diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8421 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.0473597903539947,
+  "eval_steps": 500,
+  "global_step": 6000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 31.329082489013672,
+      "learning_rate": 5e-06,
+      "loss": 9.5746,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 23.84173583984375,
+      "learning_rate": 1e-05,
+      "loss": 9.1184,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 20.422855377197266,
+      "learning_rate": 1.5e-05,
+      "loss": 8.3474,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 13.185832977294922,
+      "learning_rate": 2e-05,
+      "loss": 7.5227,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 9.406152725219727,
+      "learning_rate": 2.5e-05,
+      "loss": 6.9373,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.582286357879639,
+      "learning_rate": 3e-05,
+      "loss": 6.4761,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.742688179016113,
+      "learning_rate": 3.5e-05,
+      "loss": 6.141,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.44151496887207,
+      "learning_rate": 4e-05,
+      "loss": 5.7703,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.459256172180176,
+      "learning_rate": 4.5e-05,
+      "loss": 5.3558,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.325704574584961,
+      "learning_rate": 5e-05,
+      "loss": 4.937,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.220489978790283,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 4.6292,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.544135093688965,
+      "learning_rate": 6e-05,
+      "loss": 4.4433,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.351686954498291,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 4.2298,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.968824863433838,
+      "learning_rate": 7e-05,
+      "loss": 4.0201,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.30537486076355,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 3.8438,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.717099666595459,
+      "learning_rate": 8e-05,
+      "loss": 3.7233,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.0127949714660645,
+      "learning_rate": 8.5e-05,
+      "loss": 3.5801,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.6356096267700195,
+      "learning_rate": 9e-05,
+      "loss": 3.5695,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.9147191047668457,
+      "learning_rate": 9.5e-05,
+      "loss": 3.5812,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.6216323375701904,
+      "learning_rate": 0.0001,
+      "loss": 3.3526,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.657348871231079,
+      "learning_rate": 9.999999961506464e-05,
+      "loss": 3.3275,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.350072145462036,
+      "learning_rate": 9.999999846025851e-05,
+      "loss": 3.3834,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.2300047874450684,
+      "learning_rate": 9.999999653558166e-05,
+      "loss": 3.2307,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.2206859588623047,
+      "learning_rate": 9.999999384103412e-05,
+      "loss": 3.1758,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.009891986846924,
+      "learning_rate": 9.999999037661592e-05,
+      "loss": 3.1717,
+      "step": 125
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.015596389770508,
+      "learning_rate": 9.999998614232713e-05,
+      "loss": 3.0934,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.1175801753997803,
+      "learning_rate": 9.99999811381678e-05,
+      "loss": 3.1363,
+      "step": 135
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.8993569612503052,
+      "learning_rate": 9.9999975364138e-05,
+      "loss": 3.076,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.0575289726257324,
+      "learning_rate": 9.999996882023784e-05,
+      "loss": 3.0822,
+      "step": 145
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.9871021509170532,
+      "learning_rate": 9.99999615064674e-05,
+      "loss": 3.1702,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.984358549118042,
+      "learning_rate": 9.99999534228268e-05,
+      "loss": 3.0491,
+      "step": 155
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.8417689800262451,
+      "learning_rate": 9.999994456931619e-05,
+      "loss": 3.0723,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.925311803817749,
+      "learning_rate": 9.999993494593566e-05,
+      "loss": 2.9627,
+      "step": 165
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7795164585113525,
+      "learning_rate": 9.999992455268539e-05,
+      "loss": 2.9241,
+      "step": 170
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.826548457145691,
+      "learning_rate": 9.999991338956554e-05,
+      "loss": 2.9123,
+      "step": 175
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.8433538675308228,
+      "learning_rate": 9.999990145657626e-05,
+      "loss": 2.8614,
+      "step": 180
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.304640293121338,
+      "learning_rate": 9.999988875371776e-05,
+      "loss": 2.8874,
+      "step": 185
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.775341272354126,
+      "learning_rate": 9.999987528099021e-05,
+      "loss": 2.9374,
+      "step": 190
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7687417268753052,
+      "learning_rate": 9.999986103839384e-05,
+      "loss": 2.8623,
+      "step": 195
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.026808738708496,
+      "learning_rate": 9.999984602592884e-05,
+      "loss": 2.9046,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7305485010147095,
+      "learning_rate": 9.999983024359549e-05,
+      "loss": 2.8702,
+      "step": 205
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7887104749679565,
+      "learning_rate": 9.9999813691394e-05,
+      "loss": 2.858,
+      "step": 210
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6602907180786133,
+      "learning_rate": 9.999979636932461e-05,
+      "loss": 2.8076,
+      "step": 215
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.8972954750061035,
+      "learning_rate": 9.999977827738762e-05,
+      "loss": 2.8045,
+      "step": 220
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.594986081123352,
+      "learning_rate": 9.999975941558328e-05,
+      "loss": 2.8634,
+      "step": 225
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5907307863235474,
+      "learning_rate": 9.999973978391191e-05,
+      "loss": 2.7785,
+      "step": 230
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6393113136291504,
+      "learning_rate": 9.99997193823738e-05,
+      "loss": 2.8777,
+      "step": 235
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5739319324493408,
+      "learning_rate": 9.999969821096924e-05,
+      "loss": 2.8323,
+      "step": 240
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.646419644355774,
+      "learning_rate": 9.99996762696986e-05,
+      "loss": 2.7639,
+      "step": 245
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.711586356163025,
+      "learning_rate": 9.999965355856218e-05,
+      "loss": 2.8389,
+      "step": 250
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.555133581161499,
+      "learning_rate": 9.999963007756035e-05,
+      "loss": 2.8512,
+      "step": 255
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5264123678207397,
+      "learning_rate": 9.999960582669346e-05,
+      "loss": 2.7201,
+      "step": 260
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5675524473190308,
+      "learning_rate": 9.99995808059619e-05,
+      "loss": 2.7726,
+      "step": 265
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.802937388420105,
+      "learning_rate": 9.999955501536603e-05,
+      "loss": 2.6741,
+      "step": 270
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5660523176193237,
+      "learning_rate": 9.999952845490626e-05,
+      "loss": 2.7189,
+      "step": 275
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4223791360855103,
+      "learning_rate": 9.999950112458302e-05,
+      "loss": 2.7207,
+      "step": 280
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4202953577041626,
+      "learning_rate": 9.99994730243967e-05,
+      "loss": 2.6489,
+      "step": 285
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5310500860214233,
+      "learning_rate": 9.999944415434774e-05,
+      "loss": 2.6701,
+      "step": 290
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.226552724838257,
+      "learning_rate": 9.99994145144366e-05,
+      "loss": 2.7658,
+      "step": 295
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4547815322875977,
+      "learning_rate": 9.999938410466371e-05,
+      "loss": 2.7918,
+      "step": 300
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.258984088897705,
+      "learning_rate": 9.999935292502958e-05,
+      "loss": 2.7489,
+      "step": 305
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.385048747062683,
+      "learning_rate": 9.999932097553463e-05,
+      "loss": 2.6871,
+      "step": 310
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.427621603012085,
+      "learning_rate": 9.999928825617941e-05,
+      "loss": 2.7442,
+      "step": 315
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.351544737815857,
+      "learning_rate": 9.999925476696439e-05,
+      "loss": 2.7264,
+      "step": 320
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.369752287864685,
+      "learning_rate": 9.99992205078901e-05,
+      "loss": 2.6754,
+      "step": 325
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.8152669668197632,
+      "learning_rate": 9.999918547895705e-05,
+      "loss": 2.7205,
+      "step": 330
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3544217348098755,
+      "learning_rate": 9.999914968016579e-05,
+      "loss": 2.7823,
+      "step": 335
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5886297225952148,
+      "learning_rate": 9.999911311151689e-05,
+      "loss": 2.6905,
+      "step": 340
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3437013626098633,
+      "learning_rate": 9.999907577301089e-05,
+      "loss": 2.6386,
+      "step": 345
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2984123229980469,
+      "learning_rate": 9.999903766464837e-05,
+      "loss": 2.7183,
+      "step": 350
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2885876893997192,
+      "learning_rate": 9.99989987864299e-05,
+      "loss": 2.6193,
+      "step": 355
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3513743877410889,
+      "learning_rate": 9.999895913835613e-05,
+      "loss": 2.6461,
+      "step": 360
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4668092727661133,
+      "learning_rate": 9.999891872042762e-05,
+      "loss": 2.6742,
+      "step": 365
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4826070070266724,
+      "learning_rate": 9.9998877532645e-05,
+      "loss": 2.6457,
+      "step": 370
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.376685619354248,
+      "learning_rate": 9.999883557500892e-05,
+      "loss": 2.6139,
+      "step": 375
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.330474853515625,
+      "learning_rate": 9.999879284752003e-05,
+      "loss": 2.6985,
+      "step": 380
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4162460565567017,
+      "learning_rate": 9.999874935017897e-05,
+      "loss": 2.6534,
+      "step": 385
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2565540075302124,
+      "learning_rate": 9.999870508298641e-05,
+      "loss": 2.618,
+      "step": 390
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2994046211242676,
+      "learning_rate": 9.999866004594303e-05,
+      "loss": 2.6473,
+      "step": 395
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2812210321426392,
+      "learning_rate": 9.999861423904954e-05,
+      "loss": 2.7013,
+      "step": 400
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.26098895072937,
+      "learning_rate": 9.999856766230665e-05,
+      "loss": 2.5837,
+      "step": 405
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2872728109359741,
+      "learning_rate": 9.999852031571506e-05,
+      "loss": 2.6799,
+      "step": 410
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4405131340026855,
+      "learning_rate": 9.999847219927549e-05,
+      "loss": 2.6446,
+      "step": 415
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3099191188812256,
+      "learning_rate": 9.999842331298872e-05,
+      "loss": 2.6428,
+      "step": 420
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2911869287490845,
+      "learning_rate": 9.999837365685544e-05,
+      "loss": 2.7084,
+      "step": 425
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3066972494125366,
+      "learning_rate": 9.999832323087649e-05,
+      "loss": 2.6417,
+      "step": 430
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2133055925369263,
+      "learning_rate": 9.999827203505259e-05,
+      "loss": 2.6594,
+      "step": 435
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2466551065444946,
+      "learning_rate": 9.999822006938455e-05,
+      "loss": 2.6107,
+      "step": 440
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.312983751296997,
+      "learning_rate": 9.999816733387316e-05,
+      "loss": 2.5724,
+      "step": 445
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2931486368179321,
+      "learning_rate": 9.999811382851924e-05,
+      "loss": 2.5689,
+      "step": 450
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4380275011062622,
+      "learning_rate": 9.999805955332362e-05,
+      "loss": 2.5487,
+      "step": 455
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.199204683303833,
+      "learning_rate": 9.999800450828711e-05,
+      "loss": 2.6876,
+      "step": 460
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.8874166011810303,
+      "learning_rate": 9.99979486934106e-05,
+      "loss": 2.6115,
+      "step": 465
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3101069927215576,
+      "learning_rate": 9.99978921086949e-05,
+      "loss": 2.6109,
+      "step": 470
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1679068803787231,
+      "learning_rate": 9.999783475414093e-05,
+      "loss": 2.6207,
+      "step": 475
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1897002458572388,
+      "learning_rate": 9.999777662974952e-05,
+      "loss": 2.6305,
+      "step": 480
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1695494651794434,
+      "learning_rate": 9.999771773552162e-05,
+      "loss": 2.5409,
+      "step": 485
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3282781839370728,
+      "learning_rate": 9.999765807145811e-05,
+      "loss": 2.6162,
+      "step": 490
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4140591621398926,
+      "learning_rate": 9.99975976375599e-05,
+      "loss": 2.6242,
+      "step": 495
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2444924116134644,
+      "learning_rate": 9.999753643382793e-05,
+      "loss": 2.6244,
+      "step": 500
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2308467626571655,
+      "learning_rate": 9.999747446026315e-05,
+      "loss": 2.6031,
+      "step": 505
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4245002269744873,
+      "learning_rate": 9.999741171686649e-05,
+      "loss": 2.5394,
+      "step": 510
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.218187689781189,
+      "learning_rate": 9.999734820363895e-05,
+      "loss": 2.533,
+      "step": 515
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.175797462463379,
+      "learning_rate": 9.99972839205815e-05,
+      "loss": 2.5326,
+      "step": 520
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2117722034454346,
+      "learning_rate": 9.999721886769511e-05,
+      "loss": 2.5349,
+      "step": 525
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3010272979736328,
+      "learning_rate": 9.999715304498078e-05,
+      "loss": 2.5283,
+      "step": 530
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.224746584892273,
+      "learning_rate": 9.999708645243955e-05,
+      "loss": 2.45,
+      "step": 535
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.236034631729126,
+      "learning_rate": 9.999701909007242e-05,
+      "loss": 2.6755,
+      "step": 540
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1827280521392822,
+      "learning_rate": 9.999695095788044e-05,
+      "loss": 2.576,
+      "step": 545
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3072748184204102,
+      "learning_rate": 9.999688205586468e-05,
+      "loss": 2.5304,
+      "step": 550
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1770302057266235,
+      "learning_rate": 9.999681238402615e-05,
+      "loss": 2.5676,
+      "step": 555
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1951831579208374,
+      "learning_rate": 9.999674194236597e-05,
+      "loss": 2.4993,
+      "step": 560
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.137583613395691,
+      "learning_rate": 9.999667073088519e-05,
+      "loss": 2.5471,
+      "step": 565
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1907936334609985,
+      "learning_rate": 9.999659874958493e-05,
+      "loss": 2.5733,
+      "step": 570
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2016326189041138,
+      "learning_rate": 9.99965259984663e-05,
+      "loss": 2.5704,
+      "step": 575
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1379742622375488,
+      "learning_rate": 9.999645247753039e-05,
+      "loss": 2.5828,
+      "step": 580
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.0949510335922241,
+      "learning_rate": 9.999637818677836e-05,
+      "loss": 2.5425,
+      "step": 585
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.219426155090332,
+      "learning_rate": 9.999630312621136e-05,
+      "loss": 2.5915,
+      "step": 590
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1658213138580322,
+      "learning_rate": 9.999622729583051e-05,
+      "loss": 2.6148,
+      "step": 595
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.169696569442749,
+      "learning_rate": 9.999615069563701e-05,
+      "loss": 2.5523,
+      "step": 600
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3456404209136963,
+      "learning_rate": 9.999607332563205e-05,
+      "loss": 2.582,
+      "step": 605
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1582688093185425,
+      "learning_rate": 9.999599518581676e-05,
+      "loss": 2.5467,
+      "step": 610
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2877380847930908,
+      "learning_rate": 9.999591627619242e-05,
+      "loss": 2.5339,
+      "step": 615
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1368844509124756,
+      "learning_rate": 9.999583659676019e-05,
+      "loss": 2.5285,
+      "step": 620
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1442762613296509,
+      "learning_rate": 9.999575614752133e-05,
+      "loss": 2.5173,
+      "step": 625
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.05731999874115,
+      "learning_rate": 9.999567492847706e-05,
+      "loss": 2.5671,
+      "step": 630
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1240289211273193,
+      "learning_rate": 9.999559293962864e-05,
+      "loss": 2.6078,
+      "step": 635
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1165677309036255,
+      "learning_rate": 9.99955101809773e-05,
+      "loss": 2.5451,
+      "step": 640
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1120537519454956,
+      "learning_rate": 9.999542665252437e-05,
+      "loss": 2.5496,
+      "step": 645
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1016920804977417,
+      "learning_rate": 9.999534235427109e-05,
+      "loss": 2.5893,
+      "step": 650
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1280940771102905,
+      "learning_rate": 9.999525728621879e-05,
+      "loss": 2.4667,
+      "step": 655
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.132176399230957,
+      "learning_rate": 9.999517144836876e-05,
+      "loss": 2.4725,
+      "step": 660
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1237218379974365,
+      "learning_rate": 9.999508484072233e-05,
+      "loss": 2.5278,
+      "step": 665
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0854023694992065,
+      "learning_rate": 9.999499746328083e-05,
+      "loss": 2.5136,
+      "step": 670
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0707095861434937,
+      "learning_rate": 9.99949093160456e-05,
+      "loss": 2.5501,
+      "step": 675
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0843431949615479,
+      "learning_rate": 9.9994820399018e-05,
+      "loss": 2.5286,
+      "step": 680
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.014254570007324,
+      "learning_rate": 9.999473071219941e-05,
+      "loss": 2.5821,
+      "step": 685
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.097906470298767,
+      "learning_rate": 9.999464025559121e-05,
+      "loss": 2.5924,
+      "step": 690
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1643943786621094,
+      "learning_rate": 9.999454902919478e-05,
+      "loss": 2.5123,
+      "step": 695
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1262152194976807,
+      "learning_rate": 9.999445703301154e-05,
+      "loss": 2.4778,
+      "step": 700
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.106859803199768,
+      "learning_rate": 9.999436426704287e-05,
+      "loss": 2.4995,
+      "step": 705
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5095537900924683,
+      "learning_rate": 9.999427073129026e-05,
+      "loss": 2.5481,
+      "step": 710
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1243454217910767,
+      "learning_rate": 9.999417642575509e-05,
+      "loss": 2.5466,
+      "step": 715
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0687675476074219,
+      "learning_rate": 9.999408135043886e-05,
+      "loss": 2.5672,
+      "step": 720
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1190004348754883,
+      "learning_rate": 9.999398550534298e-05,
+      "loss": 2.5277,
+      "step": 725
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0488616228103638,
+      "learning_rate": 9.999388889046899e-05,
+      "loss": 2.504,
+      "step": 730
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1087074279785156,
+      "learning_rate": 9.999379150581833e-05,
+      "loss": 2.5751,
+      "step": 735
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.015958547592163,
+      "learning_rate": 9.999369335139251e-05,
+      "loss": 2.4042,
+      "step": 740
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1038602590560913,
+      "learning_rate": 9.999359442719306e-05,
+      "loss": 2.5094,
+      "step": 745
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0758789777755737,
+      "learning_rate": 9.999349473322149e-05,
+      "loss": 2.5773,
+      "step": 750
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.040269374847412,
+      "learning_rate": 9.999339426947932e-05,
+      "loss": 2.5112,
+      "step": 755
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0394530296325684,
+      "learning_rate": 9.999329303596812e-05,
+      "loss": 2.4741,
+      "step": 760
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0501093864440918,
+      "learning_rate": 9.999319103268945e-05,
+      "loss": 2.4422,
+      "step": 765
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0786670446395874,
+      "learning_rate": 9.999308825964485e-05,
+      "loss": 2.4747,
+      "step": 770
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.075774908065796,
