9283750653266907,
+      "learning_rate": 7.355237515616732e-05,
+      "loss": 3.3903,
+      "step": 1720
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.0188536643981934,
+      "learning_rate": 7.340518318451914e-05,
+      "loss": 3.4892,
+      "step": 1725
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.9486240744590759,
+      "learning_rate": 7.325773100214893e-05,
+      "loss": 3.5478,
+      "step": 1730
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.8894400596618652,
+      "learning_rate": 7.311002024837899e-05,
+      "loss": 3.4897,
+      "step": 1735
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.0249364376068115,
+      "learning_rate": 7.296205256540633e-05,
+      "loss": 3.395,
+      "step": 1740
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.9660959839820862,
+      "learning_rate": 7.281382959828443e-05,
+      "loss": 3.4896,
+      "step": 1745
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.8965651392936707,
+      "learning_rate": 7.26653529949049e-05,
+      "loss": 3.4113,
+      "step": 1750
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.916313886642456,
+      "learning_rate": 7.25166244059792e-05,
+      "loss": 3.4342,
+      "step": 1755
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.918171226978302,
+      "learning_rate": 7.236764548502029e-05,
+      "loss": 3.4073,
+      "step": 1760
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.9626057147979736,
+      "learning_rate": 7.221841788832421e-05,
+      "loss": 3.5081,
+      "step": 1765
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.8918098211288452,
+      "learning_rate": 7.206894327495173e-05,
+      "loss": 3.4852,
+      "step": 1770
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.9311957359313965,
+      "learning_rate": 7.191922330670982e-05,
+      "loss": 3.4472,
+      "step": 1775
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.917389988899231,
+      "learning_rate": 7.176925964813326e-05,
+      "loss": 3.5054,
+      "step": 1780
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.9579627513885498,
+      "learning_rate": 7.161905396646607e-05,
+      "loss": 3.5408,
+      "step": 1785
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.9102108478546143,
+      "learning_rate": 7.146860793164299e-05,
+      "loss": 3.453,
+      "step": 1790
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.8844248652458191,
+      "learning_rate": 7.131792321627098e-05,
+      "loss": 3.4266,
+      "step": 1795
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.8853104114532471,
+      "learning_rate": 7.116700149561048e-05,
+      "loss": 3.4452,
+      "step": 1800
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.905546247959137,
+      "learning_rate": 7.101584444755696e-05,
+      "loss": 3.5243,
+      "step": 1805
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.021131157875061,
+      "learning_rate": 7.086445375262212e-05,
+      "loss": 3.5432,
+      "step": 1810
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.9376331567764282,
+      "learning_rate": 7.071283109391528e-05,
+      "loss": 3.4106,
+      "step": 1815
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.8878265023231506,
+      "learning_rate": 7.056097815712466e-05,
+      "loss": 3.4425,
+      "step": 1820
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.9961305260658264,
+      "learning_rate": 7.040889663049862e-05,
+      "loss": 3.4492,
+      "step": 1825
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.8520357608795166,
+      "learning_rate": 7.025658820482693e-05,
+      "loss": 3.4443,
+      "step": 1830
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.9509548544883728,
+      "learning_rate": 7.010405457342192e-05,
+      "loss": 3.4016,
+      "step": 1835
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.9308119416236877,
+      "learning_rate": 6.995129743209967e-05,
+      "loss": 3.4314,
+      "step": 1840
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.8559417724609375,
+      "learning_rate": 6.97983184791612e-05,
+      "loss": 3.4727,
+      "step": 1845
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.8854596614837646,
+      "learning_rate": 6.964511941537355e-05,
+      "loss": 3.3541,
+      "step": 1850
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.8448160886764526,
+      "learning_rate": 6.949170194395083e-05,
+      "loss": 3.4456,
+      "step": 1855
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.885260283946991,
+      "learning_rate": 6.933806777053536e-05,
+      "loss": 3.4313,
+      "step": 1860
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.8721843957901001,
+      "learning_rate": 6.918421860317872e-05,
+      "loss": 3.4689,
+      "step": 1865
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.830544114112854,
+      "learning_rate": 6.903015615232263e-05,
+      "loss": 3.4405,
+      "step": 1870
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.8577764630317688,
+      "learning_rate": 6.887588213078012e-05,
+      "loss": 3.4241,
+      "step": 1875
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.9850843548774719,
+      "learning_rate": 6.87213982537163e-05,
+      "loss": 3.3351,
+      "step": 1880
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.9125412106513977,
+      "learning_rate": 6.856670623862943e-05,
+      "loss": 3.4431,
+      "step": 1885
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.8923556208610535,
+      "learning_rate": 6.841180780533179e-05,
+      "loss": 3.4541,
+      "step": 1890
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.9279801249504089,
+      "learning_rate": 6.82567046759305e-05,
+      "loss": 3.4647,
+      "step": 1895
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.9199705123901367,
+      "learning_rate": 6.810139857480844e-05,
+      "loss": 3.4653,
+      "step": 1900
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.0211433172225952,
+      "learning_rate": 6.794589122860509e-05,
+      "loss": 3.4191,
+      "step": 1905
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.0141521692276,
+      "learning_rate": 6.779018436619725e-05,
+      "loss": 3.4482,
+      "step": 1910
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.0456479787826538,
+      "learning_rate": 6.763427971867992e-05,
+      "loss": 3.4434,
+      "step": 1915
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.862886905670166,
+      "learning_rate": 6.747817901934699e-05,
+      "loss": 3.3638,
+      "step": 1920
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.9483063817024231,
+      "learning_rate": 6.732188400367197e-05,
+      "loss": 3.4011,
+      "step": 1925
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.8820633888244629,
+      "learning_rate": 6.716539640928871e-05,
+      "loss": 3.4679,
+      "step": 1930
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.8584756851196289,
+      "learning_rate": 6.70087179759721e-05,
+      "loss": 3.3725,
+      "step": 1935
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.8979827761650085,
+      "learning_rate": 6.685185044561874e-05,
+      "loss": 3.4385,
+      "step": 1940
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.8375410437583923,
+      "learning_rate": 6.669479556222747e-05,
+      "loss": 3.4367,
+      "step": 1945
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.8942875266075134,
+      "learning_rate": 6.653755507188013e-05,
+      "loss": 3.4236,
+      "step": 1950
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.966738224029541,
+      "learning_rate": 6.638013072272205e-05,
+      "loss": 3.328,
+      "step": 1955
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.9720861315727234,
+      "learning_rate": 6.622252426494259e-05,
+      "loss": 3.4291,
+      "step": 1960
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.92670738697052,
+      "learning_rate": 6.606473745075581e-05,
+      "loss": 3.4255,
+      "step": 1965
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.8368898630142212,
+      "learning_rate": 6.590677203438084e-05,
+      "loss": 3.4261,
+      "step": 1970
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.9312162399291992,
+      "learning_rate": 6.574862977202252e-05,
+      "loss": 3.3538,
+      "step": 1975
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.9195160269737244,
+      "learning_rate": 6.559031242185174e-05,
+      "loss": 3.4648,
+      "step": 1980
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.8352183699607849,
+      "learning_rate": 6.543182174398597e-05,
+      "loss": 3.4051,
+      "step": 1985
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.8951199650764465,
+      "learning_rate": 6.52731595004697e-05,
+      "loss": 3.4039,
+      "step": 1990
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.9216262698173523,
+      "learning_rate": 6.51143274552548e-05,
+      "loss": 3.3785,
+      "step": 1995
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.011792778968811,
+      "learning_rate": 6.495532737418098e-05,
+      "loss": 3.4595,
+      "step": 2000
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.8370119333267212,
+      "learning_rate": 6.479616102495605e-05,
