diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,16599 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 11838,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.8175293207168579,
+      "learning_rate": 0.0001999999119654754,
+      "loss": 1.7625,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.461701363325119,
+      "learning_rate": 0.00019999964786205653,
+      "loss": 1.5481,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.5782307386398315,
+      "learning_rate": 0.00019999920769020845,
+      "loss": 1.466,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.6779885292053223,
+      "learning_rate": 0.00019999859145070615,
+      "loss": 1.2965,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.5275429487228394,
+      "learning_rate": 0.00019999779914463462,
+      "loss": 1.1209,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.4957067668437958,
+      "learning_rate": 0.0001999968307733889,
+      "loss": 1.0367,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.41212332248687744,
+      "learning_rate": 0.000199995686338674,
+      "loss": 1.1318,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.6483787894248962,
+      "learning_rate": 0.00019999436584250483,
+      "loss": 1.1682,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.5552895665168762,
+      "learning_rate": 0.00019999286928720647,
+      "loss": 1.1751,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.5129882097244263,
+      "learning_rate": 0.00019999119667541386,
+      "loss": 1.1218,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.504658043384552,
+      "learning_rate": 0.00019998934801007193,
+      "loss": 1.0083,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5507149696350098,
+      "learning_rate": 0.00019998732329443562,
+      "loss": 1.0734,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.4575830101966858,
+      "learning_rate": 0.00019998512253206982,
+      "loss": 1.1557,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5136922001838684,
+      "learning_rate": 0.0001999827457268494,
+      "loss": 1.0223,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.4587676227092743,
+      "learning_rate": 0.00019998019288295922,
+      "loss": 1.1057,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5710563063621521,
+      "learning_rate": 0.00019997746400489397,
+      "loss": 1.0945,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.4434487223625183,
+      "learning_rate": 0.00019997455909745844,
+      "loss": 1.1251,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5528855323791504,
+      "learning_rate": 0.00019997147816576717,
+      "loss": 1.1266,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5259051322937012,
+      "learning_rate": 0.00019996822121524485,
+      "loss": 1.1532,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.532082736492157,
+      "learning_rate": 0.00019996478825162585,
+      "loss": 1.3226,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5315178632736206,
+      "learning_rate": 0.00019996117928095463,
+      "loss": 1.0409,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5802783370018005,
+      "learning_rate": 0.00019995739430958545,
+      "loss": 1.1499,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5138378739356995,
+      "learning_rate": 0.00019995343334418245,
+      "loss": 1.0474,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.4681589603424072,
+      "learning_rate": 0.0001999492963917197,
+      "loss": 1.1214,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5345844626426697,
+      "learning_rate": 0.00019994498345948108,
+      "loss": 1.0918,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5528503060340881,
+      "learning_rate": 0.00019994049455506033,
+      "loss": 1.178,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5040786266326904,
+      "learning_rate": 0.00019993582968636097,
+      "loss": 0.9658,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.4512442350387573,
+      "learning_rate": 0.0001999309888615965,
+      "loss": 1.1075,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.464190810918808,
+      "learning_rate": 0.00019992597208929,
+      "loss": 1.08,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5537208914756775,
+      "learning_rate": 0.00019992077937827456,
+      "loss": 1.1566,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.42293214797973633,
+      "learning_rate": 0.00019991541073769283,
+      "loss": 1.0393,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.4625384211540222,
+      "learning_rate": 0.0001999098661769974,
+      "loss": 1.0042,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.4168233871459961,
+      "learning_rate": 0.0001999041457059505,
+      "loss": 1.0598,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.46117836236953735,
+      "learning_rate": 0.0001998982493346241,
+      "loss": 1.0951,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5079285502433777,
+      "learning_rate": 0.0001998921770733999,
+      "loss": 1.043,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5561312437057495,
+      "learning_rate": 0.00019988592893296927,
+      "loss": 0.9438,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5467772483825684,
+      "learning_rate": 0.00019987950492433325,
+      "loss": 0.9862,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.41703835129737854,
+      "learning_rate": 0.0001998729050588025,
+      "loss": 1.0838,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5075678825378418,
+      "learning_rate": 0.0001998661293479974,
+      "loss": 0.9902,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.50690096616745,
+      "learning_rate": 0.00019985917780384786,
+      "loss": 1.0627,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5631605386734009,
+      "learning_rate": 0.00019985205043859336,
+      "loss": 1.0876,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6694210171699524,
+      "learning_rate": 0.00019984474726478303,
+      "loss": 1.1046,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5434871912002563,
+      "learning_rate": 0.00019983726829527547,
+      "loss": 1.0263,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6015032529830933,
+      "learning_rate": 0.00019982961354323887,
+      "loss": 1.0756,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.4603692591190338,
+      "learning_rate": 0.00019982178302215082,
+      "loss": 1.0387,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5583211183547974,
+      "learning_rate": 0.00019981377674579845,
+      "loss": 0.9821,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7076869606971741,
+      "learning_rate": 0.00019980559472827843,
+      "loss": 1.1412,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5172699689865112,
+      "learning_rate": 0.00019979723698399665,
+      "loss": 1.0096,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5216347575187683,
+      "learning_rate": 0.00019978870352766853,
+      "loss": 0.946,
+      "step": 245
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.490485817193985,
+      "learning_rate": 0.0001997799943743189,
+      "loss": 1.0325,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5536062717437744,
+      "learning_rate": 0.00019977110953928182,
+      "loss": 0.9317,
+      "step": 255
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5582374930381775,
+      "learning_rate": 0.0001997620490382008,
+      "loss": 0.954,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.47851961851119995,
+      "learning_rate": 0.0001997528128870285,
+      "loss": 1.0212,
+      "step": 265
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5606931447982788,
+      "learning_rate": 0.00019974340110202697,
+      "loss": 1.0405,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.611594021320343,
+      "learning_rate": 0.00019973381369976746,
+      "loss": 1.0257,
+      "step": 275
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.46861839294433594,
+      "learning_rate": 0.00019972405069713041,
+      "loss": 0.9963,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6010798811912537,
+      "learning_rate": 0.00019971411211130543,
+      "loss": 1.1186,
+      "step": 285
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.427389919757843,
+      "learning_rate": 0.00019970399795979132,
+      "loss": 1.0869,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.46258246898651123,
+      "learning_rate": 0.00019969370826039592,
+      "loss": 0.9047,
+      "step": 295
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5251803994178772,
+      "learning_rate": 0.00019968324303123625,
+      "loss": 0.9507,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5841838121414185,
+      "learning_rate": 0.00019967260229073836,
+      "loss": 0.9455,
+      "step": 305
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6464457511901855,
+      "learning_rate": 0.00019966178605763726,
+      "loss": 1.1312,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5451318621635437,
+      "learning_rate": 0.00019965079435097698,
+      "loss": 1.0751,
+      "step": 315
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6237244009971619,
+      "learning_rate": 0.00019963962719011055,
+      "loss": 1.0577,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4929693043231964,
+      "learning_rate": 0.00019962828459469984,
+      "loss": 0.995,
+      "step": 325
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.529183566570282,
+      "learning_rate": 0.0001996167665847157,
+      "loss": 1.0397,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5752055644989014,
+      "learning_rate": 0.00019960507318043775,
+      "loss": 1.0677,
+      "step": 335
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5159496665000916,
+      "learning_rate": 0.00019959320440245443,
+      "loss": 1.0188,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5186464786529541,
+      "learning_rate": 0.00019958116027166307,
+      "loss": 1.0477,
+      "step": 345
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5070178508758545,
+      "learning_rate": 0.00019956894080926958,
+      "loss": 1.0221,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6111320853233337,
+      "learning_rate": 0.00019955654603678866,
+      "loss": 1.0038,
+      "step": 355
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5824604630470276,
+      "learning_rate": 0.0001995439759760437,
+      "loss": 1.1021,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5628572702407837,
+      "learning_rate": 0.00019953123064916665,
+      "loss": 1.0429,
+      "step": 365
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5787538290023804,
+      "learning_rate": 0.00019951831007859814,
+      "loss": 1.0226,
+      "step": 370
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.46730467677116394,
+      "learning_rate": 0.00019950521428708723,
+      "loss": 0.9446,
+      "step": 375
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5635990500450134,
+      "learning_rate": 0.0001994919432976916,
+      "loss": 1.081,
+      "step": 380
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.47697439789772034,
+      "learning_rate": 0.00019947849713377734,
+      "loss": 1.0084,
+      "step": 385
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.566819965839386,
+      "learning_rate": 0.00019946487581901895,
+      "loss": 0.9866,
+      "step": 390
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.585972785949707,
+      "learning_rate": 0.00019945107937739944,
+      "loss": 1.0689,
+      "step": 395
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5571792125701904,
+      "learning_rate": 0.00019943710783320998,
+      "loss": 0.9717,
+      "step": 400
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.506241500377655,
+      "learning_rate": 0.00019942296121105017,
+      "loss": 0.9637,
+      "step": 405
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5207057595252991,
+      "learning_rate": 0.00019940863953582787,
+      "loss": 0.9639,
+      "step": 410
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4859829246997833,
+      "learning_rate": 0.00019939414283275906,
+      "loss": 0.9717,
+      "step": 415
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.599038302898407,
+      "learning_rate": 0.00019937947112736796,
+      "loss": 1.1271,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6036107540130615,
+      "learning_rate": 0.00019936462444548693,
+      "loss": 0.9301,
+      "step": 425
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6932292580604553,
+      "learning_rate": 0.00019934960281325635,
+      "loss": 1.0803,
+      "step": 430
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6190482974052429,
+      "learning_rate": 0.0001993344062571247,
+      "loss": 0.9878,
+      "step": 435
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4994017481803894,
+      "learning_rate": 0.00019931903480384838,
+      "loss": 0.9301,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4884631633758545,
+      "learning_rate": 0.00019930348848049177,
+      "loss": 0.9366,
+      "step": 445
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5066394209861755,
+      "learning_rate": 0.00019928776731442712,
+      "loss": 0.9241,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5030893683433533,
+      "learning_rate": 0.00019927187133333456,
+      "loss": 0.9532,
+      "step": 455
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5532776713371277,
+      "learning_rate": 0.00019925580056520198,
+      "loss": 0.8736,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4686765968799591,
+      "learning_rate": 0.00019923955503832504,
+      "loss": 0.9661,
+      "step": 465
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5450586676597595,
+      "learning_rate": 0.00019922313478130713,
+      "loss": 0.9508,
+      "step": 470
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.536483645439148,
+      "learning_rate": 0.00019920653982305911,
+      "loss": 1.128,
+      "step": 475
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6655276417732239,
+      "learning_rate": 0.0001991897701927997,
+      "loss": 1.0245,
+      "step": 480
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6011082530021667,
+      "learning_rate": 0.00019917282592005496,
+      "loss": 1.1497,
+      "step": 485
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9368462562561035,
+      "learning_rate": 0.0001991557070346585,
+      "loss": 0.8886,
+      "step": 490
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6440569162368774,
+      "learning_rate": 0.00019913841356675142,
+      "loss": 1.1014,
+      "step": 495
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5625852942466736,
+      "learning_rate": 0.00019912094554678215,
+      "loss": 0.8747,
+      "step": 500
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5758554339408875,
+      "learning_rate": 0.00019910330300550646,
+      "loss": 1.0431,
+      "step": 505
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6677930355072021,
+      "learning_rate": 0.00019908548597398742,
+      "loss": 0.7627,
+      "step": 510
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4941328763961792,
+      "learning_rate": 0.0001990674944835953,
+      "loss": 0.9874,
+      "step": 515
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5404765009880066,
+      "learning_rate": 0.00019904932856600752,
+      "loss": 1.087,
+      "step": 520
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5204087495803833,
+      "learning_rate": 0.00019903098825320867,
+      "loss": 1.0357,
+      "step": 525
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.45817357301712036,
+      "learning_rate": 0.00019901247357749036,
+      "loss": 1.017,
+      "step": 530
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7364314794540405,
+      "learning_rate": 0.0001989937845714512,
+      "loss": 1.0032,
+      "step": 535
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4314422905445099,
+      "learning_rate": 0.00019897492126799674,
+      "loss": 0.8679,
+      "step": 540
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5871337652206421,
+      "learning_rate": 0.00019895588370033942,
+      "loss": 0.9531,
+      "step": 545
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5464487671852112,
+      "learning_rate": 0.00019893667190199848,
+      "loss": 1.074,
+      "step": 550
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5099916458129883,
+      "learning_rate": 0.00019891728590680003,
+      "loss": 1.026,
+      "step": 555
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5275132656097412,
+      "learning_rate": 0.00019889772574887673,
+      "loss": 1.0716,
+      "step": 560
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6441323161125183,
+      "learning_rate": 0.000198877991462668,
+      "loss": 1.0636,
+      "step": 565
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6186425089836121,
+      "learning_rate": 0.00019885808308291977,
+      "loss": 0.9923,
+      "step": 570
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6667066216468811,
+      "learning_rate": 0.00019883800064468462,
+      "loss": 1.0927,
+      "step": 575
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5096490383148193,
+      "learning_rate": 0.0001988177441833214,
+      "loss": 0.9456,
+      "step": 580
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.581665575504303,
+      "learning_rate": 0.00019879731373449554,
+      "loss": 1.1691,
+      "step": 585
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6705063581466675,
+      "learning_rate": 0.00019877670933417872,
+      "loss": 1.0357,
+      "step": 590
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5527098774909973,
+      "learning_rate": 0.0001987559310186489,
+      "loss": 0.9919,
+      "step": 595
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5956196188926697,
+      "learning_rate": 0.0001987349788244903,
+      "loss": 1.1586,
+      "step": 600
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5246900916099548,
+      "learning_rate": 0.0001987138527885932,
+      "loss": 1.0395,
+      "step": 605
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6794516444206238,
+      "learning_rate": 0.00019869255294815402,
+      "loss": 1.0114,
+      "step": 610
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6502415537834167,
+      "learning_rate": 0.00019867107934067523,
+      "loss": 1.0978,
+      "step": 615
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4909512996673584,
+      "learning_rate": 0.00019864943200396517,
+      "loss": 0.9087,
+      "step": 620
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6279435753822327,
+      "learning_rate": 0.0001986276109761381,
+      "loss": 1.0851,
+      "step": 625
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5172310471534729,
+      "learning_rate": 0.0001986056162956141,
+      "loss": 1.0416,
+      "step": 630
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.49973011016845703,
+      "learning_rate": 0.00019858344800111898,
+      "loss": 1.1246,
+      "step": 635
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5211731195449829,
+      "learning_rate": 0.0001985611061316843,
+      "loss": 0.9146,
+      "step": 640
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6123340725898743,
+      "learning_rate": 0.0001985385907266471,
+      "loss": 1.1233,
+      "step": 645
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5667193531990051,
+      "learning_rate": 0.00019851590182565012,
+      "loss": 1.0207,
+      "step": 650
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5176669359207153,
+      "learning_rate": 0.0001984930394686414,
+      "loss": 0.8969,
+      "step": 655
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5587745308876038,
+      "learning_rate": 0.00019847000369587457,
+      "loss": 0.9042,
+      "step": 660
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6178166270256042,
+      "learning_rate": 0.00019844679454790844,
+      "loss": 1.1374,
+      "step": 665
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.7185156941413879,
+      "learning_rate": 0.00019842341206560712,
+      "loss": 1.0347,
+      "step": 670
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6146171689033508,
+      "learning_rate": 0.00019839985629013999,
+      "loss": 0.9015,
+      "step": 675
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.550493061542511,
+      "learning_rate": 0.00019837612726298143,
+      "loss": 1.0746,
+      "step": 680
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5300818681716919,
+      "learning_rate": 0.0001983522250259109,
+      "loss": 1.0232,
+      "step": 685
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.7765306234359741,
+      "learning_rate": 0.0001983281496210129,
+      "loss": 1.0948,
+      "step": 690
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6479122638702393,
+      "learning_rate": 0.00019830390109067673,
+      "loss": 1.0204,
+      "step": 695
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6021077036857605,
+      "learning_rate": 0.00019827947947759653,
+      "loss": 0.9591,
+      "step": 700
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.7223843336105347,
+      "learning_rate": 0.0001982548848247712,
+      "loss": 1.0684,
+      "step": 705
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6151075959205627,
+      "learning_rate": 0.00019823011717550438,
+      "loss": 1.0078,
+      "step": 710
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5454387068748474,
+      "learning_rate": 0.0001982051765734042,
+      "loss": 0.9134,
+      "step": 715
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5072502493858337,
+      "learning_rate": 0.00019818006306238328,
+      "loss": 0.9802,
+      "step": 720
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5242615938186646,
+      "learning_rate": 0.0001981547766866588,
+      "loss": 1.0727,
+      "step": 725
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.573273241519928,
+      "learning_rate": 0.00019812931749075223,
+      "loss": 0.941,
+      "step": 730
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6620715856552124,
+      "learning_rate": 0.00019810368551948936,
+      "loss": 0.7937,
+      "step": 735
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5468537211418152,
+      "learning_rate": 0.00019807788081800012,
+      "loss": 0.9955,
+      "step": 740
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4930030405521393,
+      "learning_rate": 0.00019805190343171857,
+      "loss": 1.0195,
+      "step": 745
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6010162234306335,
+      "learning_rate": 0.00019802575340638296,
+      "loss": 0.8513,
+      "step": 750
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6166960000991821,
+      "learning_rate": 0.0001979994307880353,
+      "loss": 1.0954,
+      "step": 755
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6688129901885986,
+      "learning_rate": 0.00019797293562302158,
+      "loss": 0.9794,
+      "step": 760
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6365656852722168,
+      "learning_rate": 0.00019794626795799158,
+      "loss": 1.0153,
+      "step": 765
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5212395787239075,
+      "learning_rate": 0.00019791942783989889,
+      "loss": 0.9277,
+      "step": 770
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.7180500626564026,
+      "learning_rate": 0.00019789241531600053,
+      "loss": 1.001,
+      "step": 775
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5612985491752625,
+      "learning_rate": 0.00019786523043385727,
+      "loss": 1.0079,
+      "step": 780
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5919773578643799,
+      "learning_rate": 0.00019783787324133324,
+      "loss": 0.8491,
+      "step": 785
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5292619466781616,
+      "learning_rate": 0.00019781034378659604,
+      "loss": 0.8185,
+      "step": 790
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.45849528908729553,
+      "learning_rate": 0.00019778264211811646,
+      "loss": 0.8061,
+      "step": 795
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6099018454551697,
+      "learning_rate": 0.0001977547682846686,
+      "loss": 1.0933,
+      "step": 800
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4419599175453186,
+      "learning_rate": 0.00019772672233532964,
+      "loss": 1.0046,
+      "step": 805
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5074923634529114,
+      "learning_rate": 0.0001976985043194798,
+      "loss": 0.9954,
+      "step": 810
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5581572651863098,
+      "learning_rate": 0.00019767011428680227,
+      "loss": 0.8782,
+      "step": 815
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.603863537311554,
+      "learning_rate": 0.00019764155228728315,
+      "loss": 0.8936,
+      "step": 820
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6839099526405334,
+      "learning_rate": 0.0001976128183712113,
+      "loss": 0.98,
+      "step": 825
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6504345536231995,
+      "learning_rate": 0.00019758391258917814,
+      "loss": 0.9985,
+      "step": 830
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5674350261688232,
+      "learning_rate": 0.0001975548349920779,
+      "loss": 1.0153,
+      "step": 835
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5308940410614014,
+      "learning_rate": 0.00019752558563110724,
+      "loss": 1.0934,
+      "step": 840
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6243263483047485,
+      "learning_rate": 0.0001974961645577652,
+      "loss": 1.059,
+      "step": 845
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5877118706703186,
+      "learning_rate": 0.00019746657182385314,
+      "loss": 0.9244,
+      "step": 850
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6568809151649475,
+      "learning_rate": 0.00019743680748147478,
+      "loss": 0.8841,
+      "step": 855
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5473036170005798,
+      "learning_rate": 0.00019740687158303585,
+      "loss": 0.8777,
+      "step": 860
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6813970804214478,
+      "learning_rate": 0.0001973767641812443,
+      "loss": 1.0363,
+      "step": 865
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6258594393730164,
+      "learning_rate": 0.00019734648532910982,