+      "learning_rate": 9.999298471683593e-05,
+      "loss": 2.4402,
+      "step": 775
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0835989713668823,
+      "learning_rate": 9.999288040426429e-05,
+      "loss": 2.5132,
+      "step": 780
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0708087682724,
+      "learning_rate": 9.999277532193151e-05,
+      "loss": 2.4411,
+      "step": 785
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.047179937362671,
+      "learning_rate": 9.999266946983923e-05,
+      "loss": 2.5162,
+      "step": 790
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0772581100463867,
+      "learning_rate": 9.999256284798906e-05,
+      "loss": 2.5647,
+      "step": 795
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0744632482528687,
+      "learning_rate": 9.999245545638267e-05,
+      "loss": 2.5847,
+      "step": 800
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0552600622177124,
+      "learning_rate": 9.999234729502167e-05,
+      "loss": 2.5091,
+      "step": 805
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0266720056533813,
+      "learning_rate": 9.999223836390777e-05,
+      "loss": 2.472,
+      "step": 810
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0497556924819946,
+      "learning_rate": 9.999212866304262e-05,
+      "loss": 2.4728,
+      "step": 815
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.2190697193145752,
+      "learning_rate": 9.999201819242794e-05,
+      "loss": 2.4791,
+      "step": 820
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0361429452896118,
+      "learning_rate": 9.99919069520654e-05,
+      "loss": 2.4556,
+      "step": 825
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0319561958312988,
+      "learning_rate": 9.99917949419567e-05,
+      "loss": 2.5034,
+      "step": 830
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9958091974258423,
+      "learning_rate": 9.999168216210362e-05,
+      "loss": 2.4668,
+      "step": 835
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0275272130966187,
+      "learning_rate": 9.999156861250784e-05,
+      "loss": 2.4014,
+      "step": 840
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0396531820297241,
+      "learning_rate": 9.999145429317114e-05,
+      "loss": 2.4858,
+      "step": 845
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0468873977661133,
+      "learning_rate": 9.999133920409527e-05,
+      "loss": 2.4661,
+      "step": 850
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.065082311630249,
+      "learning_rate": 9.999122334528199e-05,
+      "loss": 2.4305,
+      "step": 855
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0413296222686768,
+      "learning_rate": 9.999110671673313e-05,
+      "loss": 2.4377,
+      "step": 860
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9800443053245544,
+      "learning_rate": 9.999098931845044e-05,
+      "loss": 2.4452,
+      "step": 865
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.987051248550415,
+      "learning_rate": 9.999087115043573e-05,
+      "loss": 2.4676,
+      "step": 870
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9938090443611145,
+      "learning_rate": 9.999075221269083e-05,
+      "loss": 2.4948,
+      "step": 875
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.056393027305603,
+      "learning_rate": 9.999063250521759e-05,
+      "loss": 2.4873,
+      "step": 880
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0409207344055176,
+      "learning_rate": 9.999051202801781e-05,
+      "loss": 2.4029,
+      "step": 885
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0096546411514282,
+      "learning_rate": 9.999039078109341e-05,
+      "loss": 2.4359,
+      "step": 890
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0127755403518677,
+      "learning_rate": 9.999026876444617e-05,
+      "loss": 2.4543,
+      "step": 895
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0116196870803833,
+      "learning_rate": 9.999014597807805e-05,
+      "loss": 2.488,
+      "step": 900
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0375676155090332,
+      "learning_rate": 9.99900224219909e-05,
+      "loss": 2.4887,
+      "step": 905
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.030221700668335,
+      "learning_rate": 9.998989809618662e-05,
+      "loss": 2.4656,
+      "step": 910
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.014849305152893,
+      "learning_rate": 9.998977300066715e-05,
+      "loss": 2.4754,
+      "step": 915
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.043725609779358,
+      "learning_rate": 9.998964713543437e-05,
+      "loss": 2.4659,
+      "step": 920
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9985063672065735,
+      "learning_rate": 9.998952050049026e-05,
+      "loss": 2.4761,
+      "step": 925
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0001811981201172,
+      "learning_rate": 9.998939309583677e-05,
+      "loss": 2.4965,
+      "step": 930
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.010023832321167,
+      "learning_rate": 9.998926492147585e-05,
+      "loss": 2.3596,
+      "step": 935
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9906774163246155,
+      "learning_rate": 9.998913597740945e-05,
+      "loss": 2.4105,
+      "step": 940
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.009761095046997,
+      "learning_rate": 9.99890062636396e-05,
+      "loss": 2.4864,
+      "step": 945
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1004563570022583,
+      "learning_rate": 9.998887578016827e-05,
+      "loss": 2.3609,
+      "step": 950
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.982509434223175,
+      "learning_rate": 9.998874452699747e-05,
+      "loss": 2.4159,
+      "step": 955
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9928018450737,
+      "learning_rate": 9.998861250412922e-05,
+      "loss": 2.4256,
+      "step": 960
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0368977785110474,
+      "learning_rate": 9.998847971156556e-05,
+      "loss": 2.4668,
+      "step": 965
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9629600644111633,
+      "learning_rate": 9.998834614930855e-05,
+      "loss": 2.4797,
+      "step": 970
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9998363852500916,
+      "learning_rate": 9.99882118173602e-05,
+      "loss": 2.4697,
+      "step": 975
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9921650290489197,
+      "learning_rate": 9.998807671572263e-05,
+      "loss": 2.4362,
+      "step": 980
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9890091419219971,
+      "learning_rate": 9.998794084439789e-05,
+      "loss": 2.4213,
+      "step": 985
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0788167715072632,
+      "learning_rate": 9.998780420338809e-05,
+      "loss": 2.4712,
+      "step": 990
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9879705905914307,
+      "learning_rate": 9.998766679269531e-05,
+      "loss": 2.4646,
+      "step": 995
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.00596022605896,
+      "learning_rate": 9.998752861232168e-05,
+      "loss": 2.4293,
+      "step": 1000
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0193803310394287,
+      "learning_rate": 9.998738966226933e-05,
+      "loss": 2.432,
+      "step": 1005
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9599147439002991,
+      "learning_rate": 9.998724994254041e-05,
+      "loss": 2.4526,
+      "step": 1010
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9678425788879395,
+      "learning_rate": 9.998710945313705e-05,
+      "loss": 2.4572,
+      "step": 1015
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0561902523040771,
+      "learning_rate": 9.998696819406141e-05,
+      "loss": 2.4401,
+      "step": 1020
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.075734257698059,
+      "learning_rate": 9.99868261653157e-05,
+      "loss": 2.404,
+      "step": 1025
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9181925058364868,
+      "learning_rate": 9.998668336690206e-05,
+      "loss": 2.4252,
+      "step": 1030
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0398027896881104,
+      "learning_rate": 9.998653979882271e-05,
+      "loss": 2.4517,
+      "step": 1035
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9753462076187134,
+      "learning_rate": 9.998639546107989e-05,
+      "loss": 2.4114,
+      "step": 1040
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9526083469390869,
+      "learning_rate": 9.998625035367577e-05,
+      "loss": 2.3745,
+      "step": 1045
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.8385837078094482,
+      "learning_rate": 9.998610447661261e-05,
+      "loss": 2.3777,
+      "step": 1050
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9161767363548279,
+      "learning_rate": 9.998595782989266e-05,
+      "loss": 2.455,
+      "step": 1055
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0112615823745728,
+      "learning_rate": 9.998581041351818e-05,
+      "loss": 2.4468,
+      "step": 1060
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9608880281448364,
+      "learning_rate": 9.998566222749142e-05,
+      "loss": 2.4577,
+      "step": 1065
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.021729588508606,
+      "learning_rate": 9.998551327181468e-05,
+      "loss": 2.2928,
+      "step": 1070
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9443703293800354,
+      "learning_rate": 9.998536354649026e-05,
+      "loss": 2.4964,
+      "step": 1075
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9725671410560608,
+      "learning_rate": 9.998521305152045e-05,
+      "loss": 2.5256,
+      "step": 1080
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0634089708328247,
+      "learning_rate": 9.998506178690755e-05,
+      "loss": 2.4312,
+      "step": 1085
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9096149206161499,
+      "learning_rate": 9.998490975265393e-05,
+      "loss": 2.4464,
+      "step": 1090
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9670140147209167,
+      "learning_rate": 9.99847569487619e-05,
+      "loss": 2.3921,
+      "step": 1095
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.9685615301132202,
+      "learning_rate": 9.998460337523383e-05,
+      "loss": 2.3908,
+      "step": 1100
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9297245740890503,
+      "learning_rate": 9.998444903207207e-05,
+      "loss": 2.4063,
+      "step": 1105
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.7635539770126343,
+      "learning_rate": 9.998429391927902e-05,
+      "loss": 2.4087,
+      "step": 1110
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9922268986701965,
+      "learning_rate": 9.998413803685705e-05,
+      "loss": 2.453,
+      "step": 1115
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9371087551116943,
+      "learning_rate": 9.998398138480855e-05,
+      "loss": 2.4178,
+      "step": 1120
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1819846630096436,
+      "learning_rate": 9.998382396313597e-05,
+      "loss": 2.3912,
+      "step": 1125
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9539498090744019,
+      "learning_rate": 9.998366577184168e-05,
+      "loss": 2.4663,
+      "step": 1130
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9802814722061157,
+      "learning_rate": 9.998350681092817e-05,
+      "loss": 2.3312,
+      "step": 1135
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4821765422821045,
+      "learning_rate": 9.998334708039785e-05,
+      "loss": 2.434,
+      "step": 1140
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9477716684341431,
+      "learning_rate": 9.998318658025318e-05,
+      "loss": 2.3748,
+      "step": 1145
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9432247877120972,
+      "learning_rate": 9.998302531049667e-05,
+      "loss": 2.4137,
+      "step": 1150
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9842959642410278,
+      "learning_rate": 9.998286327113076e-05,
+      "loss": 2.4715,
+      "step": 1155
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9596757292747498,
+      "learning_rate": 9.998270046215797e-05,
+      "loss": 2.4174,
+      "step": 1160
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9663878083229065,
+      "learning_rate": 9.998253688358078e-05,
+      "loss": 2.3988,
+      "step": 1165
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9495770335197449,
+      "learning_rate": 9.998237253540174e-05,
+      "loss": 2.4401,
+      "step": 1170
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9514854550361633,
+      "learning_rate": 9.998220741762338e-05,
+      "loss": 2.3711,
+      "step": 1175
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9852288961410522,
+      "learning_rate": 9.998204153024821e-05,
+      "loss": 2.4197,
+      "step": 1180
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9039212465286255,
+      "learning_rate": 9.998187487327882e-05,
+      "loss": 2.3845,
+      "step": 1185
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9594272375106812,
+      "learning_rate": 9.998170744671775e-05,
+      "loss": 2.3254,
+      "step": 1190
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.037550449371338,
+      "learning_rate": 9.998153925056758e-05,
+      "loss": 2.3713,
+      "step": 1195
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.01638662815094,
+      "learning_rate": 9.998137028483092e-05,
+      "loss": 2.3188,
+      "step": 1200
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9161309599876404,
+      "learning_rate": 9.998120054951035e-05,
+      "loss": 2.4133,
+      "step": 1205
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0021674633026123,
+      "learning_rate": 9.99810300446085e-05,
+      "loss": 2.39,
+      "step": 1210
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.949479877948761,
+      "learning_rate": 9.9980858770128e-05,
+      "loss": 2.4021,
+      "step": 1215
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.976344883441925,
+      "learning_rate": 9.998068672607146e-05,
+      "loss": 2.4418,
+      "step": 1220
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0196268558502197,
+      "learning_rate": 9.998051391244155e-05,
+      "loss": 2.3922,
+      "step": 1225
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9341482520103455,
+      "learning_rate": 9.998034032924092e-05,
+      "loss": 2.4349,
+      "step": 1230
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8798526525497437,
+      "learning_rate": 9.998016597647225e-05,
+      "loss": 2.4089,
+      "step": 1235
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9425748586654663,
+      "learning_rate": 9.997999085413823e-05,
+      "loss": 2.4471,
+      "step": 1240
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.922503650188446,
+      "learning_rate": 9.997981496224153e-05,
+      "loss": 2.2993,
+      "step": 1245
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8759360313415527,
+      "learning_rate": 9.99796383007849e-05,
+      "loss": 2.3464,
+      "step": 1250
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9183030128479004,
+      "learning_rate": 9.997946086977103e-05,
+      "loss": 2.3771,
+      "step": 1255
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8992334008216858,
+      "learning_rate": 9.997928266920266e-05,
+      "loss": 2.4432,
+      "step": 1260
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9862260222434998,
+      "learning_rate": 9.997910369908254e-05,
+      "loss": 2.4241,
+      "step": 1265
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9254174828529358,
+      "learning_rate": 9.99789239594134e-05,
+      "loss": 2.3828,
+      "step": 1270
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9690923690795898,
+      "learning_rate": 9.997874345019805e-05,
+      "loss": 2.3298,
+      "step": 1275
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9351096153259277,
+      "learning_rate": 9.997856217143921e-05,
+      "loss": 2.3329,
+      "step": 1280
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8824529051780701,
+      "learning_rate": 9.997838012313974e-05,
+      "loss": 2.3097,
+      "step": 1285
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9723858833312988,
+      "learning_rate": 9.99781973053024e-05,
+      "loss": 2.3911,
+      "step": 1290
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0103102922439575,
+      "learning_rate": 9.997801371793001e-05,
+      "loss": 2.2349,
+      "step": 1295
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9164794087409973,
+      "learning_rate": 9.997782936102542e-05,
+      "loss": 2.2733,
+      "step": 1300
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9511635899543762,
+      "learning_rate": 9.997764423459144e-05,
+      "loss": 2.3434,
+      "step": 1305
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9471263289451599,
+      "learning_rate": 9.997745833863092e-05,
+      "loss": 2.3328,
+      "step": 1310
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.930119514465332,
+      "learning_rate": 9.997727167314676e-05,
+      "loss": 2.4094,
+      "step": 1315
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.931616485118866,
+      "learning_rate": 9.997708423814178e-05,
+      "loss": 2.3733,
+      "step": 1320
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9242105484008789,
+      "learning_rate": 9.99768960336189e-05,
+      "loss": 2.4189,
+      "step": 1325
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9103718996047974,
+      "learning_rate": 9.997670705958102e-05,
+      "loss": 2.4026,
+      "step": 1330
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9525967240333557,
+      "learning_rate": 9.997651731603104e-05,
+      "loss": 2.3936,
+      "step": 1335
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8929914236068726,
+      "learning_rate": 9.997632680297188e-05,
+      "loss": 2.4072,
+      "step": 1340
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4306888580322266,
+      "learning_rate": 9.997613552040648e-05,
+      "loss": 2.3504,
+      "step": 1345
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9343785643577576,
+      "learning_rate": 9.997594346833779e-05,
+      "loss": 2.3951,
+      "step": 1350
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9039192199707031,
+      "learning_rate": 9.997575064676875e-05,
+      "loss": 2.3611,
+      "step": 1355
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8957434296607971,
+      "learning_rate": 9.997555705570232e-05,
+      "loss": 2.3512,
+      "step": 1360
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.486773133277893,
+      "learning_rate": 9.997536269514152e-05,
+      "loss": 2.3746,
+      "step": 1365
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9405444860458374,
+      "learning_rate": 9.99751675650893e-05,
+      "loss": 2.3057,
+      "step": 1370
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8957177400588989,
+      "learning_rate": 9.99749716655487e-05,
+      "loss": 2.318,
+      "step": 1375
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9588918089866638,
+      "learning_rate": 9.997477499652273e-05,
+      "loss": 2.4122,
+      "step": 1380
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8862133622169495,
+      "learning_rate": 9.99745775580144e-05,
+      "loss": 2.3261,
+      "step": 1385
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9151259064674377,
+      "learning_rate": 9.997437935002676e-05,
+      "loss": 2.342,
+      "step": 1390
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9340454339981079,
+      "learning_rate": 9.997418037256285e-05,
+      "loss": 2.362,
+      "step": 1395
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9024438261985779,
+      "learning_rate": 9.997398062562575e-05,
+      "loss": 2.4309,
+      "step": 1400
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8933719396591187,
+      "learning_rate": 9.997378010921856e-05,
+      "loss": 2.3361,
+      "step": 1405
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.925423264503479,
+      "learning_rate": 9.99735788233443e-05,
+      "loss": 2.3994,
+      "step": 1410
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9387218356132507,
+      "learning_rate": 9.997337676800613e-05,
+      "loss": 2.3632,
+      "step": 1415
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8598355650901794,
+      "learning_rate": 9.997317394320713e-05,
+      "loss": 2.3605,
+      "step": 1420
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9040392637252808,
+      "learning_rate": 9.997297034895044e-05,
+      "loss": 2.4563,
+      "step": 1425
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.931604266166687,
+      "learning_rate": 9.99727659852392e-05,
+      "loss": 2.3421,
+      "step": 1430
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.891230046749115,
+      "learning_rate": 9.997256085207652e-05,
+      "loss": 2.3513,
+      "step": 1435
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.036242961883545,
+      "learning_rate": 9.997235494946557e-05,
+      "loss": 2.3867,
+      "step": 1440
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.957940399646759,
+      "learning_rate": 9.997214827740955e-05,
+      "loss": 2.3335,
+      "step": 1445
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9209109544754028,
+      "learning_rate": 9.997194083591164e-05,
+      "loss": 2.3764,
+      "step": 1450
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8931889533996582,
+      "learning_rate": 9.9971732624975e-05,
+      "loss": 2.3714,
+      "step": 1455
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9236003756523132,
+      "learning_rate": 9.997152364460287e-05,
+      "loss": 2.3431,
+      "step": 1460
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8814219832420349,
+      "learning_rate": 9.997131389479844e-05,
+      "loss": 2.3522,
+      "step": 1465
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8789190053939819,
+      "learning_rate": 9.997110337556494e-05,
+      "loss": 2.3048,
+      "step": 1470
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8852155804634094,
+      "learning_rate": 9.997089208690564e-05,
+      "loss": 2.4125,
+      "step": 1475
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8692938089370728,
+      "learning_rate": 9.997068002882377e-05,
+      "loss": 2.3274,
+      "step": 1480
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9112242460250854,
+      "learning_rate": 9.997046720132261e-05,
+      "loss": 2.3634,
+      "step": 1485
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8992840647697449,
+      "learning_rate": 9.997025360440541e-05,
+      "loss": 2.4428,
+      "step": 1490
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8762646317481995,
+      "learning_rate": 9.997003923807551e-05,
+      "loss": 2.3352,
+      "step": 1495
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9114954471588135,
+      "learning_rate": 9.996982410233615e-05,
+      "loss": 2.4336,
+      "step": 1500
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8913519978523254,
+      "learning_rate": 9.996960819719068e-05,
+      "loss": 2.4055,
+      "step": 1505
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9203553795814514,
+      "learning_rate": 9.996939152264242e-05,
+      "loss": 2.3756,
+      "step": 1510
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9276959896087646,
+      "learning_rate": 9.99691740786947e-05,
+      "loss": 2.4349,
+      "step": 1515
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.923149824142456,
+      "learning_rate": 9.996895586535086e-05,
+      "loss": 2.3607,
+      "step": 1520
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8756529092788696,