+      "loss": 3.4424,
+      "step": 2005
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.8701230883598328,
+      "learning_rate": 6.463683017713638e-05,
+      "loss": 3.4074,
+      "step": 2010
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.8995570540428162,
+      "learning_rate": 6.447733660210715e-05,
+      "loss": 3.3802,
+      "step": 2015
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.8354151248931885,
+      "learning_rate": 6.431768207306272e-05,
+      "loss": 3.3636,
+      "step": 2020
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.857018232345581,
+      "learning_rate": 6.415786836498684e-05,
+      "loss": 3.3024,
+      "step": 2025
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.9143330454826355,
+      "learning_rate": 6.399789725463298e-05,
+      "loss": 3.4235,
+      "step": 2030
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.8741365671157837,
+      "learning_rate": 6.383777052050458e-05,
+      "loss": 3.4025,
+      "step": 2035
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.823043942451477,
+      "learning_rate": 6.367748994283518e-05,
+      "loss": 3.3333,
+      "step": 2040
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8715491890907288,
+      "learning_rate": 6.351705730356877e-05,
+      "loss": 3.3439,
+      "step": 2045
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8645880222320557,
+      "learning_rate": 6.335647438633987e-05,
+      "loss": 3.3991,
+      "step": 2050
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.91361403465271,
+      "learning_rate": 6.319574297645374e-05,
+      "loss": 3.3544,
+      "step": 2055
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.904449462890625,
+      "learning_rate": 6.303486486086654e-05,
+      "loss": 3.4209,
+      "step": 2060
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.9177110195159912,
+      "learning_rate": 6.287384182816546e-05,
+      "loss": 3.3704,
+      "step": 2065
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8553934097290039,
+      "learning_rate": 6.271267566854883e-05,
+      "loss": 3.423,
+      "step": 2070
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8918532133102417,
+      "learning_rate": 6.255136817380618e-05,
+      "loss": 3.3414,
+      "step": 2075
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8273350596427917,
+      "learning_rate": 6.23899211372984e-05,
+      "loss": 3.3959,
+      "step": 2080
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8525533080101013,
+      "learning_rate": 6.222833635393772e-05,
+      "loss": 3.3691,
+      "step": 2085
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8700625896453857,
+      "learning_rate": 6.206661562016782e-05,
+      "loss": 3.4058,
+      "step": 2090
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.8108021020889282,
+      "learning_rate": 6.190476073394382e-05,
+      "loss": 3.387,
+      "step": 2095
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.8776712417602539,
+      "learning_rate": 6.17427734947123e-05,
+      "loss": 3.3676,
+      "step": 2100
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.9267117381095886,
+      "learning_rate": 6.158065570339127e-05,
+      "loss": 3.3566,
+      "step": 2105
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.0094754695892334,
+      "learning_rate": 6.141840916235021e-05,
+      "loss": 3.367,
+      "step": 2110
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.1877857446670532,
+      "learning_rate": 6.125603567539001e-05,
+      "loss": 3.3563,
+      "step": 2115
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.8413233757019043,
+      "learning_rate": 6.109353704772284e-05,
+      "loss": 3.2802,
+      "step": 2120
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.9599123001098633,
+      "learning_rate": 6.0930915085952164e-05,
+      "loss": 3.452,
+      "step": 2125
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.8594855666160583,
+      "learning_rate": 6.076817159805267e-05,
+      "loss": 3.2819,
+      "step": 2130
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.8785787224769592,
+      "learning_rate": 6.06053083933501e-05,
+      "loss": 3.3245,
+      "step": 2135
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.8580414652824402,
+      "learning_rate": 6.044232728250116e-05,
+      "loss": 3.2992,
+      "step": 2140
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.8790245652198792,
+      "learning_rate": 6.027923007747339e-05,
+      "loss": 3.3385,
+      "step": 2145
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.8154867887496948,
+      "learning_rate": 6.011601859152506e-05,
+      "loss": 3.3793,
+      "step": 2150
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.8092056512832642,
+      "learning_rate": 5.995269463918495e-05,
+      "loss": 3.3127,
+      "step": 2155
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.855643630027771,
+      "learning_rate": 5.97892600362322e-05,
+      "loss": 3.3378,
+      "step": 2160
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.847649872303009,
+      "learning_rate": 5.962571659967614e-05,
+      "loss": 3.3644,
+      "step": 2165
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.9203967452049255,
+      "learning_rate": 5.946206614773606e-05,
+      "loss": 3.3344,
+      "step": 2170
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.9248597025871277,
+      "learning_rate": 5.929831049982103e-05,
+      "loss": 3.3794,
+      "step": 2175
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.8023829460144043,
+      "learning_rate": 5.9134451476509633e-05,
+      "loss": 3.4343,
+      "step": 2180
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.8594465851783752,
+      "learning_rate": 5.897049089952974e-05,
+      "loss": 3.4572,
+      "step": 2185
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.8430312871932983,
+      "learning_rate": 5.880643059173826e-05,
+      "loss": 3.3531,
+      "step": 2190
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.9222793579101562,
+      "learning_rate": 5.864227237710093e-05,
+      "loss": 3.3767,
+      "step": 2195
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.89219069480896,
+      "learning_rate": 5.847801808067189e-05,
+      "loss": 3.3839,
+      "step": 2200
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.9887323975563049,
+      "learning_rate": 5.831366952857357e-05,
+      "loss": 3.3403,
+      "step": 2205
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7898027300834656,
+      "learning_rate": 5.814922854797622e-05,
+      "loss": 3.3232,
+      "step": 2210
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.9959586262702942,
+      "learning_rate": 5.798469696707775e-05,
+      "loss": 3.3977,
+      "step": 2215
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.9313060641288757,
+      "learning_rate": 5.782007661508331e-05,
+      "loss": 3.3479,
+      "step": 2220
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.8173262476921082,
+      "learning_rate": 5.765536932218495e-05,
+      "loss": 3.3965,
+      "step": 2225
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.8597461581230164,
+      "learning_rate": 5.7490576919541315e-05,
+      "loss": 3.3328,
+      "step": 2230
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.8330791592597961,
+      "learning_rate": 5.732570123925729e-05,
+      "loss": 3.404,
+      "step": 2235
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.8566576242446899,
+      "learning_rate": 5.7160744114363593e-05,
+      "loss": 3.4414,
+      "step": 2240
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.141120433807373,
+      "learning_rate": 5.699570737879641e-05,
+      "loss": 3.2698,
+      "step": 2245
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.8277701735496521,
+      "learning_rate": 5.683059286737702e-05,
+      "loss": 3.3993,
+      "step": 2250
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.8419466614723206,
+      "learning_rate": 5.666540241579139e-05,
+      "loss": 3.334,
+      "step": 2255
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.8160429000854492,
+      "learning_rate": 5.6500137860569766e-05,
+      "loss": 3.3605,
+      "step": 2260
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.9068493843078613,
+      "learning_rate": 5.633480103906624e-05,
+      "loss": 3.3465,
+      "step": 2265
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.8464723229408264,
+      "learning_rate": 5.616939378943834e-05,
+      "loss": 3.3015,
+      "step": 2270
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.8291537761688232,
+      "learning_rate": 5.6003917950626595e-05,
+      "loss": 3.3815,
+      "step": 2275
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.8295892477035522,
+      "learning_rate": 5.583837536233407e-05,
+      "loss": 3.3028,
+      "step": 2280
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.8144384026527405,
+      "learning_rate": 5.567276786500596e-05,
+      "loss": 3.2974,
+      "step": 2285
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.8660650253295898,
+      "learning_rate": 5.5507097299809054e-05,
+      "loss": 3.3467,