+      "loss": 0.9106,
+      "step": 870
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5775337815284729,
+      "learning_rate": 0.00019731603507994416,
+      "loss": 1.0125,
+      "step": 875
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5688369870185852,
+      "learning_rate": 0.00019728541348736084,
+      "loss": 0.9422,
+      "step": 880
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6406494975090027,
+      "learning_rate": 0.00019725462060527489,
+      "loss": 1.0349,
+      "step": 885
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5844593048095703,
+      "learning_rate": 0.00019722365648790313,
+      "loss": 0.8833,
+      "step": 890
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5722345113754272,
+      "learning_rate": 0.00019719252118976374,
+      "loss": 0.9489,
+      "step": 895
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.49358856678009033,
+      "learning_rate": 0.00019716121476567639,
+      "loss": 0.9482,
+      "step": 900
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.777001142501831,
+      "learning_rate": 0.00019712973727076195,
+      "loss": 1.0312,
+      "step": 905
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5142468810081482,
+      "learning_rate": 0.0001970980887604426,
+      "loss": 0.9216,
+      "step": 910
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5355789065361023,
+      "learning_rate": 0.0001970662692904415,
+      "loss": 0.9777,
+      "step": 915
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6558182239532471,
+      "learning_rate": 0.000197034278916783,
+      "loss": 1.035,
+      "step": 920
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5900774598121643,
+      "learning_rate": 0.00019700211769579213,
+      "loss": 0.9305,
+      "step": 925
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6186517477035522,
+      "learning_rate": 0.00019696978568409495,
+      "loss": 1.0441,
+      "step": 930
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5638003349304199,
+      "learning_rate": 0.000196937282938618,
+      "loss": 0.9227,
+      "step": 935
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.574424147605896,
+      "learning_rate": 0.0001969046095165887,
+      "loss": 1.0394,
+      "step": 940
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5076507925987244,
+      "learning_rate": 0.0001968717654755347,
+      "loss": 0.9493,
+      "step": 945
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5305479168891907,
+      "learning_rate": 0.00019683875087328427,
+      "loss": 0.9148,
+      "step": 950
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.7044605612754822,
+      "learning_rate": 0.0001968055657679659,
+      "loss": 0.9093,
+      "step": 955
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6829433441162109,
+      "learning_rate": 0.00019677221021800824,
+      "loss": 0.9498,
+      "step": 960
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.691895067691803,
+      "learning_rate": 0.00019673868428214016,
+      "loss": 1.002,
+      "step": 965
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5967201590538025,
+      "learning_rate": 0.00019670498801939044,
+      "loss": 0.9357,
+      "step": 970
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6142829060554504,
+      "learning_rate": 0.0001966711214890877,
+      "loss": 1.0207,
+      "step": 975
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5852668881416321,
+      "learning_rate": 0.0001966370847508605,
+      "loss": 1.0109,
+      "step": 980
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.572533369064331,
+      "learning_rate": 0.00019660287786463698,
+      "loss": 0.8034,
+      "step": 985
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5755594968795776,
+      "learning_rate": 0.00019656850089064484,
+      "loss": 1.0993,
+      "step": 990
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6088036298751831,
+      "learning_rate": 0.00019653395388941137,
+      "loss": 0.9437,
+      "step": 995
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5511387586593628,
+      "learning_rate": 0.00019649923692176304,
+      "loss": 0.9882,
+      "step": 1000
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5409395098686218,
+      "learning_rate": 0.00019646435004882576,
+      "loss": 1.038,
+      "step": 1005
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.629104495048523,
+      "learning_rate": 0.00019642929333202452,
+      "loss": 0.9842,
+      "step": 1010
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5766791105270386,
+      "learning_rate": 0.00019639406683308336,
+      "loss": 1.0024,
+      "step": 1015
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4806615114212036,
+      "learning_rate": 0.00019635867061402516,
+      "loss": 0.9551,
+      "step": 1020
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5499693751335144,
+      "learning_rate": 0.00019632310473717172,
+      "loss": 0.9023,
+      "step": 1025
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6102238893508911,
+      "learning_rate": 0.00019628736926514365,
+      "loss": 0.9405,
+      "step": 1030
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5422660708427429,
+      "learning_rate": 0.00019625146426085994,
+      "loss": 1.0331,
+      "step": 1035
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5721043944358826,
+      "learning_rate": 0.00019621538978753823,
+      "loss": 0.9257,
+      "step": 1040
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6052049994468689,
+      "learning_rate": 0.00019617914590869452,
+      "loss": 0.9577,
+      "step": 1045
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5600439310073853,
+      "learning_rate": 0.00019614273268814305,
+      "loss": 0.8236,
+      "step": 1050
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5360490083694458,
+      "learning_rate": 0.00019610615018999622,
+      "loss": 1.0483,
+      "step": 1055
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6123297810554504,
+      "learning_rate": 0.0001960693984786645,
+      "loss": 0.9726,
+      "step": 1060
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5726840496063232,
+      "learning_rate": 0.00019603247761885629,
+      "loss": 1.0083,
+      "step": 1065
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5117479562759399,
+      "learning_rate": 0.00019599538767557775,
+      "loss": 0.8722,
+      "step": 1070
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5932919979095459,
+      "learning_rate": 0.00019595812871413281,
+      "loss": 1.0538,
+      "step": 1075
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5998363494873047,
+      "learning_rate": 0.00019592070080012302,
+      "loss": 0.9474,
+      "step": 1080
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5751232504844666,
+      "learning_rate": 0.00019588310399944726,
+      "loss": 0.9461,
+      "step": 1085
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6181055903434753,
+      "learning_rate": 0.00019584533837830196,
+      "loss": 0.9216,
+      "step": 1090
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5857065916061401,
+      "learning_rate": 0.00019580740400318062,
+      "loss": 0.8235,
+      "step": 1095
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6219877600669861,
+      "learning_rate": 0.00019576930094087396,
+      "loss": 1.0098,
+      "step": 1100
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5622178316116333,
+      "learning_rate": 0.00019573102925846968,
+      "loss": 0.8896,
+      "step": 1105
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4788983464241028,
+      "learning_rate": 0.00019569258902335236,
+      "loss": 0.9877,
+      "step": 1110
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5626773238182068,
+      "learning_rate": 0.00019565398030320336,
+      "loss": 0.9384,
+      "step": 1115
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6275051236152649,
+      "learning_rate": 0.0001956152031660007,
+      "loss": 0.9299,
+      "step": 1120
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5800061821937561,
+      "learning_rate": 0.00019557625768001886,
+      "loss": 0.825,
+      "step": 1125
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.717424213886261,
+      "learning_rate": 0.00019553714391382887,
+      "loss": 0.93,
+      "step": 1130
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5426806211471558,
+      "learning_rate": 0.0001954978619362979,
+      "loss": 1.007,
+      "step": 1135
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.578883945941925,
+      "learning_rate": 0.00019545841181658943,
+      "loss": 1.0774,
+      "step": 1140
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5970021486282349,
+      "learning_rate": 0.0001954187936241628,
+      "loss": 0.9603,
+      "step": 1145
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.669302761554718,
+      "learning_rate": 0.00019537900742877344,
+      "loss": 1.0047,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5312142968177795,
+      "learning_rate": 0.00019533905330047256,
+      "loss": 0.9533,
+      "step": 1155
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5459814667701721,
+      "learning_rate": 0.000195298931309607,
+      "loss": 1.0127,
+      "step": 1160
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6206276416778564,
+      "learning_rate": 0.00019525864152681913,
+      "loss": 0.8151,
+      "step": 1165
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6036674380302429,
+      "learning_rate": 0.00019521818402304681,
+      "loss": 0.9973,
+      "step": 1170
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5393249988555908,
+      "learning_rate": 0.0001951775588695232,
+      "loss": 1.0176,
+      "step": 1175
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6075353026390076,
+      "learning_rate": 0.0001951367661377766,
+      "loss": 1.0152,
+      "step": 1180
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5007694363594055,
+      "learning_rate": 0.00019509580589963034,
+      "loss": 0.9276,
+      "step": 1185
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5564121007919312,
+      "learning_rate": 0.0001950546782272028,
+      "loss": 0.9572,
+      "step": 1190
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5426845550537109,
+      "learning_rate": 0.00019501338319290708,
+      "loss": 0.8693,
+      "step": 1195
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6294403076171875,
+      "learning_rate": 0.00019497192086945093,
+      "loss": 1.0606,
+      "step": 1200
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5286447405815125,
+      "learning_rate": 0.00019493029132983662,
+      "loss": 0.9979,
+      "step": 1205
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4923780858516693,
+      "learning_rate": 0.00019488849464736096,
+      "loss": 1.07,
+      "step": 1210
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5261374711990356,
+      "learning_rate": 0.00019484653089561494,
+      "loss": 1.0666,
+      "step": 1215
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5969892740249634,
+      "learning_rate": 0.00019480440014848377,
+      "loss": 0.8119,
+      "step": 1220
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5572890043258667,
+      "learning_rate": 0.00019476210248014656,
+      "loss": 0.9422,
+      "step": 1225
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7353805303573608,
+      "learning_rate": 0.0001947196379650765,
+      "loss": 0.9641,
+      "step": 1230
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5229889750480652,
+      "learning_rate": 0.00019467700667804048,
+      "loss": 0.9063,
+      "step": 1235
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4695720672607422,
+      "learning_rate": 0.00019463420869409893,
+      "loss": 1.0855,
+      "step": 1240
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5980749726295471,
+      "learning_rate": 0.00019459124408860586,
+      "loss": 0.9487,
+      "step": 1245
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5264121294021606,
+      "learning_rate": 0.0001945481129372087,
+      "loss": 0.9755,
+      "step": 1250
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5308972001075745,
+      "learning_rate": 0.000194504815315848,
+      "loss": 0.9853,
+      "step": 1255
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5365180373191833,
+      "learning_rate": 0.0001944613513007575,
+      "loss": 0.9656,
+      "step": 1260
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5717142820358276,
+      "learning_rate": 0.00019441772096846384,
+      "loss": 0.952,
+      "step": 1265
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6359343528747559,
+      "learning_rate": 0.0001943739243957866,
+      "loss": 0.8685,
+      "step": 1270
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4931330382823944,
+      "learning_rate": 0.00019432996165983797,
+      "loss": 0.9423,
+      "step": 1275
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6247501373291016,
+      "learning_rate": 0.00019428583283802265,
+      "loss": 1.0109,
+      "step": 1280
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5827012062072754,
+      "learning_rate": 0.0001942415380080379,
+      "loss": 0.9511,
+      "step": 1285
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5535582304000854,
+      "learning_rate": 0.00019419707724787323,
+      "loss": 0.9364,
+      "step": 1290
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5828814506530762,
+      "learning_rate": 0.00019415245063581025,
+      "loss": 0.9613,
+      "step": 1295
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5350412726402283,
+      "learning_rate": 0.00019410765825042257,
+      "loss": 0.9448,
+      "step": 1300
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7406845092773438,
+      "learning_rate": 0.00019406270017057576,
+      "loss": 0.9062,
+      "step": 1305
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5344066023826599,
+      "learning_rate": 0.00019401757647542707,
+      "loss": 0.8728,
+      "step": 1310
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6511924862861633,
+      "learning_rate": 0.00019397228724442537,
+      "loss": 0.871,
+      "step": 1315
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5950039625167847,
+      "learning_rate": 0.00019392683255731096,
+      "loss": 1.0393,
+      "step": 1320
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8080087304115295,
+      "learning_rate": 0.00019388121249411553,
+      "loss": 1.0526,
+      "step": 1325
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5649018883705139,
+      "learning_rate": 0.0001938354271351618,
+      "loss": 0.9657,
+      "step": 1330
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6691799163818359,
+      "learning_rate": 0.00019378947656106373,
+      "loss": 0.9644,
+      "step": 1335
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6107922792434692,
+      "learning_rate": 0.00019374336085272595,
+      "loss": 0.9491,
+      "step": 1340
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5547528862953186,
+      "learning_rate": 0.000193697080091344,
+      "loss": 1.004,
+      "step": 1345
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6795949935913086,
+      "learning_rate": 0.000193650634358404,
+      "loss": 1.0367,
+      "step": 1350
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6405168175697327,
+      "learning_rate": 0.00019360402373568247,
+      "loss": 0.9986,
+      "step": 1355
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5481475591659546,
+      "learning_rate": 0.0001935572483052463,
+      "loss": 1.0394,
+      "step": 1360
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5024700164794922,
+      "learning_rate": 0.00019351030814945255,
+      "loss": 0.9751,
+      "step": 1365
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5783774852752686,
+      "learning_rate": 0.0001934632033509483,
+      "loss": 0.7837,
+      "step": 1370
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6434618830680847,
+      "learning_rate": 0.00019341593399267053,
+      "loss": 0.8806,
+      "step": 1375
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6476423740386963,
+      "learning_rate": 0.00019336850015784594,
+      "loss": 1.0671,
+      "step": 1380
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6362334489822388,
+      "learning_rate": 0.00019332090192999087,
+      "loss": 0.9831,
+      "step": 1385
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5181714296340942,
+      "learning_rate": 0.000193273139392911,
+      "loss": 0.8729,
+      "step": 1390
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5226086378097534,
+      "learning_rate": 0.0001932252126307014,
+      "loss": 0.9101,
+      "step": 1395
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5152550339698792,
+      "learning_rate": 0.00019317712172774632,
+      "loss": 0.8719,
+      "step": 1400
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5058336853981018,
+      "learning_rate": 0.00019312886676871888,
+      "loss": 0.8942,
+      "step": 1405
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5755054354667664,
+      "learning_rate": 0.00019308044783858115,
+      "loss": 0.9656,
+      "step": 1410
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6656663417816162,
+      "learning_rate": 0.0001930318650225839,
+      "loss": 1.0673,
+      "step": 1415
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5695021152496338,
+      "learning_rate": 0.0001929831184062664,
+      "loss": 1.0599,
+      "step": 1420
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.520157516002655,
+      "learning_rate": 0.0001929342080754564,
+      "loss": 1.0071,
+      "step": 1425
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5740432739257812,
+      "learning_rate": 0.00019288513411626983,
+      "loss": 0.9698,
+      "step": 1430
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5819209218025208,
+      "learning_rate": 0.00019283589661511072,
+      "loss": 0.8553,
+      "step": 1435
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6635054349899292,
+      "learning_rate": 0.0001927864956586711,
+      "loss": 0.9225,
+      "step": 1440
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6044415235519409,
+      "learning_rate": 0.00019273693133393076,
+      "loss": 1.0187,
+      "step": 1445
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5304946899414062,
+      "learning_rate": 0.00019268720372815713,
+      "loss": 0.9037,
+      "step": 1450
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5266116857528687,
+      "learning_rate": 0.00019263731292890515,
+      "loss": 0.9544,
+      "step": 1455
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5947398543357849,
+      "learning_rate": 0.00019258725902401703,
+      "loss": 1.1737,
+      "step": 1460
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5687413215637207,
+      "learning_rate": 0.00019253704210162224,
+      "loss": 0.8613,
+      "step": 1465
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6292214393615723,
+      "learning_rate": 0.00019248666225013726,
+      "loss": 1.0017,
+      "step": 1470
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5356031656265259,
+      "learning_rate": 0.00019243611955826537,
+      "loss": 0.855,
+      "step": 1475
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6292675733566284,
+      "learning_rate": 0.00019238541411499663,
+      "loss": 0.9568,
+      "step": 1480
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.562177836894989,
+      "learning_rate": 0.0001923345460096076,
+      "loss": 0.9332,
+      "step": 1485
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6126812696456909,
+      "learning_rate": 0.00019228351533166134,
+      "loss": 1.0263,
+      "step": 1490
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6032052636146545,
+      "learning_rate": 0.000192232322171007,
+      "loss": 0.9851,
+      "step": 1495
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.5387076735496521,
+      "learning_rate": 0.00019218096661777992,
+      "loss": 0.913,
+      "step": 1500
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.5938889980316162,
+      "learning_rate": 0.00019212944876240137,
+      "loss": 0.9152,
+      "step": 1505
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.5894052982330322,
+      "learning_rate": 0.00019207776869557833,
+      "loss": 0.9353,
+      "step": 1510
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6678389310836792,
+      "learning_rate": 0.00019202592650830337,
+      "loss": 0.992,
+      "step": 1515
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.5318301916122437,
+      "learning_rate": 0.00019197392229185453,
+      "loss": 0.9352,
+      "step": 1520
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6742917895317078,
+      "learning_rate": 0.0001919217561377952,
+      "loss": 0.8734,
+      "step": 1525
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.7230522632598877,
+      "learning_rate": 0.0001918694281379738,
+      "loss": 1.0778,
+      "step": 1530
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6975288391113281,
+      "learning_rate": 0.0001918169383845237,
+      "loss": 1.0108,
+      "step": 1535
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.5289484858512878,
+      "learning_rate": 0.0001917642869698632,
+      "loss": 0.9754,
+      "step": 1540
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.5557776093482971,
+      "learning_rate": 0.0001917114739866951,
+      "loss": 1.016,
+      "step": 1545
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.5662341117858887,
+      "learning_rate": 0.00019165849952800667,
+      "loss": 0.7972,
+      "step": 1550
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6539111733436584,
+      "learning_rate": 0.0001916053636870696,
+      "loss": 1.0211,
+      "step": 1555
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6270111203193665,
+      "learning_rate": 0.00019155206655743965,
+      "loss": 0.9429,
+      "step": 1560
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.633138120174408,
+      "learning_rate": 0.00019149860823295656,
+      "loss": 1.1118,
+      "step": 1565
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.5337046384811401,
+      "learning_rate": 0.00019144498880774386,
+      "loss": 0.8627,
+      "step": 1570
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.7054827809333801,
+      "learning_rate": 0.00019139120837620882,
+      "loss": 1.0627,
+      "step": 1575
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.7774983644485474,
+      "learning_rate": 0.00019133726703304208,
+      "loss": 0.9882,
+      "step": 1580
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.571898877620697,
+      "learning_rate": 0.00019128316487321772,
+      "loss": 0.988,
+      "step": 1585
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.5544953942298889,
+      "learning_rate": 0.00019122890199199284,
+      "loss": 0.8627,
+      "step": 1590
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6004731059074402,
+      "learning_rate": 0.0001911744784849076,
+      "loss": 0.8951,
+      "step": 1595
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.526996910572052,
+      "learning_rate": 0.00019111989444778492,
+      "loss": 1.012,
+      "step": 1600
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.68979811668396,
+      "learning_rate": 0.00019106514997673047,
+      "loss": 0.9303,
+      "step": 1605
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6508918404579163,
+      "learning_rate": 0.00019101024516813224,
+      "loss": 1.0252,
+      "step": 1610
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.5442551374435425,
+      "learning_rate": 0.00019095518011866063,
+      "loss": 1.0041,
+      "step": 1615
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6386384963989258,
+      "learning_rate": 0.0001908999549252682,
+      "loss": 1.0107,
+      "step": 1620
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6549636125564575,
+      "learning_rate": 0.0001908445696851893,
+      "loss": 0.9851,
+      "step": 1625
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.5537642240524292,
+      "learning_rate": 0.00019078902449594032,
+      "loss": 0.9206,
+      "step": 1630
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.5848385095596313,
+      "learning_rate": 0.00019073331945531908,
+      "loss": 0.9487,
+      "step": 1635
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6486941576004028,
+      "learning_rate": 0.00019067745466140495,
+      "loss": 0.8375,
+      "step": 1640
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.5679529905319214,
+      "learning_rate": 0.0001906214302125586,
+      "loss": 0.99,
+      "step": 1645
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.5364441275596619,
+      "learning_rate": 0.00019056524620742157,
+      "loss": 0.9085,
+      "step": 1650
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.5058781504631042,
+      "learning_rate": 0.00019050890274491665,
+      "loss": 0.9123,
+      "step": 1655
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.820630669593811,
+      "learning_rate": 0.00019045239992424717,
+      "loss": 1.0015,
+      "step": 1660
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.5033766031265259,
+      "learning_rate": 0.00019039573784489716,
+      "loss": 0.937,
+      "step": 1665
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.585077166557312,
+      "learning_rate": 0.00019033891660663098,
+      "loss": 1.0317,
+      "step": 1670
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.5304557681083679,
+      "learning_rate": 0.00019028193630949323,
+      "loss": 0.9043,
+      "step": 1675
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6207611560821533,
+      "learning_rate": 0.00019022479705380857,
+      "loss": 0.9307,
+      "step": 1680
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7342296242713928,
+      "learning_rate": 0.0001901674989401816,
+      "loss": 0.8608,
+      "step": 1685
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6580296158790588,
+      "learning_rate": 0.00019011004206949652,
+      "loss": 0.9218,
+      "step": 1690
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.5700413584709167,
+      "learning_rate": 0.00019005242654291708,
+      "loss": 1.0436,
+      "step": 1695
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6400768756866455,
+      "learning_rate": 0.00018999465246188644,
+      "loss": 1.0718,