+      "learning_rate": 9.996873688261428e-05,
+      "loss": 2.4384,
+      "step": 1525
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9622218012809753,
+      "learning_rate": 9.996851713048831e-05,
+      "loss": 2.4017,
+      "step": 1530
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9224346280097961,
+      "learning_rate": 9.996829660897637e-05,
+      "loss": 2.3625,
+      "step": 1535
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8947101831436157,
+      "learning_rate": 9.996807531808181e-05,
+      "loss": 2.2832,
+      "step": 1540
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8814922571182251,
+      "learning_rate": 9.996785325780806e-05,
+      "loss": 2.4508,
+      "step": 1545
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8913693428039551,
+      "learning_rate": 9.996763042815853e-05,
+      "loss": 2.3847,
+      "step": 1550
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8914551138877869,
+      "learning_rate": 9.996740682913668e-05,
+      "loss": 2.3383,
+      "step": 1555
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8685865998268127,
+      "learning_rate": 9.996718246074591e-05,
+      "loss": 2.3442,
+      "step": 1560
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8809703588485718,
+      "learning_rate": 9.99669573229897e-05,
+      "loss": 2.3206,
+      "step": 1565
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8730484843254089,
+      "learning_rate": 9.996673141587151e-05,
+      "loss": 2.3652,
+      "step": 1570
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8531500101089478,
+      "learning_rate": 9.996650473939482e-05,
+      "loss": 2.4256,
+      "step": 1575
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8794611096382141,
+      "learning_rate": 9.996627729356314e-05,
+      "loss": 2.3105,
+      "step": 1580
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8566188216209412,
+      "learning_rate": 9.996604907837993e-05,
+      "loss": 2.221,
+      "step": 1585
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9095826148986816,
+      "learning_rate": 9.996582009384874e-05,
+      "loss": 2.3664,
+      "step": 1590
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8489287495613098,
+      "learning_rate": 9.996559033997307e-05,
+      "loss": 2.3443,
+      "step": 1595
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8541042804718018,
+      "learning_rate": 9.996535981675648e-05,
+      "loss": 2.3405,
+      "step": 1600
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9183443784713745,
+      "learning_rate": 9.996512852420251e-05,
+      "loss": 2.381,
+      "step": 1605
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.2018860578536987,
+      "learning_rate": 9.996489646231471e-05,
+      "loss": 2.277,
+      "step": 1610
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9906625747680664,
+      "learning_rate": 9.996466363109667e-05,
+      "loss": 2.3901,
+      "step": 1615
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8886979222297668,
+      "learning_rate": 9.996443003055196e-05,
+      "loss": 2.3238,
+      "step": 1620
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9703171849250793,
+      "learning_rate": 9.99641956606842e-05,
+      "loss": 2.2804,
+      "step": 1625
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8842098712921143,
+      "learning_rate": 9.996396052149698e-05,
+      "loss": 2.4093,
+      "step": 1630
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9199352860450745,
+      "learning_rate": 9.996372461299392e-05,
+      "loss": 2.3657,
+      "step": 1635
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8715570569038391,
+      "learning_rate": 9.996348793517866e-05,
+      "loss": 2.3488,
+      "step": 1640
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8484712839126587,
+      "learning_rate": 9.996325048805484e-05,
+      "loss": 2.3421,
+      "step": 1645
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8742313981056213,
+      "learning_rate": 9.996301227162612e-05,
+      "loss": 2.3296,
+      "step": 1650
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8930214047431946,
+      "learning_rate": 9.996277328589618e-05,
+      "loss": 2.3889,
+      "step": 1655
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8672399520874023,
+      "learning_rate": 9.996253353086866e-05,
+      "loss": 2.3434,
+      "step": 1660
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8865512609481812,
+      "learning_rate": 9.996229300654729e-05,
+      "loss": 2.3529,
+      "step": 1665
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9091736674308777,
+      "learning_rate": 9.996205171293576e-05,
+      "loss": 2.3181,
+      "step": 1670
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8516930937767029,
+      "learning_rate": 9.996180965003778e-05,
+      "loss": 2.3538,
+      "step": 1675
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8434582352638245,
+      "learning_rate": 9.996156681785708e-05,
+      "loss": 2.2755,
+      "step": 1680
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9584338068962097,
+      "learning_rate": 9.996132321639741e-05,
+      "loss": 2.3937,
+      "step": 1685
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9319642186164856,
+      "learning_rate": 9.996107884566252e-05,
+      "loss": 2.3859,
+      "step": 1690
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9255611300468445,
+      "learning_rate": 9.996083370565616e-05,
+      "loss": 2.2856,
+      "step": 1695
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9114521741867065,
+      "learning_rate": 9.99605877963821e-05,
+      "loss": 2.3893,
+      "step": 1700
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8871684074401855,
+      "learning_rate": 9.996034111784413e-05,
+      "loss": 2.3134,
+      "step": 1705
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8890218734741211,
+      "learning_rate": 9.996009367004607e-05,
+      "loss": 2.3891,
+      "step": 1710
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8816906213760376,
+      "learning_rate": 9.995984545299171e-05,
+      "loss": 2.413,
+      "step": 1715
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9005905389785767,
+      "learning_rate": 9.995959646668488e-05,
+      "loss": 2.3572,
+      "step": 1720
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.196043610572815,
+      "learning_rate": 9.995934671112941e-05,
+      "loss": 2.3063,
+      "step": 1725
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8865216970443726,
+      "learning_rate": 9.995909618632915e-05,
+      "loss": 2.3993,
+      "step": 1730
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8638002872467041,
+      "learning_rate": 9.995884489228794e-05,
+      "loss": 2.2831,
+      "step": 1735
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8383564949035645,
+      "learning_rate": 9.995859282900967e-05,
+      "loss": 2.3231,
+      "step": 1740
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8859516382217407,
+      "learning_rate": 9.995833999649821e-05,
+      "loss": 2.3404,
+      "step": 1745
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8711836338043213,
+      "learning_rate": 9.995808639475746e-05,
+      "loss": 2.3636,
+      "step": 1750
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8318776488304138,
+      "learning_rate": 9.995783202379131e-05,
+      "loss": 2.2478,
+      "step": 1755
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.928284227848053,
+      "learning_rate": 9.99575768836037e-05,
+      "loss": 2.3557,
+      "step": 1760
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8805647492408752,
+      "learning_rate": 9.995732097419855e-05,
+      "loss": 2.4188,
+      "step": 1765
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9802470803260803,
+      "learning_rate": 9.995706429557978e-05,
+      "loss": 2.3085,
+      "step": 1770
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8908125162124634,
+      "learning_rate": 9.995680684775137e-05,
+      "loss": 2.3487,
+      "step": 1775
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8908265829086304,
+      "learning_rate": 9.995654863071729e-05,
+      "loss": 2.3833,
+      "step": 1780
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9045591950416565,
+      "learning_rate": 9.995628964448147e-05,
+      "loss": 2.4213,
+      "step": 1785
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8972904086112976,
+      "learning_rate": 9.995602988904793e-05,
+      "loss": 2.3666,
+      "step": 1790
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8666526079177856,
+      "learning_rate": 9.995576936442068e-05,
+      "loss": 2.3501,
+      "step": 1795
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8966271281242371,
+      "learning_rate": 9.995550807060371e-05,
+      "loss": 2.3306,
+      "step": 1800
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.897575855255127,
+      "learning_rate": 9.995524600760105e-05,
+      "loss": 2.2659,
+      "step": 1805
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9191930294036865,
+      "learning_rate": 9.995498317541674e-05,
+      "loss": 2.4903,
+      "step": 1810
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8635687232017517,
+      "learning_rate": 9.995471957405483e-05,
+      "loss": 2.3546,
+      "step": 1815
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9872962832450867,
+      "learning_rate": 9.995445520351935e-05,
+      "loss": 2.4088,
+      "step": 1820
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8429505228996277,
+      "learning_rate": 9.995419006381442e-05,
+      "loss": 2.3235,
+      "step": 1825
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8187404870986938,
+      "learning_rate": 9.995392415494407e-05,
+      "loss": 2.2618,
+      "step": 1830
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8425337076187134,
+      "learning_rate": 9.995365747691243e-05,
+      "loss": 2.3128,
+      "step": 1835
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8289220333099365,
+      "learning_rate": 9.995339002972359e-05,
+      "loss": 2.2759,
+      "step": 1840
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8320723176002502,
+      "learning_rate": 9.995312181338168e-05,
+      "loss": 2.3006,
+      "step": 1845
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8397169709205627,
+      "learning_rate": 9.995285282789082e-05,
+      "loss": 2.3272,
+      "step": 1850
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0048141479492188,
+      "learning_rate": 9.995258307325515e-05,
+      "loss": 2.359,
+      "step": 1855
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.871519923210144,
+      "learning_rate": 9.995231254947883e-05,
+      "loss": 2.4668,
+      "step": 1860
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8374744653701782,
+      "learning_rate": 9.995204125656603e-05,
+      "loss": 2.2553,
+      "step": 1865
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8343096971511841,
+      "learning_rate": 9.995176919452092e-05,
+      "loss": 2.3257,
+      "step": 1870
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8547272086143494,
+      "learning_rate": 9.995149636334769e-05,
+      "loss": 2.269,
+      "step": 1875
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8535609841346741,
+      "learning_rate": 9.995122276305054e-05,
+      "loss": 2.366,
+      "step": 1880
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8598992228507996,
+      "learning_rate": 9.995094839363368e-05,
+      "loss": 2.3656,
+      "step": 1885
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8206318616867065,
+      "learning_rate": 9.995067325510132e-05,
+      "loss": 2.2898,
+      "step": 1890
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8444651961326599,
+      "learning_rate": 9.995039734745774e-05,
+      "loss": 2.2783,
+      "step": 1895
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8685657978057861,
+      "learning_rate": 9.995012067070715e-05,
+      "loss": 2.3033,
+      "step": 1900
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.90837562084198,
+      "learning_rate": 9.994984322485383e-05,
+      "loss": 2.4353,
+      "step": 1905
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8722414970397949,
+      "learning_rate": 9.994956500990205e-05,
+      "loss": 2.2394,
+      "step": 1910
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9344310760498047,
+      "learning_rate": 9.994928602585608e-05,
+      "loss": 2.3639,
+      "step": 1915
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8679055571556091,
+      "learning_rate": 9.994900627272023e-05,
+      "loss": 2.3595,
+      "step": 1920
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8151097893714905,
+      "learning_rate": 9.99487257504988e-05,
+      "loss": 2.2945,
+      "step": 1925
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.845343828201294,
+      "learning_rate": 9.994844445919612e-05,
+      "loss": 2.3663,
+      "step": 1930
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8670045137405396,
+      "learning_rate": 9.994816239881648e-05,
+      "loss": 2.2881,
+      "step": 1935
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8487751483917236,
+      "learning_rate": 9.99478795693643e-05,
+      "loss": 2.349,
+      "step": 1940
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9012033343315125,
+      "learning_rate": 9.994759597084384e-05,
+      "loss": 2.3283,
+      "step": 1945
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.9928873777389526,
+      "learning_rate": 9.994731160325955e-05,
+      "loss": 2.4312,
+      "step": 1950
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8887792229652405,
+      "learning_rate": 9.994702646661577e-05,
+      "loss": 2.33,
+      "step": 1955
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8843797445297241,
+      "learning_rate": 9.994674056091689e-05,
+      "loss": 2.3083,
+      "step": 1960
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9022790193557739,
+      "learning_rate": 9.994645388616732e-05,
+      "loss": 2.3563,
+      "step": 1965
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9043310284614563,
+      "learning_rate": 9.994616644237148e-05,
+      "loss": 2.3488,
+      "step": 1970
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9122346639633179,
+      "learning_rate": 9.994587822953378e-05,
+      "loss": 2.3276,
+      "step": 1975
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8918317556381226,
+      "learning_rate": 9.994558924765866e-05,
+      "loss": 2.3935,
+      "step": 1980
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8877593278884888,
+      "learning_rate": 9.99452994967506e-05,
+      "loss": 2.3241,
+      "step": 1985
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.930396318435669,
+      "learning_rate": 9.994500897681401e-05,
+      "loss": 2.3767,
+      "step": 1990
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.866046130657196,
+      "learning_rate": 9.994471768785339e-05,
+      "loss": 2.3039,
+      "step": 1995
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8440557718276978,
+      "learning_rate": 9.994442562987324e-05,
+      "loss": 2.3333,
+      "step": 2000
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9192540645599365,
+      "learning_rate": 9.994413280287803e-05,
+      "loss": 2.2898,
+      "step": 2005
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8495872616767883,
+      "learning_rate": 9.994383920687229e-05,
+      "loss": 2.3459,
+      "step": 2010
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8004748821258545,
+      "learning_rate": 9.994354484186053e-05,
+      "loss": 2.2632,
+      "step": 2015
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8296611905097961,
+      "learning_rate": 9.994324970784727e-05,
+      "loss": 2.2513,
+      "step": 2020
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.400059461593628,
+      "learning_rate": 9.994295380483709e-05,
+      "loss": 2.3631,
+      "step": 2025
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8156818151473999,
+      "learning_rate": 9.99426571328345e-05,
+      "loss": 2.2694,
+      "step": 2030
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.0410442352294922,
+      "learning_rate": 9.994235969184411e-05,
+      "loss": 2.2858,
+      "step": 2035
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.1258503198623657,
+      "learning_rate": 9.994206148187046e-05,
+      "loss": 2.3036,
+      "step": 2040
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8512048125267029,
+      "learning_rate": 9.994176250291818e-05,
+      "loss": 2.2902,
+      "step": 2045
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.1432678699493408,
+      "learning_rate": 9.994146275499186e-05,
+      "loss": 2.3792,
+      "step": 2050
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8748124241828918,
+      "learning_rate": 9.994116223809611e-05,
+      "loss": 2.2246,
+      "step": 2055
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.812293529510498,
+      "learning_rate": 9.994086095223557e-05,
+      "loss": 2.3156,
+      "step": 2060
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8857740759849548,
+      "learning_rate": 9.994055889741485e-05,
+      "loss": 2.267,
+      "step": 2065
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8101369142532349,
+      "learning_rate": 9.994025607363862e-05,
+      "loss": 2.3441,
+      "step": 2070
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8808976411819458,
+      "learning_rate": 9.993995248091155e-05,
+      "loss": 2.3833,
+      "step": 2075
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8417057394981384,
+      "learning_rate": 9.993964811923831e-05,
+      "loss": 2.2536,
+      "step": 2080
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8599678874015808,
+      "learning_rate": 9.993934298862358e-05,
+      "loss": 2.3639,
+      "step": 2085
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8209399580955505,
+      "learning_rate": 9.993903708907205e-05,
+      "loss": 2.3666,
+      "step": 2090
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8604609370231628,
+      "learning_rate": 9.993873042058845e-05,
+      "loss": 2.292,
+      "step": 2095
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8243845701217651,
+      "learning_rate": 9.993842298317749e-05,
+      "loss": 2.3765,
+      "step": 2100
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8566551804542542,
+      "learning_rate": 9.993811477684392e-05,
+      "loss": 2.3135,
+      "step": 2105
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8709605932235718,
+      "learning_rate": 9.993780580159246e-05,
+      "loss": 2.3159,
+      "step": 2110
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8219949007034302,
+      "learning_rate": 9.993749605742788e-05,
+      "loss": 2.2951,
+      "step": 2115
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8679947853088379,
+      "learning_rate": 9.993718554435495e-05,
+      "loss": 2.3384,
+      "step": 2120
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8880906701087952,
+      "learning_rate": 9.993687426237846e-05,
+      "loss": 2.3425,
+      "step": 2125
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8346153497695923,
+      "learning_rate": 9.993656221150318e-05,
+      "loss": 2.3219,
+      "step": 2130
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8416732549667358,
+      "learning_rate": 9.993624939173394e-05,
+      "loss": 2.3155,
+      "step": 2135
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8757607936859131,
+      "learning_rate": 9.993593580307553e-05,
+      "loss": 2.3201,
+      "step": 2140
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9076690673828125,
+      "learning_rate": 9.99356214455328e-05,
+      "loss": 2.2905,
+      "step": 2145
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8474802374839783,
+      "learning_rate": 9.993530631911059e-05,
+      "loss": 2.2839,
+      "step": 2150
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9619282484054565,
+      "learning_rate": 9.993499042381373e-05,
+      "loss": 2.2312,
+      "step": 2155
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8291295766830444,
+      "learning_rate": 9.993467375964711e-05,
+      "loss": 2.333,
+      "step": 2160
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8207452297210693,
+      "learning_rate": 9.99343563266156e-05,
+      "loss": 2.3016,
+      "step": 2165
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8051577806472778,
+      "learning_rate": 9.993403812472408e-05,
+      "loss": 2.2785,
+      "step": 2170
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8279198408126831,
+      "learning_rate": 9.993371915397745e-05,
+      "loss": 2.2124,
+      "step": 2175
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8633261919021606,
+      "learning_rate": 9.993339941438062e-05,
+      "loss": 2.3205,
+      "step": 2180
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8566678166389465,
+      "learning_rate": 9.993307890593853e-05,
+      "loss": 2.365,
+      "step": 2185
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8351194858551025,
+      "learning_rate": 9.993275762865609e-05,
+      "loss": 2.2302,
+      "step": 2190
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.929428219795227,
+      "learning_rate": 9.993243558253826e-05,
+      "loss": 2.3459,
+      "step": 2195
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8406592011451721,
+      "learning_rate": 9.993211276759e-05,
+      "loss": 2.2434,
+      "step": 2200
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9025505185127258,
+      "learning_rate": 9.993178918381628e-05,
+      "loss": 2.2704,
+      "step": 2205
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8673206567764282,
+      "learning_rate": 9.993146483122207e-05,
+      "loss": 2.2624,
+      "step": 2210
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8158714771270752,
+      "learning_rate": 9.993113970981238e-05,
+      "loss": 2.2947,
+      "step": 2215
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8454875946044922,
+      "learning_rate": 9.993081381959221e-05,
+      "loss": 2.2843,
+      "step": 2220
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8135572075843811,
+      "learning_rate": 9.99304871605666e-05,
+      "loss": 2.34,
+      "step": 2225
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8534564971923828,
+      "learning_rate": 9.993015973274053e-05,
+      "loss": 2.2564,
+      "step": 2230
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8532238602638245,
+      "learning_rate": 9.992983153611908e-05,
+      "loss": 2.3687,
+      "step": 2235
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8580873608589172,
+      "learning_rate": 9.99295025707073e-05,
+      "loss": 2.2579,
+      "step": 2240
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8617312908172607,
+      "learning_rate": 9.992917283651024e-05,
+      "loss": 2.3325,
+      "step": 2245
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8575713634490967,
+      "learning_rate": 9.992884233353298e-05,
+      "loss": 2.3637,
+      "step": 2250
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8337883949279785,
+      "learning_rate": 9.992851106178063e-05,
+      "loss": 2.3338,
+      "step": 2255
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8462134599685669,
+      "learning_rate": 9.992817902125826e-05,
+      "loss": 2.3401,
+      "step": 2260
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8727317452430725,
+      "learning_rate": 9.9927846211971e-05,
+      "loss": 2.3479,
+      "step": 2265
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8276302814483643,
+      "learning_rate": 9.992751263392397e-05,