+      "step": 2290
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7973735332489014,
+      "learning_rate": 5.534136550861133e-05,
+      "loss": 3.3473,
+      "step": 2295
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7773838639259338,
+      "learning_rate": 5.5175574333961465e-05,
+      "loss": 3.3118,
+      "step": 2300
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.8147760629653931,
+      "learning_rate": 5.500972561906832e-05,
+      "loss": 3.3131,
+      "step": 2305
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.8618188500404358,
+      "learning_rate": 5.484382120778048e-05,
+      "loss": 3.3739,
+      "step": 2310
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.8593544960021973,
+      "learning_rate": 5.467786294456575e-05,
+      "loss": 3.3894,
+      "step": 2315
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.8763341903686523,
+      "learning_rate": 5.451185267449061e-05,
+      "loss": 3.4009,
+      "step": 2320
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.09891939163208,
+      "learning_rate": 5.43457922431998e-05,
+      "loss": 3.3172,
+      "step": 2325
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7956613898277283,
+      "learning_rate": 5.417968349689566e-05,
+      "loss": 3.1964,
+      "step": 2330
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8214170932769775,
+      "learning_rate": 5.401352828231772e-05,
+      "loss": 3.3333,
+      "step": 2335
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8590761423110962,
+      "learning_rate": 5.384732844672211e-05,
+      "loss": 3.3274,
+      "step": 2340
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8010850548744202,
+      "learning_rate": 5.368108583786107e-05,
+      "loss": 3.3073,
+      "step": 2345
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8691250681877136,
+      "learning_rate": 5.3514802303962344e-05,
+      "loss": 3.3787,
+      "step": 2350
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.9491872191429138,
+      "learning_rate": 5.334847969370868e-05,
+      "loss": 3.3521,
+      "step": 2355
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8348716497421265,
+      "learning_rate": 5.3182119856217284e-05,
+      "loss": 3.3439,
+      "step": 2360
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.874321699142456,
+      "learning_rate": 5.3015724641019214e-05,
+      "loss": 3.2895,
+      "step": 2365
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8504311442375183,
+      "learning_rate": 5.284929589803884e-05,
+      "loss": 3.2222,
+      "step": 2370
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.860460102558136,
+      "learning_rate": 5.2682835477573336e-05,
+      "loss": 3.2721,
+      "step": 2375
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8044112920761108,
+      "learning_rate": 5.2516345230271965e-05,
+      "loss": 3.3099,
+      "step": 2380
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.8295807242393494,
+      "learning_rate": 5.234982700711569e-05,
+      "loss": 3.3845,
+      "step": 2385
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.8794777393341064,
+      "learning_rate": 5.218328265939643e-05,
+      "loss": 3.3037,
+      "step": 2390
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.8412184715270996,
+      "learning_rate": 5.201671403869657e-05,
+      "loss": 3.2339,
+      "step": 2395
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.8259831070899963,
+      "learning_rate": 5.1850122996868366e-05,
+      "loss": 3.3287,
+      "step": 2400
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.8123874068260193,
+      "learning_rate": 5.168351138601334e-05,
+      "loss": 3.2838,
+      "step": 2405
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.8798362016677856,
+      "learning_rate": 5.1516881058461675e-05,
+      "loss": 3.3506,
+      "step": 2410
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.8345399498939514,
+      "learning_rate": 5.135023386675166e-05,
+      "loss": 3.2879,
+      "step": 2415
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.8196049332618713,
+      "learning_rate": 5.118357166360906e-05,
+      "loss": 3.3534,
+      "step": 2420
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.8316165804862976,
+      "learning_rate": 5.101689630192655e-05,
+      "loss": 3.3233,
+      "step": 2425
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.8012190461158752,
+      "learning_rate": 5.085020963474307e-05,
+      "loss": 3.3246,
+      "step": 2430
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.8753196597099304,
+      "learning_rate": 5.068351351522329e-05,
+      "loss": 3.3063,
+      "step": 2435
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.8270070552825928,
+      "learning_rate": 5.0516809796636935e-05,
+      "loss": 3.2703,
+      "step": 2440
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.9107401371002197,
+      "learning_rate": 5.035010033233821e-05,
+      "loss": 3.3625,
+      "step": 2445
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.9239689707756042,
+      "learning_rate": 5.018338697574523e-05,
+      "loss": 3.2771,
+      "step": 2450
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7925752997398376,
+      "learning_rate": 5.0016671580319354e-05,
+      "loss": 3.2949,
+      "step": 2455
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7893439531326294,
+      "learning_rate": 4.984995599954461e-05,
+      "loss": 3.315,
+      "step": 2460
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.8512730598449707,
+      "learning_rate": 4.968324208690712e-05,
+      "loss": 3.3766,
+      "step": 2465
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.793056309223175,
+      "learning_rate": 4.951653169587441e-05,
+      "loss": 3.2775,
+      "step": 2470
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.8076874613761902,
+      "learning_rate": 4.93498266798749e-05,
+      "loss": 3.2826,
+      "step": 2475
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7923920154571533,
+      "learning_rate": 4.918312889227722e-05,
+      "loss": 3.3037,
+      "step": 2480
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.8230359554290771,
+      "learning_rate": 4.901644018636966e-05,
+      "loss": 3.2542,
+      "step": 2485
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.801092267036438,
+      "learning_rate": 4.8849762415339526e-05,
+      "loss": 3.3019,
+      "step": 2490
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.8467086553573608,
+      "learning_rate": 4.868309743225256e-05,
+      "loss": 3.3187,
+      "step": 2495
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.8135638236999512,
+      "learning_rate": 4.851644709003233e-05,
+      "loss": 3.2581,
+      "step": 2500
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.8063775300979614,
+      "learning_rate": 4.834981324143964e-05,
+      "loss": 3.3253,
+      "step": 2505
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.8042765855789185,
+      "learning_rate": 4.818319773905191e-05,
+      "loss": 3.3437,
+      "step": 2510
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7903698682785034,
+      "learning_rate": 4.801660243524261e-05,
+      "loss": 3.2981,
+      "step": 2515
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.9083207249641418,
+      "learning_rate": 4.7850029182160626e-05,
+      "loss": 3.3335,
+      "step": 2520
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.8911808133125305,
+      "learning_rate": 4.768347983170973e-05,
+      "loss": 3.281,
+      "step": 2525
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.8923286199569702,
+      "learning_rate": 4.7516956235527884e-05,
+      "loss": 3.3236,
+      "step": 2530
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.8573455214500427,
+      "learning_rate": 4.735046024496682e-05,
+      "loss": 3.3264,
+      "step": 2535
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7962297201156616,
+      "learning_rate": 4.7183993711071286e-05,
+      "loss": 3.2597,
+      "step": 2540
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.8235583305358887,
+      "learning_rate": 4.7017558484558554e-05,
+      "loss": 3.3059,
+      "step": 2545
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7752333879470825,
+      "learning_rate": 4.6851156415797844e-05,
+      "loss": 3.2725,
+      "step": 2550
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.8234376907348633,
+      "learning_rate": 4.6684789354789746e-05,
+      "loss": 3.2395,
+      "step": 2555
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.8423200249671936,
+      "learning_rate": 4.651845915114563e-05,
+      "loss": 3.2337,
+      "step": 2560
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.8219478726387024,
+      "learning_rate": 4.6352167654067095e-05,
+      "loss": 3.2808,
+      "step": 2565
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.8123370409011841,
+      "learning_rate": 4.618591671232544e-05,
+      "loss": 3.3276,
+      "step": 2570
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8480009436607361,
+      "learning_rate": 4.601970817424106e-05,
+      "loss": 3.3076,
+      "step": 2575