+      "step": 1700
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.5975671410560608,
+      "learning_rate": 0.00018993671992812683,
+      "loss": 0.9152,
+      "step": 1705
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.5661541223526001,
+      "learning_rate": 0.00018987862904363954,
+      "loss": 1.0337,
+      "step": 1710
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7099330425262451,
+      "learning_rate": 0.00018982037991070462,
+      "loss": 0.9411,
+      "step": 1715
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.5101850628852844,
+      "learning_rate": 0.00018976197263188079,
+      "loss": 0.9404,
+      "step": 1720
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.5761948823928833,
+      "learning_rate": 0.00018970340731000516,
+      "loss": 0.9742,
+      "step": 1725
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6656047105789185,
+      "learning_rate": 0.00018964468404819313,
+      "loss": 0.8846,
+      "step": 1730
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6487800478935242,
+      "learning_rate": 0.00018958580294983822,
+      "loss": 1.0617,
+      "step": 1735
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.549220621585846,
+      "learning_rate": 0.00018952676411861184,
+      "loss": 0.943,
+      "step": 1740
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.5477379560470581,
+      "learning_rate": 0.00018946756765846304,
+      "loss": 0.9236,
+      "step": 1745
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.618018627166748,
+      "learning_rate": 0.00018940821367361847,
+      "loss": 0.9085,
+      "step": 1750
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7567707300186157,
+      "learning_rate": 0.00018934870226858217,
+      "loss": 0.9261,
+      "step": 1755
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6278589963912964,
+      "learning_rate": 0.0001892890335481353,
+      "loss": 0.8587,
+      "step": 1760
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.5153677463531494,
+      "learning_rate": 0.00018922920761733596,
+      "loss": 0.9223,
+      "step": 1765
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.5698601007461548,
+      "learning_rate": 0.00018916922458151914,
+      "loss": 0.9802,
+      "step": 1770
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.548460066318512,
+      "learning_rate": 0.0001891090845462964,
+      "loss": 0.8717,
+      "step": 1775
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.5761381983757019,
+      "learning_rate": 0.00018904878761755569,
+      "loss": 0.8788,
+      "step": 1780
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6153483986854553,
+      "learning_rate": 0.0001889883339014613,
+      "loss": 0.9666,
+      "step": 1785
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6116446256637573,
+      "learning_rate": 0.00018892772350445345,
+      "loss": 0.9085,
+      "step": 1790
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6375726461410522,
+      "learning_rate": 0.00018886695653324832,
+      "loss": 0.9871,
+      "step": 1795
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6123708486557007,
+      "learning_rate": 0.00018880603309483776,
+      "loss": 0.8711,
+      "step": 1800
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6581029295921326,
+      "learning_rate": 0.00018874495329648908,
+      "loss": 1.0394,
+      "step": 1805
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6889798641204834,
+      "learning_rate": 0.00018868371724574488,
+      "loss": 0.8321,
+      "step": 1810
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.5388063788414001,
+      "learning_rate": 0.00018862232505042288,
+      "loss": 0.9827,
+      "step": 1815
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7022605538368225,
+      "learning_rate": 0.00018856077681861578,
+      "loss": 0.9111,
+      "step": 1820
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6853319406509399,
+      "learning_rate": 0.0001884990726586909,
+      "loss": 0.9423,
+      "step": 1825
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.5777489542961121,
+      "learning_rate": 0.00018843721267929023,
+      "loss": 0.8929,
+      "step": 1830
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8301356434822083,
+      "learning_rate": 0.00018837519698933002,
+      "loss": 1.0039,
+      "step": 1835
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6395474076271057,
+      "learning_rate": 0.00018831302569800073,
+      "loss": 0.8821,
+      "step": 1840
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7763014435768127,
+      "learning_rate": 0.00018825069891476671,
+      "loss": 0.9198,
+      "step": 1845
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.583583652973175,
+      "learning_rate": 0.00018818821674936623,
+      "loss": 0.9855,
+      "step": 1850
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.5729549527168274,
+      "learning_rate": 0.00018812557931181093,
+      "loss": 0.9374,
+      "step": 1855
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6349241733551025,
+      "learning_rate": 0.000188062786712386,
+      "loss": 0.9622,
+      "step": 1860
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.49465417861938477,
+      "learning_rate": 0.00018799983906164983,
+      "loss": 0.9116,
+      "step": 1865
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.678386390209198,
+      "learning_rate": 0.00018793673647043364,
+      "loss": 0.9196,
+      "step": 1870
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6616655588150024,
+      "learning_rate": 0.00018787347904984165,
+      "loss": 1.0232,
+      "step": 1875
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.512854278087616,
+      "learning_rate": 0.00018781006691125053,
+      "loss": 0.8166,
+      "step": 1880
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6445545554161072,
+      "learning_rate": 0.0001877465001663095,
+      "loss": 1.0117,
+      "step": 1885
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7158140540122986,
+      "learning_rate": 0.0001876827789269399,
+      "loss": 0.9406,
+      "step": 1890
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6912984848022461,
+      "learning_rate": 0.0001876189033053351,
+      "loss": 1.0287,
+      "step": 1895
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7598811388015747,
+      "learning_rate": 0.00018755487341396028,
+      "loss": 1.0113,
+      "step": 1900
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6084948778152466,
+      "learning_rate": 0.00018749068936555228,
+      "loss": 1.0907,
+      "step": 1905
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.625379741191864,
+      "learning_rate": 0.00018742635127311935,
+      "loss": 0.9019,
+      "step": 1910
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.5288641452789307,
+      "learning_rate": 0.00018736185924994096,
+      "loss": 0.8875,
+      "step": 1915
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.5389487147331238,
+      "learning_rate": 0.00018729721340956758,
+      "loss": 0.9304,
+      "step": 1920
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6083305478096008,
+      "learning_rate": 0.0001872324138658206,
+      "loss": 0.966,
+      "step": 1925
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.5934799909591675,
+      "learning_rate": 0.00018716746073279184,
+      "loss": 0.8495,
+      "step": 1930
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.5639356970787048,
+      "learning_rate": 0.00018710235412484373,
+      "loss": 0.9915,
+      "step": 1935
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.5736001133918762,
+      "learning_rate": 0.00018703709415660887,
+      "loss": 0.8608,
+      "step": 1940
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8066298365592957,
+      "learning_rate": 0.00018697168094298984,
+      "loss": 0.9448,
+      "step": 1945
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.5450537204742432,
+      "learning_rate": 0.00018690611459915908,
+      "loss": 0.9935,
+      "step": 1950
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6609448194503784,
+      "learning_rate": 0.00018684039524055862,
+      "loss": 1.0216,
+      "step": 1955
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6248445510864258,
+      "learning_rate": 0.0001867745229828999,
+      "loss": 0.9785,
+      "step": 1960
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7231089472770691,
+      "learning_rate": 0.00018670849794216355,
+      "loss": 0.8399,
+      "step": 1965
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5523176193237305,
+      "learning_rate": 0.00018664232023459933,
+      "loss": 1.0047,
+      "step": 1970
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.662384569644928,
+      "learning_rate": 0.00018657598997672562,
+      "loss": 1.1134,
+      "step": 1975
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6732993125915527,
+      "learning_rate": 0.00018650950728532948,
+      "loss": 1.0003,
+      "step": 1980
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6265495419502258,
+      "learning_rate": 0.00018644287227746636,
+      "loss": 0.9312,
+      "step": 1985
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6594348549842834,
+      "learning_rate": 0.0001863760850704599,
+      "loss": 1.1153,
+      "step": 1990
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5969790816307068,
+      "learning_rate": 0.0001863091457819017,
+      "loss": 0.87,
+      "step": 1995
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6349937319755554,
+      "learning_rate": 0.00018624205452965112,
+      "loss": 1.0112,
+      "step": 2000
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5987355709075928,
+      "learning_rate": 0.00018617481143183508,
+      "loss": 1.0038,
+      "step": 2005
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6250818371772766,
+      "learning_rate": 0.00018610741660684784,
+      "loss": 0.9675,
+      "step": 2010
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6271193623542786,
+      "learning_rate": 0.00018603987017335092,
+      "loss": 0.9785,
+      "step": 2015
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7074173092842102,
+      "learning_rate": 0.0001859721722502726,
+      "loss": 0.8998,
+      "step": 2020
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5771471261978149,
+      "learning_rate": 0.000185904322956808,
+      "loss": 0.9635,
+      "step": 2025
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5951381325721741,
+      "learning_rate": 0.0001858363224124187,
+      "loss": 0.8828,
+      "step": 2030
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8378333449363708,
+      "learning_rate": 0.0001857681707368326,
+      "loss": 1.0037,
+      "step": 2035
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5882629752159119,
+      "learning_rate": 0.0001856998680500438,
+      "loss": 0.993,
+      "step": 2040
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.571729302406311,
+      "learning_rate": 0.00018563141447231211,
+      "loss": 1.0208,
+      "step": 2045
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5603118538856506,
+      "learning_rate": 0.0001855628101241631,
+      "loss": 0.9141,
+      "step": 2050
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6984039545059204,
+      "learning_rate": 0.00018549405512638783,
+      "loss": 1.0206,
+      "step": 2055
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5990726947784424,
+      "learning_rate": 0.00018542514960004253,
+      "loss": 0.9379,
+      "step": 2060
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6375360488891602,
+      "learning_rate": 0.0001853560936664485,
+      "loss": 0.9176,
+      "step": 2065
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.629737377166748,
+      "learning_rate": 0.00018528688744719193,
+      "loss": 0.8489,
+      "step": 2070
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.60643470287323,
+      "learning_rate": 0.0001852175310641235,
+      "loss": 0.9451,
+      "step": 2075
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6629886031150818,
+      "learning_rate": 0.00018514802463935834,
+      "loss": 1.0165,
+      "step": 2080
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.49199485778808594,
+      "learning_rate": 0.00018507836829527574,
+      "loss": 0.9393,
+      "step": 2085
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6570359468460083,
+      "learning_rate": 0.000185008562154519,
+      "loss": 0.924,
+      "step": 2090
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6161836981773376,
+      "learning_rate": 0.00018493860633999508,
+      "loss": 0.8037,
+      "step": 2095
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.697126567363739,
+      "learning_rate": 0.00018486850097487457,
+      "loss": 0.9855,
+      "step": 2100
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6809502243995667,
+      "learning_rate": 0.00018479824618259128,
+      "loss": 1.0517,
+      "step": 2105
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.61480313539505,
+      "learning_rate": 0.0001847278420868422,
+      "loss": 0.9386,
+      "step": 2110
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.5264391899108887,
+      "learning_rate": 0.00018465728881158708,
+      "loss": 0.8437,
+      "step": 2115
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.559967577457428,
+      "learning_rate": 0.00018458658648104844,
+      "loss": 1.0144,
+      "step": 2120
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.5891844034194946,
+      "learning_rate": 0.00018451573521971123,
+      "loss": 0.9301,
+      "step": 2125
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6855583190917969,
+      "learning_rate": 0.00018444473515232256,
+      "loss": 0.8999,
+      "step": 2130
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.547498881816864,
+      "learning_rate": 0.00018437358640389158,
+      "loss": 0.8687,
+      "step": 2135
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.565273642539978,
+      "learning_rate": 0.00018430228909968921,
+      "loss": 0.9559,
+      "step": 2140
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.5726015567779541,
+      "learning_rate": 0.00018423084336524793,
+      "loss": 0.9186,
+      "step": 2145
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6429528594017029,
+      "learning_rate": 0.00018415924932636157,
+      "loss": 0.7512,
+      "step": 2150
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6005961298942566,
+      "learning_rate": 0.0001840875071090851,
+      "loss": 0.9153,
+      "step": 2155
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.5854561924934387,
+      "learning_rate": 0.00018401561683973434,
+      "loss": 0.9366,
+      "step": 2160
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6333120465278625,
+      "learning_rate": 0.0001839435786448858,
+      "loss": 0.9863,
+      "step": 2165
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6162759065628052,
+      "learning_rate": 0.00018387139265137642,
+      "loss": 0.9348,
+      "step": 2170
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.5720029473304749,
+      "learning_rate": 0.00018379905898630345,
+      "loss": 0.8502,
+      "step": 2175
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.5871778130531311,
+      "learning_rate": 0.00018372657777702406,
+      "loss": 0.9709,
+      "step": 2180
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6709279417991638,
+      "learning_rate": 0.00018365394915115517,
+      "loss": 1.149,
+      "step": 2185
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.5620378851890564,
+      "learning_rate": 0.0001835811732365734,
+      "loss": 0.9973,
+      "step": 2190
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.758850634098053,
+      "learning_rate": 0.00018350825016141457,
+      "loss": 0.9039,
+      "step": 2195
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6607004404067993,
+      "learning_rate": 0.00018343518005407367,
+      "loss": 0.8812,
+      "step": 2200
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6985838413238525,
+      "learning_rate": 0.0001833619630432045,
+      "loss": 1.0202,
+      "step": 2205
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.809180736541748,
+      "learning_rate": 0.00018328859925771958,
+      "loss": 0.9662,
+      "step": 2210
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6806530952453613,
+      "learning_rate": 0.0001832150888267898,
+      "loss": 0.8784,
+      "step": 2215
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.5656675100326538,
+      "learning_rate": 0.00018314143187984433,
+      "loss": 0.905,
+      "step": 2220
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.694346010684967,
+      "learning_rate": 0.00018306762854657023,
+      "loss": 1.007,
+      "step": 2225
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6591660976409912,
+      "learning_rate": 0.00018299367895691234,
+      "loss": 0.8863,
+      "step": 2230
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6169084310531616,
+      "learning_rate": 0.00018291958324107298,
+      "loss": 0.8985,
+      "step": 2235
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7212360501289368,
+      "learning_rate": 0.00018284534152951176,
+      "loss": 0.9613,
+      "step": 2240
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.5280630588531494,
+      "learning_rate": 0.00018277095395294538,
+      "loss": 0.9924,
+      "step": 2245
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.5625287890434265,
+      "learning_rate": 0.00018269642064234733,
+      "loss": 0.8689,
+      "step": 2250
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6884534955024719,
+      "learning_rate": 0.0001826217417289477,
+      "loss": 0.9303,
+      "step": 2255
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.5620412230491638,
+      "learning_rate": 0.00018254691734423295,
+      "loss": 0.8627,
+      "step": 2260
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.5788276195526123,
+      "learning_rate": 0.00018247194761994567,
+      "loss": 1.0121,
+      "step": 2265
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.5471813082695007,
+      "learning_rate": 0.00018239683268808432,
+      "loss": 1.0408,
+      "step": 2270
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.5952843427658081,
+      "learning_rate": 0.00018232157268090307,
+      "loss": 1.019,
+      "step": 2275
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.5923483967781067,
+      "learning_rate": 0.00018224616773091147,
+      "loss": 0.9152,
+      "step": 2280
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7464591860771179,
+      "learning_rate": 0.00018217061797087434,
+      "loss": 0.8466,
+      "step": 2285
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7097582221031189,
+      "learning_rate": 0.00018209492353381138,
+      "loss": 0.9672,
+      "step": 2290
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6305751800537109,
+      "learning_rate": 0.00018201908455299707,
+      "loss": 0.9242,
+      "step": 2295
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.657504141330719,
+      "learning_rate": 0.00018194310116196043,
+      "loss": 0.9943,
+      "step": 2300
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.5786561965942383,
+      "learning_rate": 0.00018186697349448463,
+      "loss": 0.9458,
+      "step": 2305
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6431105732917786,
+      "learning_rate": 0.000181790701684607,
+      "loss": 0.9829,
+      "step": 2310
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6315768957138062,
+      "learning_rate": 0.0001817142858666185,
+      "loss": 1.018,
+      "step": 2315
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6315834522247314,
+      "learning_rate": 0.00018163772617506383,
+      "loss": 0.9245,
+      "step": 2320
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6845858097076416,
+      "learning_rate": 0.00018156102274474086,
+      "loss": 0.9734,
+      "step": 2325
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5911626815795898,
+      "learning_rate": 0.00018148417571070056,
+      "loss": 0.8466,
+      "step": 2330
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7361390590667725,
+      "learning_rate": 0.00018140718520824684,
+      "loss": 0.8586,
+      "step": 2335
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6635876297950745,
+      "learning_rate": 0.0001813300513729361,
+      "loss": 1.0269,
+      "step": 2340
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6086567640304565,
+      "learning_rate": 0.0001812527743405772,
+      "loss": 0.9184,
+      "step": 2345
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5980998277664185,
+      "learning_rate": 0.00018117535424723102,
+      "loss": 0.9805,
+      "step": 2350
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5727255344390869,
+      "learning_rate": 0.0001810977912292104,
+      "loss": 1.0672,
+      "step": 2355
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6550287008285522,
+      "learning_rate": 0.00018102008542307982,
+      "loss": 1.0177,
+      "step": 2360
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5286814570426941,
+      "learning_rate": 0.00018094223696565512,
+      "loss": 0.7283,
+      "step": 2365
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7307307124137878,
+      "learning_rate": 0.0001808642459940034,
+      "loss": 0.9459,
+      "step": 2370
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7078204154968262,
+      "learning_rate": 0.0001807861126454426,
+      "loss": 0.9447,
+      "step": 2375
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6107788681983948,
+      "learning_rate": 0.00018070783705754134,
+      "loss": 1.0948,
+      "step": 2380
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5652554035186768,
+      "learning_rate": 0.00018062941936811868,
+      "loss": 0.9937,
+      "step": 2385
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5260435342788696,
+      "learning_rate": 0.00018055085971524398,
+      "loss": 0.9008,
+      "step": 2390
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6040837168693542,
+      "learning_rate": 0.0001804721582372364,
+      "loss": 0.9693,
+      "step": 2395
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5115237832069397,
+      "learning_rate": 0.00018039331507266492,
+      "loss": 0.9628,
+      "step": 2400
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6985922455787659,
+      "learning_rate": 0.00018031433036034793,
+      "loss": 0.9271,
+      "step": 2405
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.9261385798454285,
+      "learning_rate": 0.0001802352042393531,
+      "loss": 0.898,
+      "step": 2410
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5070507526397705,
+      "learning_rate": 0.00018015593684899702,
+      "loss": 0.9087,
+      "step": 2415
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5075459480285645,
+      "learning_rate": 0.000180076528328845,
+      "loss": 0.909,
+      "step": 2420
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5567365288734436,
+      "learning_rate": 0.0001799969788187109,
+      "loss": 0.912,
+      "step": 2425
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6508195996284485,
+      "learning_rate": 0.0001799172884586568,
+      "loss": 0.8513,
+      "step": 2430
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7663083672523499,
+      "learning_rate": 0.0001798374573889927,
+      "loss": 0.9042,
+      "step": 2435
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5899822115898132,
+      "learning_rate": 0.00017975748575027646,
+      "loss": 0.9742,
+      "step": 2440
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5543133616447449,
+      "learning_rate": 0.00017967737368331337,
+      "loss": 0.9616,
+      "step": 2445
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5752557516098022,
+      "learning_rate": 0.00017959712132915599,
+      "loss": 0.9027,
+      "step": 2450
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7131261825561523,
+      "learning_rate": 0.00017951672882910385,
+      "loss": 0.9206,
+      "step": 2455
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7709518671035767,
+      "learning_rate": 0.0001794361963247033,
+      "loss": 0.934,
+      "step": 2460
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5232674479484558,
+      "learning_rate": 0.00017935552395774708,
+      "loss": 0.9652,
+      "step": 2465
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6537283658981323,
+      "learning_rate": 0.00017927471187027436,
+      "loss": 0.9757,
+      "step": 2470
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5960981249809265,
+      "learning_rate": 0.00017919376020457018,
+      "loss": 0.9607,
+      "step": 2475
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5479128956794739,
+      "learning_rate": 0.0001791126691031653,
+      "loss": 0.9325,
+      "step": 2480
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8228976726531982,
+      "learning_rate": 0.00017903143870883615,
+      "loss": 0.9168,
+      "step": 2485
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6928779482841492,
+      "learning_rate": 0.00017895006916460426,
+      "loss": 0.9469,
+      "step": 2490
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6461488008499146,
+      "learning_rate": 0.00017886856061373623,
+      "loss": 0.8883,
+      "step": 2495
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5686217546463013,
+      "learning_rate": 0.00017878691319974337,
+      "loss": 0.8932,
+      "step": 2500
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6429763436317444,
+      "learning_rate": 0.00017870512706638148,
+      "loss": 0.9782,
+      "step": 2505
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.645390510559082,
+      "learning_rate": 0.0001786232023576507,
+      "loss": 0.9944,
+      "step": 2510
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6108589768409729,
+      "learning_rate": 0.00017854113921779509,
+      "loss": 1.0078,
+      "step": 2515
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5913373231887817,
+      "learning_rate": 0.00017845893779130237,
+      "loss": 0.8979,
+      "step": 2520
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.711692750453949,
+      "learning_rate": 0.00017837659822290386,
+      "loss": 1.014,
+      "step": 2525
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5642191767692566,