+      "loss": 2.2906,
+      "step": 2270
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7732294797897339,
+      "learning_rate": 9.992717828712232e-05,
+      "loss": 2.328,
+      "step": 2275
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8584631681442261,
+      "learning_rate": 9.992684317157117e-05,
+      "loss": 2.2645,
+      "step": 2280
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8632560968399048,
+      "learning_rate": 9.992650728727571e-05,
+      "loss": 2.2294,
+      "step": 2285
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8182496428489685,
+      "learning_rate": 9.99261706342411e-05,
+      "loss": 2.3341,
+      "step": 2290
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8008748888969421,
+      "learning_rate": 9.99258332124725e-05,
+      "loss": 2.3168,
+      "step": 2295
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8059354424476624,
+      "learning_rate": 9.992549502197514e-05,
+      "loss": 2.2787,
+      "step": 2300
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7874477505683899,
+      "learning_rate": 9.992515606275423e-05,
+      "loss": 2.2365,
+      "step": 2305
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8032822608947754,
+      "learning_rate": 9.992481633481495e-05,
+      "loss": 2.2618,
+      "step": 2310
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8162176609039307,
+      "learning_rate": 9.992447583816257e-05,
+      "loss": 2.3203,
+      "step": 2315
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8731299638748169,
+      "learning_rate": 9.992413457280231e-05,
+      "loss": 2.3718,
+      "step": 2320
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8222412467002869,
+      "learning_rate": 9.992379253873943e-05,
+      "loss": 2.4746,
+      "step": 2325
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.845116913318634,
+      "learning_rate": 9.99234497359792e-05,
+      "loss": 2.272,
+      "step": 2330
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8091548681259155,
+      "learning_rate": 9.992310616452689e-05,
+      "loss": 2.3423,
+      "step": 2335
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.780876100063324,
+      "learning_rate": 9.99227618243878e-05,
+      "loss": 2.2751,
+      "step": 2340
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7881205677986145,
+      "learning_rate": 9.992241671556724e-05,
+      "loss": 2.3766,
+      "step": 2345
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8151964545249939,
+      "learning_rate": 9.99220708380705e-05,
+      "loss": 2.277,
+      "step": 2350
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8407646417617798,
+      "learning_rate": 9.99217241919029e-05,
+      "loss": 2.2392,
+      "step": 2355
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8737926483154297,
+      "learning_rate": 9.992137677706982e-05,
+      "loss": 2.243,
+      "step": 2360
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8274579644203186,
+      "learning_rate": 9.99210285935766e-05,
+      "loss": 2.1786,
+      "step": 2365
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8261242508888245,
+      "learning_rate": 9.992067964142856e-05,
+      "loss": 2.3601,
+      "step": 2370
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.526295781135559,
+      "learning_rate": 9.992032992063109e-05,
+      "loss": 2.2648,
+      "step": 2375
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8177210092544556,
+      "learning_rate": 9.99199794311896e-05,
+      "loss": 2.3406,
+      "step": 2380
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7912571430206299,
+      "learning_rate": 9.991962817310947e-05,
+      "loss": 2.2192,
+      "step": 2385
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8298333287239075,
+      "learning_rate": 9.991927614639612e-05,
+      "loss": 2.3671,
+      "step": 2390
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8209797739982605,
+      "learning_rate": 9.991892335105493e-05,
+      "loss": 2.2852,
+      "step": 2395
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8209306001663208,
+      "learning_rate": 9.991856978709137e-05,
+      "loss": 2.2801,
+      "step": 2400
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.780633807182312,
+      "learning_rate": 9.99182154545109e-05,
+      "loss": 2.3464,
+      "step": 2405
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8129504323005676,
+      "learning_rate": 9.991786035331893e-05,
+      "loss": 2.1754,
+      "step": 2410
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.84320467710495,
+      "learning_rate": 9.991750448352096e-05,
+      "loss": 2.2367,
+      "step": 2415
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8007714748382568,
+      "learning_rate": 9.991714784512243e-05,
+      "loss": 2.2758,
+      "step": 2420
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8000234365463257,
+      "learning_rate": 9.991679043812887e-05,
+      "loss": 2.2287,
+      "step": 2425
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8236909508705139,
+      "learning_rate": 9.99164322625458e-05,
+      "loss": 2.271,
+      "step": 2430
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8246254324913025,
+      "learning_rate": 9.991607331837867e-05,
+      "loss": 2.2803,
+      "step": 2435
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8431694507598877,
+      "learning_rate": 9.991571360563306e-05,
+      "loss": 2.2784,
+      "step": 2440
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8336653709411621,
+      "learning_rate": 9.99153531243145e-05,
+      "loss": 2.2698,
+      "step": 2445
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7936285734176636,
+      "learning_rate": 9.991499187442852e-05,
+      "loss": 2.2273,
+      "step": 2450
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.817367434501648,
+      "learning_rate": 9.99146298559807e-05,
+      "loss": 2.289,
+      "step": 2455
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8004292249679565,
+      "learning_rate": 9.991426706897661e-05,
+      "loss": 2.3296,
+      "step": 2460
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.827859103679657,
+      "learning_rate": 9.991390351342185e-05,
+      "loss": 2.321,
+      "step": 2465
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8595304489135742,
+      "learning_rate": 9.991353918932198e-05,
+      "loss": 2.268,
+      "step": 2470
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7970768809318542,
+      "learning_rate": 9.991317409668264e-05,
+      "loss": 2.3562,
+      "step": 2475
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7910333871841431,
+      "learning_rate": 9.991280823550945e-05,
+      "loss": 2.3061,
+      "step": 2480
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8261690735816956,
+      "learning_rate": 9.991244160580803e-05,
+      "loss": 2.2656,
+      "step": 2485
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8216850161552429,
+      "learning_rate": 9.991207420758404e-05,
+      "loss": 2.2904,
+      "step": 2490
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8099186420440674,
+      "learning_rate": 9.991170604084314e-05,
+      "loss": 2.2554,
+      "step": 2495
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8381199836730957,
+      "learning_rate": 9.991133710559097e-05,
+      "loss": 2.3241,
+      "step": 2500
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7938763499259949,
+      "learning_rate": 9.991096740183324e-05,
+      "loss": 2.1962,
+      "step": 2505
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8465071320533752,
+      "learning_rate": 9.991059692957563e-05,
+      "loss": 2.2658,
+      "step": 2510
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8084853887557983,
+      "learning_rate": 9.991022568882384e-05,
+      "loss": 2.3321,
+      "step": 2515
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7942716479301453,
+      "learning_rate": 9.99098536795836e-05,
+      "loss": 2.2557,
+      "step": 2520
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7877028584480286,
+      "learning_rate": 9.990948090186063e-05,
+      "loss": 2.1841,
+      "step": 2525
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8037130832672119,
+      "learning_rate": 9.990910735566065e-05,
+      "loss": 2.2726,
+      "step": 2530
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7703967690467834,
+      "learning_rate": 9.990873304098946e-05,
+      "loss": 2.2236,
+      "step": 2535
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9438284039497375,
+      "learning_rate": 9.990835795785278e-05,
+      "loss": 2.1883,
+      "step": 2540
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7936575412750244,
+      "learning_rate": 9.990798210625642e-05,
+      "loss": 2.2094,
+      "step": 2545
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.775001585483551,
+      "learning_rate": 9.990760548620613e-05,
+      "loss": 2.3506,
+      "step": 2550
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.824762225151062,
+      "learning_rate": 9.990722809770773e-05,
+      "loss": 2.2906,
+      "step": 2555
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8164494037628174,
+      "learning_rate": 9.990684994076702e-05,
+      "loss": 2.2605,
+      "step": 2560
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8562635183334351,
+      "learning_rate": 9.990647101538983e-05,
+      "loss": 2.2908,
+      "step": 2565
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8207958340644836,
+      "learning_rate": 9.990609132158201e-05,
+      "loss": 2.1955,
+      "step": 2570
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8010104298591614,
+      "learning_rate": 9.990571085934938e-05,
+      "loss": 2.2179,
+      "step": 2575
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8539251685142517,
+      "learning_rate": 9.990532962869783e-05,
+      "loss": 2.3503,
+      "step": 2580
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8188031911849976,
+      "learning_rate": 9.990494762963318e-05,
+      "loss": 2.2657,
+      "step": 2585
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7854506373405457,
+      "learning_rate": 9.990456486216136e-05,
+      "loss": 2.2736,
+      "step": 2590
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7967750430107117,
+      "learning_rate": 9.990418132628824e-05,
+      "loss": 2.2286,
+      "step": 2595
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7982003688812256,
+      "learning_rate": 9.990379702201974e-05,
+      "loss": 2.3146,
+      "step": 2600
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7923963665962219,
+      "learning_rate": 9.990341194936173e-05,
+      "loss": 2.2377,
+      "step": 2605
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.0643680095672607,
+      "learning_rate": 9.990302610832021e-05,
+      "loss": 2.2354,
+      "step": 2610
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8070834875106812,
+      "learning_rate": 9.990263949890108e-05,
+      "loss": 2.3313,
+      "step": 2615
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.841102659702301,
+      "learning_rate": 9.990225212111029e-05,
+      "loss": 2.2734,
+      "step": 2620
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.820033848285675,
+      "learning_rate": 9.990186397495383e-05,
+      "loss": 2.3702,
+      "step": 2625
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7915102243423462,
+      "learning_rate": 9.990147506043766e-05,
+      "loss": 2.3064,
+      "step": 2630
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7722065448760986,
+      "learning_rate": 9.990108537756775e-05,
+      "loss": 2.3332,
+      "step": 2635
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9008132219314575,
+      "learning_rate": 9.990069492635015e-05,
+      "loss": 2.3351,
+      "step": 2640
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8250614404678345,
+      "learning_rate": 9.990030370679082e-05,
+      "loss": 2.2763,
+      "step": 2645
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7958188056945801,
+      "learning_rate": 9.989991171889581e-05,
+      "loss": 2.2712,
+      "step": 2650
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8424177765846252,
+      "learning_rate": 9.989951896267115e-05,
+      "loss": 2.2136,
+      "step": 2655
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7857699990272522,
+      "learning_rate": 9.989912543812289e-05,
+      "loss": 2.3042,
+      "step": 2660
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8058030009269714,
+      "learning_rate": 9.989873114525709e-05,
+      "loss": 2.2178,
+      "step": 2665
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8010798096656799,
+      "learning_rate": 9.989833608407981e-05,
+      "loss": 2.389,
+      "step": 2670
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8370392322540283,
+      "learning_rate": 9.989794025459715e-05,
+      "loss": 2.3176,
+      "step": 2675
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8398842215538025,
+      "learning_rate": 9.989754365681519e-05,
+      "loss": 2.2286,
+      "step": 2680
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7836652398109436,
+      "learning_rate": 9.989714629074006e-05,
+      "loss": 2.2143,
+      "step": 2685
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7818453907966614,
+      "learning_rate": 9.989674815637784e-05,
+      "loss": 2.2741,
+      "step": 2690
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.0922948122024536,
+      "learning_rate": 9.989634925373468e-05,
+      "loss": 2.2101,
+      "step": 2695
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8325046896934509,
+      "learning_rate": 9.989594958281673e-05,
+      "loss": 2.2935,
+      "step": 2700
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.789572536945343,
+      "learning_rate": 9.989554914363015e-05,
+      "loss": 2.172,
+      "step": 2705
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7880695462226868,
+      "learning_rate": 9.989514793618109e-05,
+      "loss": 2.274,
+      "step": 2710
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7893670797348022,
+      "learning_rate": 9.989474596047572e-05,
+      "loss": 2.2899,
+      "step": 2715
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7785813808441162,
+      "learning_rate": 9.989434321652024e-05,
+      "loss": 2.248,
+      "step": 2720
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7924873232841492,
+      "learning_rate": 9.989393970432085e-05,
+      "loss": 2.2716,
+      "step": 2725
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.1108251810073853,
+      "learning_rate": 9.989353542388378e-05,
+      "loss": 2.3403,
+      "step": 2730
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.823577344417572,
+      "learning_rate": 9.989313037521524e-05,
+      "loss": 2.2802,
+      "step": 2735
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8067580461502075,
+      "learning_rate": 9.989272455832145e-05,
+      "loss": 2.218,
+      "step": 2740
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8310951590538025,
+      "learning_rate": 9.989231797320869e-05,
+      "loss": 2.4018,
+      "step": 2745
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7966132164001465,
+      "learning_rate": 9.989191061988319e-05,
+      "loss": 2.2987,
+      "step": 2750
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8315147161483765,
+      "learning_rate": 9.989150249835125e-05,
+      "loss": 2.3248,
+      "step": 2755
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7670814394950867,
+      "learning_rate": 9.989109360861913e-05,
+      "loss": 2.3181,
+      "step": 2760
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7536638379096985,
+      "learning_rate": 9.989068395069313e-05,
+      "loss": 2.1507,
+      "step": 2765
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8170732259750366,
+      "learning_rate": 9.989027352457958e-05,
+      "loss": 2.1863,
+      "step": 2770
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8143478631973267,
+      "learning_rate": 9.988986233028478e-05,
+      "loss": 2.3212,
+      "step": 2775
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.825748860836029,
+      "learning_rate": 9.988945036781507e-05,
+      "loss": 2.268,
+      "step": 2780
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7642278671264648,
+      "learning_rate": 9.988903763717679e-05,
+      "loss": 2.3349,
+      "step": 2785
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8048906326293945,
+      "learning_rate": 9.988862413837628e-05,
+      "loss": 2.1889,
+      "step": 2790
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.852315366268158,
+      "learning_rate": 9.988820987141993e-05,
+      "loss": 2.328,
+      "step": 2795
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8227264881134033,
+      "learning_rate": 9.988779483631409e-05,
+      "loss": 2.2805,
+      "step": 2800
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9187013506889343,
+      "learning_rate": 9.98873790330652e-05,
+      "loss": 2.2808,
+      "step": 2805
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8761085867881775,
+      "learning_rate": 9.988696246167961e-05,
+      "loss": 2.3223,
+      "step": 2810
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7833371162414551,
+      "learning_rate": 9.988654512216378e-05,
+      "loss": 2.185,
+      "step": 2815
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7843089699745178,
+      "learning_rate": 9.988612701452409e-05,
+      "loss": 2.282,
+      "step": 2820
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8061702251434326,
+      "learning_rate": 9.988570813876701e-05,
+      "loss": 2.3109,
+      "step": 2825
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8324564695358276,
+      "learning_rate": 9.988528849489897e-05,
+      "loss": 2.2572,
+      "step": 2830
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8403586149215698,
+      "learning_rate": 9.988486808292646e-05,
+      "loss": 2.2328,
+      "step": 2835
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7963917851448059,
+      "learning_rate": 9.988444690285592e-05,
+      "loss": 2.2712,
+      "step": 2840
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8057711124420166,
+      "learning_rate": 9.988402495469385e-05,
+      "loss": 2.2066,
+      "step": 2845
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7924182415008545,
+      "learning_rate": 9.988360223844673e-05,
+      "loss": 2.3149,
+      "step": 2850
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7841897010803223,
+      "learning_rate": 9.98831787541211e-05,
+      "loss": 2.3088,
+      "step": 2855
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8609781265258789,
+      "learning_rate": 9.988275450172346e-05,
+      "loss": 2.2234,
+      "step": 2860
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7985591888427734,
+      "learning_rate": 9.988232948126037e-05,
+      "loss": 2.3678,
+      "step": 2865
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8035890460014343,
+      "learning_rate": 9.988190369273833e-05,
+      "loss": 2.291,
+      "step": 2870
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8113985657691956,
+      "learning_rate": 9.988147713616391e-05,
+      "loss": 2.2554,
+      "step": 2875
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8097103834152222,
+      "learning_rate": 9.988104981154371e-05,
+      "loss": 2.2727,
+      "step": 2880
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7664754986763,
+      "learning_rate": 9.988062171888429e-05,
+      "loss": 2.2988,
+      "step": 2885
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9402970671653748,
+      "learning_rate": 9.988019285819221e-05,
+      "loss": 2.3045,
+      "step": 2890
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7949076890945435,
+      "learning_rate": 9.987976322947411e-05,
+      "loss": 2.2295,
+      "step": 2895
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7814761996269226,
+      "learning_rate": 9.98793328327366e-05,
+      "loss": 2.2537,
+      "step": 2900
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7619650363922119,
+      "learning_rate": 9.98789016679863e-05,
+      "loss": 2.2395,
+      "step": 2905
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8340640664100647,
+      "learning_rate": 9.987846973522985e-05,
+      "loss": 2.3355,
+      "step": 2910
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8053739070892334,
+      "learning_rate": 9.98780370344739e-05,
+      "loss": 2.2696,
+      "step": 2915
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7935998439788818,
+      "learning_rate": 9.987760356572512e-05,
+      "loss": 2.298,
+      "step": 2920
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7855820059776306,
+      "learning_rate": 9.987716932899018e-05,
+      "loss": 2.312,
+      "step": 2925
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7881000638008118,
+      "learning_rate": 9.987673432427576e-05,
+      "loss": 2.2306,
+      "step": 2930
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8305438756942749,
+      "learning_rate": 9.987629855158855e-05,
+      "loss": 2.216,
+      "step": 2935
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7545214891433716,
+      "learning_rate": 9.98758620109353e-05,
+      "loss": 2.2598,
+      "step": 2940
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7717337012290955,
+      "learning_rate": 9.98754247023227e-05,
+      "loss": 2.2676,
+      "step": 2945
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7839546203613281,
+      "learning_rate": 9.987498662575747e-05,
+      "loss": 2.244,
+      "step": 2950
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7877485752105713,
+      "learning_rate": 9.987454778124639e-05,
+      "loss": 2.263,
+      "step": 2955
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7731161713600159,
+      "learning_rate": 9.987410816879619e-05,
+      "loss": 2.3108,
+      "step": 2960
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8078286051750183,
+      "learning_rate": 9.987366778841365e-05,
+      "loss": 2.2229,
+      "step": 2965
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7988601922988892,
+      "learning_rate": 9.987322664010554e-05,
+      "loss": 2.3318,
+      "step": 2970
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7419587969779968,
+      "learning_rate": 9.987278472387867e-05,
+      "loss": 2.2145,
+      "step": 2975
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7820016145706177,
+      "learning_rate": 9.987234203973986e-05,
+      "loss": 2.2917,
+      "step": 2980
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8232306241989136,
+      "learning_rate": 9.987189858769586e-05,
+      "loss": 2.2397,
+      "step": 2985
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7810310125350952,
+      "learning_rate": 9.987145436775357e-05,
+      "loss": 2.2911,
+      "step": 2990
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7752991914749146,
+      "learning_rate": 9.98710093799198e-05,
+      "loss": 2.2516,
+      "step": 2995
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7823570966720581,
+      "learning_rate": 9.987056362420138e-05,
+      "loss": 2.2863,
+      "step": 3000
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8193786144256592,
+      "learning_rate": 9.98701171006052e-05,
+      "loss": 2.244,
+      "step": 3005
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8298470377922058,
+      "learning_rate": 9.986966980913817e-05,
+      "loss": 2.3603,
+      "step": 3010
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7937309145927429,
+      "learning_rate": 9.986922174980709e-05,
+      "loss": 2.1724,
+      "step": 3015
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8074214458465576,