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8245264291763306,
+      "learning_rate": 4.585354388766292e-05,
+      "loss": 3.2825,
+      "step": 2580
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8287261128425598,
+      "learning_rate": 4.568742569994802e-05,
+      "loss": 3.2332,
+      "step": 2585
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.9042434096336365,
+      "learning_rate": 4.552135545794086e-05,
+      "loss": 3.3015,
+      "step": 2590
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8049812316894531,
+      "learning_rate": 4.535533500795288e-05,
+      "loss": 3.2932,
+      "step": 2595
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8048073649406433,
+      "learning_rate": 4.5189366195741953e-05,
+      "loss": 3.2472,
+      "step": 2600
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8750132322311401,
+      "learning_rate": 4.502345086649186e-05,
+      "loss": 3.3242,
+      "step": 2605
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.7771335244178772,
+      "learning_rate": 4.485759086479179e-05,
+      "loss": 3.2696,
+      "step": 2610
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8382053375244141,
+      "learning_rate": 4.469178803461579e-05,
+      "loss": 3.2338,
+      "step": 2615
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8180707097053528,
+      "learning_rate": 4.4526044219302326e-05,
+      "loss": 3.2866,
+      "step": 2620
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.847093403339386,
+      "learning_rate": 4.4360361261533745e-05,
+      "loss": 3.2347,
+      "step": 2625
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.7841566801071167,
+      "learning_rate": 4.419474100331579e-05,
+      "loss": 3.2444,
+      "step": 2630
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.7948038578033447,
+      "learning_rate": 4.402918528595715e-05,
+      "loss": 3.2591,
+      "step": 2635
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.8009411096572876,
+      "learning_rate": 4.386369595004896e-05,
+      "loss": 3.2533,
+      "step": 2640
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.8153753876686096,
+      "learning_rate": 4.3698274835444354e-05,
+      "loss": 3.2453,
+      "step": 2645
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.8134960532188416,
+      "learning_rate": 4.3532923781238e-05,
+      "loss": 3.2249,
+      "step": 2650
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.8260480165481567,
+      "learning_rate": 4.336764462574566e-05,
+      "loss": 3.2123,
+      "step": 2655
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.8051034808158875,
+      "learning_rate": 4.320243920648376e-05,
+      "loss": 3.2513,
+      "step": 2660
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.8382444381713867,
+      "learning_rate": 4.303730936014894e-05,
+      "loss": 3.2913,
+      "step": 2665
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.8280849456787109,
+      "learning_rate": 4.287225692259765e-05,
+      "loss": 3.2828,
+      "step": 2670
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.8976969718933105,
+      "learning_rate": 4.270728372882575e-05,
+      "loss": 3.2859,
+      "step": 2675
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.8455418944358826,
+      "learning_rate": 4.254239161294804e-05,
+      "loss": 3.1686,
+      "step": 2680
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.9356967806816101,
+      "learning_rate": 4.237758240817802e-05,
+      "loss": 3.2828,
+      "step": 2685
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.8209810256958008,
+      "learning_rate": 4.2212857946807336e-05,
+      "loss": 3.1977,
+      "step": 2690
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.8756744265556335,
+      "learning_rate": 4.2048220060185516e-05,
+      "loss": 3.2809,
+      "step": 2695
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.8613160252571106,
+      "learning_rate": 4.188367057869957e-05,
+      "loss": 3.2154,
+      "step": 2700
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.8148147463798523,
+      "learning_rate": 4.171921133175365e-05,
+      "loss": 3.2539,
+      "step": 2705
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.899475634098053,
+      "learning_rate": 4.155484414774872e-05,
+      "loss": 3.208,
+      "step": 2710
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7842448949813843,
+      "learning_rate": 4.139057085406221e-05,
+      "loss": 3.2039,
+      "step": 2715
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.8090781569480896,
+      "learning_rate": 4.1226393277027726e-05,
+      "loss": 3.3005,
+      "step": 2720
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.8377348184585571,
+      "learning_rate": 4.106231324191471e-05,
+      "loss": 3.3182,
+      "step": 2725
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.8332144618034363,
+      "learning_rate": 4.089833257290817e-05,
+      "loss": 3.258,
+      "step": 2730
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.8233649134635925,
+      "learning_rate": 4.073445309308842e-05,
+      "loss": 3.3097,
+      "step": 2735
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.8969298005104065,
+      "learning_rate": 4.0570676624410756e-05,
+      "loss": 3.2348,
+      "step": 2740
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.823154091835022,
+      "learning_rate": 4.040700498768525e-05,
+      "loss": 3.2481,
+      "step": 2745
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.8446685075759888,
+      "learning_rate": 4.024344000255648e-05,
+      "loss": 3.2539,
+      "step": 2750
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.77149498462677,
+      "learning_rate": 4.0079983487483313e-05,
+      "loss": 3.2859,
+      "step": 2755
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.8215575814247131,
+      "learning_rate": 3.9916637259718683e-05,
+      "loss": 3.3178,
+      "step": 2760
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7828111052513123,
+      "learning_rate": 3.9753403135289396e-05,
+      "loss": 3.3045,
+      "step": 2765
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.8224146962165833,
+      "learning_rate": 3.9590282928975914e-05,
+      "loss": 3.2101,
+      "step": 2770
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7711217999458313,
+      "learning_rate": 3.942727845429221e-05,
+      "loss": 3.2301,
+      "step": 2775
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7876419425010681,
+      "learning_rate": 3.926439152346558e-05,
+      "loss": 3.1997,
+      "step": 2780
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.8607501983642578,
+      "learning_rate": 3.910162394741653e-05,
+      "loss": 3.289,
+      "step": 2785
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.8112248182296753,
+      "learning_rate": 3.893897753573861e-05,
+      "loss": 3.284,
+      "step": 2790
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.8297057747840881,
+      "learning_rate": 3.877645409667829e-05,
+      "loss": 3.1787,
+      "step": 2795
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.8352723121643066,
+      "learning_rate": 3.861405543711491e-05,
+      "loss": 3.2283,
+      "step": 2800
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.8486620783805847,
+      "learning_rate": 3.8451783362540507e-05,
+      "loss": 3.3196,
+      "step": 2805
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.8207599520683289,
+      "learning_rate": 3.828963967703983e-05,
+      "loss": 3.2317,
+      "step": 2810
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.8350016474723816,
+      "learning_rate": 3.8127626183270223e-05,
+      "loss": 3.2321,
+      "step": 2815
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.8970246911048889,
+      "learning_rate": 3.796574468244161e-05,
+      "loss": 3.1897,
+      "step": 2820
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.8711051940917969,
+      "learning_rate": 3.7803996974296444e-05,
+      "loss": 3.2698,
+      "step": 2825
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.8610997200012207,
+      "learning_rate": 3.7642384857089776e-05,
+      "loss": 3.2274,
+      "step": 2830
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7913540601730347,
+      "learning_rate": 3.748091012756915e-05,
+      "loss": 3.2757,
+      "step": 2835
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7936437726020813,
+      "learning_rate": 3.731957458095467e-05,
+      "loss": 3.2132,
+      "step": 2840
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.8374015092849731,
+      "learning_rate": 3.71583800109191e-05,
+      "loss": 3.2772,
+      "step": 2845
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.8306594491004944,
+      "learning_rate": 3.699732820956784e-05,
+      "loss": 3.2765,
+      "step": 2850
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.9261174201965332,
+      "learning_rate": 3.6836420967419057e-05,
+      "loss": 3.1874,
+      "step": 2855
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7884053587913513,
+      "learning_rate": 3.6675660073383745e-05,
+      "loss": 3.2205,
+      "step": 2860
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.8375781774520874,