+      "learning_rate": 0.00017829412065757398,
+      "loss": 1.0521,
+      "step": 2530
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5885759592056274,
+      "learning_rate": 0.0001782115052405303,
+      "loss": 0.9956,
+      "step": 2535
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6232182383537292,
+      "learning_rate": 0.00017812875211723291,
+      "loss": 0.9338,
+      "step": 2540
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6030421257019043,
+      "learning_rate": 0.00017804586143338455,
+      "loss": 0.9516,
+      "step": 2545
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7771492004394531,
+      "learning_rate": 0.00017796283333492997,
+      "loss": 1.0976,
+      "step": 2550
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5960396528244019,
+      "learning_rate": 0.00017787966796805596,
+      "loss": 0.9811,
+      "step": 2555
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6035930514335632,
+      "learning_rate": 0.00017779636547919102,
+      "loss": 0.9693,
+      "step": 2560
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6715995669364929,
+      "learning_rate": 0.00017771292601500505,
+      "loss": 1.0308,
+      "step": 2565
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5918522477149963,
+      "learning_rate": 0.00017762934972240913,
+      "loss": 0.8584,
+      "step": 2570
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5988922715187073,
+      "learning_rate": 0.0001775456367485552,
+      "loss": 0.8162,
+      "step": 2575
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5941436886787415,
+      "learning_rate": 0.00017746178724083593,
+      "loss": 0.8879,
+      "step": 2580
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.697307288646698,
+      "learning_rate": 0.00017737780134688435,
+      "loss": 1.0048,
+      "step": 2585
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5817750692367554,
+      "learning_rate": 0.00017729367921457363,
+      "loss": 0.9586,
+      "step": 2590
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6275402307510376,
+      "learning_rate": 0.00017720942099201678,
+      "loss": 0.9415,
+      "step": 2595
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5959211587905884,
+      "learning_rate": 0.00017712502682756646,
+      "loss": 0.8882,
+      "step": 2600
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5991825461387634,
+      "learning_rate": 0.00017704049686981471,
+      "loss": 1.0279,
+      "step": 2605
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6534688472747803,
+      "learning_rate": 0.0001769558312675926,
+      "loss": 0.9533,
+      "step": 2610
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.56687331199646,
+      "learning_rate": 0.00017687103016997003,
+      "loss": 0.8215,
+      "step": 2615
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7316619753837585,
+      "learning_rate": 0.0001767860937262555,
+      "loss": 0.9756,
+      "step": 2620
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.580111026763916,
+      "learning_rate": 0.0001767010220859958,
+      "loss": 0.909,
+      "step": 2625
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6079258918762207,
+      "learning_rate": 0.00017661581539897577,
+      "loss": 0.8558,
+      "step": 2630
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5412486791610718,
+      "learning_rate": 0.000176530473815218,
+      "loss": 0.9592,
+      "step": 2635
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5887441635131836,
+      "learning_rate": 0.00017644499748498263,
+      "loss": 0.8299,
+      "step": 2640
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.62538743019104,
+      "learning_rate": 0.000176359386558767,
+      "loss": 0.8315,
+      "step": 2645
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4921663999557495,
+      "learning_rate": 0.00017627364118730544,
+      "loss": 0.7318,
+      "step": 2650
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6708532571792603,
+      "learning_rate": 0.00017618776152156901,
+      "loss": 0.9884,
+      "step": 2655
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5910497903823853,
+      "learning_rate": 0.00017610174771276525,
+      "loss": 0.905,
+      "step": 2660
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6253485679626465,
+      "learning_rate": 0.0001760155999123378,
+      "loss": 0.9262,
+      "step": 2665
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6524810194969177,
+      "learning_rate": 0.0001759293182719664,
+      "loss": 1.0608,
+      "step": 2670
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5854840874671936,
+      "learning_rate": 0.00017584290294356616,
+      "loss": 0.9515,
+      "step": 2675
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.645991325378418,
+      "learning_rate": 0.00017575635407928784,
+      "loss": 0.9688,
+      "step": 2680
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.56763756275177,
+      "learning_rate": 0.00017566967183151714,
+      "loss": 0.9068,
+      "step": 2685
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5318624377250671,
+      "learning_rate": 0.00017558285635287465,
+      "loss": 0.9443,
+      "step": 2690
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7761218547821045,
+      "learning_rate": 0.00017549590779621563,
+      "loss": 0.9784,
+      "step": 2695
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6218341588973999,
+      "learning_rate": 0.00017540882631462954,
+      "loss": 0.9407,
+      "step": 2700
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6817795634269714,
+      "learning_rate": 0.00017532161206143993,
+      "loss": 0.9117,
+      "step": 2705
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6102976202964783,
+      "learning_rate": 0.0001752342651902041,
+      "loss": 0.9124,
+      "step": 2710
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5011729001998901,
+      "learning_rate": 0.00017514678585471284,
+      "loss": 0.8357,
+      "step": 2715
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.8257537484169006,
+      "learning_rate": 0.00017505917420899018,
+      "loss": 0.8757,
+      "step": 2720
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6271941065788269,
+      "learning_rate": 0.00017497143040729314,
+      "loss": 0.9015,
+      "step": 2725
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6705794334411621,
+      "learning_rate": 0.0001748835546041114,
+      "loss": 0.9526,
+      "step": 2730
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5870561003684998,
+      "learning_rate": 0.000174795546954167,
+      "loss": 0.9221,
+      "step": 2735
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6101082563400269,
+      "learning_rate": 0.00017470740761241422,
+      "loss": 0.886,
+      "step": 2740
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6358086466789246,
+      "learning_rate": 0.00017461913673403915,
+      "loss": 1.0226,
+      "step": 2745
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4912102222442627,
+      "learning_rate": 0.00017453073447445952,
+      "loss": 0.8895,
+      "step": 2750
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6197881698608398,
+      "learning_rate": 0.0001744422009893243,
+      "loss": 0.8461,
+      "step": 2755
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6709913015365601,
+      "learning_rate": 0.00017435353643451357,
+      "loss": 0.9809,
+      "step": 2760
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6350414156913757,
+      "learning_rate": 0.00017426474096613812,
+      "loss": 0.9783,
+      "step": 2765
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5429471135139465,
+      "learning_rate": 0.00017417581474053938,
+      "loss": 0.8892,
+      "step": 2770
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6633498668670654,
+      "learning_rate": 0.00017408675791428886,
+      "loss": 1.0242,
+      "step": 2775
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5957458019256592,
+      "learning_rate": 0.00017399757064418805,
+      "loss": 0.8556,
+      "step": 2780
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6234441995620728,
+      "learning_rate": 0.00017390825308726817,
+      "loss": 1.1179,
+      "step": 2785
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6558435559272766,
+      "learning_rate": 0.00017381880540078974,
+      "loss": 1.0413,
+      "step": 2790
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.618256151676178,
+      "learning_rate": 0.0001737292277422425,
+      "loss": 0.9414,
+      "step": 2795
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.5883316397666931,
+      "learning_rate": 0.000173639520269345,
+      "loss": 0.8783,
+      "step": 2800
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7670148015022278,
+      "learning_rate": 0.0001735496831400443,
+      "loss": 0.8774,
+      "step": 2805
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7569119334220886,
+      "learning_rate": 0.00017345971651251576,
+      "loss": 1.0018,
+      "step": 2810
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6054137945175171,
+      "learning_rate": 0.00017336962054516277,
+      "loss": 0.9388,
+      "step": 2815
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7382391691207886,
+      "learning_rate": 0.0001732793953966165,
+      "loss": 0.8181,
+      "step": 2820
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6390617489814758,
+      "learning_rate": 0.00017318904122573542,
+      "loss": 0.9499,
+      "step": 2825
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.606440007686615,
+      "learning_rate": 0.00017309855819160535,
+      "loss": 0.8691,
+      "step": 2830
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.5001400709152222,
+      "learning_rate": 0.00017300794645353884,
+      "loss": 0.9139,
+      "step": 2835
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.627774178981781,
+      "learning_rate": 0.00017291720617107516,
+      "loss": 0.9617,
+      "step": 2840
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6031743288040161,
+      "learning_rate": 0.00017282633750397984,
+      "loss": 0.9019,
+      "step": 2845
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6420722007751465,
+      "learning_rate": 0.0001727353406122445,
+      "loss": 1.001,
+      "step": 2850
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7209452390670776,
+      "learning_rate": 0.00017264421565608648,
+      "loss": 1.0402,
+      "step": 2855
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6446462869644165,
+      "learning_rate": 0.00017255296279594862,
+      "loss": 0.9832,
+      "step": 2860
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6604474782943726,
+      "learning_rate": 0.000172461582192499,
+      "loss": 0.9241,
+      "step": 2865
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6114985942840576,
+      "learning_rate": 0.00017237007400663053,
+      "loss": 0.8937,
+      "step": 2870
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6553681492805481,
+      "learning_rate": 0.0001722784383994608,
+      "loss": 0.9854,
+      "step": 2875
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.5967362523078918,
+      "learning_rate": 0.00017218667553233182,
+      "loss": 0.9486,
+      "step": 2880
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.5432732105255127,
+      "learning_rate": 0.00017209478556680957,
+      "loss": 0.9295,
+      "step": 2885
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.579897940158844,
+      "learning_rate": 0.00017200276866468375,
+      "loss": 0.8984,
+      "step": 2890
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.5317714214324951,
+      "learning_rate": 0.0001719106249879678,
+      "loss": 0.8364,
+      "step": 2895
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7333313226699829,
+      "learning_rate": 0.00017181835469889812,
+      "loss": 0.9368,
+      "step": 2900
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.663776159286499,
+      "learning_rate": 0.00017172595795993413,
+      "loss": 0.9061,
+      "step": 2905
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6138433814048767,
+      "learning_rate": 0.0001716334349337579,
+      "loss": 0.8793,
+      "step": 2910
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.7783580422401428,
+      "learning_rate": 0.00017154078578327387,
+      "loss": 0.9532,
+      "step": 2915
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6159131526947021,
+      "learning_rate": 0.00017144801067160844,
+      "loss": 0.96,
+      "step": 2920
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6060130596160889,
+      "learning_rate": 0.00017135510976211,
+      "loss": 0.8548,
+      "step": 2925
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6942853927612305,
+      "learning_rate": 0.0001712620832183482,
+      "loss": 0.9283,
+      "step": 2930
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.489607036113739,
+      "learning_rate": 0.00017116893120411398,
+      "loss": 0.9536,
+      "step": 2935
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6349008679389954,
+      "learning_rate": 0.00017107565388341925,
+      "loss": 0.8852,
+      "step": 2940
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6873804926872253,
+      "learning_rate": 0.0001709822514204965,
+      "loss": 0.915,
+      "step": 2945
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.5652430653572083,
+      "learning_rate": 0.00017088872397979854,
+      "loss": 0.9261,
+      "step": 2950
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6061018109321594,
+      "learning_rate": 0.00017079507172599828,
+      "loss": 0.9668,
+      "step": 2955
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.5947713851928711,
+      "learning_rate": 0.00017070129482398832,
+      "loss": 0.9097,
+      "step": 2960
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6620383858680725,
+      "learning_rate": 0.00017060739343888076,
+      "loss": 0.9397,
+      "step": 2965
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.7726333141326904,
+      "learning_rate": 0.00017051336773600686,
+      "loss": 0.9941,
+      "step": 2970
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6241410970687866,
+      "learning_rate": 0.00017041921788091684,
+      "loss": 0.9903,
+      "step": 2975
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6833446025848389,
+      "learning_rate": 0.0001703249440393794,
+      "loss": 1.0074,
+      "step": 2980
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6586493253707886,
+      "learning_rate": 0.0001702305463773816,
+      "loss": 0.8399,
+      "step": 2985
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6630938053131104,
+      "learning_rate": 0.00017013602506112853,
+      "loss": 0.8064,
+      "step": 2990
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6912885904312134,
+      "learning_rate": 0.00017004138025704298,
+      "loss": 0.957,
+      "step": 2995
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6541727185249329,
+      "learning_rate": 0.00016994661213176512,
+      "loss": 0.9641,
+      "step": 3000
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6963586211204529,
+      "learning_rate": 0.00016985172085215235,
+      "loss": 0.9264,
+      "step": 3005
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6190909147262573,
+      "learning_rate": 0.00016975670658527875,
+      "loss": 0.9477,
+      "step": 3010
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.7124261260032654,
+      "learning_rate": 0.00016966156949843513,
+      "loss": 0.8615,
+      "step": 3015
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.5455434918403625,
+      "learning_rate": 0.0001695663097591284,
+      "loss": 0.8761,
+      "step": 3020
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6497594118118286,
+      "learning_rate": 0.00016947092753508147,
+      "loss": 0.9232,
+      "step": 3025
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6465092301368713,
+      "learning_rate": 0.00016937542299423294,
+      "loss": 0.9114,
+      "step": 3030
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.5808166861534119,
+      "learning_rate": 0.00016927979630473677,
+      "loss": 0.9503,
+      "step": 3035
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.5648312568664551,
+      "learning_rate": 0.0001691840476349619,
+      "loss": 0.9362,
+      "step": 3040
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.714506983757019,
+      "learning_rate": 0.00016908817715349217,
+      "loss": 0.9906,
+      "step": 3045
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.5561381578445435,
+      "learning_rate": 0.00016899218502912578,
+      "loss": 0.8099,
+      "step": 3050
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.5070639848709106,
+      "learning_rate": 0.00016889607143087516,
+      "loss": 1.008,
+      "step": 3055
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6084549427032471,
+      "learning_rate": 0.0001687998365279666,
+      "loss": 1.0026,
+      "step": 3060
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.633202850818634,
+      "learning_rate": 0.00016870348048984,
+      "loss": 0.9752,
+      "step": 3065
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.7638620138168335,
+      "learning_rate": 0.0001686070034861485,
+      "loss": 0.7598,
+      "step": 3070
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6333314776420593,
+      "learning_rate": 0.0001685104056867583,
+      "loss": 0.9456,
+      "step": 3075
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6526273488998413,
+      "learning_rate": 0.00016841368726174812,
+      "loss": 0.9099,
+      "step": 3080
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6543826460838318,
+      "learning_rate": 0.00016831684838140927,
+      "loss": 0.9487,
+      "step": 3085
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.5424016714096069,
+      "learning_rate": 0.00016821988921624499,
+      "loss": 0.8815,
+      "step": 3090
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6248295307159424,
+      "learning_rate": 0.00016812280993697037,
+      "loss": 0.9128,
+      "step": 3095
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6224409341812134,
+      "learning_rate": 0.000168025610714512,
+      "loss": 0.9597,
+      "step": 3100
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6363341212272644,
+      "learning_rate": 0.0001679282917200076,
+      "loss": 0.9314,
+      "step": 3105
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6113150119781494,
+      "learning_rate": 0.00016783085312480585,
+      "loss": 1.0532,
+      "step": 3110
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6174533367156982,
+      "learning_rate": 0.00016773329510046586,
+      "loss": 1.0273,
+      "step": 3115
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6600282788276672,
+      "learning_rate": 0.0001676356178187572,
+      "loss": 0.958,
+      "step": 3120
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.5333192944526672,
+      "learning_rate": 0.0001675378214516593,
+      "loss": 0.9824,
+      "step": 3125
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.745349109172821,
+      "learning_rate": 0.00016743990617136128,
+      "loss": 1.0886,
+      "step": 3130
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6083282232284546,
+      "learning_rate": 0.00016734187215026167,
+      "loss": 1.0196,
+      "step": 3135
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5257988572120667,
+      "learning_rate": 0.000167243719560968,
+      "loss": 0.813,
+      "step": 3140
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6657525300979614,
+      "learning_rate": 0.00016714544857629666,
+      "loss": 0.9309,
+      "step": 3145
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.7205111980438232,
+      "learning_rate": 0.00016704705936927244,
+      "loss": 0.8655,
+      "step": 3150
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.0305330753326416,
+      "learning_rate": 0.00016694855211312818,
+      "loss": 1.0213,
+      "step": 3155
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.7004939317703247,
+      "learning_rate": 0.00016684992698130476,
+      "loss": 0.9316,
+      "step": 3160
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.663589358329773,
+      "learning_rate": 0.00016675118414745052,
+      "loss": 0.8924,
+      "step": 3165
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5968930125236511,
+      "learning_rate": 0.000166652323785421,
+      "loss": 0.926,
+      "step": 3170
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6623892784118652,
+      "learning_rate": 0.00016655334606927865,
+      "loss": 0.937,
+      "step": 3175
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.7068220376968384,
+      "learning_rate": 0.00016645425117329268,
+      "loss": 1.0194,
+      "step": 3180
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5999795794487,
+      "learning_rate": 0.0001663550392719385,
+      "loss": 0.9363,
+      "step": 3185
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5697855949401855,
+      "learning_rate": 0.00016625571053989754,
+      "loss": 0.8442,
+      "step": 3190
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5499809384346008,
+      "learning_rate": 0.00016615626515205695,
+      "loss": 0.8832,
+      "step": 3195
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.780519425868988,
+      "learning_rate": 0.00016605670328350932,
+      "loss": 0.9941,
+      "step": 3200
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.710145890712738,
+      "learning_rate": 0.00016595702510955227,
+      "loss": 1.0341,
+      "step": 3205
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.7597246766090393,
+      "learning_rate": 0.00016585723080568817,
+      "loss": 0.8778,
+      "step": 3210
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.662965714931488,
+      "learning_rate": 0.00016575732054762397,
+      "loss": 0.8733,
+      "step": 3215
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6497308015823364,
+      "learning_rate": 0.00016565729451127067,
+      "loss": 0.8922,
+      "step": 3220
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.7098811864852905,
+      "learning_rate": 0.00016555715287274318,
+      "loss": 0.8309,
+      "step": 3225
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5802999138832092,
+      "learning_rate": 0.00016545689580835994,
+      "loss": 0.9676,
+      "step": 3230
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6192752718925476,
+      "learning_rate": 0.00016535652349464254,
+      "loss": 0.8422,
+      "step": 3235
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.604736864566803,
+      "learning_rate": 0.00016525603610831566,
+      "loss": 0.8476,
+      "step": 3240
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6643623113632202,
+      "learning_rate": 0.0001651554338263064,
+      "loss": 1.0216,
+      "step": 3245
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.7391632199287415,
+      "learning_rate": 0.0001650547168257443,
+      "loss": 0.864,
+      "step": 3250
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.7031934857368469,
+      "learning_rate": 0.0001649538852839608,
+      "loss": 0.9264,
+      "step": 3255
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.7183375954627991,
+      "learning_rate": 0.00016485293937848903,
+      "loss": 0.9041,
+      "step": 3260
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.7928013801574707,
+      "learning_rate": 0.0001647518792870635,
+      "loss": 1.0241,
+      "step": 3265
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.7501674294471741,
+      "learning_rate": 0.00016465070518761977,
+      "loss": 0.9598,
+      "step": 3270
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.7421630024909973,
+      "learning_rate": 0.00016454941725829405,
+      "loss": 0.9176,
+      "step": 3275
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6832653284072876,
+      "learning_rate": 0.0001644480156774231,
+      "loss": 0.8812,
+      "step": 3280
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6840265989303589,
+      "learning_rate": 0.0001643465006235437,
+      "loss": 0.9848,
+      "step": 3285
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6492328643798828,
+      "learning_rate": 0.00016424487227539243,
+      "loss": 0.9241,
+      "step": 3290
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.5492807626724243,
+      "learning_rate": 0.00016414313081190537,
+      "loss": 0.8956,
+      "step": 3295
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6606726050376892,
+      "learning_rate": 0.00016404127641221774,
+      "loss": 0.8719,
+      "step": 3300
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6319718360900879,
+      "learning_rate": 0.00016393930925566358,
+      "loss": 0.9931,
+      "step": 3305
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6968238949775696,
+      "learning_rate": 0.00016383722952177557,
+      "loss": 0.9831,
+      "step": 3310
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.7028079032897949,
+      "learning_rate": 0.00016373503739028448,
+      "loss": 1.0404,
+      "step": 3315
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.7885096073150635,
+      "learning_rate": 0.00016363273304111902,
+      "loss": 0.9998,
+      "step": 3320
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.7284576892852783,
+      "learning_rate": 0.00016353031665440547,
+      "loss": 0.8875,
+      "step": 3325
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.9729665517807007,
+      "learning_rate": 0.00016342778841046745,
+      "loss": 0.9376,
+      "step": 3330
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4611142575740814,
+      "learning_rate": 0.00016332514848982542,
+      "loss": 0.8302,
+      "step": 3335
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6475715041160583,
+      "learning_rate": 0.00016322239707319648,
+      "loss": 0.9361,
+      "step": 3340
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.771458625793457,
+      "learning_rate": 0.00016311953434149413,
+      "loss": 0.8875,
+      "step": 3345
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6289612054824829,
+      "learning_rate": 0.0001630165604758278,
+      "loss": 0.8626,
+      "step": 3350
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.659892737865448,
+      "learning_rate": 0.00016291347565750255,
+      "loss": 0.8926,
+      "step": 3355