+      "learning_rate": 9.986877292261893e-05,
+      "loss": 2.2534,
+      "step": 3020
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7816885113716125,
+      "learning_rate": 9.986832332758058e-05,
+      "loss": 2.216,
+      "step": 3025
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7590488195419312,
+      "learning_rate": 9.986787296469895e-05,
+      "loss": 2.3154,
+      "step": 3030
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7819816470146179,
+      "learning_rate": 9.986742183398099e-05,
+      "loss": 2.2801,
+      "step": 3035
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8010467886924744,
+      "learning_rate": 9.986696993543366e-05,
+      "loss": 2.2478,
+      "step": 3040
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7891252040863037,
+      "learning_rate": 9.986651726906387e-05,
+      "loss": 2.2389,
+      "step": 3045
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7613987326622009,
+      "learning_rate": 9.986606383487862e-05,
+      "loss": 2.1539,
+      "step": 3050
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7475893497467041,
+      "learning_rate": 9.98656096328849e-05,
+      "loss": 2.3225,
+      "step": 3055
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7872675657272339,
+      "learning_rate": 9.98651546630897e-05,
+      "loss": 2.2741,
+      "step": 3060
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7816417813301086,
+      "learning_rate": 9.986469892550001e-05,
+      "loss": 2.1496,
+      "step": 3065
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7538018226623535,
+      "learning_rate": 9.986424242012287e-05,
+      "loss": 2.3071,
+      "step": 3070
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7815521955490112,
+      "learning_rate": 9.986378514696528e-05,
+      "loss": 2.2423,
+      "step": 3075
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7846755385398865,
+      "learning_rate": 9.98633271060343e-05,
+      "loss": 2.2525,
+      "step": 3080
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7789773941040039,
+      "learning_rate": 9.9862868297337e-05,
+      "loss": 2.3,
+      "step": 3085
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7960700392723083,
+      "learning_rate": 9.98624087208804e-05,
+      "loss": 2.3114,
+      "step": 3090
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8385229706764221,
+      "learning_rate": 9.98619483766716e-05,
+      "loss": 2.292,
+      "step": 3095
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7653298377990723,
+      "learning_rate": 9.98614872647177e-05,
+      "loss": 2.2651,
+      "step": 3100
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7650991678237915,
+      "learning_rate": 9.986102538502579e-05,
+      "loss": 2.3293,
+      "step": 3105
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7574900984764099,
+      "learning_rate": 9.986056273760297e-05,
+      "loss": 2.2519,
+      "step": 3110
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7572404146194458,
+      "learning_rate": 9.986009932245637e-05,
+      "loss": 2.2711,
+      "step": 3115
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7852334976196289,
+      "learning_rate": 9.985963513959313e-05,
+      "loss": 2.2171,
+      "step": 3120
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7899317145347595,
+      "learning_rate": 9.985917018902041e-05,
+      "loss": 2.2474,
+      "step": 3125
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8059554100036621,
+      "learning_rate": 9.985870447074535e-05,
+      "loss": 2.2828,
+      "step": 3130
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7786118984222412,
+      "learning_rate": 9.985823798477512e-05,
+      "loss": 2.2882,
+      "step": 3135
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7965154647827148,
+      "learning_rate": 9.985777073111691e-05,
+      "loss": 2.2776,
+      "step": 3140
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7785694599151611,
+      "learning_rate": 9.985730270977791e-05,
+      "loss": 2.2168,
+      "step": 3145
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8186535239219666,
+      "learning_rate": 9.985683392076534e-05,
+      "loss": 2.2973,
+      "step": 3150
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7959901690483093,
+      "learning_rate": 9.98563643640864e-05,
+      "loss": 2.2776,
+      "step": 3155
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8005150556564331,
+      "learning_rate": 9.985589403974833e-05,
+      "loss": 2.2982,
+      "step": 3160
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7597604393959045,
+      "learning_rate": 9.985542294775838e-05,
+      "loss": 2.239,
+      "step": 3165
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7775261402130127,
+      "learning_rate": 9.985495108812377e-05,
+      "loss": 2.2397,
+      "step": 3170
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7719820737838745,
+      "learning_rate": 9.985447846085181e-05,
+      "loss": 2.207,
+      "step": 3175
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7705333232879639,
+      "learning_rate": 9.985400506594974e-05,
+      "loss": 2.1883,
+      "step": 3180
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7781678438186646,
+      "learning_rate": 9.985353090342487e-05,
+      "loss": 2.2988,
+      "step": 3185
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7830728888511658,
+      "learning_rate": 9.985305597328449e-05,
+      "loss": 2.2043,
+      "step": 3190
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7484901547431946,
+      "learning_rate": 9.985258027553593e-05,
+      "loss": 2.2068,
+      "step": 3195
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7883464694023132,
+      "learning_rate": 9.98521038101865e-05,
+      "loss": 2.2344,
+      "step": 3200
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7928439378738403,
+      "learning_rate": 9.985162657724354e-05,
+      "loss": 2.187,
+      "step": 3205
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7933062314987183,
+      "learning_rate": 9.98511485767144e-05,
+      "loss": 2.2176,
+      "step": 3210
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7998464703559875,
+      "learning_rate": 9.985066980860642e-05,
+      "loss": 2.3402,
+      "step": 3215
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7712178230285645,
+      "learning_rate": 9.9850190272927e-05,
+      "loss": 2.2215,
+      "step": 3220
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7754117846488953,
+      "learning_rate": 9.984970996968352e-05,
+      "loss": 2.1784,
+      "step": 3225
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8206325173377991,
+      "learning_rate": 9.984922889888335e-05,
+      "loss": 2.3218,
+      "step": 3230
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7611032724380493,
+      "learning_rate": 9.984874706053393e-05,
+      "loss": 2.3104,
+      "step": 3235
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7420569062232971,
+      "learning_rate": 9.984826445464266e-05,
+      "loss": 2.2426,
+      "step": 3240
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7493786811828613,
+      "learning_rate": 9.984778108121698e-05,
+      "loss": 2.2349,
+      "step": 3245
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.782757043838501,
+      "learning_rate": 9.984729694026431e-05,
+      "loss": 2.3479,
+      "step": 3250
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7833341360092163,
+      "learning_rate": 9.984681203179214e-05,
+      "loss": 2.2467,
+      "step": 3255
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7433379888534546,
+      "learning_rate": 9.984632635580792e-05,
+      "loss": 2.1899,
+      "step": 3260
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7744027376174927,
+      "learning_rate": 9.984583991231913e-05,
+      "loss": 2.2238,
+      "step": 3265
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7829711437225342,
+      "learning_rate": 9.984535270133324e-05,
+      "loss": 2.3362,
+      "step": 3270
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.791251003742218,
+      "learning_rate": 9.984486472285777e-05,
+      "loss": 2.2392,
+      "step": 3275
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7698483467102051,
+      "learning_rate": 9.984437597690024e-05,
+      "loss": 2.2562,
+      "step": 3280
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7463228702545166,
+      "learning_rate": 9.984388646346817e-05,
+      "loss": 2.2256,
+      "step": 3285
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8019618391990662,
+      "learning_rate": 9.984339618256909e-05,
+      "loss": 2.2388,
+      "step": 3290
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7785443663597107,
+      "learning_rate": 9.984290513421057e-05,
+      "loss": 2.2414,
+      "step": 3295
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.830256462097168,
+      "learning_rate": 9.984241331840015e-05,
+      "loss": 2.2923,
+      "step": 3300
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7782317996025085,
+      "learning_rate": 9.984192073514538e-05,
+      "loss": 2.2192,
+      "step": 3305
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7628448009490967,
+      "learning_rate": 9.98414273844539e-05,
+      "loss": 2.278,
+      "step": 3310
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7749215960502625,
+      "learning_rate": 9.984093326633325e-05,
+      "loss": 2.3104,
+      "step": 3315
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.792353630065918,
+      "learning_rate": 9.98404383807911e-05,
+      "loss": 2.2161,
+      "step": 3320
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.784152626991272,
+      "learning_rate": 9.983994272783502e-05,
+      "loss": 2.1858,
+      "step": 3325
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7461652755737305,
+      "learning_rate": 9.983944630747266e-05,
+      "loss": 2.2439,
+      "step": 3330
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7818230986595154,
+      "learning_rate": 9.983894911971167e-05,
+      "loss": 2.309,
+      "step": 3335
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8173055648803711,
+      "learning_rate": 9.983845116455969e-05,
+      "loss": 2.2567,
+      "step": 3340
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7806358337402344,
+      "learning_rate": 9.983795244202439e-05,
+      "loss": 2.255,
+      "step": 3345
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7607512474060059,
+      "learning_rate": 9.983745295211347e-05,
+      "loss": 2.2527,
+      "step": 3350
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7981928586959839,
+      "learning_rate": 9.983695269483459e-05,
+      "loss": 2.2624,
+      "step": 3355
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7384070754051208,
+      "learning_rate": 9.983645167019546e-05,
+      "loss": 2.2858,
+      "step": 3360
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7685624957084656,
+      "learning_rate": 9.983594987820381e-05,
+      "loss": 2.2855,
+      "step": 3365
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8180063366889954,
+      "learning_rate": 9.983544731886735e-05,
+      "loss": 2.3476,
+      "step": 3370
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7558609247207642,
+      "learning_rate": 9.983494399219386e-05,
+      "loss": 2.2556,
+      "step": 3375
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.752285361289978,
+      "learning_rate": 9.983443989819103e-05,
+      "loss": 2.2518,
+      "step": 3380
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7406316995620728,
+      "learning_rate": 9.983393503686665e-05,
+      "loss": 2.2274,
+      "step": 3385
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7262222766876221,
+      "learning_rate": 9.983342940822849e-05,
+      "loss": 2.2292,
+      "step": 3390
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7691320776939392,
+      "learning_rate": 9.983292301228432e-05,
+      "loss": 2.2499,
+      "step": 3395
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7926821112632751,
+      "learning_rate": 9.983241584904197e-05,
+      "loss": 2.3299,
+      "step": 3400
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7649868130683899,
+      "learning_rate": 9.983190791850925e-05,
+      "loss": 2.2269,
+      "step": 3405
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7810275554656982,
+      "learning_rate": 9.983139922069394e-05,
+      "loss": 2.2288,
+      "step": 3410
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7733004093170166,
+      "learning_rate": 9.983088975560391e-05,
+      "loss": 2.2697,
+      "step": 3415
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7984127998352051,
+      "learning_rate": 9.983037952324699e-05,
+      "loss": 2.337,
+      "step": 3420
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7867530584335327,
+      "learning_rate": 9.982986852363102e-05,
+      "loss": 2.27,
+      "step": 3425
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7704403400421143,
+      "learning_rate": 9.98293567567639e-05,
+      "loss": 2.239,
+      "step": 3430
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7712245583534241,
+      "learning_rate": 9.982884422265349e-05,
+      "loss": 2.2647,
+      "step": 3435
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7278101444244385,
+      "learning_rate": 9.982833092130768e-05,
+      "loss": 2.2548,
+      "step": 3440
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8156419396400452,
+      "learning_rate": 9.982781685273438e-05,
+      "loss": 2.2152,
+      "step": 3445
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8223443627357483,
+      "learning_rate": 9.982730201694153e-05,
+      "loss": 2.2136,
+      "step": 3450
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8217001557350159,
+      "learning_rate": 9.982678641393701e-05,
+      "loss": 2.3087,
+      "step": 3455
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7830549478530884,
+      "learning_rate": 9.98262700437288e-05,
+      "loss": 2.3242,
+      "step": 3460
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7626489400863647,
+      "learning_rate": 9.982575290632481e-05,
+      "loss": 2.3058,
+      "step": 3465
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7531270980834961,
+      "learning_rate": 9.982523500173304e-05,
+      "loss": 2.2203,
+      "step": 3470
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0710747241973877,
+      "learning_rate": 9.982471632996144e-05,
+      "loss": 2.3252,
+      "step": 3475
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7447677850723267,
+      "learning_rate": 9.9824196891018e-05,
+      "loss": 2.2634,
+      "step": 3480
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7643184661865234,
+      "learning_rate": 9.982367668491075e-05,
+      "loss": 2.2562,
+      "step": 3485
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7577258348464966,
+      "learning_rate": 9.982315571164765e-05,
+      "loss": 2.1823,
+      "step": 3490
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7602296471595764,
+      "learning_rate": 9.982263397123675e-05,
+      "loss": 2.1932,
+      "step": 3495
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7266128659248352,
+      "learning_rate": 9.982211146368608e-05,
+      "loss": 2.2212,
+      "step": 3500
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7796376943588257,
+      "learning_rate": 9.982158818900369e-05,
+      "loss": 2.1652,
+      "step": 3505
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7639287114143372,
+      "learning_rate": 9.982106414719762e-05,
+      "loss": 2.221,
+      "step": 3510
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7654688358306885,
+      "learning_rate": 9.982053933827595e-05,
+      "loss": 2.2607,
+      "step": 3515
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7109196186065674,
+      "learning_rate": 9.982001376224677e-05,
+      "loss": 2.2756,
+      "step": 3520
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7907342314720154,
+      "learning_rate": 9.981948741911816e-05,
+      "loss": 2.2816,
+      "step": 3525
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7449715733528137,
+      "learning_rate": 9.981896030889821e-05,
+      "loss": 2.2252,
+      "step": 3530
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7606565952301025,
+      "learning_rate": 9.981843243159507e-05,
+      "loss": 2.2615,
+      "step": 3535
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7656440734863281,
+      "learning_rate": 9.981790378721684e-05,
+      "loss": 2.251,
+      "step": 3540
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7695711255073547,
+      "learning_rate": 9.98173743757717e-05,
+      "loss": 2.2089,
+      "step": 3545
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7780543565750122,
+      "learning_rate": 9.981684419726775e-05,
+      "loss": 2.2451,
+      "step": 3550
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7883675694465637,
+      "learning_rate": 9.981631325171318e-05,
+      "loss": 2.2775,
+      "step": 3555
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7771449089050293,
+      "learning_rate": 9.981578153911616e-05,
+      "loss": 2.233,
+      "step": 3560
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7928645610809326,
+      "learning_rate": 9.981524905948487e-05,
+      "loss": 2.245,
+      "step": 3565
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7615565061569214,
+      "learning_rate": 9.981471581282753e-05,
+      "loss": 2.2276,
+      "step": 3570
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7577892541885376,
+      "learning_rate": 9.981418179915234e-05,
+      "loss": 2.2243,
+      "step": 3575
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7521548271179199,
+      "learning_rate": 9.981364701846752e-05,
+      "loss": 2.1065,
+      "step": 3580
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7811530828475952,
+      "learning_rate": 9.981311147078128e-05,
+      "loss": 2.2488,
+      "step": 3585
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7479802370071411,
+      "learning_rate": 9.981257515610192e-05,
+      "loss": 2.2849,
+      "step": 3590
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7747140526771545,
+      "learning_rate": 9.981203807443764e-05,
+      "loss": 2.2868,
+      "step": 3595
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.771148681640625,
+      "learning_rate": 9.981150022579676e-05,
+      "loss": 2.2633,
+      "step": 3600
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7430528402328491,
+      "learning_rate": 9.981096161018754e-05,
+      "loss": 2.2196,
+      "step": 3605
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7625300884246826,
+      "learning_rate": 9.981042222761826e-05,
+      "loss": 2.2146,
+      "step": 3610
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7585175633430481,
+      "learning_rate": 9.980988207809726e-05,
+      "loss": 2.2839,
+      "step": 3615
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.757308840751648,
+      "learning_rate": 9.98093411616328e-05,
+      "loss": 2.2727,
+      "step": 3620
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7561479210853577,
+      "learning_rate": 9.980879947823327e-05,
+      "loss": 2.3001,
+      "step": 3625
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7386552095413208,
+      "learning_rate": 9.980825702790696e-05,
+      "loss": 2.189,
+      "step": 3630
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7596527934074402,
+      "learning_rate": 9.980771381066227e-05,
+      "loss": 2.1982,
+      "step": 3635
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7550732493400574,
+      "learning_rate": 9.980716982650753e-05,
+      "loss": 2.286,
+      "step": 3640
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7760612368583679,
+      "learning_rate": 9.980662507545113e-05,
+      "loss": 2.1867,
+      "step": 3645
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.800132691860199,
+      "learning_rate": 9.980607955750146e-05,
+      "loss": 2.2847,
+      "step": 3650
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7580264806747437,
+      "learning_rate": 9.98055332726669e-05,
+      "loss": 2.2757,
+      "step": 3655
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8093569874763489,
+      "learning_rate": 9.980498622095588e-05,
+      "loss": 2.2556,
+      "step": 3660
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7657761573791504,
+      "learning_rate": 9.980443840237681e-05,
+      "loss": 2.2277,
+      "step": 3665
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8395922780036926,
+      "learning_rate": 9.980388981693816e-05,
+      "loss": 2.3037,
+      "step": 3670
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7619218826293945,
+      "learning_rate": 9.980334046464832e-05,
+      "loss": 2.2442,
+      "step": 3675
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7523065209388733,
+      "learning_rate": 9.980279034551579e-05,
+      "loss": 2.2516,
+      "step": 3680
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7730770111083984,
+      "learning_rate": 9.980223945954902e-05,
+      "loss": 2.196,
+      "step": 3685
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.782738208770752,
+      "learning_rate": 9.980168780675651e-05,
+      "loss": 2.2418,
+      "step": 3690
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7697749137878418,
+      "learning_rate": 9.980113538714675e-05,
+      "loss": 2.2477,
+      "step": 3695
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7952249050140381,
+      "learning_rate": 9.980058220072825e-05,
+      "loss": 2.2447,
+      "step": 3700
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7860794067382812,
+      "learning_rate": 9.98000282475095e-05,
+      "loss": 2.2161,
+      "step": 3705
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7611528635025024,
+      "learning_rate": 9.979947352749906e-05,
+      "loss": 2.2347,
+      "step": 3710
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7546845078468323,
+      "learning_rate": 9.979891804070545e-05,
+      "loss": 2.3018,
+      "step": 3715
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7348127961158752,
+      "learning_rate": 9.979836178713724e-05,
+      "loss": 2.2417,
+      "step": 3720
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7471240758895874,
+      "learning_rate": 9.979780476680298e-05,
+      "loss": 2.2534,
+      "step": 3725
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7498319149017334,
+      "learning_rate": 9.979724697971125e-05,
+      "loss": 2.2542,
+      "step": 3730
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7646072506904602,
+      "learning_rate": 9.979668842587065e-05,
+      "loss": 2.2482,
+      "step": 3735
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7263892292976379,
+      "learning_rate": 9.979612910528976e-05,
+      "loss": 2.2907,
+      "step": 3740
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7579857707023621,
+      "learning_rate": 9.979556901797723e-05,
+      "loss": 2.2396,
+      "step": 3745
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7285515069961548,
+      "learning_rate": 9.979500816394165e-05,
+      "loss": 2.1767,
+      "step": 3750
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7444717288017273,
+      "learning_rate": 9.979444654319165e-05,
+      "loss": 2.2789,
+      "step": 3755
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7371863722801208,
+      "learning_rate": 9.979388415573591e-05,
+      "loss": 2.2708,
+      "step": 3760
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7589312195777893,
+      "learning_rate": 9.979332100158307e-05,