+      "learning_rate": 3.6515047314745856e-05,
+      "loss": 3.2302,
+      "step": 2865
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.8226302862167358,
+      "learning_rate": 3.6354584477142437e-05,
+      "loss": 3.3306,
+      "step": 2870
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.8241636157035828,
+      "learning_rate": 3.6194273344543736e-05,
+      "loss": 3.2692,
+      "step": 2875
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.8509169220924377,
+      "learning_rate": 3.6034115699233425e-05,
+      "loss": 3.2319,
+      "step": 2880
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.8542621731758118,
+      "learning_rate": 3.5874113321788736e-05,
+      "loss": 3.1808,
+      "step": 2885
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7962069511413574,
+      "learning_rate": 3.571426799106071e-05,
+      "loss": 3.2203,
+      "step": 2890
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.8124511241912842,
+      "learning_rate": 3.555458148415437e-05,
+      "loss": 3.2235,
+      "step": 2895
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7960987687110901,
+      "learning_rate": 3.539505557640901e-05,
+      "loss": 3.1952,
+      "step": 2900
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7768739461898804,
+      "learning_rate": 3.523569204137843e-05,
+      "loss": 3.2864,
+      "step": 2905
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.8407379388809204,
+      "learning_rate": 3.5076492650811246e-05,
+      "loss": 3.2544,
+      "step": 2910
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8421855568885803,
+      "learning_rate": 3.491745917463113e-05,
+      "loss": 3.1736,
+      "step": 2915
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8619421720504761,
+      "learning_rate": 3.475859338091721e-05,
+      "loss": 3.2242,
+      "step": 2920
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8203493356704712,
+      "learning_rate": 3.4599897035884374e-05,
+      "loss": 3.2295,
+      "step": 2925
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7925184965133667,
+      "learning_rate": 3.444137190386363e-05,
+      "loss": 3.1904,
+      "step": 2930
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.9476016759872437,
+      "learning_rate": 3.4283019747282514e-05,
+      "loss": 3.2148,
+      "step": 2935
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8012914657592773,
+      "learning_rate": 3.412484232664545e-05,
+      "loss": 3.2104,
+      "step": 2940
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8429858684539795,
+      "learning_rate": 3.396684140051424e-05,
+      "loss": 3.1776,
+      "step": 2945
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8614286184310913,
+      "learning_rate": 3.3809018725488466e-05,
+      "loss": 3.2487,
+      "step": 2950
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8177609443664551,
+      "learning_rate": 3.365137605618598e-05,
+      "loss": 3.3414,
+      "step": 2955
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7997239232063293,
+      "learning_rate": 3.3493915145223395e-05,
+      "loss": 3.1999,
+      "step": 2960
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7842835187911987,
+      "learning_rate": 3.3336637743196584e-05,
+      "loss": 3.1948,
+      "step": 2965
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.8240490555763245,
+      "learning_rate": 3.317954559866126e-05,
+      "loss": 3.2576,
+      "step": 2970
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.8564560413360596,
+      "learning_rate": 3.302264045811344e-05,
+      "loss": 3.1953,
+      "step": 2975
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.8260619044303894,
+      "learning_rate": 3.286592406597021e-05,
+      "loss": 3.2912,
+      "step": 2980
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.8823931813240051,
+      "learning_rate": 3.270939816455012e-05,
+      "loss": 3.1892,
+      "step": 2985
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.829565703868866,
+      "learning_rate": 3.255306449405395e-05,
+      "loss": 3.2287,
+      "step": 2990
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.8487275838851929,
+      "learning_rate": 3.2396924792545304e-05,
+      "loss": 3.2321,
+      "step": 2995
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.8765036463737488,
+      "learning_rate": 3.224098079593132e-05,
+      "loss": 3.2274,
+      "step": 3000
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.8339765071868896,
+      "learning_rate": 3.2085234237943354e-05,
+      "loss": 3.1882,
+      "step": 3005
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.8532525897026062,
+      "learning_rate": 3.19296868501177e-05,
+      "loss": 3.3398,
+      "step": 3010
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7803841233253479,
+      "learning_rate": 3.177434036177636e-05,
+      "loss": 3.204,
+      "step": 3015
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.849578320980072,
+      "learning_rate": 3.1619196500007804e-05,
+      "loss": 3.2008,
+      "step": 3020
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.9354525208473206,
+      "learning_rate": 3.146425698964776e-05,
+      "loss": 3.1633,
+      "step": 3025
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.750751256942749,
+      "learning_rate": 3.1309523553260046e-05,
+      "loss": 3.2002,
+      "step": 3030
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.8016756772994995,
+      "learning_rate": 3.115499791111743e-05,
+      "loss": 3.2153,
+      "step": 3035
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7778719663619995,
+      "learning_rate": 3.10006817811825e-05,
+      "loss": 3.1818,
+      "step": 3040
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7951830625534058,
+      "learning_rate": 3.084657687908855e-05,
+      "loss": 3.1904,
+      "step": 3045
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.8168771862983704,
+      "learning_rate": 3.069268491812052e-05,
+      "loss": 3.177,
+      "step": 3050
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.8381624817848206,
+      "learning_rate": 3.0539007609195934e-05,
+      "loss": 3.2262,
+      "step": 3055
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.8873036503791809,
+      "learning_rate": 3.0385546660845908e-05,
+      "loss": 3.0912,
+      "step": 3060
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.9288731813430786,
+      "learning_rate": 3.0232303779196132e-05,
+      "loss": 3.2078,
+      "step": 3065
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.8773919343948364,
+      "learning_rate": 3.0079280667947885e-05,
+      "loss": 3.1059,
+      "step": 3070
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.823991596698761,
+      "learning_rate": 2.9926479028359132e-05,
+      "loss": 3.2226,
+      "step": 3075
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7931019067764282,
+      "learning_rate": 2.97739005592256e-05,
+      "loss": 3.1763,
+      "step": 3080
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7744354009628296,
+      "learning_rate": 2.962154695686187e-05,
+      "loss": 3.146,
+      "step": 3085
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.8355756402015686,
+      "learning_rate": 2.9469419915082536e-05,
+      "loss": 3.1799,
+      "step": 3090
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.8058030009269714,
+      "learning_rate": 2.9317521125183368e-05,
+      "loss": 3.2193,
+      "step": 3095
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.8332851529121399,
+      "learning_rate": 2.9165852275922524e-05,
+      "loss": 3.2367,
+      "step": 3100
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.8169847726821899,
+      "learning_rate": 2.901441505350174e-05,
+      "loss": 3.2343,
+      "step": 3105
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.8637244701385498,
+      "learning_rate": 2.886321114154762e-05,
+      "loss": 3.3044,
+      "step": 3110
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7913427948951721,
+      "learning_rate": 2.87122422210929e-05,
+      "loss": 3.2248,
+      "step": 3115
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.8639921545982361,
+      "learning_rate": 2.8561509970557736e-05,
+      "loss": 3.2188,
+      "step": 3120
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.801016092300415,
+      "learning_rate": 2.8411016065731146e-05,
+      "loss": 3.2085,
+      "step": 3125
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7943453788757324,
+      "learning_rate": 2.826076217975222e-05,
+      "loss": 3.1371,
+      "step": 3130
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7858323454856873,
+      "learning_rate": 2.8110749983091632e-05,
+      "loss": 3.1532,
+      "step": 3135
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.8061541318893433,
+      "learning_rate": 2.7960981143533053e-05,
+      "loss": 3.1412,
+      "step": 3140
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7628043293952942,
+      "learning_rate": 2.781145732615457e-05,
+      "loss": 3.1437,
+      "step": 3145
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.8349359035491943,