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.5767649412155151,
+      "learning_rate": 0.00016281028006801887,
+      "loss": 0.9045,
+      "step": 3360
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.5347493886947632,
+      "learning_rate": 0.0001627069738890723,
+      "loss": 0.83,
+      "step": 3365
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6171711087226868,
+      "learning_rate": 0.00016260355730255297,
+      "loss": 0.8179,
+      "step": 3370
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.7306368350982666,
+      "learning_rate": 0.0001625000304905455,
+      "loss": 0.9047,
+      "step": 3375
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5920233130455017,
+      "learning_rate": 0.00016239639363532858,
+      "loss": 0.9198,
+      "step": 3380
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6208842992782593,
+      "learning_rate": 0.00016229264691937462,
+      "loss": 0.9004,
+      "step": 3385
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5397033095359802,
+      "learning_rate": 0.00016218879052534949,
+      "loss": 1.021,
+      "step": 3390
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6394155025482178,
+      "learning_rate": 0.0001620848246361122,
+      "loss": 0.9549,
+      "step": 3395
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6426846981048584,
+      "learning_rate": 0.0001619807494347144,
+      "loss": 0.8002,
+      "step": 3400
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6755315065383911,
+      "learning_rate": 0.0001618765651044004,
+      "loss": 0.8586,
+      "step": 3405
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6798297762870789,
+      "learning_rate": 0.00016177227182860647,
+      "loss": 0.8918,
+      "step": 3410
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6921480298042297,
+      "learning_rate": 0.00016166786979096088,
+      "loss": 0.8891,
+      "step": 3415
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.521027684211731,
+      "learning_rate": 0.00016156335917528325,
+      "loss": 0.8009,
+      "step": 3420
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6095343232154846,
+      "learning_rate": 0.00016145874016558443,
+      "loss": 0.9309,
+      "step": 3425
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6683770418167114,
+      "learning_rate": 0.00016135401294606618,
+      "loss": 0.9387,
+      "step": 3430
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.9191362261772156,
+      "learning_rate": 0.0001612491777011206,
+      "loss": 0.7803,
+      "step": 3435
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.7780088186264038,
+      "learning_rate": 0.00016114423461533026,
+      "loss": 0.9855,
+      "step": 3440
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5860457420349121,
+      "learning_rate": 0.00016103918387346732,
+      "loss": 0.8831,
+      "step": 3445
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.7714948654174805,
+      "learning_rate": 0.00016093402566049367,
+      "loss": 1.0446,
+      "step": 3450
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6364210844039917,
+      "learning_rate": 0.0001608287601615604,
+      "loss": 0.8378,
+      "step": 3455
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6578925251960754,
+      "learning_rate": 0.00016072338756200746,
+      "loss": 0.9447,
+      "step": 3460
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5990071296691895,
+      "learning_rate": 0.00016061790804736332,
+      "loss": 0.9604,
+      "step": 3465
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6787405610084534,
+      "learning_rate": 0.00016051232180334485,
+      "loss": 0.8815,
+      "step": 3470
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5929312705993652,
+      "learning_rate": 0.00016040662901585674,
+      "loss": 0.917,
+      "step": 3475
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6015139818191528,
+      "learning_rate": 0.00016030082987099123,
+      "loss": 0.948,
+      "step": 3480
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6716448068618774,
+      "learning_rate": 0.00016019492455502787,
+      "loss": 0.9984,
+      "step": 3485
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6419260501861572,
+      "learning_rate": 0.00016008891325443317,
+      "loss": 0.9077,
+      "step": 3490
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.7909138202667236,
+      "learning_rate": 0.0001599827961558602,
+      "loss": 0.9885,
+      "step": 3495
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.889897346496582,
+      "learning_rate": 0.00015987657344614835,
+      "loss": 0.8763,
+      "step": 3500
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6767850518226624,
+      "learning_rate": 0.0001597702453123229,
+      "loss": 0.935,
+      "step": 3505
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6611208915710449,
+      "learning_rate": 0.00015966381194159482,
+      "loss": 0.9498,
+      "step": 3510
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.7573800683021545,
+      "learning_rate": 0.0001595572735213603,
+      "loss": 0.8646,
+      "step": 3515
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6432080268859863,
+      "learning_rate": 0.00015945063023920056,
+      "loss": 1.0834,
+      "step": 3520
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6003210544586182,
+      "learning_rate": 0.00015934388228288138,
+      "loss": 0.8283,
+      "step": 3525
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.7602962255477905,
+      "learning_rate": 0.00015923702984035288,
+      "loss": 0.913,
+      "step": 3530
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6201465725898743,
+      "learning_rate": 0.00015913007309974916,
+      "loss": 0.9628,
+      "step": 3535
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6988320350646973,
+      "learning_rate": 0.00015902301224938792,
+      "loss": 0.7929,
+      "step": 3540
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.7519627809524536,
+      "learning_rate": 0.00015891584747777018,
+      "loss": 0.966,
+      "step": 3545
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.7300755977630615,
+      "learning_rate": 0.00015880857897357994,
+      "loss": 0.9138,
+      "step": 3550
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.682937502861023,
+      "learning_rate": 0.00015870120692568383,
+      "loss": 0.9415,
+      "step": 3555
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6300337314605713,
+      "learning_rate": 0.00015859373152313078,
+      "loss": 0.957,
+      "step": 3560
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6108503937721252,
+      "learning_rate": 0.00015848615295515175,
+      "loss": 0.8324,
+      "step": 3565
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.7553655505180359,
+      "learning_rate": 0.00015837847141115927,
+      "loss": 0.8505,
+      "step": 3570
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6497313380241394,
+      "learning_rate": 0.00015827068708074724,
+      "loss": 0.8825,
+      "step": 3575
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.8078399896621704,
+      "learning_rate": 0.00015816280015369045,
+      "loss": 0.9456,
+      "step": 3580
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6483976244926453,
+      "learning_rate": 0.00015805481081994444,
+      "loss": 1.0213,
+      "step": 3585
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6289666891098022,
+      "learning_rate": 0.00015794671926964497,
+      "loss": 0.8518,
+      "step": 3590
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.577955424785614,
+      "learning_rate": 0.00015783852569310785,
+      "loss": 0.8843,
+      "step": 3595
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.5437608361244202,
+      "learning_rate": 0.00015773023028082842,
+      "loss": 0.8382,
+      "step": 3600
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6673296093940735,
+      "learning_rate": 0.00015762183322348144,
+      "loss": 0.9322,
+      "step": 3605
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.7770671844482422,
+      "learning_rate": 0.0001575133347119205,
+      "loss": 1.0315,
+      "step": 3610
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.7004712820053101,
+      "learning_rate": 0.00015740473493717802,
+      "loss": 0.8411,
+      "step": 3615
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.595714807510376,
+      "learning_rate": 0.00015729603409046447,
+      "loss": 0.9721,
+      "step": 3620
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.5591194033622742,
+      "learning_rate": 0.00015718723236316846,
+      "loss": 0.9229,
+      "step": 3625
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.7704307436943054,
+      "learning_rate": 0.0001570783299468562,
+      "loss": 0.9729,
+      "step": 3630
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.535123884677887,
+      "learning_rate": 0.000156969327033271,
+      "loss": 0.8177,
+      "step": 3635
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.59185791015625,
+      "learning_rate": 0.00015686022381433337,
+      "loss": 0.8999,
+      "step": 3640
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.5567179322242737,
+      "learning_rate": 0.00015675102048214027,
+      "loss": 0.8699,
+      "step": 3645
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6161515712738037,
+      "learning_rate": 0.000156641717228965,
+      "loss": 0.9498,
+      "step": 3650
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.8555355668067932,
+      "learning_rate": 0.00015653231424725671,
+      "loss": 0.9246,
+      "step": 3655
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.5881509780883789,
+      "learning_rate": 0.00015642281172964024,
+      "loss": 0.8512,
+      "step": 3660
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.5532040596008301,
+      "learning_rate": 0.0001563132098689156,
+      "loss": 0.8884,
+      "step": 3665
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6380072832107544,
+      "learning_rate": 0.00015620350885805774,
+      "loss": 0.8272,
+      "step": 3670
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6400458812713623,
+      "learning_rate": 0.00015609370889021617,
+      "loss": 0.9215,
+      "step": 3675
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6543514132499695,
+      "learning_rate": 0.00015598381015871472,
+      "loss": 0.8601,
+      "step": 3680
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.5944799184799194,
+      "learning_rate": 0.000155873812857051,
+      "loss": 0.8123,
+      "step": 3685
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6395347714424133,
+      "learning_rate": 0.0001557637171788962,
+      "loss": 0.8114,
+      "step": 3690
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6879188418388367,
+      "learning_rate": 0.00015565352331809473,
+      "loss": 0.8653,
+      "step": 3695
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6839675307273865,
+      "learning_rate": 0.0001555432314686639,
+      "loss": 0.868,
+      "step": 3700
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6724333167076111,
+      "learning_rate": 0.00015543284182479352,
+      "loss": 0.9477,
+      "step": 3705
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.5998671054840088,
+      "learning_rate": 0.00015532235458084554,
+      "loss": 0.8883,
+      "step": 3710
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.7111408710479736,
+      "learning_rate": 0.00015521176993135388,
+      "loss": 0.9072,
+      "step": 3715
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6554681658744812,
+      "learning_rate": 0.00015510108807102383,
+      "loss": 0.9257,
+      "step": 3720
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6711482405662537,
+      "learning_rate": 0.00015499030919473186,
+      "loss": 0.9075,
+      "step": 3725
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6953566670417786,
+      "learning_rate": 0.00015487943349752533,
+      "loss": 0.9182,
+      "step": 3730
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.5130366086959839,
+      "learning_rate": 0.00015476846117462204,
+      "loss": 0.9534,
+      "step": 3735
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.7102193236351013,
+      "learning_rate": 0.00015465739242140987,
+      "loss": 0.8824,
+      "step": 3740
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6412502527236938,
+      "learning_rate": 0.0001545462274334465,
+      "loss": 0.9035,
+      "step": 3745
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.5918426513671875,
+      "learning_rate": 0.00015443496640645915,
+      "loss": 0.9179,
+      "step": 3750
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6007682681083679,
+      "learning_rate": 0.00015432360953634397,
+      "loss": 0.9842,
+      "step": 3755
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.7916061878204346,
+      "learning_rate": 0.00015421215701916596,
+      "loss": 0.9593,
+      "step": 3760
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.8170971870422363,
+      "learning_rate": 0.00015410060905115852,
+      "loss": 0.8976,
+      "step": 3765
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6541162729263306,
+      "learning_rate": 0.0001539889658287231,
+      "loss": 0.929,
+      "step": 3770
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.5680133104324341,
+      "learning_rate": 0.00015387722754842885,
+      "loss": 0.9046,
+      "step": 3775
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.7456454634666443,
+      "learning_rate": 0.0001537653944070123,
+      "loss": 0.9795,
+      "step": 3780
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.571003794670105,
+      "learning_rate": 0.00015365346660137702,
+      "loss": 0.9768,
+      "step": 3785
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6417176127433777,
+      "learning_rate": 0.0001535414443285932,
+      "loss": 0.8871,
+      "step": 3790
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.7255185842514038,
+      "learning_rate": 0.0001534293277858974,
+      "loss": 0.9026,
+      "step": 3795
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6499106287956238,
+      "learning_rate": 0.00015331711717069216,
+      "loss": 0.9168,
+      "step": 3800
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6854183077812195,
+      "learning_rate": 0.0001532048126805456,
+      "loss": 1.0304,
+      "step": 3805
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.618883490562439,
+      "learning_rate": 0.00015309241451319126,
+      "loss": 0.8583,
+      "step": 3810
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.613739550113678,
+      "learning_rate": 0.00015297992286652745,
+      "loss": 0.932,
+      "step": 3815
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.8370459675788879,
+      "learning_rate": 0.0001528673379386172,
+      "loss": 1.0117,
+      "step": 3820
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.5930368304252625,
+      "learning_rate": 0.0001527546599276876,
+      "loss": 0.9328,
+      "step": 3825
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.5250547528266907,
+      "learning_rate": 0.00015264188903212991,
+      "loss": 0.8252,
+      "step": 3830
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.7919967174530029,
+      "learning_rate": 0.00015252902545049866,
+      "loss": 0.9526,
+      "step": 3835
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.7373855710029602,
+      "learning_rate": 0.00015241606938151177,
+      "loss": 0.8234,
+      "step": 3840
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.8167823553085327,
+      "learning_rate": 0.00015230302102404986,
+      "loss": 0.9571,
+      "step": 3845
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6310274004936218,
+      "learning_rate": 0.0001521898805771561,
+      "loss": 1.0162,
+      "step": 3850
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7180379629135132,
+      "learning_rate": 0.0001520766482400358,
+      "loss": 0.8484,
+      "step": 3855
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5930597186088562,
+      "learning_rate": 0.0001519633242120561,
+      "loss": 0.8296,
+      "step": 3860
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6432483792304993,
+      "learning_rate": 0.0001518499086927455,
+      "loss": 0.659,
+      "step": 3865
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.611929714679718,
+      "learning_rate": 0.00015173640188179363,
+      "loss": 0.8142,
+      "step": 3870
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6128609776496887,
+      "learning_rate": 0.00015162280397905086,
+      "loss": 0.9515,
+      "step": 3875
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6003267765045166,
+      "learning_rate": 0.00015150911518452793,
+      "loss": 0.9143,
+      "step": 3880
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6806387305259705,
+      "learning_rate": 0.00015139533569839565,
+      "loss": 0.8744,
+      "step": 3885
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7915828824043274,
+      "learning_rate": 0.00015128146572098442,
+      "loss": 0.9343,
+      "step": 3890
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5177380442619324,
+      "learning_rate": 0.00015116750545278408,
+      "loss": 0.876,
+      "step": 3895
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7162129878997803,
+      "learning_rate": 0.00015105345509444336,
+      "loss": 0.8494,
+      "step": 3900
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7143672108650208,
+      "learning_rate": 0.00015093931484676967,
+      "loss": 0.8876,
+      "step": 3905
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6772016286849976,
+      "learning_rate": 0.00015082508491072864,
+      "loss": 0.9315,
+      "step": 3910
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6652882099151611,
+      "learning_rate": 0.00015071076548744386,
+      "loss": 0.9208,
+      "step": 3915
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.743068277835846,
+      "learning_rate": 0.00015059635677819636,
+      "loss": 1.0118,
+      "step": 3920
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6887498497962952,
+      "learning_rate": 0.00015048185898442463,
+      "loss": 0.8486,
+      "step": 3925
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.8008378744125366,
+      "learning_rate": 0.00015036727230772367,
+      "loss": 0.9585,
+      "step": 3930
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7885955572128296,
+      "learning_rate": 0.00015025259694984524,
+      "loss": 0.9206,
+      "step": 3935
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5348213315010071,
+      "learning_rate": 0.0001501378331126972,
+      "loss": 0.9397,
+      "step": 3940
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6782258152961731,
+      "learning_rate": 0.00015002298099834303,
+      "loss": 0.8909,
+      "step": 3945
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7077736854553223,
+      "learning_rate": 0.00014990804080900185,
+      "loss": 0.9807,
+      "step": 3950
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7034088969230652,
+      "learning_rate": 0.0001497930127470477,
+      "loss": 1.0362,
+      "step": 3955
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5973007082939148,
+      "learning_rate": 0.00014967789701500944,
+      "loss": 0.9119,
+      "step": 3960
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7092791199684143,
+      "learning_rate": 0.00014956269381557024,
+      "loss": 0.8564,
+      "step": 3965
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.5920495986938477,
+      "learning_rate": 0.00014944740335156724,
+      "loss": 0.9436,
+      "step": 3970
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.7596008777618408,
+      "learning_rate": 0.0001493320258259913,
+      "loss": 1.0043,
+      "step": 3975
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6508396863937378,
+      "learning_rate": 0.00014921656144198652,
+      "loss": 0.8831,
+      "step": 3980
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.5840360522270203,
+      "learning_rate": 0.00014910101040284992,
+      "loss": 0.9207,
+      "step": 3985
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.7185476422309875,
+      "learning_rate": 0.00014898537291203117,
+      "loss": 0.8468,
+      "step": 3990
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6996331810951233,
+      "learning_rate": 0.00014886964917313207,
+      "loss": 1.0592,
+      "step": 3995
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.599885106086731,
+      "learning_rate": 0.00014875383938990627,
+      "loss": 0.8687,
+      "step": 4000
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.7138045430183411,
+      "learning_rate": 0.00014863794376625904,
+      "loss": 0.8685,
+      "step": 4005
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6436852216720581,
+      "learning_rate": 0.00014852196250624662,
+      "loss": 0.9592,
+      "step": 4010
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6193510293960571,
+      "learning_rate": 0.00014840589581407616,
+      "loss": 0.928,
+      "step": 4015
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.7025611996650696,
+      "learning_rate": 0.00014828974389410516,
+      "loss": 0.9371,
+      "step": 4020
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.7858377695083618,
+      "learning_rate": 0.0001481735069508412,
+      "loss": 0.8292,
+      "step": 4025
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6432536840438843,
+      "learning_rate": 0.00014805718518894157,
+      "loss": 0.8523,
+      "step": 4030
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6322362422943115,
+      "learning_rate": 0.00014794077881321292,
+      "loss": 0.9446,
+      "step": 4035
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6036275625228882,
+      "learning_rate": 0.0001478242880286108,
+      "loss": 0.9576,
+      "step": 4040
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.611020565032959,
+      "learning_rate": 0.00014770771304023942,
+      "loss": 0.8126,
+      "step": 4045
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.5951793789863586,
+      "learning_rate": 0.00014759105405335132,
+      "loss": 0.8255,
+      "step": 4050
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6406493782997131,
+      "learning_rate": 0.00014747431127334678,
+      "loss": 0.9738,
+      "step": 4055
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.7525074481964111,
+      "learning_rate": 0.0001473574849057738,
+      "loss": 1.0128,
+      "step": 4060
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.5915870070457458,
+      "learning_rate": 0.00014724057515632738,
+      "loss": 0.9119,
+      "step": 4065
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.8320327997207642,
+      "learning_rate": 0.00014712358223084942,
+      "loss": 0.8072,
+      "step": 4070
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.5680391192436218,
+      "learning_rate": 0.00014700650633532827,
+      "loss": 0.8269,
+      "step": 4075
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.7768528461456299,
+      "learning_rate": 0.00014688934767589833,
+      "loss": 0.9261,
+      "step": 4080
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6114219427108765,
+      "learning_rate": 0.00014677210645883977,
+      "loss": 0.9493,
+      "step": 4085
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5603370070457458,
+      "learning_rate": 0.00014665478289057805,
+      "loss": 0.876,
+      "step": 4090
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.718364417552948,
+      "learning_rate": 0.00014653737717768367,
+      "loss": 0.9818,
+      "step": 4095
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6249436140060425,
+      "learning_rate": 0.00014641988952687177,
+      "loss": 0.9,
+      "step": 4100
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5478988885879517,
+      "learning_rate": 0.0001463023201450017,
+      "loss": 0.8955,
+      "step": 4105
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6322693824768066,
+      "learning_rate": 0.00014618466923907678,
+      "loss": 0.996,
+      "step": 4110
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5370429754257202,
+      "learning_rate": 0.00014606693701624385,
+      "loss": 0.8568,
+      "step": 4115
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5983506441116333,
+      "learning_rate": 0.0001459491236837929,
+      "loss": 0.8785,
+      "step": 4120
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.490751177072525,
+      "learning_rate": 0.00014583122944915672,
+      "loss": 0.9129,
+      "step": 4125
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6703299283981323,
+      "learning_rate": 0.00014571325451991066,
+      "loss": 0.9941,
+      "step": 4130
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6486307978630066,
+      "learning_rate": 0.00014559519910377193,
+      "loss": 0.794,
+      "step": 4135
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6234001517295837,
+      "learning_rate": 0.0001454770634085997,
+      "loss": 0.9608,
+      "step": 4140
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6722338199615479,
+      "learning_rate": 0.00014535884764239424,
+      "loss": 1.0127,
+      "step": 4145
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.7016012668609619,
+      "learning_rate": 0.00014524055201329704,
+      "loss": 0.9298,
+      "step": 4150
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.536430835723877,
+      "learning_rate": 0.00014512217672959003,
+      "loss": 0.895,
+      "step": 4155
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6761835217475891,
+      "learning_rate": 0.00014500372199969546,
+      "loss": 0.9557,
+      "step": 4160
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6276063323020935,
+      "learning_rate": 0.00014488518803217542,
+      "loss": 1.0102,
+      "step": 4165
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6394610404968262,
+      "learning_rate": 0.0001447665750357316,
+      "loss": 0.8033,
+      "step": 4170
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.7304995656013489,
+      "learning_rate": 0.00014464788321920472,
+      "loss": 0.9172,
+      "step": 4175
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6288149952888489,
+      "learning_rate": 0.00014452911279157435,
+      "loss": 0.8689,
+      "step": 4180
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6916682720184326,