+      "loss": 2.2467,
+      "step": 3765
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7504525184631348,
+      "learning_rate": 9.979275708074179e-05,
+      "loss": 2.2393,
+      "step": 3770
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.773160994052887,
+      "learning_rate": 9.979219239322078e-05,
+      "loss": 2.2324,
+      "step": 3775
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7567590475082397,
+      "learning_rate": 9.979162693902871e-05,
+      "loss": 2.2069,
+      "step": 3780
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7856701016426086,
+      "learning_rate": 9.97910607181743e-05,
+      "loss": 2.234,
+      "step": 3785
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7490012049674988,
+      "learning_rate": 9.979049373066626e-05,
+      "loss": 2.3488,
+      "step": 3790
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7411119341850281,
+      "learning_rate": 9.978992597651334e-05,
+      "loss": 2.1875,
+      "step": 3795
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7689605951309204,
+      "learning_rate": 9.978935745572424e-05,
+      "loss": 2.2317,
+      "step": 3800
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7453191876411438,
+      "learning_rate": 9.978878816830776e-05,
+      "loss": 2.2762,
+      "step": 3805
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.763738214969635,
+      "learning_rate": 9.978821811427263e-05,
+      "loss": 2.2001,
+      "step": 3810
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7537261843681335,
+      "learning_rate": 9.978764729362766e-05,
+      "loss": 2.2793,
+      "step": 3815
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7751771807670593,
+      "learning_rate": 9.978707570638162e-05,
+      "loss": 2.1271,
+      "step": 3820
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7752400636672974,
+      "learning_rate": 9.978650335254332e-05,
+      "loss": 2.2727,
+      "step": 3825
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7720538377761841,
+      "learning_rate": 9.978593023212155e-05,
+      "loss": 2.2371,
+      "step": 3830
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.753073513507843,
+      "learning_rate": 9.978535634512515e-05,
+      "loss": 2.2649,
+      "step": 3835
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7765339016914368,
+      "learning_rate": 9.978478169156298e-05,
+      "loss": 2.1382,
+      "step": 3840
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7568251490592957,
+      "learning_rate": 9.978420627144385e-05,
+      "loss": 2.228,
+      "step": 3845
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7640216946601868,
+      "learning_rate": 9.978363008477665e-05,
+      "loss": 2.207,
+      "step": 3850
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7376381754875183,
+      "learning_rate": 9.978305313157022e-05,
+      "loss": 2.1794,
+      "step": 3855
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7352732419967651,
+      "learning_rate": 9.978247541183348e-05,
+      "loss": 2.2451,
+      "step": 3860
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.76202392578125,
+      "learning_rate": 9.978189692557529e-05,
+      "loss": 2.2863,
+      "step": 3865
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7009533047676086,
+      "learning_rate": 9.978131767280456e-05,
+      "loss": 2.1976,
+      "step": 3870
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7791014313697815,
+      "learning_rate": 9.978073765353026e-05,
+      "loss": 2.2846,
+      "step": 3875
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7256622910499573,
+      "learning_rate": 9.978015686776125e-05,
+      "loss": 2.2758,
+      "step": 3880
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7650802135467529,
+      "learning_rate": 9.977957531550652e-05,
+      "loss": 2.2256,
+      "step": 3885
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7623248100280762,
+      "learning_rate": 9.977899299677502e-05,
+      "loss": 2.2491,
+      "step": 3890
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7370533347129822,
+      "learning_rate": 9.97784099115757e-05,
+      "loss": 2.0975,
+      "step": 3895
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7421417832374573,
+      "learning_rate": 9.977782605991754e-05,
+      "loss": 2.2133,
+      "step": 3900
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.77610844373703,
+      "learning_rate": 9.977724144180953e-05,
+      "loss": 2.2068,
+      "step": 3905
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7804012298583984,
+      "learning_rate": 9.977665605726067e-05,
+      "loss": 2.0725,
+      "step": 3910
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7572207450866699,
+      "learning_rate": 9.977606990628001e-05,
+      "loss": 2.1981,
+      "step": 3915
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7459365725517273,
+      "learning_rate": 9.977548298887651e-05,
+      "loss": 2.1891,
+      "step": 3920
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7796589136123657,
+      "learning_rate": 9.977489530505927e-05,
+      "loss": 2.2018,
+      "step": 3925
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7565658092498779,
+      "learning_rate": 9.97743068548373e-05,
+      "loss": 2.1974,
+      "step": 3930
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7224230170249939,
+      "learning_rate": 9.977371763821966e-05,
+      "loss": 2.2469,
+      "step": 3935
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7682448029518127,
+      "learning_rate": 9.977312765521545e-05,
+      "loss": 2.2173,
+      "step": 3940
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7504734396934509,
+      "learning_rate": 9.977253690583372e-05,
+      "loss": 2.1913,
+      "step": 3945
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7540750503540039,
+      "learning_rate": 9.97719453900836e-05,
+      "loss": 2.2571,
+      "step": 3950
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7593798041343689,
+      "learning_rate": 9.977135310797418e-05,
+      "loss": 2.1802,
+      "step": 3955
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7199345827102661,
+      "learning_rate": 9.977076005951458e-05,
+      "loss": 2.2444,
+      "step": 3960
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7365019917488098,
+      "learning_rate": 9.977016624471392e-05,
+      "loss": 2.2854,
+      "step": 3965
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7885620594024658,
+      "learning_rate": 9.976957166358137e-05,
+      "loss": 2.2839,
+      "step": 3970
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7510995864868164,
+      "learning_rate": 9.976897631612607e-05,
+      "loss": 2.2101,
+      "step": 3975
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.799024760723114,
+      "learning_rate": 9.97683802023572e-05,
+      "loss": 2.184,
+      "step": 3980
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7352229952812195,
+      "learning_rate": 9.976778332228389e-05,
+      "loss": 2.204,
+      "step": 3985
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7800156474113464,
+      "learning_rate": 9.97671856759154e-05,
+      "loss": 2.1849,
+      "step": 3990
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7751713991165161,
+      "learning_rate": 9.976658726326088e-05,
+      "loss": 2.3133,
+      "step": 3995
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7398266792297363,
+      "learning_rate": 9.976598808432957e-05,
+      "loss": 2.2627,
+      "step": 4000
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7849751114845276,
+      "learning_rate": 9.976538813913068e-05,
+      "loss": 2.2719,
+      "step": 4005
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7781943082809448,
+      "learning_rate": 9.976478742767347e-05,
+      "loss": 2.2343,
+      "step": 4010
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.766110360622406,
+      "learning_rate": 9.976418594996716e-05,
+      "loss": 2.2475,
+      "step": 4015
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7706555128097534,
+      "learning_rate": 9.976358370602103e-05,
+      "loss": 2.2772,
+      "step": 4020
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7289323210716248,
+      "learning_rate": 9.976298069584438e-05,
+      "loss": 2.1866,
+      "step": 4025
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7644367814064026,
+      "learning_rate": 9.976237691944643e-05,
+      "loss": 2.2517,
+      "step": 4030
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.739638090133667,
+      "learning_rate": 9.976177237683653e-05,
+      "loss": 2.2255,
+      "step": 4035
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7123497128486633,
+      "learning_rate": 9.976116706802395e-05,
+      "loss": 2.2047,
+      "step": 4040
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7563219666481018,
+      "learning_rate": 9.976056099301805e-05,
+      "loss": 2.2056,
+      "step": 4045
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.731116533279419,
+      "learning_rate": 9.975995415182812e-05,
+      "loss": 2.1929,
+      "step": 4050
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7657923102378845,
+      "learning_rate": 9.975934654446355e-05,
+      "loss": 2.1764,
+      "step": 4055
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7605503797531128,
+      "learning_rate": 9.975873817093367e-05,
+      "loss": 2.2487,
+      "step": 4060
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7371482849121094,
+      "learning_rate": 9.975812903124782e-05,
+      "loss": 2.184,
+      "step": 4065
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7549729943275452,
+      "learning_rate": 9.975751912541544e-05,
+      "loss": 2.1623,
+      "step": 4070
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7648780941963196,
+      "learning_rate": 9.975690845344588e-05,
+      "loss": 2.2834,
+      "step": 4075
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8522197604179382,
+      "learning_rate": 9.975629701534854e-05,
+      "loss": 2.251,
+      "step": 4080
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7165658473968506,
+      "learning_rate": 9.975568481113285e-05,
+      "loss": 2.2056,
+      "step": 4085
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7588106393814087,
+      "learning_rate": 9.975507184080823e-05,
+      "loss": 2.3164,
+      "step": 4090
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7857149839401245,
+      "learning_rate": 9.975445810438413e-05,
+      "loss": 2.1528,
+      "step": 4095
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7805386781692505,
+      "learning_rate": 9.975384360186999e-05,
+      "loss": 2.2305,
+      "step": 4100
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7365330457687378,
+      "learning_rate": 9.975322833327527e-05,
+      "loss": 2.2156,
+      "step": 4105
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7389424443244934,
+      "learning_rate": 9.975261229860944e-05,
+      "loss": 2.2889,
+      "step": 4110
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7772135138511658,
+      "learning_rate": 9.975199549788199e-05,
+      "loss": 2.2685,
+      "step": 4115
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7588873505592346,
+      "learning_rate": 9.975137793110243e-05,
+      "loss": 2.2344,
+      "step": 4120
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3929243087768555,
+      "learning_rate": 9.975075959828026e-05,
+      "loss": 2.2368,
+      "step": 4125
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.738547682762146,
+      "learning_rate": 9.975014049942499e-05,
+      "loss": 2.2135,
+      "step": 4130
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.9665803909301758,
+      "learning_rate": 9.974952063454615e-05,
+      "loss": 2.2553,
+      "step": 4135
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7391039729118347,
+      "learning_rate": 9.974890000365332e-05,
+      "loss": 2.1701,
+      "step": 4140
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7538667917251587,
+      "learning_rate": 9.9748278606756e-05,
+      "loss": 2.2263,
+      "step": 4145
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.739439070224762,
+      "learning_rate": 9.974765644386382e-05,
+      "loss": 2.2006,
+      "step": 4150
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7388566732406616,
+      "learning_rate": 9.974703351498631e-05,
+      "loss": 2.1782,
+      "step": 4155
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.732667863368988,
+      "learning_rate": 9.974640982013307e-05,
+      "loss": 2.2286,
+      "step": 4160
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7351599931716919,
+      "learning_rate": 9.974578535931374e-05,
+      "loss": 2.2321,
+      "step": 4165
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7515677809715271,
+      "learning_rate": 9.974516013253789e-05,
+      "loss": 2.2649,
+      "step": 4170
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7446742057800293,
+      "learning_rate": 9.974453413981517e-05,
+      "loss": 2.1648,
+      "step": 4175
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6835678815841675,
+      "learning_rate": 9.974390738115521e-05,
+      "loss": 2.2378,
+      "step": 4180
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7197744846343994,
+      "learning_rate": 9.974327985656767e-05,
+      "loss": 2.2344,
+      "step": 4185
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.733710765838623,
+      "learning_rate": 9.974265156606221e-05,
+      "loss": 2.2465,
+      "step": 4190
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7016286253929138,
+      "learning_rate": 9.974202250964849e-05,
+      "loss": 2.1308,
+      "step": 4195
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7656921148300171,
+      "learning_rate": 9.974139268733623e-05,
+      "loss": 2.1522,
+      "step": 4200
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7712021470069885,
+      "learning_rate": 9.974076209913508e-05,
+      "loss": 2.1873,
+      "step": 4205
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7752746939659119,
+      "learning_rate": 9.974013074505477e-05,
+      "loss": 2.0976,
+      "step": 4210
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7568255662918091,
+      "learning_rate": 9.973949862510503e-05,
+      "loss": 2.1743,
+      "step": 4215
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7724996209144592,
+      "learning_rate": 9.97388657392956e-05,
+      "loss": 2.216,
+      "step": 4220
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7261400818824768,
+      "learning_rate": 9.97382320876362e-05,
+      "loss": 2.1538,
+      "step": 4225
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7269912362098694,
+      "learning_rate": 9.973759767013662e-05,
+      "loss": 2.3339,
+      "step": 4230
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7358632683753967,
+      "learning_rate": 9.973696248680659e-05,
+      "loss": 2.2754,
+      "step": 4235
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7370445132255554,
+      "learning_rate": 9.97363265376559e-05,
+      "loss": 2.1991,
+      "step": 4240
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7441362738609314,
+      "learning_rate": 9.973568982269436e-05,
+      "loss": 2.2,
+      "step": 4245
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.746447741985321,
+      "learning_rate": 9.973505234193177e-05,
+      "loss": 2.0682,
+      "step": 4250
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7665226459503174,
+      "learning_rate": 9.973441409537793e-05,
+      "loss": 2.2071,
+      "step": 4255
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7432265281677246,
+      "learning_rate": 9.973377508304269e-05,
+      "loss": 2.1387,
+      "step": 4260
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.779838502407074,
+      "learning_rate": 9.973313530493586e-05,
+      "loss": 2.2272,
+      "step": 4265
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7383325695991516,
+      "learning_rate": 9.973249476106732e-05,
+      "loss": 2.1959,
+      "step": 4270
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7830000519752502,
+      "learning_rate": 9.97318534514469e-05,
+      "loss": 2.2368,
+      "step": 4275
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7459074258804321,
+      "learning_rate": 9.97312113760845e-05,
+      "loss": 2.2536,
+      "step": 4280
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7600255012512207,
+      "learning_rate": 9.973056853499001e-05,
+      "loss": 2.2564,
+      "step": 4285
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7546904683113098,
+      "learning_rate": 9.972992492817331e-05,
+      "loss": 2.2651,
+      "step": 4290
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8085007667541504,
+      "learning_rate": 9.972928055564432e-05,
+      "loss": 2.1846,
+      "step": 4295
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7899944186210632,
+      "learning_rate": 9.972863541741297e-05,
+      "loss": 2.2867,
+      "step": 4300
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7350949048995972,
+      "learning_rate": 9.972798951348918e-05,
+      "loss": 2.2018,
+      "step": 4305
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.758342444896698,
+      "learning_rate": 9.972734284388288e-05,
+      "loss": 2.3106,
+      "step": 4310
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7310266494750977,
+      "learning_rate": 9.972669540860406e-05,
+      "loss": 2.238,
+      "step": 4315
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7507831454277039,
+      "learning_rate": 9.972604720766268e-05,
+      "loss": 2.1405,
+      "step": 4320
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7043246030807495,
+      "learning_rate": 9.97253982410687e-05,
+      "loss": 2.229,
+      "step": 4325
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7222868800163269,
+      "learning_rate": 9.972474850883213e-05,
+      "loss": 2.2353,
+      "step": 4330
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7711742520332336,
+      "learning_rate": 9.972409801096299e-05,
+      "loss": 2.2429,
+      "step": 4335
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.752962052822113,
+      "learning_rate": 9.972344674747125e-05,
+      "loss": 2.2839,
+      "step": 4340
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7373786568641663,
+      "learning_rate": 9.972279471836698e-05,
+      "loss": 2.2444,
+      "step": 4345
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7498070001602173,
+      "learning_rate": 9.972214192366021e-05,
+      "loss": 2.1517,
+      "step": 4350
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7440705895423889,
+      "learning_rate": 9.972148836336098e-05,
+      "loss": 2.1799,
+      "step": 4355
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7498330473899841,
+      "learning_rate": 9.972083403747937e-05,
+      "loss": 2.2475,
+      "step": 4360
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7158123254776001,
+      "learning_rate": 9.972017894602542e-05,
+      "loss": 2.2189,
+      "step": 4365
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7591242790222168,
+      "learning_rate": 9.971952308900925e-05,
+      "loss": 2.177,
+      "step": 4370
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7281230688095093,
+      "learning_rate": 9.971886646644096e-05,
+      "loss": 2.171,
+      "step": 4375
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7447307109832764,
+      "learning_rate": 9.971820907833065e-05,
+      "loss": 2.2622,
+      "step": 4380
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.743661105632782,
+      "learning_rate": 9.971755092468843e-05,
+      "loss": 2.2115,
+      "step": 4385
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7463720440864563,
+      "learning_rate": 9.971689200552445e-05,
+      "loss": 2.2601,
+      "step": 4390
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7044093012809753,
+      "learning_rate": 9.971623232084886e-05,
+      "loss": 2.2291,
+      "step": 4395
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7776219248771667,
+      "learning_rate": 9.971557187067179e-05,
+      "loss": 2.163,
+      "step": 4400
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7467032670974731,
+      "learning_rate": 9.971491065500345e-05,
+      "loss": 2.1908,
+      "step": 4405
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.779761016368866,
+      "learning_rate": 9.9714248673854e-05,
+      "loss": 2.1773,
+      "step": 4410
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7087513208389282,
+      "learning_rate": 9.971358592723361e-05,
+      "loss": 2.2293,
+      "step": 4415
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7501884698867798,
+      "learning_rate": 9.97129224151525e-05,
+      "loss": 2.2595,
+      "step": 4420
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.752416729927063,
+      "learning_rate": 9.971225813762094e-05,
+      "loss": 2.2467,
+      "step": 4425
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7428842186927795,
+      "learning_rate": 9.971159309464906e-05,
+      "loss": 2.168,
+      "step": 4430
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6953173279762268,
+      "learning_rate": 9.971092728624718e-05,
+      "loss": 2.1657,
+      "step": 4435
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7553196549415588,
+      "learning_rate": 9.97102607124255e-05,
+      "loss": 2.1946,
+      "step": 4440
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7362040281295776,
+      "learning_rate": 9.970959337319433e-05,
+      "loss": 2.1821,
+      "step": 4445
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7337948679924011,
+      "learning_rate": 9.970892526856392e-05,
+      "loss": 2.1704,
+      "step": 4450
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7240247130393982,
+      "learning_rate": 9.970825639854454e-05,
+      "loss": 2.1616,
+      "step": 4455
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7261995077133179,
+      "learning_rate": 9.970758676314652e-05,
+      "loss": 2.0955,
+      "step": 4460
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7293042540550232,
+      "learning_rate": 9.970691636238017e-05,
+      "loss": 2.2682,
+      "step": 4465
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7412492036819458,
+      "learning_rate": 9.970624519625578e-05,
+      "loss": 2.2326,
+      "step": 4470
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7806591987609863,
+      "learning_rate": 9.970557326478371e-05,
+      "loss": 2.2344,
+      "step": 4475
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7624565362930298,
+      "learning_rate": 9.97049005679743e-05,
+      "loss": 2.1923,
+      "step": 4480
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7771266102790833,
+      "learning_rate": 9.970422710583791e-05,
+      "loss": 2.2161,
+      "step": 4485
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.703076958656311,
+      "learning_rate": 9.970355287838491e-05,
+      "loss": 2.1819,
+      "step": 4490
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.748218834400177,
+      "learning_rate": 9.970287788562567e-05,
+      "loss": 2.1974,
+      "step": 4495
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7302867770195007,
+      "learning_rate": 9.97022021275706e-05,
+      "loss": 2.1732,
+      "step": 4500
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7248017191886902,
+      "learning_rate": 9.97015256042301e-05,
+      "loss": 2.1867,
+      "step": 4505
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7811912298202515,
+      "learning_rate": 9.97008483156146e-05,
+      "loss": 2.2075,
+      "step": 4510
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6772184371948242,