+      "learning_rate": 2.7662180193310218e-05,
+      "loss": 3.1472,
+      "step": 3150
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.8226621150970459,
+      "learning_rate": 2.751315140461145e-05,
+      "loss": 3.1925,
+      "step": 3155
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.8105551600456238,
+      "learning_rate": 2.7364372616908744e-05,
+      "loss": 3.1886,
+      "step": 3160
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.8078845143318176,
+      "learning_rate": 2.7215845484273152e-05,
+      "loss": 3.2221,
+      "step": 3165
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.8181054592132568,
+      "learning_rate": 2.7067571657977893e-05,
+      "loss": 3.1477,
+      "step": 3170
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7936989068984985,
+      "learning_rate": 2.691955278648003e-05,
+      "loss": 3.1467,
+      "step": 3175
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7987035512924194,
+      "learning_rate": 2.6771790515402112e-05,
+      "loss": 3.155,
+      "step": 3180
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.8322426080703735,
+      "learning_rate": 2.6624286487513916e-05,
+      "loss": 3.2001,
+      "step": 3185
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.8069536685943604,
+      "learning_rate": 2.6477042342714137e-05,
+      "loss": 3.2272,
+      "step": 3190
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.8421223163604736,
+      "learning_rate": 2.633005971801219e-05,
+      "loss": 3.1936,
+      "step": 3195
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.8561714887619019,
+      "learning_rate": 2.6183340247510013e-05,
+      "loss": 3.12,
+      "step": 3200
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.8508149981498718,
+      "learning_rate": 2.6036885562383856e-05,
+      "loss": 3.1496,
+      "step": 3205
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.8741605281829834,
+      "learning_rate": 2.5890697290866206e-05,
+      "loss": 3.2406,
+      "step": 3210
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7964335083961487,
+      "learning_rate": 2.5744777058227642e-05,
+      "loss": 3.1666,
+      "step": 3215
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.8591488003730774,
+      "learning_rate": 2.5599126486758777e-05,
+      "loss": 3.2426,
+      "step": 3220
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.8198850154876709,
+      "learning_rate": 2.5453747195752243e-05,
+      "loss": 3.1771,
+      "step": 3225
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.8103959560394287,
+      "learning_rate": 2.530864080148464e-05,
+      "loss": 3.229,
+      "step": 3230
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.8377269506454468,
+      "learning_rate": 2.5163808917198615e-05,
+      "loss": 3.139,
+      "step": 3235
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7943127155303955,
+      "learning_rate": 2.501925315308492e-05,
+      "loss": 3.2239,
+      "step": 3240
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7915096879005432,
+      "learning_rate": 2.4874975116264477e-05,
+      "loss": 3.2592,
+      "step": 3245
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.791080117225647,
+      "learning_rate": 2.4730976410770534e-05,
+      "loss": 3.1514,
+      "step": 3250
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8079975247383118,
+      "learning_rate": 2.458725863753084e-05,
+      "loss": 3.1366,
+      "step": 3255
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7989469766616821,
+      "learning_rate": 2.4443823394349834e-05,
+      "loss": 3.2059,
+      "step": 3260
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7954628467559814,
+      "learning_rate": 2.430067227589088e-05,
+      "loss": 3.2649,
+      "step": 3265
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7923185229301453,
+      "learning_rate": 2.4157806873658517e-05,
+      "loss": 3.14,
+      "step": 3270
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7876720428466797,
+      "learning_rate": 2.401522877598087e-05,
+      "loss": 3.154,
+      "step": 3275
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8176397681236267,
+      "learning_rate": 2.3872939567991827e-05,
+      "loss": 3.2069,
+      "step": 3280
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7727697491645813,
+      "learning_rate": 2.373094083161353e-05,
+      "loss": 3.099,
+      "step": 3285
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7817246913909912,
+      "learning_rate": 2.358923414553877e-05,
+      "loss": 3.1946,
+      "step": 3290
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7991405725479126,
+      "learning_rate": 2.3447821085213405e-05,
+      "loss": 3.1937,
+      "step": 3295
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.8209265470504761,
+      "learning_rate": 2.3306703222818878e-05,
+      "loss": 3.233,
+      "step": 3300
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.8106775283813477,
+      "learning_rate": 2.3165882127254705e-05,
+      "loss": 3.2158,
+      "step": 3305
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.8128020167350769,
+      "learning_rate": 2.302535936412108e-05,
+      "loss": 3.2527,
+      "step": 3310
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.8070123791694641,
+      "learning_rate": 2.2885136495701415e-05,
+      "loss": 3.206,
+      "step": 3315
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.8676778674125671,
+      "learning_rate": 2.274521508094501e-05,
+      "loss": 3.1197,
+      "step": 3320
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.8604967594146729,
+      "learning_rate": 2.2605596675449698e-05,
+      "loss": 3.2578,
+      "step": 3325
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.836823582649231,
+      "learning_rate": 2.246628283144457e-05,
+      "loss": 3.1424,
+      "step": 3330
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7955985069274902,
+      "learning_rate": 2.232727509777269e-05,
+      "loss": 3.2224,
+      "step": 3335
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.777150571346283,
+      "learning_rate": 2.2188575019873932e-05,
+      "loss": 3.2291,
+      "step": 3340
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.8456872701644897,
+      "learning_rate": 2.2050184139767704e-05,
+      "loss": 3.1922,
+      "step": 3345
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7916918396949768,
+      "learning_rate": 2.191210399603591e-05,
+      "loss": 3.1397,
+      "step": 3350
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.8304871320724487,
+      "learning_rate": 2.1774336123805772e-05,
+      "loss": 3.106,
+      "step": 3355
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.8173920512199402,
+      "learning_rate": 2.1636882054732776e-05,
+      "loss": 3.1667,
+      "step": 3360
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.8554221987724304,
+      "learning_rate": 2.1499743316983684e-05,
+      "loss": 3.2484,
+      "step": 3365
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.8113805651664734,
+      "learning_rate": 2.1362921435219473e-05,
+      "loss": 3.1443,
+      "step": 3370
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.8250212669372559,
+      "learning_rate": 2.1226417930578464e-05,
+      "loss": 3.2164,
+      "step": 3375
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.804509699344635,
+      "learning_rate": 2.109023432065935e-05,
+      "loss": 3.1842,
+      "step": 3380
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.8785647749900818,
+      "learning_rate": 2.095437211950434e-05,
+      "loss": 3.1888,
+      "step": 3385
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7957813739776611,
+      "learning_rate": 2.0818832837582352e-05,
+      "loss": 3.1795,
+      "step": 3390
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.8057026267051697,
+      "learning_rate": 2.068361798177218e-05,
+      "loss": 3.2627,
+      "step": 3395
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7766295671463013,
+      "learning_rate": 2.0548729055345778e-05,
+      "loss": 3.2093,
+      "step": 3400
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.785518229007721,
+      "learning_rate": 2.0414167557951514e-05,
+      "loss": 3.2074,
+      "step": 3405
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.840319812297821,
+      "learning_rate": 2.0279934985597527e-05,
+      "loss": 3.2023,
+      "step": 3410
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.8056073188781738,
+      "learning_rate": 2.0146032830635054e-05,
+      "loss": 3.2114,
+      "step": 3415
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.167738676071167,
+      "learning_rate": 2.001246258174192e-05,
+      "loss": 3.2406,
+      "step": 3420
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7821819186210632,
+      "learning_rate": 1.9879225723905886e-05,
+      "loss": 3.1548,
+      "step": 3425
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7894221544265747,
+      "learning_rate": 1.9746323738408203e-05,
+      "loss": 3.1988,
+      "step": 3430
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.8630125522613525,
+      "learning_rate": 1.9613758102807117e-05,