+      "learning_rate": 0.0001444102639619585,
+      "loss": 0.9401,
+      "step": 4185
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.709947407245636,
+      "learning_rate": 0.00014429133693961304,
+      "loss": 0.9208,
+      "step": 4190
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5847663283348083,
+      "learning_rate": 0.0001441723319339318,
+      "loss": 1.025,
+      "step": 4195
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6856701374053955,
+      "learning_rate": 0.00014405324915444572,
+      "loss": 0.8597,
+      "step": 4200
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.5357947945594788,
+      "learning_rate": 0.00014393408881082265,
+      "loss": 0.916,
+      "step": 4205
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6699780821800232,
+      "learning_rate": 0.00014381485111286714,
+      "loss": 0.8708,
+      "step": 4210
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6534074544906616,
+      "learning_rate": 0.00014369553627051982,
+      "loss": 0.9067,
+      "step": 4215
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.7450271844863892,
+      "learning_rate": 0.0001435761444938573,
+      "loss": 0.9721,
+      "step": 4220
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6965327858924866,
+      "learning_rate": 0.00014345667599309142,
+      "loss": 0.909,
+      "step": 4225
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6757059097290039,
+      "learning_rate": 0.0001433371309785693,
+      "loss": 0.929,
+      "step": 4230
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.8920908570289612,
+      "learning_rate": 0.0001432175096607727,
+      "loss": 0.957,
+      "step": 4235
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6721371412277222,
+      "learning_rate": 0.00014309781225031778,
+      "loss": 0.9679,
+      "step": 4240
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.5155514478683472,
+      "learning_rate": 0.00014297803895795455,
+      "loss": 0.9219,
+      "step": 4245
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.7284682393074036,
+      "learning_rate": 0.00014285818999456676,
+      "loss": 0.8668,
+      "step": 4250
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6953817009925842,
+      "learning_rate": 0.0001427382655711713,
+      "loss": 0.9595,
+      "step": 4255
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.5577906966209412,
+      "learning_rate": 0.000142618265898918,
+      "loss": 0.8521,
+      "step": 4260
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6189967393875122,
+      "learning_rate": 0.00014249819118908915,
+      "loss": 0.9177,
+      "step": 4265
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6514103412628174,
+      "learning_rate": 0.00014237804165309913,
+      "loss": 0.9944,
+      "step": 4270
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.5756633281707764,
+      "learning_rate": 0.0001422578175024941,
+      "loss": 0.9258,
+      "step": 4275
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.5479834675788879,
+      "learning_rate": 0.00014213751894895154,
+      "loss": 0.9599,
+      "step": 4280
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6857898235321045,
+      "learning_rate": 0.00014201714620428,
+      "loss": 0.9585,
+      "step": 4285
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.7923577427864075,
+      "learning_rate": 0.00014189669948041863,
+      "loss": 0.899,
+      "step": 4290
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.7185238003730774,
+      "learning_rate": 0.00014177617898943683,
+      "loss": 0.8412,
+      "step": 4295
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6681710481643677,
+      "learning_rate": 0.00014165558494353385,
+      "loss": 0.9985,
+      "step": 4300
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.7086035013198853,
+      "learning_rate": 0.00014153491755503853,
+      "loss": 1.0187,
+      "step": 4305
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.806403636932373,
+      "learning_rate": 0.00014141417703640875,
+      "loss": 0.9801,
+      "step": 4310
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.7036173343658447,
+      "learning_rate": 0.0001412933636002312,
+      "loss": 0.8904,
+      "step": 4315
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6700453162193298,
+      "learning_rate": 0.00014117247745922101,
+      "loss": 0.955,
+      "step": 4320
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6932754516601562,
+      "learning_rate": 0.00014105151882622122,
+      "loss": 0.8445,
+      "step": 4325
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6237407326698303,
+      "learning_rate": 0.00014093048791420252,
+      "loss": 0.7346,
+      "step": 4330
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.7434117197990417,
+      "learning_rate": 0.00014080938493626286,
+      "loss": 0.9259,
+      "step": 4335
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.8616409301757812,
+      "learning_rate": 0.00014068821010562718,
+      "loss": 0.9227,
+      "step": 4340
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.8434512615203857,
+      "learning_rate": 0.00014056696363564682,
+      "loss": 0.9353,
+      "step": 4345
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6269782781600952,
+      "learning_rate": 0.00014044564573979925,
+      "loss": 0.9082,
+      "step": 4350
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.7604944705963135,
+      "learning_rate": 0.0001403242566316878,
+      "loss": 0.9702,
+      "step": 4355
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.5997477769851685,
+      "learning_rate": 0.0001402027965250411,
+      "loss": 0.9641,
+      "step": 4360
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.7834346294403076,
+      "learning_rate": 0.00014008126563371274,
+      "loss": 0.9176,
+      "step": 4365
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.734910249710083,
+      "learning_rate": 0.0001399596641716811,
+      "loss": 1.0047,
+      "step": 4370
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6782670617103577,
+      "learning_rate": 0.0001398379923530487,
+      "loss": 0.8528,
+      "step": 4375
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.7385916113853455,
+      "learning_rate": 0.0001397162503920419,
+      "loss": 0.9069,
+      "step": 4380
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.7298621535301208,
+      "learning_rate": 0.00013959443850301061,
+      "loss": 0.8389,
+      "step": 4385
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.65964674949646,
+      "learning_rate": 0.00013947255690042795,
+      "loss": 0.9972,
+      "step": 4390
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6677912473678589,
+      "learning_rate": 0.00013935060579888962,
+      "loss": 0.8049,
+      "step": 4395
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.5836318731307983,
+      "learning_rate": 0.00013922858541311382,
+      "loss": 0.8543,
+      "step": 4400
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6943433284759521,
+      "learning_rate": 0.00013910649595794058,
+      "loss": 0.9893,
+      "step": 4405
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.5648766160011292,
+      "learning_rate": 0.00013898433764833178,
+      "loss": 0.8044,
+      "step": 4410
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.5739986896514893,
+      "learning_rate": 0.00013886211069937034,
+      "loss": 1.0803,
+      "step": 4415
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6268223524093628,
+      "learning_rate": 0.00013873981532626007,
+      "loss": 0.8361,
+      "step": 4420
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.9593233466148376,
+      "learning_rate": 0.00013861745174432525,
+      "loss": 0.9178,
+      "step": 4425
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.622471272945404,
+      "learning_rate": 0.00013849502016901035,
+      "loss": 0.9039,
+      "step": 4430
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6628968715667725,
+      "learning_rate": 0.00013837252081587938,
+      "loss": 0.9192,
+      "step": 4435
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.7090737223625183,
+      "learning_rate": 0.0001382499539006159,
+      "loss": 0.8897,
+      "step": 4440
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6674631237983704,
+      "learning_rate": 0.00013812731963902224,
+      "loss": 0.864,
+      "step": 4445
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.8230262398719788,
+      "learning_rate": 0.0001380046182470194,
+      "loss": 0.8783,
+      "step": 4450
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6914052367210388,
+      "learning_rate": 0.0001378818499406465,
+      "loss": 0.9236,
+      "step": 4455
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.672724187374115,
+      "learning_rate": 0.00013775901493606063,
+      "loss": 0.8779,
+      "step": 4460
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6276265978813171,
+      "learning_rate": 0.0001376361134495361,
+      "loss": 0.9381,
+      "step": 4465
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6782820820808411,
+      "learning_rate": 0.0001375131456974645,
+      "loss": 0.8978,
+      "step": 4470
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.711754322052002,
+      "learning_rate": 0.0001373901118963539,
+      "loss": 0.9513,
+      "step": 4475
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.5582084655761719,
+      "learning_rate": 0.00013726701226282885,
+      "loss": 0.9309,
+      "step": 4480
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.8418028950691223,
+      "learning_rate": 0.00013714384701362956,
+      "loss": 0.963,
+      "step": 4485
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6022467017173767,
+      "learning_rate": 0.000137020616365612,
+      "loss": 0.7766,
+      "step": 4490
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.8316811323165894,
+      "learning_rate": 0.0001368973205357472,
+      "loss": 0.9274,
+      "step": 4495
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.7477040886878967,
+      "learning_rate": 0.00013677395974112094,
+      "loss": 0.9448,
+      "step": 4500
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.8349044919013977,
+      "learning_rate": 0.00013665053419893337,
+      "loss": 0.8744,
+      "step": 4505
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.7245243787765503,
+      "learning_rate": 0.0001365270441264987,
+      "loss": 0.8248,
+      "step": 4510
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.5794562101364136,
+      "learning_rate": 0.00013640348974124474,
+      "loss": 0.8832,
+      "step": 4515
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.7254599928855896,
+      "learning_rate": 0.0001362798712607125,
+      "loss": 0.9828,
+      "step": 4520
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.7025798559188843,
+      "learning_rate": 0.00013615618890255589,
+      "loss": 0.916,
+      "step": 4525
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6569787263870239,
+      "learning_rate": 0.0001360324428845412,
+      "loss": 0.9896,
+      "step": 4530
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6779191493988037,
+      "learning_rate": 0.00013590863342454693,
+      "loss": 0.8824,
+      "step": 4535
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.5817300081253052,
+      "learning_rate": 0.0001357847607405632,
+      "loss": 0.8415,
+      "step": 4540
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.7714138627052307,
+      "learning_rate": 0.00013566082505069143,
+      "loss": 0.9976,
+      "step": 4545
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.5546395778656006,
+      "learning_rate": 0.00013553682657314412,
+      "loss": 0.8445,
+      "step": 4550
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6671056747436523,
+      "learning_rate": 0.00013541276552624405,
+      "loss": 0.8457,
+      "step": 4555
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5993417501449585,
+      "learning_rate": 0.00013528864212842444,
+      "loss": 0.8144,
+      "step": 4560
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.847915530204773,
+      "learning_rate": 0.00013516445659822815,
+      "loss": 0.9305,
+      "step": 4565
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7033442258834839,
+      "learning_rate": 0.00013504020915430746,
+      "loss": 0.9296,
+      "step": 4570
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5511518716812134,
+      "learning_rate": 0.00013491590001542367,
+      "loss": 0.8731,
+      "step": 4575
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7261431813240051,
+      "learning_rate": 0.00013479152940044665,
+      "loss": 0.9634,
+      "step": 4580
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7022355198860168,
+      "learning_rate": 0.00013466709752835466,
+      "loss": 0.9566,
+      "step": 4585
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7419948577880859,
+      "learning_rate": 0.00013454260461823365,
+      "loss": 0.8646,
+      "step": 4590
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7006118893623352,
+      "learning_rate": 0.00013441805088927706,
+      "loss": 0.9215,
+      "step": 4595
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5963191390037537,
+      "learning_rate": 0.00013429343656078555,
+      "loss": 0.8042,
+      "step": 4600
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5838218927383423,
+      "learning_rate": 0.0001341687618521663,
+      "loss": 0.8131,
+      "step": 4605
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.750098705291748,
+      "learning_rate": 0.00013404402698293294,
+      "loss": 0.8713,
+      "step": 4610
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6926030516624451,
+      "learning_rate": 0.00013391923217270497,
+      "loss": 0.9185,
+      "step": 4615
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6723347902297974,
+      "learning_rate": 0.00013379437764120738,
+      "loss": 0.8885,
+      "step": 4620
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6616873145103455,
+      "learning_rate": 0.00013366946360827037,
+      "loss": 0.9178,
+      "step": 4625
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7160006165504456,
+      "learning_rate": 0.00013354449029382893,
+      "loss": 0.7647,
+      "step": 4630
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7113422751426697,
+      "learning_rate": 0.00013341945791792238,
+      "loss": 0.9592,
+      "step": 4635
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7037850618362427,
+      "learning_rate": 0.00013329436670069395,
+      "loss": 0.8489,
+      "step": 4640
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6213445067405701,
+      "learning_rate": 0.0001331692168623907,
+      "loss": 0.9049,
+      "step": 4645
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6793341636657715,
+      "learning_rate": 0.00013304400862336263,
+      "loss": 0.7977,
+      "step": 4650
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7716271877288818,
+      "learning_rate": 0.00013291874220406274,
+      "loss": 0.9861,
+      "step": 4655
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6577031016349792,
+      "learning_rate": 0.00013279341782504645,
+      "loss": 0.9055,
+      "step": 4660
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6821931004524231,
+      "learning_rate": 0.00013266803570697116,
+      "loss": 0.8969,
+      "step": 4665
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6055404543876648,
+      "learning_rate": 0.00013254259607059605,
+      "loss": 0.9834,
+      "step": 4670
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6545614004135132,
+      "learning_rate": 0.0001324170991367814,
+      "loss": 0.8664,
+      "step": 4675
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6915416121482849,
+      "learning_rate": 0.0001322915451264885,
+      "loss": 0.879,
+      "step": 4680
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6689884066581726,
+      "learning_rate": 0.00013216593426077918,
+      "loss": 0.9359,
+      "step": 4685
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7084957361221313,
+      "learning_rate": 0.00013204026676081517,
+      "loss": 0.9283,
+      "step": 4690
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.8455063700675964,
+      "learning_rate": 0.0001319145428478581,
+      "loss": 0.8446,
+      "step": 4695
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6622646450996399,
+      "learning_rate": 0.0001317887627432689,
+      "loss": 0.7989,
+      "step": 4700
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5944858193397522,
+      "learning_rate": 0.00013166292666850734,
+      "loss": 0.8847,
+      "step": 4705
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7730399966239929,
+      "learning_rate": 0.00013153703484513186,
+      "loss": 0.9543,
+      "step": 4710
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6421423554420471,
+      "learning_rate": 0.00013141108749479898,
+      "loss": 0.8916,
+      "step": 4715
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6441932916641235,
+      "learning_rate": 0.00013128508483926298,
+      "loss": 0.8689,
+      "step": 4720
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.8238367438316345,
+      "learning_rate": 0.00013115902710037554,
+      "loss": 0.8758,
+      "step": 4725
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.634405791759491,
+      "learning_rate": 0.00013103291450008533,
+      "loss": 0.8721,
+      "step": 4730
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.8031609654426575,
+      "learning_rate": 0.00013090674726043766,
+      "loss": 0.8235,
+      "step": 4735
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6754260063171387,
+      "learning_rate": 0.0001307805256035739,
+      "loss": 0.8455,
+      "step": 4740
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7197476029396057,
+      "learning_rate": 0.00013065424975173135,
+      "loss": 0.8448,
+      "step": 4745
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.9335188269615173,
+      "learning_rate": 0.00013052791992724275,
+      "loss": 0.9965,
+      "step": 4750
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5767316222190857,
+      "learning_rate": 0.00013040153635253575,
+      "loss": 0.9352,
+      "step": 4755
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6993480324745178,
+      "learning_rate": 0.00013027509925013275,
+      "loss": 0.9312,
+      "step": 4760
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7889806628227234,
+      "learning_rate": 0.00013014860884265036,
+      "loss": 0.961,
+      "step": 4765
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6709961891174316,
+      "learning_rate": 0.000130022065352799,
+      "loss": 0.8681,
+      "step": 4770
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7165763974189758,
+      "learning_rate": 0.00012989546900338264,
+      "loss": 0.9219,
+      "step": 4775
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.8095049262046814,
+      "learning_rate": 0.00012976882001729823,
+      "loss": 0.8742,
+      "step": 4780
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7213955521583557,
+      "learning_rate": 0.00012964211861753543,
+      "loss": 0.826,
+      "step": 4785
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7863954305648804,
+      "learning_rate": 0.00012951536502717623,
+      "loss": 0.8956,
+      "step": 4790
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.7114418745040894,
+      "learning_rate": 0.00012938855946939443,
+      "loss": 0.9338,
+      "step": 4795
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.7426603436470032,
+      "learning_rate": 0.0001292617021674554,
+      "loss": 0.9268,
+      "step": 4800
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.7291275858879089,
+      "learning_rate": 0.00012913479334471557,
+      "loss": 0.919,
+      "step": 4805
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6532735824584961,
+      "learning_rate": 0.0001290078332246221,
+      "loss": 0.9307,
+      "step": 4810
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6177434325218201,
+      "learning_rate": 0.0001288808220307125,
+      "loss": 0.9095,
+      "step": 4815
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6966139674186707,
+      "learning_rate": 0.0001287537599866141,
+      "loss": 0.7596,
+      "step": 4820
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.7126907110214233,
+      "learning_rate": 0.00012862664731604388,
+      "loss": 0.8657,
+      "step": 4825
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5501668453216553,
+      "learning_rate": 0.0001284994842428079,
+      "loss": 0.8841,
+      "step": 4830
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.7405208945274353,
+      "learning_rate": 0.00012837227099080098,
+      "loss": 1.028,
+      "step": 4835
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.674213171005249,
+      "learning_rate": 0.00012824500778400627,
+      "loss": 0.8924,
+      "step": 4840
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6531869769096375,
+      "learning_rate": 0.00012811769484649492,
+      "loss": 0.9886,
+      "step": 4845
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.7602865695953369,
+      "learning_rate": 0.0001279903324024256,
+      "loss": 0.9032,
+      "step": 4850
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6832072734832764,
+      "learning_rate": 0.0001278629206760441,
+      "loss": 0.8728,
+      "step": 4855
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6938978433609009,
+      "learning_rate": 0.0001277354598916831,
+      "loss": 0.977,
+      "step": 4860
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6449591517448425,
+      "learning_rate": 0.00012760795027376158,
+      "loss": 0.8576,
+      "step": 4865
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.7969632744789124,
+      "learning_rate": 0.00012748039204678446,
+      "loss": 0.9209,
+      "step": 4870
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6520712971687317,
+      "learning_rate": 0.00012735278543534243,
+      "loss": 0.8105,
+      "step": 4875
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.7720639705657959,
+      "learning_rate": 0.00012722513066411103,
+      "loss": 0.9094,
+      "step": 4880
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6899474859237671,
+      "learning_rate": 0.00012709742795785097,
+      "loss": 0.8983,
+      "step": 4885
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.7263126969337463,
+      "learning_rate": 0.00012696967754140714,
+      "loss": 0.9146,
+      "step": 4890
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.0260558128356934,
+      "learning_rate": 0.00012684187963970847,
+      "loss": 0.9221,
+      "step": 4895
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6547735333442688,
+      "learning_rate": 0.00012671403447776753,
+      "loss": 0.8363,
+      "step": 4900
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6629674434661865,
+      "learning_rate": 0.00012658614228068003,
+      "loss": 0.8795,
+      "step": 4905
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6511769890785217,
+      "learning_rate": 0.00012645820327362466,
+      "loss": 1.0364,
+      "step": 4910
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.8055089712142944,
+      "learning_rate": 0.0001263302176818623,
+      "loss": 1.0052,
+      "step": 4915
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6773073673248291,
+      "learning_rate": 0.000126202185730736,
+      "loss": 1.0077,
+      "step": 4920
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.669937789440155,
+      "learning_rate": 0.00012607410764567045,
+      "loss": 0.8325,
+      "step": 4925
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7064899206161499,
+      "learning_rate": 0.00012594598365217144,
+      "loss": 0.8735,
+      "step": 4930
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.8059523105621338,
+      "learning_rate": 0.00012581781397582567,
+      "loss": 0.9312,
+      "step": 4935
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6511587500572205,
+      "learning_rate": 0.00012568959884230036,
+      "loss": 0.8524,
+      "step": 4940
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.623746931552887,
+      "learning_rate": 0.0001255613384773426,
+      "loss": 0.887,
+      "step": 4945
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.593783438205719,
+      "learning_rate": 0.0001254330331067792,
+      "loss": 1.0126,
+      "step": 4950
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6896436214447021,
+      "learning_rate": 0.00012530468295651617,
+      "loss": 1.0481,
+      "step": 4955
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6880872249603271,
+      "learning_rate": 0.00012517628825253852,
+      "loss": 0.9641,
+      "step": 4960
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6226217150688171,
+      "learning_rate": 0.00012504784922090945,
+      "loss": 0.9193,
+      "step": 4965
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6571257710456848,
+      "learning_rate": 0.00012491936608777045,
+      "loss": 0.7745,
+      "step": 4970
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6938855648040771,
+      "learning_rate": 0.00012479083907934052,
+      "loss": 0.8427,
+      "step": 4975
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6288595199584961,
+      "learning_rate": 0.00012466226842191587,
+      "loss": 0.8178,
+      "step": 4980
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6910018920898438,
+      "learning_rate": 0.00012453365434186975,
+      "loss": 0.9649,
+      "step": 4985
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.766521692276001,
+      "learning_rate": 0.00012440499706565164,
+      "loss": 0.8424,
+      "step": 4990
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6587792038917542,
+      "learning_rate": 0.00012427629681978724,
+      "loss": 0.7928,
+      "step": 4995
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6307712197303772,
+      "learning_rate": 0.00012414755383087785,
+      "loss": 0.8884,
+      "step": 5000
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6760114431381226,
+      "learning_rate": 0.0001240187683256,
+      "loss": 0.8033,
+      "step": 5005
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6159674525260925,
+      "learning_rate": 0.00012388994053070512,
+      "loss": 0.7698,
+      "step": 5010