+      "learning_rate": 9.970017026173448e-05,
+      "loss": 2.2844,
+      "step": 4515
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.765239417552948,
+      "learning_rate": 9.969949144260022e-05,
+      "loss": 2.1604,
+      "step": 4520
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7914250493049622,
+      "learning_rate": 9.969881185822227e-05,
+      "loss": 2.1678,
+      "step": 4525
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7274059057235718,
+      "learning_rate": 9.969813150861111e-05,
+      "loss": 2.1975,
+      "step": 4530
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7635666728019714,
+      "learning_rate": 9.969745039377717e-05,
+      "loss": 2.2579,
+      "step": 4535
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.775330126285553,
+      "learning_rate": 9.969676851373098e-05,
+      "loss": 2.1931,
+      "step": 4540
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7226781845092773,
+      "learning_rate": 9.969608586848301e-05,
+      "loss": 2.2533,
+      "step": 4545
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7443475127220154,
+      "learning_rate": 9.969540245804377e-05,
+      "loss": 2.1798,
+      "step": 4550
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7397128939628601,
+      "learning_rate": 9.96947182824238e-05,
+      "loss": 2.2813,
+      "step": 4555
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7504158616065979,
+      "learning_rate": 9.969403334163364e-05,
+      "loss": 2.2387,
+      "step": 4560
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7849494218826294,
+      "learning_rate": 9.969334763568383e-05,
+      "loss": 2.2444,
+      "step": 4565
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7473179697990417,
+      "learning_rate": 9.969266116458491e-05,
+      "loss": 2.2922,
+      "step": 4570
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7466105222702026,
+      "learning_rate": 9.969197392834747e-05,
+      "loss": 2.1976,
+      "step": 4575
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7327406406402588,
+      "learning_rate": 9.969128592698207e-05,
+      "loss": 2.174,
+      "step": 4580
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7635794878005981,
+      "learning_rate": 9.969059716049934e-05,
+      "loss": 2.234,
+      "step": 4585
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7294145822525024,
+      "learning_rate": 9.968990762890986e-05,
+      "loss": 2.1978,
+      "step": 4590
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7107588052749634,
+      "learning_rate": 9.968921733222425e-05,
+      "loss": 2.202,
+      "step": 4595
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7241854667663574,
+      "learning_rate": 9.968852627045313e-05,
+      "loss": 2.1966,
+      "step": 4600
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7373654842376709,
+      "learning_rate": 9.968783444360716e-05,
+      "loss": 2.1689,
+      "step": 4605
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7428138256072998,
+      "learning_rate": 9.968714185169698e-05,
+      "loss": 2.1982,
+      "step": 4610
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.752863347530365,
+      "learning_rate": 9.968644849473326e-05,
+      "loss": 2.2568,
+      "step": 4615
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7259964942932129,
+      "learning_rate": 9.968575437272667e-05,
+      "loss": 2.2312,
+      "step": 4620
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7335218787193298,
+      "learning_rate": 9.968505948568792e-05,
+      "loss": 2.2533,
+      "step": 4625
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7347412109375,
+      "learning_rate": 9.968436383362766e-05,
+      "loss": 2.1224,
+      "step": 4630
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7354990839958191,
+      "learning_rate": 9.968366741655663e-05,
+      "loss": 2.1607,
+      "step": 4635
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.731635570526123,
+      "learning_rate": 9.968297023448558e-05,
+      "loss": 2.1559,
+      "step": 4640
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7349185347557068,
+      "learning_rate": 9.96822722874252e-05,
+      "loss": 2.1669,
+      "step": 4645
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7441719174385071,
+      "learning_rate": 9.968157357538625e-05,
+      "loss": 2.1522,
+      "step": 4650
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7515274882316589,
+      "learning_rate": 9.968087409837951e-05,
+      "loss": 2.1608,
+      "step": 4655
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.73979252576828,
+      "learning_rate": 9.968017385641574e-05,
+      "loss": 2.1875,
+      "step": 4660
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7343612313270569,
+      "learning_rate": 9.967947284950569e-05,
+      "loss": 2.1909,
+      "step": 4665
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7157119512557983,
+      "learning_rate": 9.967877107766021e-05,
+      "loss": 2.225,
+      "step": 4670
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.8495104908943176,
+      "learning_rate": 9.967806854089006e-05,
+      "loss": 2.2664,
+      "step": 4675
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.8175004124641418,
+      "learning_rate": 9.967736523920607e-05,
+      "loss": 2.2109,
+      "step": 4680
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.705700695514679,
+      "learning_rate": 9.967666117261907e-05,
+      "loss": 2.2268,
+      "step": 4685
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6927974224090576,
+      "learning_rate": 9.96759563411399e-05,
+      "loss": 2.1329,
+      "step": 4690
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7129865884780884,
+      "learning_rate": 9.967525074477943e-05,
+      "loss": 2.2262,
+      "step": 4695
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6956382393836975,
+      "learning_rate": 9.96745443835485e-05,
+      "loss": 2.1643,
+      "step": 4700
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7136667966842651,
+      "learning_rate": 9.967383725745799e-05,
+      "loss": 2.1797,
+      "step": 4705
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7251401543617249,
+      "learning_rate": 9.96731293665188e-05,
+      "loss": 2.2441,
+      "step": 4710
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.709603488445282,
+      "learning_rate": 9.967242071074181e-05,
+      "loss": 2.1961,
+      "step": 4715
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.8045169115066528,
+      "learning_rate": 9.967171129013797e-05,
+      "loss": 2.1369,
+      "step": 4720
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7314881682395935,
+      "learning_rate": 9.967100110471817e-05,
+      "loss": 2.2243,
+      "step": 4725
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7059980630874634,
+      "learning_rate": 9.967029015449334e-05,
+      "loss": 2.1868,
+      "step": 4730
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7522546052932739,
+      "learning_rate": 9.966957843947444e-05,
+      "loss": 2.1538,
+      "step": 4735
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.8023698329925537,
+      "learning_rate": 9.966886595967243e-05,
+      "loss": 2.1869,
+      "step": 4740
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.667367160320282,
+      "learning_rate": 9.966815271509828e-05,
+      "loss": 2.2086,
+      "step": 4745
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7070870399475098,
+      "learning_rate": 9.966743870576297e-05,
+      "loss": 2.2467,
+      "step": 4750
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7457792162895203,
+      "learning_rate": 9.96667239316775e-05,
+      "loss": 2.1841,
+      "step": 4755
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7340573072433472,
+      "learning_rate": 9.966600839285286e-05,
+      "loss": 2.3083,
+      "step": 4760
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7258387804031372,
+      "learning_rate": 9.966529208930008e-05,
+      "loss": 2.2487,
+      "step": 4765
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7208442687988281,
+      "learning_rate": 9.96645750210302e-05,
+      "loss": 2.1048,
+      "step": 4770
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7116550803184509,
+      "learning_rate": 9.966385718805423e-05,
+      "loss": 2.1457,
+      "step": 4775
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7189687490463257,
+      "learning_rate": 9.966313859038325e-05,
+      "loss": 2.2203,
+      "step": 4780
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7161012291908264,
+      "learning_rate": 9.966241922802832e-05,
+      "loss": 2.1971,
+      "step": 4785
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7251018285751343,
+      "learning_rate": 9.96616991010005e-05,
+      "loss": 2.3208,
+      "step": 4790
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7236688733100891,
+      "learning_rate": 9.96609782093109e-05,
+      "loss": 2.3052,
+      "step": 4795
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7215144634246826,
+      "learning_rate": 9.96602565529706e-05,
+      "loss": 2.2356,
+      "step": 4800
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7379950881004333,
+      "learning_rate": 9.965953413199071e-05,
+      "loss": 2.1893,
+      "step": 4805
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7421005964279175,
+      "learning_rate": 9.965881094638238e-05,
+      "loss": 2.1676,
+      "step": 4810
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7644197940826416,
+      "learning_rate": 9.965808699615674e-05,
+      "loss": 2.1421,
+      "step": 4815
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7171550393104553,
+      "learning_rate": 9.965736228132492e-05,
+      "loss": 2.1181,
+      "step": 4820
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6936640739440918,
+      "learning_rate": 9.965663680189807e-05,
+      "loss": 2.1951,
+      "step": 4825
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6827539205551147,
+      "learning_rate": 9.965591055788738e-05,
+      "loss": 2.1462,
+      "step": 4830
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7244593501091003,
+      "learning_rate": 9.965518354930405e-05,
+      "loss": 2.2108,
+      "step": 4835
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7006765604019165,
+      "learning_rate": 9.965445577615922e-05,
+      "loss": 2.1656,
+      "step": 4840
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7224099040031433,
+      "learning_rate": 9.965372723846415e-05,
+      "loss": 2.1642,
+      "step": 4845
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7207775712013245,
+      "learning_rate": 9.965299793623003e-05,
+      "loss": 2.1911,
+      "step": 4850
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7262431383132935,
+      "learning_rate": 9.96522678694681e-05,
+      "loss": 2.1947,
+      "step": 4855
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7530291676521301,
+      "learning_rate": 9.965153703818959e-05,
+      "loss": 2.255,
+      "step": 4860
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7480681538581848,
+      "learning_rate": 9.965080544240575e-05,
+      "loss": 2.179,
+      "step": 4865
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.8119124174118042,
+      "learning_rate": 9.965007308212786e-05,
+      "loss": 2.2205,
+      "step": 4870
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7498155832290649,
+      "learning_rate": 9.964933995736721e-05,
+      "loss": 2.1818,
+      "step": 4875
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6956478953361511,
+      "learning_rate": 9.964860606813504e-05,
+      "loss": 2.177,
+      "step": 4880
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7450419664382935,
+      "learning_rate": 9.96478714144427e-05,
+      "loss": 2.1414,
+      "step": 4885
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7240822315216064,
+      "learning_rate": 9.964713599630146e-05,
+      "loss": 2.2461,
+      "step": 4890
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7414240837097168,
+      "learning_rate": 9.964639981372267e-05,
+      "loss": 2.1905,
+      "step": 4895
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7259507179260254,
+      "learning_rate": 9.964566286671766e-05,
+      "loss": 2.216,
+      "step": 4900
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.795586347579956,
+      "learning_rate": 9.964492515529777e-05,
+      "loss": 2.2785,
+      "step": 4905
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7391549944877625,
+      "learning_rate": 9.964418667947438e-05,
+      "loss": 2.2236,
+      "step": 4910
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7140523791313171,
+      "learning_rate": 9.964344743925884e-05,
+      "loss": 2.197,
+      "step": 4915
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7355964183807373,
+      "learning_rate": 9.964270743466253e-05,
+      "loss": 2.2051,
+      "step": 4920
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7651568055152893,
+      "learning_rate": 9.964196666569686e-05,
+      "loss": 2.2379,
+      "step": 4925
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.790097713470459,
+      "learning_rate": 9.964122513237322e-05,
+      "loss": 2.0967,
+      "step": 4930
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.8111740946769714,
+      "learning_rate": 9.964048283470305e-05,
+      "loss": 2.2531,
+      "step": 4935
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.76014244556427,
+      "learning_rate": 9.963973977269776e-05,
+      "loss": 2.2152,
+      "step": 4940
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6939969658851624,
+      "learning_rate": 9.963899594636879e-05,
+      "loss": 2.1971,
+      "step": 4945
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.737716019153595,
+      "learning_rate": 9.96382513557276e-05,
+      "loss": 2.2663,
+      "step": 4950
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.723233163356781,
+      "learning_rate": 9.963750600078565e-05,
+      "loss": 2.1476,
+      "step": 4955
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7083010673522949,
+      "learning_rate": 9.963675988155443e-05,
+      "loss": 2.139,
+      "step": 4960
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6773269176483154,
+      "learning_rate": 9.963601299804541e-05,
+      "loss": 2.2573,
+      "step": 4965
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7193275094032288,
+      "learning_rate": 9.963526535027012e-05,
+      "loss": 2.2147,
+      "step": 4970
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7541141510009766,
+      "learning_rate": 9.963451693824003e-05,
+      "loss": 2.2555,
+      "step": 4975
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7327399849891663,
+      "learning_rate": 9.963376776196669e-05,
+      "loss": 2.2937,
+      "step": 4980
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7227214574813843,
+      "learning_rate": 9.963301782146163e-05,
+      "loss": 2.2705,
+      "step": 4985
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7133594155311584,
+      "learning_rate": 9.96322671167364e-05,
+      "loss": 2.2672,
+      "step": 4990
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7374505400657654,
+      "learning_rate": 9.963151564780256e-05,
+      "loss": 2.1706,
+      "step": 4995
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7275349497795105,
+      "learning_rate": 9.963076341467165e-05,
+      "loss": 2.1282,
+      "step": 5000
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7133877873420715,
+      "learning_rate": 9.963001041735531e-05,
+      "loss": 2.2,
+      "step": 5005
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7409718036651611,
+      "learning_rate": 9.962925665586509e-05,
+      "loss": 2.1943,
+      "step": 5010
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7413187623023987,
+      "learning_rate": 9.962850213021262e-05,
+      "loss": 2.2458,
+      "step": 5015
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.8329206705093384,
+      "learning_rate": 9.96277468404095e-05,
+      "loss": 2.2,
+      "step": 5020
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7446252107620239,
+      "learning_rate": 9.962699078646735e-05,
+      "loss": 2.1952,
+      "step": 5025
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7112839221954346,
+      "learning_rate": 9.962623396839785e-05,
+      "loss": 2.1961,
+      "step": 5030
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7617784738540649,
+      "learning_rate": 9.962547638621262e-05,
+      "loss": 2.0994,
+      "step": 5035
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7698768973350525,
+      "learning_rate": 9.962471803992334e-05,
+      "loss": 2.2191,
+      "step": 5040
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7386558651924133,
+      "learning_rate": 9.962395892954168e-05,
+      "loss": 2.1476,
+      "step": 5045
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7202014923095703,
+      "learning_rate": 9.962319905507934e-05,
+      "loss": 2.1956,
+      "step": 5050
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7245826125144958,
+      "learning_rate": 9.9622438416548e-05,
+      "loss": 2.159,
+      "step": 5055
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7026588916778564,
+      "learning_rate": 9.962167701395937e-05,
+      "loss": 2.1495,
+      "step": 5060
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7569719552993774,
+      "learning_rate": 9.962091484732521e-05,
+      "loss": 2.1719,
+      "step": 5065
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7101033926010132,
+      "learning_rate": 9.962015191665722e-05,
+      "loss": 2.2019,
+      "step": 5070
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7440845370292664,
+      "learning_rate": 9.961938822196716e-05,
+      "loss": 2.1499,
+      "step": 5075
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.70537930727005,
+      "learning_rate": 9.961862376326678e-05,
+      "loss": 2.1704,
+      "step": 5080
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7123816013336182,
+      "learning_rate": 9.961785854056788e-05,
+      "loss": 2.3022,
+      "step": 5085
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7520079612731934,
+      "learning_rate": 9.961709255388221e-05,
+      "loss": 2.1517,
+      "step": 5090
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7449693083763123,
+      "learning_rate": 9.961632580322156e-05,
+      "loss": 2.1839,
+      "step": 5095
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.800480842590332,
+      "learning_rate": 9.961555828859777e-05,
+      "loss": 2.221,
+      "step": 5100
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7339938282966614,
+      "learning_rate": 9.961479001002264e-05,
+      "loss": 2.1222,
+      "step": 5105
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7168148159980774,
+      "learning_rate": 9.961402096750799e-05,
+      "loss": 2.185,
+      "step": 5110
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7237013578414917,
+      "learning_rate": 9.961325116106566e-05,
+      "loss": 2.2429,
+      "step": 5115
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.707354724407196,
+      "learning_rate": 9.961248059070751e-05,
+      "loss": 2.2248,
+      "step": 5120
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7562894821166992,
+      "learning_rate": 9.961170925644542e-05,
+      "loss": 2.2438,
+      "step": 5125
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6723471879959106,
+      "learning_rate": 9.961093715829127e-05,
+      "loss": 2.1897,
+      "step": 5130
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7189382910728455,
+      "learning_rate": 9.961016429625691e-05,
+      "loss": 2.2436,
+      "step": 5135
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7740445137023926,
+      "learning_rate": 9.960939067035426e-05,
+      "loss": 2.2018,
+      "step": 5140
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7558847069740295,
+      "learning_rate": 9.960861628059524e-05,
+      "loss": 2.2125,
+      "step": 5145
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7414452433586121,
+      "learning_rate": 9.960784112699177e-05,
+      "loss": 2.2548,
+      "step": 5150
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7136910557746887,
+      "learning_rate": 9.960706520955578e-05,
+      "loss": 2.1767,
+      "step": 5155
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6968001127243042,
+      "learning_rate": 9.960628852829923e-05,
+      "loss": 2.2076,
+      "step": 5160
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7011262774467468,
+      "learning_rate": 9.960551108323406e-05,
+      "loss": 2.1814,
+      "step": 5165
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7526137828826904,
+      "learning_rate": 9.960473287437224e-05,
+      "loss": 2.1562,
+      "step": 5170
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7345152497291565,
+      "learning_rate": 9.960395390172579e-05,
+      "loss": 2.2377,
+      "step": 5175
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7164118885993958,
+      "learning_rate": 9.960317416530665e-05,
+      "loss": 2.1862,
+      "step": 5180
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7254704833030701,
+      "learning_rate": 9.960239366512686e-05,
+      "loss": 2.2098,
+      "step": 5185
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.712552011013031,
+      "learning_rate": 9.960161240119843e-05,
+      "loss": 2.252,
+      "step": 5190
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7242210507392883,
+      "learning_rate": 9.96008303735334e-05,
+      "loss": 2.1245,
+      "step": 5195
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7383831739425659,
+      "learning_rate": 9.960004758214379e-05,
+      "loss": 2.2816,
+      "step": 5200
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7217000126838684,
+      "learning_rate": 9.959926402704165e-05,
+      "loss": 2.1724,
+      "step": 5205
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7098327875137329,
+      "learning_rate": 9.959847970823908e-05,
+      "loss": 2.1683,
+      "step": 5210
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6777403354644775,
+      "learning_rate": 9.959769462574813e-05,
+      "loss": 2.1716,
+      "step": 5215
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7004913687705994,
+      "learning_rate": 9.959690877958088e-05,
+      "loss": 2.1772,
+      "step": 5220
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6826672554016113,
+      "learning_rate": 9.959612216974946e-05,
+      "loss": 2.117,
+      "step": 5225
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7090833783149719,
+      "learning_rate": 9.959533479626596e-05,
+      "loss": 2.1637,
+      "step": 5230
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7036669850349426,
+      "learning_rate": 9.95945466591425e-05,
+      "loss": 2.1329,
+      "step": 5235
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.722649335861206,
+      "learning_rate": 9.959375775839124e-05,
+      "loss": 2.2826,
+      "step": 5240
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7024977803230286,
+      "learning_rate": 9.95929680940243e-05,
+      "loss": 2.1858,
+      "step": 5245
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7440763711929321,
+      "learning_rate": 9.959217766605384e-05,
+      "loss": 2.182,
+      "step": 5250
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6927558779716492,
+      "learning_rate": 9.959138647449205e-05,
+      "loss": 2.2524,
+      "step": 5255
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6485484838485718,
+      "learning_rate": 9.95905945193511e-05,