+      "loss": 3.1896,
+      "step": 3435
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.8015197515487671,
+      "learning_rate": 1.9481530290921474e-05,
+      "loss": 3.1377,
+      "step": 3440
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.8257816433906555,
+      "learning_rate": 1.934964177281428e-05,
+      "loss": 3.094,
+      "step": 3445
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.8061752319335938,
+      "learning_rate": 1.9218094014776434e-05,
+      "loss": 3.1906,
+      "step": 3450
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.8156401515007019,
+      "learning_rate": 1.9086888479310333e-05,
+      "loss": 3.1326,
+      "step": 3455
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.8027080297470093,
+      "learning_rate": 1.895602662511371e-05,
+      "loss": 3.242,
+      "step": 3460
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.8097110986709595,
+      "learning_rate": 1.8825509907063327e-05,
+      "loss": 3.2284,
+      "step": 3465
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7945130467414856,
+      "learning_rate": 1.8695339776198872e-05,
+      "loss": 3.138,
+      "step": 3470
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.8243803977966309,
+      "learning_rate": 1.8565517679706783e-05,
+      "loss": 3.1851,
+      "step": 3475
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.8031255602836609,
+      "learning_rate": 1.8436045060904174e-05,
+      "loss": 3.2587,
+      "step": 3480
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7686798572540283,
+      "learning_rate": 1.830692335922279e-05,
+      "loss": 3.1487,
+      "step": 3485
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.8455352783203125,
+      "learning_rate": 1.8178154010192994e-05,
+      "loss": 3.2103,
+      "step": 3490
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7771708965301514,
+      "learning_rate": 1.8049738445427822e-05,
+      "loss": 3.0721,
+      "step": 3495
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7743687033653259,
+      "learning_rate": 1.7921678092607052e-05,
+      "loss": 3.0943,
+      "step": 3500
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7884378433227539,
+      "learning_rate": 1.7793974375461352e-05,
+      "loss": 3.1499,
+      "step": 3505
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.8018225431442261,
+      "learning_rate": 1.7666628713756417e-05,
+      "loss": 3.2109,
+      "step": 3510
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7938441038131714,
+      "learning_rate": 1.7539642523277228e-05,
+      "loss": 3.1008,
+      "step": 3515
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.8159660696983337,
+      "learning_rate": 1.7413017215812273e-05,
+      "loss": 3.1646,
+      "step": 3520
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.791012704372406,
+      "learning_rate": 1.728675419913788e-05,
+      "loss": 3.3161,
+      "step": 3525
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.79207843542099,
+      "learning_rate": 1.716085487700253e-05,
+      "loss": 3.1227,
+      "step": 3530
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.8134340643882751,
+      "learning_rate": 1.703532064911131e-05,
+      "loss": 3.1784,
+      "step": 3535
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7827357649803162,
+      "learning_rate": 1.6910152911110283e-05,
+      "loss": 3.158,
+      "step": 3540
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.8012112379074097,
+      "learning_rate": 1.6785353054571024e-05,
+      "loss": 3.1053,
+      "step": 3545
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.802773118019104,
+      "learning_rate": 1.666092246697512e-05,
+      "loss": 3.1735,
+      "step": 3550
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7863219380378723,
+      "learning_rate": 1.6536862531698766e-05,
+      "loss": 3.0726,
+      "step": 3555
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7989824414253235,
+      "learning_rate": 1.6413174627997328e-05,
+      "loss": 3.2088,
+      "step": 3560
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7984068989753723,
+      "learning_rate": 1.6289860130990147e-05,
+      "loss": 3.1004,
+      "step": 3565
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7902229428291321,
+      "learning_rate": 1.6166920411645064e-05,
+      "loss": 3.125,
+      "step": 3570
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.8097918033599854,
+      "learning_rate": 1.6044356836763315e-05,
+      "loss": 3.1523,
+      "step": 3575
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7726302742958069,
+      "learning_rate": 1.5922170768964285e-05,
+      "loss": 3.1466,
+      "step": 3580
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7708826065063477,
+      "learning_rate": 1.5800363566670362e-05,
+      "loss": 3.1728,
+      "step": 3585
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.786941409111023,
+      "learning_rate": 1.5678936584091852e-05,
+      "loss": 3.183,
+      "step": 3590
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7763702869415283,
+      "learning_rate": 1.5557891171211892e-05,
+      "loss": 3.1687,
+      "step": 3595
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7879136204719543,
+      "learning_rate": 1.5437228673771465e-05,
+      "loss": 3.1369,
+      "step": 3600
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.816377580165863,
+      "learning_rate": 1.5316950433254445e-05,
+      "loss": 3.1376,
+      "step": 3605
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7834891080856323,
+      "learning_rate": 1.5197057786872649e-05,
+      "loss": 3.1835,
+      "step": 3610
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8588930368423462,
+      "learning_rate": 1.5077552067551015e-05,
+      "loss": 3.0885,
+      "step": 3615
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.9164984226226807,
+      "learning_rate": 1.4958434603912747e-05,
+      "loss": 3.2133,
+      "step": 3620
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7704552412033081,
+      "learning_rate": 1.4839706720264546e-05,
+      "loss": 3.1169,
+      "step": 3625
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8061482906341553,
+      "learning_rate": 1.4721369736581924e-05,
+      "loss": 3.2268,
+      "step": 3630
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7758923768997192,
+      "learning_rate": 1.4603424968494484e-05,
+      "loss": 3.1401,
+      "step": 3635
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7879301905632019,
+      "learning_rate": 1.448587372727132e-05,
+      "loss": 3.2194,
+      "step": 3640
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.8021620512008667,
+      "learning_rate": 1.4368717319806419e-05,
+      "loss": 3.1025,
+      "step": 3645
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.794487476348877,
+      "learning_rate": 1.4251957048604152e-05,
+      "loss": 3.1316,
+      "step": 3650
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7669387459754944,
+      "learning_rate": 1.413559421176479e-05,
+      "loss": 3.1092,
+      "step": 3655
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.8293997645378113,
+      "learning_rate": 1.4019630102970056e-05,
+      "loss": 3.167,
+      "step": 3660
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.8108026385307312,
+      "learning_rate": 1.3904066011468753e-05,
+      "loss": 3.2,
+      "step": 3665
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7944249510765076,
+      "learning_rate": 1.3788903222062433e-05,
+      "loss": 3.0714,
+      "step": 3670
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.799651563167572,
+      "learning_rate": 1.3674143015091118e-05,
+      "loss": 3.1009,
+      "step": 3675
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7957543134689331,
+      "learning_rate": 1.355978666641905e-05,
+      "loss": 3.0966,
+      "step": 3680
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.8025516271591187,
+      "learning_rate": 1.3445835447420507e-05,
+      "loss": 3.0909,
+      "step": 3685
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.8256009817123413,
+      "learning_rate": 1.3332290624965688e-05,
+      "loss": 3.2392,
+      "step": 3690
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7947035431861877,
+      "learning_rate": 1.3219153461406609e-05,
+      "loss": 3.097,
+      "step": 3695
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7943867444992065,
+      "learning_rate": 1.3106425214563078e-05,
+      "loss": 3.1614,
+      "step": 3700
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.8204525113105774,
+      "learning_rate": 1.2994107137708716e-05,
+      "loss": 3.2037,
+      "step": 3705
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.8015938997268677,
+      "learning_rate": 1.2882200479556988e-05,
+      "loss": 3.1579,
+      "step": 3710
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7741564512252808,
+      "learning_rate": 1.2770706484247397e-05,
+      "loss": 3.1801,
+      "step": 3715
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7824724316596985,
+      "learning_rate": 1.2659626391331564e-05,
+      "loss": 3.1861,