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6530436873435974,
+      "learning_rate": 0.00012376107067301912,
+      "loss": 0.9561,
+      "step": 5015
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7369070053100586,
+      "learning_rate": 0.00012363215897944187,
+      "loss": 0.9346,
+      "step": 5020
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6479840874671936,
+      "learning_rate": 0.000123503205676947,
+      "loss": 0.8693,
+      "step": 5025
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7007189989089966,
+      "learning_rate": 0.00012337421099258133,
+      "loss": 0.8471,
+      "step": 5030
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.5950848460197449,
+      "learning_rate": 0.00012324517515346467,
+      "loss": 0.8664,
+      "step": 5035
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6795806288719177,
+      "learning_rate": 0.00012311609838678905,
+      "loss": 0.9208,
+      "step": 5040
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7374679446220398,
+      "learning_rate": 0.0001229869809198188,
+      "loss": 0.9093,
+      "step": 5045
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8739282488822937,
+      "learning_rate": 0.00012285782297988984,
+      "loss": 0.8876,
+      "step": 5050
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6189777851104736,
+      "learning_rate": 0.00012272862479440922,
+      "loss": 0.8096,
+      "step": 5055
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7174555063247681,
+      "learning_rate": 0.00012259938659085504,
+      "loss": 0.9407,
+      "step": 5060
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6594709157943726,
+      "learning_rate": 0.00012247010859677576,
+      "loss": 0.8707,
+      "step": 5065
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6535437703132629,
+      "learning_rate": 0.00012234079103978993,
+      "loss": 0.9787,
+      "step": 5070
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6512961387634277,
+      "learning_rate": 0.00012221143414758572,
+      "loss": 0.921,
+      "step": 5075
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6845799684524536,
+      "learning_rate": 0.00012208203814792056,
+      "loss": 0.8912,
+      "step": 5080
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6525848507881165,
+      "learning_rate": 0.00012195260326862081,
+      "loss": 0.9101,
+      "step": 5085
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.9541882872581482,
+      "learning_rate": 0.00012182312973758118,
+      "loss": 0.9397,
+      "step": 5090
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6034583449363708,
+      "learning_rate": 0.00012169361778276451,
+      "loss": 0.9326,
+      "step": 5095
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.623876690864563,
+      "learning_rate": 0.00012156406763220128,
+      "loss": 1.0362,
+      "step": 5100
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8122617602348328,
+      "learning_rate": 0.0001214344795139892,
+      "loss": 0.8876,
+      "step": 5105
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6952608823776245,
+      "learning_rate": 0.0001213048536562928,
+      "loss": 0.9326,
+      "step": 5110
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.5716097950935364,
+      "learning_rate": 0.00012117519028734317,
+      "loss": 0.9267,
+      "step": 5115
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6167184710502625,
+      "learning_rate": 0.00012104548963543729,
+      "loss": 0.9353,
+      "step": 5120
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8996965289115906,
+      "learning_rate": 0.00012091575192893789,
+      "loss": 0.8414,
+      "step": 5125
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6290978193283081,
+      "learning_rate": 0.00012078597739627297,
+      "loss": 0.9101,
+      "step": 5130
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6652675271034241,
+      "learning_rate": 0.00012065616626593528,
+      "loss": 0.9483,
+      "step": 5135
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7232987880706787,
+      "learning_rate": 0.00012052631876648199,
+      "loss": 1.0557,
+      "step": 5140
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6346669793128967,
+      "learning_rate": 0.00012039643512653444,
+      "loss": 0.8926,
+      "step": 5145
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7574824690818787,
+      "learning_rate": 0.00012026651557477745,
+      "loss": 1.0143,
+      "step": 5150
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.6413111090660095,
+      "learning_rate": 0.00012013656033995921,
+      "loss": 1.0192,
+      "step": 5155
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.8342952132225037,
+      "learning_rate": 0.00012000656965089063,
+      "loss": 0.8629,
+      "step": 5160
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.6431555151939392,
+      "learning_rate": 0.00011987654373644506,
+      "loss": 0.8238,
+      "step": 5165
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.5522921085357666,
+      "learning_rate": 0.00011974648282555794,
+      "loss": 0.8706,
+      "step": 5170
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7213382720947266,
+      "learning_rate": 0.00011961638714722623,
+      "loss": 0.9221,
+      "step": 5175
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.0782811641693115,
+      "learning_rate": 0.00011948625693050816,
+      "loss": 0.8955,
+      "step": 5180
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.6480096578598022,
+      "learning_rate": 0.00011935609240452281,
+      "loss": 0.7231,
+      "step": 5185
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7860197424888611,
+      "learning_rate": 0.00011922589379844961,
+      "loss": 0.9138,
+      "step": 5190
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7582351565361023,
+      "learning_rate": 0.00011909566134152794,
+      "loss": 0.9187,
+      "step": 5195
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.6495263576507568,
+      "learning_rate": 0.00011896539526305694,
+      "loss": 0.8453,
+      "step": 5200
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7486453652381897,
+      "learning_rate": 0.00011883509579239482,
+      "loss": 0.9441,
+      "step": 5205
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.5885391235351562,
+      "learning_rate": 0.0001187047631589586,
+      "loss": 0.8987,
+      "step": 5210
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.709578275680542,
+      "learning_rate": 0.00011857439759222373,
+      "loss": 0.9492,
+      "step": 5215
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7070071697235107,
+      "learning_rate": 0.00011844399932172362,
+      "loss": 0.9171,
+      "step": 5220
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7506661415100098,
+      "learning_rate": 0.00011831356857704927,
+      "loss": 0.9367,
+      "step": 5225
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.6146754622459412,
+      "learning_rate": 0.00011818310558784882,
+      "loss": 0.8276,
+      "step": 5230
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7746636271476746,
+      "learning_rate": 0.00011805261058382723,
+      "loss": 0.9692,
+      "step": 5235
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7707207202911377,
+      "learning_rate": 0.0001179220837947459,
+      "loss": 1.0199,
+      "step": 5240
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7371374368667603,
+      "learning_rate": 0.00011779152545042195,
+      "loss": 0.9595,
+      "step": 5245
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7845306396484375,
+      "learning_rate": 0.00011766093578072832,
+      "loss": 0.9082,
+      "step": 5250
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7142476439476013,
+      "learning_rate": 0.00011753031501559298,
+      "loss": 0.9966,
+      "step": 5255
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7987679243087769,
+      "learning_rate": 0.00011739966338499866,
+      "loss": 0.9473,
+      "step": 5260
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.6516871452331543,
+      "learning_rate": 0.00011726898111898246,
+      "loss": 0.9249,
+      "step": 5265
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6980127692222595,
+      "learning_rate": 0.00011713826844763538,
+      "loss": 0.8802,
+      "step": 5270
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.8179835081100464,
+      "learning_rate": 0.000117007525601102,
+      "loss": 0.9815,
+      "step": 5275
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6411282420158386,
+      "learning_rate": 0.00011687675280958,
+      "loss": 0.82,
+      "step": 5280
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6856908202171326,
+      "learning_rate": 0.00011674595030331974,
+      "loss": 0.92,
+      "step": 5285
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7004132270812988,
+      "learning_rate": 0.00011661511831262401,
+      "loss": 0.9501,
+      "step": 5290
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6341331005096436,
+      "learning_rate": 0.0001164842570678475,
+      "loss": 0.8728,
+      "step": 5295
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6990119218826294,
+      "learning_rate": 0.00011635336679939624,
+      "loss": 0.8573,
+      "step": 5300
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.740168571472168,
+      "learning_rate": 0.00011622244773772755,
+      "loss": 1.1156,
+      "step": 5305
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.676529049873352,
+      "learning_rate": 0.00011609150011334937,
+      "loss": 0.9396,
+      "step": 5310
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7612829804420471,
+      "learning_rate": 0.00011596052415681992,
+      "loss": 0.9255,
+      "step": 5315
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7749308943748474,
+      "learning_rate": 0.00011582952009874737,
+      "loss": 1.0113,
+      "step": 5320
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6618918180465698,
+      "learning_rate": 0.00011569848816978924,
+      "loss": 0.8917,
+      "step": 5325
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7212786078453064,
+      "learning_rate": 0.00011556742860065226,
+      "loss": 0.854,
+      "step": 5330
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6731553077697754,
+      "learning_rate": 0.00011543634162209178,
+      "loss": 0.9732,
+      "step": 5335
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6743349432945251,
+      "learning_rate": 0.00011530522746491132,
+      "loss": 0.8256,
+      "step": 5340
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6633926630020142,
+      "learning_rate": 0.00011517408635996241,
+      "loss": 0.834,
+      "step": 5345
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6572647094726562,
+      "learning_rate": 0.00011504291853814393,
+      "loss": 0.9172,
+      "step": 5350
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6950198411941528,
+      "learning_rate": 0.00011491172423040178,
+      "loss": 0.8839,
+      "step": 5355
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6339337229728699,
+      "learning_rate": 0.00011478050366772855,
+      "loss": 0.9509,
+      "step": 5360
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7393227815628052,
+      "learning_rate": 0.00011464925708116306,
+      "loss": 0.9535,
+      "step": 5365
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6855883598327637,
+      "learning_rate": 0.00011451798470178988,
+      "loss": 0.9199,
+      "step": 5370
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6756834983825684,
+      "learning_rate": 0.0001143866867607391,
+      "loss": 0.8526,
+      "step": 5375
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6416776180267334,
+      "learning_rate": 0.0001142553634891857,
+      "loss": 0.8788,
+      "step": 5380
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7423108220100403,
+      "learning_rate": 0.00011412401511834934,
+      "loss": 0.9431,
+      "step": 5385
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6564056277275085,
+      "learning_rate": 0.00011399264187949385,
+      "loss": 0.9038,
+      "step": 5390
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6751819849014282,
+      "learning_rate": 0.00011386124400392686,
+      "loss": 0.8355,
+      "step": 5395
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7076970338821411,
+      "learning_rate": 0.0001137298217229993,
+      "loss": 0.9295,
+      "step": 5400
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.8646618723869324,
+      "learning_rate": 0.00011359837526810521,
+      "loss": 0.9086,
+      "step": 5405
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7255219221115112,
+      "learning_rate": 0.00011346690487068103,
+      "loss": 0.9413,
+      "step": 5410
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6536005139350891,
+      "learning_rate": 0.00011333541076220555,
+      "loss": 0.8992,
+      "step": 5415
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.5964111685752869,
+      "learning_rate": 0.00011320389317419908,
+      "loss": 0.8439,
+      "step": 5420
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7618805766105652,
+      "learning_rate": 0.00011307235233822345,
+      "loss": 0.9356,
+      "step": 5425
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.712472677230835,
+      "learning_rate": 0.00011294078848588136,
+      "loss": 0.8599,
+      "step": 5430
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.5790109038352966,
+      "learning_rate": 0.00011280920184881598,
+      "loss": 0.805,
+      "step": 5435
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6265473961830139,
+      "learning_rate": 0.0001126775926587107,
+      "loss": 0.8728,
+      "step": 5440
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7416415810585022,
+      "learning_rate": 0.00011254596114728859,
+      "loss": 0.8682,
+      "step": 5445
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6786869764328003,
+      "learning_rate": 0.00011241430754631194,
+      "loss": 0.9832,
+      "step": 5450
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6256182193756104,
+      "learning_rate": 0.00011228263208758206,
+      "loss": 0.8253,
+      "step": 5455
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6844419836997986,
+      "learning_rate": 0.0001121509350029386,
+      "loss": 0.778,
+      "step": 5460
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7649736404418945,
+      "learning_rate": 0.00011201921652425945,
+      "loss": 0.8338,
+      "step": 5465
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6669993996620178,
+      "learning_rate": 0.00011188747688346002,
+      "loss": 0.9117,
+      "step": 5470
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7160723209381104,
+      "learning_rate": 0.00011175571631249305,
+      "loss": 0.9093,
+      "step": 5475
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6720280647277832,
+      "learning_rate": 0.00011162393504334814,
+      "loss": 0.8271,
+      "step": 5480
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7730690836906433,
+      "learning_rate": 0.00011149213330805135,
+      "loss": 0.9728,
+      "step": 5485
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6851432919502258,
+      "learning_rate": 0.00011136031133866467,
+      "loss": 0.9282,
+      "step": 5490
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.691317081451416,
+      "learning_rate": 0.00011122846936728584,
+      "loss": 0.9255,
+      "step": 5495
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.681551456451416,
+      "learning_rate": 0.00011109660762604774,
+      "loss": 0.8205,
+      "step": 5500
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6922058463096619,
+      "learning_rate": 0.0001109647263471181,
+      "loss": 0.7905,
+      "step": 5505
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.720058798789978,
+      "learning_rate": 0.00011083282576269905,
+      "loss": 0.9443,
+      "step": 5510
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7844532132148743,
+      "learning_rate": 0.00011070090610502663,
+      "loss": 1.046,
+      "step": 5515
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7096543312072754,
+      "learning_rate": 0.00011056896760637063,
+      "loss": 0.9262,
+      "step": 5520
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.5779085755348206,
+      "learning_rate": 0.00011043701049903381,
+      "loss": 0.9182,
+      "step": 5525
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.658515214920044,
+      "learning_rate": 0.00011030503501535186,
+      "loss": 0.9063,
+      "step": 5530
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.8104361891746521,
+      "learning_rate": 0.00011017304138769272,
+      "loss": 1.0538,
+      "step": 5535
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7393267750740051,
+      "learning_rate": 0.00011004102984845635,
+      "loss": 0.855,
+      "step": 5540
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7669159173965454,
+      "learning_rate": 0.00010990900063007414,
+      "loss": 0.8585,
+      "step": 5545
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7517720460891724,
+      "learning_rate": 0.00010977695396500878,
+      "loss": 0.926,
+      "step": 5550
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7571786046028137,
+      "learning_rate": 0.00010964489008575354,
+      "loss": 0.9036,
+      "step": 5555
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7351516485214233,
+      "learning_rate": 0.00010951280922483198,
+      "loss": 0.8452,
+      "step": 5560
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6579244136810303,
+      "learning_rate": 0.0001093807116147977,
+      "loss": 0.8769,
+      "step": 5565
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6526955366134644,
+      "learning_rate": 0.00010924859748823366,
+      "loss": 1.0031,
+      "step": 5570
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.632077693939209,
+      "learning_rate": 0.00010911646707775194,
+      "loss": 0.9269,
+      "step": 5575
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6900189518928528,
+      "learning_rate": 0.00010898432061599333,
+      "loss": 0.912,
+      "step": 5580
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7273942828178406,
+      "learning_rate": 0.00010885215833562683,
+      "loss": 0.9147,
+      "step": 5585
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7947009205818176,
+      "learning_rate": 0.00010871998046934928,
+      "loss": 0.9692,
+      "step": 5590
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7955752015113831,
+      "learning_rate": 0.00010858778724988506,
+      "loss": 0.9498,
+      "step": 5595
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6939447522163391,
+      "learning_rate": 0.00010845557890998545,
+      "loss": 0.7881,
+      "step": 5600
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6717495322227478,
+      "learning_rate": 0.00010832335568242851,
+      "loss": 1.052,
+      "step": 5605
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7081969380378723,
+      "learning_rate": 0.0001081911178000183,
+      "loss": 0.8961,
+      "step": 5610
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7849963307380676,
+      "learning_rate": 0.00010805886549558484,
+      "loss": 0.782,
+      "step": 5615
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6348875164985657,
+      "learning_rate": 0.00010792659900198359,
+      "loss": 0.9086,
+      "step": 5620
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.6073439717292786,
+      "learning_rate": 0.00010779431855209478,
+      "loss": 0.8713,
+      "step": 5625
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7204303741455078,
+      "learning_rate": 0.0001076620243788234,
+      "loss": 0.9258,
+      "step": 5630
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.6874275207519531,
+      "learning_rate": 0.00010752971671509857,
+      "loss": 0.9345,
+      "step": 5635
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.6767924427986145,
+      "learning_rate": 0.00010739739579387311,
+      "loss": 0.9967,
+      "step": 5640
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7352017760276794,
+      "learning_rate": 0.00010726506184812322,
+      "loss": 0.9132,
+      "step": 5645
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.5592889189720154,
+      "learning_rate": 0.00010713271511084797,
+      "loss": 0.8286,
+      "step": 5650
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.6905889511108398,
+      "learning_rate": 0.00010700035581506908,
+      "loss": 0.9153,
+      "step": 5655
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7167163491249084,
+      "learning_rate": 0.00010686798419383027,
+      "loss": 0.8604,
+      "step": 5660
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.6322777271270752,
+      "learning_rate": 0.00010673560048019693,
+      "loss": 0.8144,
+      "step": 5665
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.6241506338119507,
+      "learning_rate": 0.0001066032049072559,
+      "loss": 0.8855,
+      "step": 5670
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7140413522720337,
+      "learning_rate": 0.00010647079770811479,
+      "loss": 0.8696,
+      "step": 5675
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.9030040502548218,
+      "learning_rate": 0.00010633837911590163,
+      "loss": 0.8033,
+      "step": 5680
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7416821718215942,
+      "learning_rate": 0.00010620594936376466,
+      "loss": 0.9086,
+      "step": 5685
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.5464077591896057,
+      "learning_rate": 0.00010607350868487165,
+      "loss": 0.8224,
+      "step": 5690
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.8908843994140625,
+      "learning_rate": 0.00010594105731240961,
+      "loss": 0.872,
+      "step": 5695
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.6819363832473755,
+      "learning_rate": 0.00010580859547958448,
+      "loss": 0.7992,
+      "step": 5700
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.9416067004203796,
+      "learning_rate": 0.00010567612341962048,
+      "loss": 0.9973,
+      "step": 5705
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7403356432914734,
+      "learning_rate": 0.00010554364136575998,
+      "loss": 0.9134,
+      "step": 5710
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.6867555379867554,
+      "learning_rate": 0.00010541114955126284,
+      "loss": 0.917,
+      "step": 5715
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.6694294214248657,
+      "learning_rate": 0.00010527864820940608,
+      "loss": 0.8229,
+      "step": 5720
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7279865741729736,
+      "learning_rate": 0.00010514613757348364,
+      "loss": 0.9539,
+      "step": 5725
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.5869626998901367,
+      "learning_rate": 0.0001050136178768057,
+      "loss": 0.8783,
+      "step": 5730
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.8429718017578125,
+      "learning_rate": 0.00010488108935269843,
+      "loss": 1.0312,
+      "step": 5735
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.6295071244239807,
+      "learning_rate": 0.00010474855223450355,
+      "loss": 0.8995,
+      "step": 5740
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.5459900498390198,
+      "learning_rate": 0.0001046160067555779,
+      "loss": 0.9544,
+      "step": 5745
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.7036649584770203,
+      "learning_rate": 0.00010448345314929301,
+      "loss": 0.947,
+      "step": 5750
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6790043115615845,
+      "learning_rate": 0.00010435089164903484,
+      "loss": 0.9549,
+      "step": 5755
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.7928381562232971,
+      "learning_rate": 0.00010421832248820309,
+      "loss": 0.9555,
+      "step": 5760
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6902924180030823,
+      "learning_rate": 0.00010408574590021101,
+      "loss": 0.9362,
+      "step": 5765
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8154725432395935,
+      "learning_rate": 0.000103953162118485,
+      "loss": 0.9168,
+      "step": 5770
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.7690364122390747,
+      "learning_rate": 0.00010382057137646401,
+      "loss": 0.984,
+      "step": 5775
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6170430183410645,
+      "learning_rate": 0.00010368797390759937,
+      "loss": 0.7565,
+      "step": 5780
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.7296398878097534,
+      "learning_rate": 0.0001035553699453541,
+      "loss": 0.971,
+      "step": 5785
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6446324586868286,
+      "learning_rate": 0.00010342275972320276,
+      "loss": 0.9776,
+      "step": 5790
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6967552304267883,
+      "learning_rate": 0.00010329014347463097,
+      "loss": 0.9204,
+      "step": 5795
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.7746196985244751,
+      "learning_rate": 0.00010315752143313479,
+      "loss": 0.887,
+      "step": 5800
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.7225843667984009,
+      "learning_rate": 0.00010302489383222065,
+      "loss": 0.8241,
+      "step": 5805
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6819685101509094,
+      "learning_rate": 0.00010289226090540473,
+      "loss": 0.7739,
+      "step": 5810
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8176074028015137,
+      "learning_rate": 0.00010275962288621251,
+      "loss": 0.9995,
+      "step": 5815
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.7226637005805969,
+      "learning_rate": 0.00010262698000817852,
+      "loss": 0.8244,
+      "step": 5820
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6456871032714844,
+      "learning_rate": 0.00010249433250484579,
+      "loss": 0.8483,
+      "step": 5825
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.7270181775093079,
+      "learning_rate": 0.00010236168060976555,
+      "loss": 0.8909,
+      "step": 5830
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8558886647224426,
+      "learning_rate": 0.00010222902455649673,
+      "loss": 0.8855,
+      "step": 5835
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.706078052520752,