+      "loss": 2.1325,
+      "step": 5260
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7544555068016052,
+      "learning_rate": 9.958980180064318e-05,
+      "loss": 2.2083,
+      "step": 5265
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7475581169128418,
+      "learning_rate": 9.958900831838051e-05,
+      "loss": 2.2095,
+      "step": 5270
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7399588227272034,
+      "learning_rate": 9.958821407257531e-05,
+      "loss": 2.0813,
+      "step": 5275
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7128239274024963,
+      "learning_rate": 9.958741906323978e-05,
+      "loss": 2.2205,
+      "step": 5280
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7056192755699158,
+      "learning_rate": 9.95866232903862e-05,
+      "loss": 2.1192,
+      "step": 5285
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7533562779426575,
+      "learning_rate": 9.958582675402677e-05,
+      "loss": 2.2307,
+      "step": 5290
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7128762602806091,
+      "learning_rate": 9.958502945417382e-05,
+      "loss": 2.1362,
+      "step": 5295
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7062804102897644,
+      "learning_rate": 9.958423139083958e-05,
+      "loss": 2.3151,
+      "step": 5300
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7992850542068481,
+      "learning_rate": 9.958343256403635e-05,
+      "loss": 2.1582,
+      "step": 5305
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7154298424720764,
+      "learning_rate": 9.958263297377643e-05,
+      "loss": 2.227,
+      "step": 5310
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7215281128883362,
+      "learning_rate": 9.958183262007215e-05,
+      "loss": 2.0689,
+      "step": 5315
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7117182612419128,
+      "learning_rate": 9.95810315029358e-05,
+      "loss": 2.2243,
+      "step": 5320
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7523080110549927,
+      "learning_rate": 9.958022962237974e-05,
+      "loss": 2.1002,
+      "step": 5325
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6958641409873962,
+      "learning_rate": 9.957942697841632e-05,
+      "loss": 2.1624,
+      "step": 5330
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.679563045501709,
+      "learning_rate": 9.957862357105785e-05,
+      "loss": 2.2211,
+      "step": 5335
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7265320420265198,
+      "learning_rate": 9.957781940031676e-05,
+      "loss": 2.1197,
+      "step": 5340
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7312959432601929,
+      "learning_rate": 9.957701446620543e-05,
+      "loss": 2.1779,
+      "step": 5345
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6817158460617065,
+      "learning_rate": 9.957620876873621e-05,
+      "loss": 2.1878,
+      "step": 5350
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7159974575042725,
+      "learning_rate": 9.957540230792152e-05,
+      "loss": 2.2154,
+      "step": 5355
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7517039179801941,
+      "learning_rate": 9.957459508377381e-05,
+      "loss": 2.1653,
+      "step": 5360
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7311538457870483,
+      "learning_rate": 9.957378709630548e-05,
+      "loss": 2.1559,
+      "step": 5365
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7014145255088806,
+      "learning_rate": 9.957297834552898e-05,
+      "loss": 2.1757,
+      "step": 5370
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7113021016120911,
+      "learning_rate": 9.957216883145675e-05,
+      "loss": 2.1984,
+      "step": 5375
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6927894949913025,
+      "learning_rate": 9.957135855410127e-05,
+      "loss": 2.2061,
+      "step": 5380
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7469661235809326,
+      "learning_rate": 9.957054751347501e-05,
+      "loss": 2.1639,
+      "step": 5385
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6951924562454224,
+      "learning_rate": 9.956973570959045e-05,
+      "loss": 2.1063,
+      "step": 5390
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7411240935325623,
+      "learning_rate": 9.956892314246012e-05,
+      "loss": 2.3004,
+      "step": 5395
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6907780170440674,
+      "learning_rate": 9.956810981209647e-05,
+      "loss": 2.1838,
+      "step": 5400
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7491859793663025,
+      "learning_rate": 9.956729571851211e-05,
+      "loss": 2.2157,
+      "step": 5405
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.730535626411438,
+      "learning_rate": 9.956648086171949e-05,
+      "loss": 2.2586,
+      "step": 5410
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7760801315307617,
+      "learning_rate": 9.956566524173121e-05,
+      "loss": 2.1833,
+      "step": 5415
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6940194368362427,
+      "learning_rate": 9.956484885855982e-05,
+      "loss": 2.1322,
+      "step": 5420
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6812244653701782,
+      "learning_rate": 9.956403171221786e-05,
+      "loss": 2.223,
+      "step": 5425
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7206534743309021,
+      "learning_rate": 9.956321380271796e-05,
+      "loss": 2.1917,
+      "step": 5430
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6982179284095764,
+      "learning_rate": 9.956239513007268e-05,
+      "loss": 2.2012,
+      "step": 5435
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7104706168174744,
+      "learning_rate": 9.956157569429464e-05,
+      "loss": 2.1663,
+      "step": 5440
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7069125175476074,
+      "learning_rate": 9.956075549539643e-05,
+      "loss": 2.1452,
+      "step": 5445
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7073805332183838,
+      "learning_rate": 9.955993453339072e-05,
+      "loss": 2.1809,
+      "step": 5450
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.76249098777771,
+      "learning_rate": 9.955911280829011e-05,
+      "loss": 2.1667,
+      "step": 5455
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.8974733352661133,
+      "learning_rate": 9.95582903201073e-05,
+      "loss": 2.1916,
+      "step": 5460
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7289822101593018,
+      "learning_rate": 9.955746706885492e-05,
+      "loss": 2.2105,
+      "step": 5465
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7234417796134949,
+      "learning_rate": 9.955664305454564e-05,
+      "loss": 2.3063,
+      "step": 5470
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7333115339279175,
+      "learning_rate": 9.955581827719216e-05,
+      "loss": 2.2581,
+      "step": 5475
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7191726565361023,
+      "learning_rate": 9.955499273680719e-05,
+      "loss": 2.1364,
+      "step": 5480
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7710874676704407,
+      "learning_rate": 9.955416643340343e-05,
+      "loss": 2.0659,
+      "step": 5485
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7640387415885925,
+      "learning_rate": 9.95533393669936e-05,
+      "loss": 2.2394,
+      "step": 5490
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7405058145523071,
+      "learning_rate": 9.955251153759044e-05,
+      "loss": 2.1685,
+      "step": 5495
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7085117697715759,
+      "learning_rate": 9.955168294520672e-05,
+      "loss": 2.223,
+      "step": 5500
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6836831569671631,
+      "learning_rate": 9.955085358985514e-05,
+      "loss": 2.2094,
+      "step": 5505
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7151683568954468,
+      "learning_rate": 9.955002347154851e-05,
+      "loss": 2.0915,
+      "step": 5510
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7524467706680298,
+      "learning_rate": 9.954919259029962e-05,
+      "loss": 2.1758,
+      "step": 5515
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7148208022117615,
+      "learning_rate": 9.954836094612123e-05,
+      "loss": 2.2079,
+      "step": 5520
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7130784392356873,
+      "learning_rate": 9.954752853902619e-05,
+      "loss": 2.198,
+      "step": 5525
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7148199677467346,
+      "learning_rate": 9.954669536902728e-05,
+      "loss": 2.2333,
+      "step": 5530
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7216602563858032,
+      "learning_rate": 9.954586143613733e-05,
+      "loss": 2.1696,
+      "step": 5535
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6996355056762695,
+      "learning_rate": 9.954502674036919e-05,
+      "loss": 2.1855,
+      "step": 5540
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7130545973777771,
+      "learning_rate": 9.954419128173572e-05,
+      "loss": 2.2598,
+      "step": 5545
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7069938778877258,
+      "learning_rate": 9.954335506024978e-05,
+      "loss": 2.168,
+      "step": 5550
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6698031425476074,
+      "learning_rate": 9.954251807592423e-05,
+      "loss": 2.1517,
+      "step": 5555
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7479323744773865,
+      "learning_rate": 9.954168032877198e-05,
+      "loss": 2.1671,
+      "step": 5560
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.702394425868988,
+      "learning_rate": 9.95408418188059e-05,
+      "loss": 2.1548,
+      "step": 5565
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7358775734901428,
+      "learning_rate": 9.954000254603894e-05,
+      "loss": 2.1658,
+      "step": 5570
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7286148071289062,
+      "learning_rate": 9.9539162510484e-05,
+      "loss": 2.3386,
+      "step": 5575
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7266163229942322,
+      "learning_rate": 9.953832171215401e-05,
+      "loss": 2.1101,
+      "step": 5580
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7392322421073914,
+      "learning_rate": 9.953748015106193e-05,
+      "loss": 2.2645,
+      "step": 5585
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7714710235595703,
+      "learning_rate": 9.953663782722069e-05,
+      "loss": 2.1822,
+      "step": 5590
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7111704349517822,
+      "learning_rate": 9.953579474064329e-05,
+      "loss": 2.1776,
+      "step": 5595
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7106512188911438,
+      "learning_rate": 9.95349508913427e-05,
+      "loss": 2.2085,
+      "step": 5600
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6965886354446411,
+      "learning_rate": 9.953410627933193e-05,
+      "loss": 2.1214,
+      "step": 5605
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7497239708900452,
+      "learning_rate": 9.953326090462397e-05,
+      "loss": 2.147,
+      "step": 5610
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.2962913513183594,
+      "learning_rate": 9.953241476723181e-05,
+      "loss": 2.3436,
+      "step": 5615
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7334530353546143,
+      "learning_rate": 9.953156786716851e-05,
+      "loss": 2.1226,
+      "step": 5620
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7144428491592407,
+      "learning_rate": 9.953072020444711e-05,
+      "loss": 2.2278,
+      "step": 5625
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7017433047294617,
+      "learning_rate": 9.952987177908066e-05,
+      "loss": 2.0956,
+      "step": 5630
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6928303837776184,
+      "learning_rate": 9.95290225910822e-05,
+      "loss": 2.2498,
+      "step": 5635
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6776155829429626,
+      "learning_rate": 9.952817264046486e-05,
+      "loss": 2.2256,
+      "step": 5640
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6662044525146484,
+      "learning_rate": 9.952732192724167e-05,
+      "loss": 2.1024,
+      "step": 5645
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6844089031219482,
+      "learning_rate": 9.952647045142576e-05,
+      "loss": 2.211,
+      "step": 5650
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7550753355026245,
+      "learning_rate": 9.952561821303021e-05,
+      "loss": 2.1727,
+      "step": 5655
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7235275506973267,
+      "learning_rate": 9.952476521206818e-05,
+      "loss": 2.2778,
+      "step": 5660
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7200480699539185,
+      "learning_rate": 9.952391144855278e-05,
+      "loss": 2.1734,
+      "step": 5665
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7026602625846863,
+      "learning_rate": 9.952305692249717e-05,
+      "loss": 2.1701,
+      "step": 5670
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6647647023200989,
+      "learning_rate": 9.952220163391451e-05,
+      "loss": 2.2151,
+      "step": 5675
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7049278020858765,
+      "learning_rate": 9.952134558281793e-05,
+      "loss": 2.256,
+      "step": 5680
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7206385135650635,
+      "learning_rate": 9.952048876922066e-05,
+      "loss": 2.0795,
+      "step": 5685
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7011607885360718,
+      "learning_rate": 9.951963119313588e-05,
+      "loss": 2.1068,
+      "step": 5690
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7563923597335815,
+      "learning_rate": 9.951877285457678e-05,
+      "loss": 2.2682,
+      "step": 5695
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7368286848068237,
+      "learning_rate": 9.951791375355659e-05,
+      "loss": 2.253,
+      "step": 5700
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7337315678596497,
+      "learning_rate": 9.951705389008852e-05,
+      "loss": 2.1121,
+      "step": 5705
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6792576909065247,
+      "learning_rate": 9.951619326418582e-05,
+      "loss": 2.2308,
+      "step": 5710
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7842636108398438,
+      "learning_rate": 9.951533187586178e-05,
+      "loss": 2.1758,
+      "step": 5715
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7006580829620361,
+      "learning_rate": 9.951446972512958e-05,
+      "loss": 2.1907,
+      "step": 5720
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7104856371879578,
+      "learning_rate": 9.951360681200256e-05,
+      "loss": 2.1451,
+      "step": 5725
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7118073105812073,
+      "learning_rate": 9.9512743136494e-05,
+      "loss": 2.2058,
+      "step": 5730
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7609933614730835,
+      "learning_rate": 9.951187869861715e-05,
+      "loss": 2.1959,
+      "step": 5735
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7531121969223022,
+      "learning_rate": 9.951101349838538e-05,
+      "loss": 2.2302,
+      "step": 5740
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7211287021636963,
+      "learning_rate": 9.9510147535812e-05,
+      "loss": 2.2106,
+      "step": 5745
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7329400181770325,
+      "learning_rate": 9.950928081091032e-05,
+      "loss": 2.1478,
+      "step": 5750
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7244143486022949,
+      "learning_rate": 9.95084133236937e-05,
+      "loss": 2.0842,
+      "step": 5755
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6806926727294922,
+      "learning_rate": 9.950754507417547e-05,
+      "loss": 2.1731,
+      "step": 5760
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7026325464248657,
+      "learning_rate": 9.950667606236906e-05,
+      "loss": 2.1561,
+      "step": 5765
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7056430578231812,
+      "learning_rate": 9.95058062882878e-05,
+      "loss": 2.1948,
+      "step": 5770
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7359597086906433,
+      "learning_rate": 9.950493575194509e-05,
+      "loss": 2.2111,
+      "step": 5775
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7660566568374634,
+      "learning_rate": 9.950406445335434e-05,
+      "loss": 2.1348,
+      "step": 5780
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7032355666160583,
+      "learning_rate": 9.950319239252896e-05,
+      "loss": 2.2291,
+      "step": 5785
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7162747979164124,
+      "learning_rate": 9.95023195694824e-05,
+      "loss": 2.1821,
+      "step": 5790
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7522682547569275,
+      "learning_rate": 9.950144598422809e-05,
+      "loss": 2.2241,
+      "step": 5795
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7413450479507446,
+      "learning_rate": 9.950057163677944e-05,
+      "loss": 2.1766,
+      "step": 5800
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7031062245368958,
+      "learning_rate": 9.949969652714997e-05,
+      "loss": 2.1084,
+      "step": 5805
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7174230217933655,
+      "learning_rate": 9.949882065535312e-05,
+      "loss": 2.1676,
+      "step": 5810
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.682476282119751,
+      "learning_rate": 9.949794402140239e-05,
+      "loss": 2.1236,
+      "step": 5815
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.766014575958252,
+      "learning_rate": 9.949706662531127e-05,
+      "loss": 2.1608,
+      "step": 5820
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7257121205329895,
+      "learning_rate": 9.949618846709327e-05,
+      "loss": 2.2428,
+      "step": 5825
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7186542749404907,
+      "learning_rate": 9.949530954676194e-05,
+      "loss": 2.313,
+      "step": 5830
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7648069858551025,
+      "learning_rate": 9.949442986433076e-05,
+      "loss": 2.1605,
+      "step": 5835
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.75149005651474,
+      "learning_rate": 9.949354941981332e-05,
+      "loss": 2.215,
+      "step": 5840
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7076643109321594,
+      "learning_rate": 9.949266821322315e-05,
+      "loss": 2.1276,
+      "step": 5845
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6976586580276489,
+      "learning_rate": 9.949178624457384e-05,
+      "loss": 2.1435,
+      "step": 5850
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7132282257080078,
+      "learning_rate": 9.949090351387895e-05,
+      "loss": 2.1747,
+      "step": 5855
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6956488490104675,
+      "learning_rate": 9.949002002115208e-05,
+      "loss": 2.2393,
+      "step": 5860
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7139861583709717,
+      "learning_rate": 9.948913576640683e-05,
+      "loss": 2.2227,
+      "step": 5865
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6706424355506897,
+      "learning_rate": 9.948825074965682e-05,
+      "loss": 2.0752,
+      "step": 5870
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.741961658000946,
+      "learning_rate": 9.948736497091569e-05,
+      "loss": 2.2841,
+      "step": 5875
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7491090893745422,
+      "learning_rate": 9.948647843019705e-05,
+      "loss": 2.1816,
+      "step": 5880
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7336971759796143,
+      "learning_rate": 9.948559112751456e-05,
+      "loss": 2.1113,
+      "step": 5885
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6787917017936707,
+      "learning_rate": 9.948470306288189e-05,
+      "loss": 2.1484,
+      "step": 5890
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7390604615211487,
+      "learning_rate": 9.948381423631273e-05,
+      "loss": 2.164,
+      "step": 5895
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.723648190498352,
+      "learning_rate": 9.948292464782072e-05,
+      "loss": 2.2314,
+      "step": 5900
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7192287445068359,
+      "learning_rate": 9.948203429741961e-05,
+      "loss": 2.1843,
+      "step": 5905
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7361121773719788,
+      "learning_rate": 9.948114318512307e-05,
+      "loss": 2.224,
+      "step": 5910
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7325870394706726,
+      "learning_rate": 9.948025131094482e-05,
+      "loss": 2.1579,
+      "step": 5915
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.721145510673523,
+      "learning_rate": 9.947935867489862e-05,
+      "loss": 2.2667,
+      "step": 5920
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6976220011711121,
+      "learning_rate": 9.94784652769982e-05,
+      "loss": 2.2537,
+      "step": 5925
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7340314984321594,
+      "learning_rate": 9.947757111725732e-05,
+      "loss": 2.2038,
+      "step": 5930
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6927967667579651,
+      "learning_rate": 9.947667619568974e-05,
+      "loss": 2.2276,
+      "step": 5935
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7019612193107605,
+      "learning_rate": 9.947578051230925e-05,
+      "loss": 2.1853,
+      "step": 5940
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6918236613273621,
+      "learning_rate": 9.947488406712962e-05,
+      "loss": 2.1048,
+      "step": 5945
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7252418398857117,
+      "learning_rate": 9.947398686016469e-05,
+      "loss": 2.2218,
+      "step": 5950
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7401813864707947,
+      "learning_rate": 9.947308889142824e-05,
+      "loss": 2.1035,
+      "step": 5955
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7462252378463745,
+      "learning_rate": 9.94721901609341e-05,
+      "loss": 2.1401,
+      "step": 5960
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6738715767860413,
+      "learning_rate": 9.947129066869614e-05,
+      "loss": 2.1997,
+      "step": 5965
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.746366024017334,
+      "learning_rate": 9.947039041472817e-05,
+      "loss": 2.117,
+      "step": 5970
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7170393466949463,
+      "learning_rate": 9.946948939904407e-05,
+      "loss": 2.1792,
+      "step": 5975
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6756740808486938,
+      "learning_rate": 9.946858762165772e-05,
+      "loss": 2.2055,
+      "step": 5980
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.680515706539154,
+      "learning_rate": 9.946768508258299e-05,
+      "loss": 2.2243,
+      "step": 5985
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7422165274620056,
+      "learning_rate": 9.946678178183377e-05,
+      "loss": 2.1693,
+      "step": 5990
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.729213297367096,
+      "learning_rate": 9.946587771942401e-05,
+      "loss": 2.2008,
+      "step": 5995
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6938876509666443,
+      "learning_rate": 9.946497289536758e-05,
+      "loss": 2.1059,
+      "step": 6000
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 126689,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 1.7487862362537984e+19,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}