+      "step": 3720
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.769964337348938,
+      "learning_rate": 1.2548961435759493e-05,
+      "loss": 3.14,
+      "step": 3725
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.8480700850486755,
+      "learning_rate": 1.2438712847865846e-05,
+      "loss": 3.1436,
+      "step": 3730
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.8069185018539429,
+      "learning_rate": 1.2328881853356244e-05,
+      "loss": 3.1581,
+      "step": 3735
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7940760254859924,
+      "learning_rate": 1.221946967329365e-05,
+      "loss": 3.132,
+      "step": 3740
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.8028521537780762,
+      "learning_rate": 1.2110477524084796e-05,
+      "loss": 3.1812,
+      "step": 3745
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7855331301689148,
+      "learning_rate": 1.2001906617466657e-05,
+      "loss": 3.1165,
+      "step": 3750
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7715919017791748,
+      "learning_rate": 1.1893758160492978e-05,
+      "loss": 3.1442,
+      "step": 3755
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.855614423751831,
+      "learning_rate": 1.1786033355520859e-05,
+      "loss": 3.0849,
+      "step": 3760
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.787558913230896,
+      "learning_rate": 1.1678733400197373e-05,
+      "loss": 3.1168,
+      "step": 3765
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.8070486783981323,
+      "learning_rate": 1.1571859487446263e-05,
+      "loss": 3.1315,
+      "step": 3770
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.8083584308624268,
+      "learning_rate": 1.1465412805454695e-05,
+      "loss": 3.1808,
+      "step": 3775
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.8175447583198547,
+      "learning_rate": 1.1359394537660011e-05,
+      "loss": 3.1449,
+      "step": 3780
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7884967923164368,
+      "learning_rate": 1.125380586273661e-05,
+      "loss": 3.2144,
+      "step": 3785
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7838531732559204,
+      "learning_rate": 1.1148647954582808e-05,
+      "loss": 3.1226,
+      "step": 3790
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.8267455697059631,
+      "learning_rate": 1.1043921982307819e-05,
+      "loss": 3.1714,
+      "step": 3795
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7771685123443604,
+      "learning_rate": 1.0939629110218735e-05,
+      "loss": 3.1241,
+      "step": 3800
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7959226965904236,
+      "learning_rate": 1.0835770497807596e-05,
+      "loss": 3.1452,
+      "step": 3805
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.8620918393135071,
+      "learning_rate": 1.0732347299738493e-05,
+      "loss": 3.2259,
+      "step": 3810
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7999968528747559,
+      "learning_rate": 1.0629360665834732e-05,
+      "loss": 3.1147,
+      "step": 3815
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7933318614959717,
+      "learning_rate": 1.052681174106604e-05,
+      "loss": 3.0951,
+      "step": 3820
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.8061344027519226,
+      "learning_rate": 1.0424701665535852e-05,
+      "loss": 3.2327,
+      "step": 3825
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.8072717785835266,
+      "learning_rate": 1.0323031574468638e-05,
+      "loss": 3.1211,
+      "step": 3830
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.8158341646194458,
+      "learning_rate": 1.0221802598197261e-05,
+      "loss": 3.1877,
+      "step": 3835
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7635796070098877,
+      "learning_rate": 1.0121015862150423e-05,
+      "loss": 3.1272,
+      "step": 3840
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7936083078384399,
+      "learning_rate": 1.0020672486840154e-05,
+      "loss": 3.1304,
+      "step": 3845
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7961178421974182,
+      "learning_rate": 9.920773587849364e-06,
+      "loss": 3.152,
+      "step": 3850
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7673249840736389,
+      "learning_rate": 9.821320275819401e-06,
+      "loss": 3.233,
+      "step": 3855
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.8225710391998291,
+      "learning_rate": 9.72231365643777e-06,
+      "loss": 3.1436,
+      "step": 3860
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.8114515542984009,
+      "learning_rate": 9.623754830425779e-06,
+      "loss": 3.1356,
+      "step": 3865
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7966388463973999,
+      "learning_rate": 9.52564489352632e-06,
+      "loss": 3.2398,
+      "step": 3870
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.8004997372627258,
+      "learning_rate": 9.427984936491702e-06,
+      "loss": 3.1899,
+      "step": 3875
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.0184743404388428,
+      "learning_rate": 9.330776045071509e-06,
+      "loss": 3.1287,
+      "step": 3880
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7918827533721924,
+      "learning_rate": 9.23401930000054e-06,
+      "loss": 3.1601,
+      "step": 3885
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.8677778840065002,
+      "learning_rate": 9.137715776986772e-06,
+      "loss": 3.0695,
+      "step": 3890
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.8687372207641602,
+      "learning_rate": 9.041866546699434e-06,
+      "loss": 3.1383,
+      "step": 3895
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.8154995441436768,
+      "learning_rate": 8.946472674757078e-06,
+      "loss": 3.1618,
+      "step": 3900
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7793681025505066,
+      "learning_rate": 8.851535221715735e-06,
+      "loss": 3.0916,
+      "step": 3905
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7917068004608154,
+      "learning_rate": 8.757055243057132e-06,
+      "loss": 3.1318,
+      "step": 3910
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.8099787831306458,
+      "learning_rate": 8.663033789176967e-06,
+      "loss": 3.1848,
+      "step": 3915
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7744637727737427,
+      "learning_rate": 8.5694719053732e-06,
+      "loss": 3.1937,
+      "step": 3920
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8059993386268616,
+      "learning_rate": 8.476370631834458e-06,
+      "loss": 3.1326,
+      "step": 3925
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8357394933700562,
+      "learning_rate": 8.383731003628452e-06,
+      "loss": 3.1179,
+      "step": 3930
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8141809105873108,
+      "learning_rate": 8.291554050690508e-06,
+      "loss": 3.1621,
+      "step": 3935
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8160678148269653,
+      "learning_rate": 8.199840797812058e-06,
+      "loss": 3.1565,
+      "step": 3940
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7965174317359924,
+      "learning_rate": 8.108592264629295e-06,
+      "loss": 3.127,
+      "step": 3945
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7999559044837952,
+      "learning_rate": 8.017809465611803e-06,
+      "loss": 3.0942,
+      "step": 3950
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8185849189758301,
+      "learning_rate": 7.927493410051324e-06,
+      "loss": 3.0918,
+      "step": 3955
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8670766353607178,
+      "learning_rate": 7.837645102050473e-06,
+      "loss": 3.1102,
+      "step": 3960
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7667662501335144,
+      "learning_rate": 7.748265540511635e-06,
+      "loss": 3.0881,
+      "step": 3965
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8150936365127563,
+      "learning_rate": 7.65935571912582e-06,
+      "loss": 3.1349,
+      "step": 3970
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7920727729797363,
+      "learning_rate": 7.5709166263616405e-06,
+      "loss": 3.152,
+      "step": 3975
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7608092427253723,
+      "learning_rate": 7.482949245454302e-06,
+      "loss": 3.1649,
+      "step": 3980
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7877283692359924,
+      "learning_rate": 7.3954545543946876e-06,
+      "loss": 3.1612,
+      "step": 3985
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7629894614219666,
+      "learning_rate": 7.308433525918468e-06,
+      "loss": 3.0384,
+      "step": 3990
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8076894879341125,
+      "learning_rate": 7.221887127495313e-06,
+      "loss": 3.1539,
+      "step": 3995
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7971545457839966,
+      "learning_rate": 7.1358163213181114e-06,
+      "loss": 3.1739,
+      "step": 4000
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 4811,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 1.0865380348133376e+19,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}