+      "learning_rate": 0.00010209636457860552,
+      "loss": 0.9707,
+      "step": 5840
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6462787985801697,
+      "learning_rate": 0.00010196370090966516,
+      "loss": 0.8618,
+      "step": 5845
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6713019013404846,
+      "learning_rate": 0.0001018310337832553,
+      "loss": 0.8665,
+      "step": 5850
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.7476385831832886,
+      "learning_rate": 0.00010169836343296162,
+      "loss": 0.885,
+      "step": 5855
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.769382655620575,
+      "learning_rate": 0.0001015656900923756,
+      "loss": 0.8916,
+      "step": 5860
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.8415050506591797,
+      "learning_rate": 0.00010143301399509395,
+      "loss": 0.9444,
+      "step": 5865
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5454522371292114,
+      "learning_rate": 0.00010130033537471815,
+      "loss": 0.8983,
+      "step": 5870
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6528340578079224,
+      "learning_rate": 0.00010116765446485423,
+      "loss": 0.9029,
+      "step": 5875
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6366670727729797,
+      "learning_rate": 0.0001010349714991122,
+      "loss": 0.7826,
+      "step": 5880
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7491776943206787,
+      "learning_rate": 0.00010090228671110568,
+      "loss": 1.0645,
+      "step": 5885
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5836167335510254,
+      "learning_rate": 0.00010076960033445155,
+      "loss": 0.883,
+      "step": 5890
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7268155217170715,
+      "learning_rate": 0.0001006369126027694,
+      "loss": 0.8976,
+      "step": 5895
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7044017910957336,
+      "learning_rate": 0.00010050422374968131,
+      "loss": 0.777,
+      "step": 5900
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6972168684005737,
+      "learning_rate": 0.00010037153400881126,
+      "loss": 0.9469,
+      "step": 5905
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7724279165267944,
+      "learning_rate": 0.00010023884361378477,
+      "loss": 0.9001,
+      "step": 5910
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7246273159980774,
+      "learning_rate": 0.00010010615279822865,
+      "loss": 0.9022,
+      "step": 5915
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7706332802772522,
+      "learning_rate": 9.99734617957703e-05,
+      "loss": 0.8842,
+      "step": 5920
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.632017195224762,
+      "learning_rate": 9.984077084003752e-05,
+      "loss": 0.9489,
+      "step": 5925
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7416812181472778,
+      "learning_rate": 9.970808016465797e-05,
+      "loss": 1.0136,
+      "step": 5930
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6422529816627502,
+      "learning_rate": 9.957539000325893e-05,
+      "loss": 1.0206,
+      "step": 5935
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.9573733806610107,
+      "learning_rate": 9.944270058946666e-05,
+      "loss": 1.0644,
+      "step": 5940
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7484020590782166,
+      "learning_rate": 9.931001215690616e-05,
+      "loss": 0.9469,
+      "step": 5945
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6472415924072266,
+      "learning_rate": 9.917732493920071e-05,
+      "loss": 0.7979,
+      "step": 5950
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6799381971359253,
+      "learning_rate": 9.90446391699714e-05,
+      "loss": 1.0575,
+      "step": 5955
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7971736192703247,
+      "learning_rate": 9.891195508283684e-05,
+      "loss": 0.8398,
+      "step": 5960
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.60069340467453,
+      "learning_rate": 9.877927291141261e-05,
+      "loss": 0.82,
+      "step": 5965
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7374914884567261,
+      "learning_rate": 9.864659288931095e-05,
+      "loss": 0.8881,
+      "step": 5970
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6885590553283691,
+      "learning_rate": 9.851391525014035e-05,
+      "loss": 0.9565,
+      "step": 5975
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7123907804489136,
+      "learning_rate": 9.838124022750502e-05,
+      "loss": 0.9049,
+      "step": 5980
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.6635785102844238,
+      "learning_rate": 9.824856805500462e-05,
+      "loss": 0.9667,
+      "step": 5985
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.63397616147995,
+      "learning_rate": 9.811589896623382e-05,
+      "loss": 0.9595,
+      "step": 5990
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.2405569553375244,
+      "learning_rate": 9.798323319478178e-05,
+      "loss": 0.7308,
+      "step": 5995
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7486843466758728,
+      "learning_rate": 9.785057097423186e-05,
+      "loss": 0.8563,
+      "step": 6000
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7018646597862244,
+      "learning_rate": 9.771791253816123e-05,
+      "loss": 0.8959,
+      "step": 6005
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.6461020112037659,
+      "learning_rate": 9.758525812014029e-05,
+      "loss": 0.8633,
+      "step": 6010
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7459319829940796,
+      "learning_rate": 9.745260795373239e-05,
+      "loss": 0.8983,
+      "step": 6015
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.6706824898719788,
+      "learning_rate": 9.731996227249347e-05,
+      "loss": 0.9012,
+      "step": 6020
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.8766348958015442,
+      "learning_rate": 9.718732130997148e-05,
+      "loss": 0.9232,
+      "step": 6025
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.6813793778419495,
+      "learning_rate": 9.705468529970613e-05,
+      "loss": 0.9458,
+      "step": 6030
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.6572619676589966,
+      "learning_rate": 9.692205447522837e-05,
+      "loss": 1.0518,
+      "step": 6035
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.8391721844673157,
+      "learning_rate": 9.678942907006002e-05,
+      "loss": 0.8991,
+      "step": 6040
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7584266662597656,
+      "learning_rate": 9.665680931771341e-05,
+      "loss": 0.949,
+      "step": 6045
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.6266197562217712,
+      "learning_rate": 9.652419545169083e-05,
+      "loss": 0.8782,
+      "step": 6050
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7364962697029114,
+      "learning_rate": 9.639158770548426e-05,
+      "loss": 0.898,
+      "step": 6055
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.8394407033920288,
+      "learning_rate": 9.625898631257492e-05,
+      "loss": 0.9739,
+      "step": 6060
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7688999772071838,
+      "learning_rate": 9.612639150643282e-05,
+      "loss": 0.8576,
+      "step": 6065
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7517114877700806,
+      "learning_rate": 9.599380352051633e-05,
+      "loss": 0.9308,
+      "step": 6070
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.5838608741760254,
+      "learning_rate": 9.586122258827193e-05,
+      "loss": 0.9275,
+      "step": 6075
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.831892192363739,
+      "learning_rate": 9.572864894313357e-05,
+      "loss": 0.8508,
+      "step": 6080
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.8587386608123779,
+      "learning_rate": 9.559608281852238e-05,
+      "loss": 0.9304,
+      "step": 6085
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7076795697212219,
+      "learning_rate": 9.546352444784632e-05,
+      "loss": 0.8549,
+      "step": 6090
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7201245427131653,
+      "learning_rate": 9.533097406449962e-05,
+      "loss": 0.953,
+      "step": 6095
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7607429027557373,
+      "learning_rate": 9.519843190186249e-05,
+      "loss": 0.923,
+      "step": 6100
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7092984914779663,
+      "learning_rate": 9.506589819330069e-05,
+      "loss": 0.8955,
+      "step": 6105
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7726672887802124,
+      "learning_rate": 9.493337317216498e-05,
+      "loss": 0.8895,
+      "step": 6110
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6731693148612976,
+      "learning_rate": 9.4800857071791e-05,
+      "loss": 0.8785,
+      "step": 6115
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6858357787132263,
+      "learning_rate": 9.466835012549855e-05,
+      "loss": 0.7771,
+      "step": 6120
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.8080103397369385,
+      "learning_rate": 9.453585256659127e-05,
+      "loss": 0.8201,
+      "step": 6125
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7510241270065308,
+      "learning_rate": 9.440336462835648e-05,
+      "loss": 0.8998,
+      "step": 6130
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7484809160232544,
+      "learning_rate": 9.42708865440644e-05,
+      "loss": 0.9297,
+      "step": 6135
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7201412916183472,
+      "learning_rate": 9.413841854696785e-05,
+      "loss": 0.8524,
+      "step": 6140
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7931877970695496,
+      "learning_rate": 9.400596087030207e-05,
+      "loss": 0.8292,
+      "step": 6145
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7390592694282532,
+      "learning_rate": 9.387351374728403e-05,
+      "loss": 0.8944,
+      "step": 6150
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6792050004005432,
+      "learning_rate": 9.3741077411112e-05,
+      "loss": 0.9021,
+      "step": 6155
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.73079913854599,
+      "learning_rate": 9.360865209496554e-05,
+      "loss": 0.8946,
+      "step": 6160
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7437912225723267,
+      "learning_rate": 9.347623803200456e-05,
+      "loss": 1.0276,
+      "step": 6165
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.5533177852630615,
+      "learning_rate": 9.334383545536918e-05,
+      "loss": 0.871,
+      "step": 6170
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6979454159736633,
+      "learning_rate": 9.321144459817952e-05,
+      "loss": 0.9913,
+      "step": 6175
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.8437588214874268,
+      "learning_rate": 9.307906569353474e-05,
+      "loss": 0.942,
+      "step": 6180
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7341928482055664,
+      "learning_rate": 9.294669897451324e-05,
+      "loss": 0.9076,
+      "step": 6185
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7048665285110474,
+      "learning_rate": 9.281434467417181e-05,
+      "loss": 0.8918,
+      "step": 6190
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6566076874732971,
+      "learning_rate": 9.268200302554533e-05,
+      "loss": 0.8851,
+      "step": 6195
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7581920623779297,
+      "learning_rate": 9.254967426164661e-05,
+      "loss": 0.8965,
+      "step": 6200
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.0405224561691284,
+      "learning_rate": 9.241735861546555e-05,
+      "loss": 0.9108,
+      "step": 6205
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6762104034423828,
+      "learning_rate": 9.228505631996905e-05,
+      "loss": 1.009,
+      "step": 6210
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7440530061721802,
+      "learning_rate": 9.215276760810061e-05,
+      "loss": 0.9062,
+      "step": 6215
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.673608124256134,
+      "learning_rate": 9.202049271277961e-05,
+      "loss": 0.8629,
+      "step": 6220
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.8083750009536743,
+      "learning_rate": 9.188823186690117e-05,
+      "loss": 0.8885,
+      "step": 6225
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5504729151725769,
+      "learning_rate": 9.175598530333582e-05,
+      "loss": 0.8642,
+      "step": 6230
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.6912506222724915,
+      "learning_rate": 9.162375325492875e-05,
+      "loss": 1.0152,
+      "step": 6235
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.667316198348999,
+      "learning_rate": 9.149153595449968e-05,
+      "loss": 0.8585,
+      "step": 6240
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7056398391723633,
+      "learning_rate": 9.135933363484236e-05,
+      "loss": 0.9625,
+      "step": 6245
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.8069510459899902,
+      "learning_rate": 9.122714652872412e-05,
+      "loss": 0.9227,
+      "step": 6250
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7210214734077454,
+      "learning_rate": 9.109497486888564e-05,
+      "loss": 0.8105,
+      "step": 6255
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.6201779246330261,
+      "learning_rate": 9.096281888804022e-05,
+      "loss": 0.9688,
+      "step": 6260
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7237870693206787,
+      "learning_rate": 9.083067881887365e-05,
+      "loss": 0.8219,
+      "step": 6265
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.721929132938385,
+      "learning_rate": 9.069855489404372e-05,
+      "loss": 0.8314,
+      "step": 6270
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.525771379470825,
+      "learning_rate": 9.056644734617975e-05,
+      "loss": 0.9326,
+      "step": 6275
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.9808894991874695,
+      "learning_rate": 9.043435640788222e-05,
+      "loss": 0.9461,
+      "step": 6280
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7714335322380066,
+      "learning_rate": 9.030228231172245e-05,
+      "loss": 0.9285,
+      "step": 6285
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7612080574035645,
+      "learning_rate": 9.0170225290242e-05,
+      "loss": 0.8963,
+      "step": 6290
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7176976203918457,
+      "learning_rate": 9.003818557595241e-05,
+      "loss": 0.8336,
+      "step": 6295
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.6766675114631653,
+      "learning_rate": 8.990616340133478e-05,
+      "loss": 0.8465,
+      "step": 6300
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7587583065032959,
+      "learning_rate": 8.977415899883928e-05,
+      "loss": 0.8742,
+      "step": 6305
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7511966228485107,
+      "learning_rate": 8.964217260088479e-05,
+      "loss": 0.821,
+      "step": 6310
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7103337049484253,
+      "learning_rate": 8.951020443985854e-05,
+      "loss": 0.9149,
+      "step": 6315
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7293053269386292,
+      "learning_rate": 8.937825474811558e-05,
+      "loss": 0.7952,
+      "step": 6320
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7592235803604126,
+      "learning_rate": 8.924632375797852e-05,
+      "loss": 0.8964,
+      "step": 6325
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5273784399032593,
+      "learning_rate": 8.911441170173698e-05,
+      "loss": 0.7355,
+      "step": 6330
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.7110022306442261,
+      "learning_rate": 8.898251881164723e-05,
+      "loss": 0.9297,
+      "step": 6335
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.7399802207946777,
+      "learning_rate": 8.88506453199319e-05,
+      "loss": 0.8804,
+      "step": 6340
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.7043013572692871,
+      "learning_rate": 8.871879145877933e-05,
+      "loss": 0.882,
+      "step": 6345
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.673186182975769,
+      "learning_rate": 8.858695746034336e-05,
+      "loss": 0.8835,
+      "step": 6350
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.710834264755249,
+      "learning_rate": 8.84551435567429e-05,
+      "loss": 0.8945,
+      "step": 6355
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.6442700624465942,
+      "learning_rate": 8.832334998006143e-05,
+      "loss": 0.8458,
+      "step": 6360
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.6050325632095337,
+      "learning_rate": 8.819157696234659e-05,
+      "loss": 0.8123,
+      "step": 6365
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8802506923675537,
+      "learning_rate": 8.805982473560996e-05,
+      "loss": 0.8687,
+      "step": 6370
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.760107696056366,
+      "learning_rate": 8.792809353182638e-05,
+      "loss": 0.9708,
+      "step": 6375
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.6955555081367493,
+      "learning_rate": 8.779638358293374e-05,
+      "loss": 0.9172,
+      "step": 6380
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5892151594161987,
+      "learning_rate": 8.766469512083251e-05,
+      "loss": 0.8301,
+      "step": 6385
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.6933249831199646,
+      "learning_rate": 8.753302837738527e-05,
+      "loss": 0.8659,
+      "step": 6390
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.7141648530960083,
+      "learning_rate": 8.740138358441648e-05,
+      "loss": 0.9506,
+      "step": 6395
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.7855405211448669,
+      "learning_rate": 8.72697609737118e-05,
+      "loss": 0.8757,
+      "step": 6400
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.6324602365493774,
+      "learning_rate": 8.713816077701792e-05,
+      "loss": 0.8307,
+      "step": 6405
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.6647676229476929,
+      "learning_rate": 8.700658322604211e-05,
+      "loss": 0.8077,
+      "step": 6410
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.7383860945701599,
+      "learning_rate": 8.687502855245169e-05,
+      "loss": 0.9212,
+      "step": 6415
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.6514068841934204,
+      "learning_rate": 8.674349698787366e-05,
+      "loss": 0.9188,
+      "step": 6420
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8000485301017761,
+      "learning_rate": 8.661198876389448e-05,
+      "loss": 0.9233,
+      "step": 6425
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5793482065200806,
+      "learning_rate": 8.64805041120594e-05,
+      "loss": 0.9497,
+      "step": 6430
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.7336552143096924,
+      "learning_rate": 8.634904326387216e-05,
+      "loss": 0.8945,
+      "step": 6435
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.9774885773658752,
+      "learning_rate": 8.621760645079468e-05,
+      "loss": 0.954,
+      "step": 6440
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.6644349694252014,
+      "learning_rate": 8.608619390424648e-05,
+      "loss": 0.669,
+      "step": 6445
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.6961809396743774,
+      "learning_rate": 8.595480585560438e-05,
+      "loss": 0.8038,
+      "step": 6450
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.7896516919136047,
+      "learning_rate": 8.582344253620208e-05,
+      "loss": 0.9361,
+      "step": 6455
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5924127101898193,
+      "learning_rate": 8.569210417732975e-05,
+      "loss": 0.6845,
+      "step": 6460
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.69370037317276,
+      "learning_rate": 8.556079101023348e-05,
+      "loss": 0.8415,
+      "step": 6465
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.8123331665992737,
+      "learning_rate": 8.542950326611525e-05,
+      "loss": 0.8595,
+      "step": 6470
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.8372954726219177,
+      "learning_rate": 8.529824117613208e-05,
+      "loss": 0.9186,
+      "step": 6475
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.8233723640441895,
+      "learning_rate": 8.516700497139589e-05,
+      "loss": 0.9458,
+      "step": 6480
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.7458873391151428,
+      "learning_rate": 8.503579488297304e-05,
+      "loss": 0.7842,
+      "step": 6485
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.7014817595481873,
+      "learning_rate": 8.490461114188383e-05,
+      "loss": 0.9634,
+      "step": 6490
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5969160795211792,
+      "learning_rate": 8.477345397910229e-05,
+      "loss": 0.9442,
+      "step": 6495
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6557583808898926,
+      "learning_rate": 8.464232362555557e-05,
+      "loss": 0.8548,
+      "step": 6500
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6432305574417114,
+      "learning_rate": 8.451122031212357e-05,
+      "loss": 0.9555,
+      "step": 6505
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6296955347061157,
+      "learning_rate": 8.438014426963874e-05,
+      "loss": 0.9523,
+      "step": 6510
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6959449052810669,
+      "learning_rate": 8.424909572888542e-05,
+      "loss": 0.8639,
+      "step": 6515
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.7091439366340637,
+      "learning_rate": 8.411807492059944e-05,
+      "loss": 0.8885,
+      "step": 6520
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.7139841914176941,
+      "learning_rate": 8.398708207546797e-05,
+      "loss": 0.9521,
+      "step": 6525
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5753167867660522,
+      "learning_rate": 8.385611742412887e-05,
+      "loss": 0.8778,
+      "step": 6530
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.8551105260848999,
+      "learning_rate": 8.372518119717027e-05,
+      "loss": 1.0675,
+      "step": 6535
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6479779481887817,
+      "learning_rate": 8.359427362513046e-05,
+      "loss": 0.8902,
+      "step": 6540
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.690642774105072,
+      "learning_rate": 8.346339493849704e-05,
+      "loss": 0.8771,
+      "step": 6545
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.7777910232543945,
+      "learning_rate": 8.333254536770696e-05,
+      "loss": 0.8605,
+      "step": 6550
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6432963013648987,
+      "learning_rate": 8.320172514314581e-05,
+      "loss": 0.8891,
+      "step": 6555
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.7652024626731873,
+      "learning_rate": 8.307093449514743e-05,
+      "loss": 0.8381,
+      "step": 6560
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6508191227912903,
+      "learning_rate": 8.294017365399377e-05,
+      "loss": 0.941,
+      "step": 6565
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.674426257610321,
+      "learning_rate": 8.280944284991418e-05,
+      "loss": 0.9117,
+      "step": 6570
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7115979790687561,
+      "learning_rate": 8.267874231308506e-05,
+      "loss": 1.003,
+      "step": 6575
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.6742101311683655,
+      "learning_rate": 8.254807227362973e-05,
+      "loss": 0.7705,
+      "step": 6580
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.6757171750068665,
+      "learning_rate": 8.241743296161759e-05,
+      "loss": 0.8727,
+      "step": 6585
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5816550850868225,
+      "learning_rate": 8.228682460706403e-05,
+      "loss": 1.0165,
+      "step": 6590
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7802658081054688,
+      "learning_rate": 8.215624743993003e-05,
+      "loss": 0.8722,
+      "step": 6595
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7401361465454102,
+      "learning_rate": 8.20257016901215e-05,
+      "loss": 0.8048,
+      "step": 6600
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.8161288499832153,
+      "learning_rate": 8.189518758748908e-05,
+      "loss": 0.8837,
+      "step": 6605
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.729274332523346,
+      "learning_rate": 8.176470536182777e-05,
+      "loss": 0.864,
+      "step": 6610
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.8532578349113464,
+      "learning_rate": 8.163425524287628e-05,
+      "loss": 0.9062,
+      "step": 6615
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.8236483931541443,
+      "learning_rate": 8.150383746031707e-05,
+      "loss": 0.9666,
+      "step": 6620
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.655390202999115,
+      "learning_rate": 8.137345224377536e-05,
+      "loss": 0.9652,
+      "step": 6625
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.8593419790267944,
+      "learning_rate": 8.124309982281914e-05,
+      "loss": 0.9262,
+      "step": 6630
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.6646649241447449,
+      "learning_rate": 8.111278042695881e-05,
+      "loss": 0.7556,
+      "step": 6635
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7062471508979797,
+      "learning_rate": 8.098249428564635e-05,
+      "loss": 0.9019,
+      "step": 6640
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7289904356002808,
+      "learning_rate": 8.08522416282754e-05,
+      "loss": 0.9905,
+      "step": 6645
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5822698473930359,
+      "learning_rate": 8.072202268418057e-05,
+      "loss": 0.8327,
+      "step": 6650
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.0038517713546753,
+      "learning_rate": 8.059183768263712e-05,
+      "loss": 0.7722,
+      "step": 6655
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.6496167182922363,
+      "learning_rate": 8.046168685286052e-05,
+      "loss": 0.8617,
+      "step": 6660
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.6460272073745728,
+      "learning_rate": 8.033157042400613e-05,
+      "loss": 0.7342,
+      "step": 6665
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7813682556152344,
+      "learning_rate": 8.02014886251687e-05,
+      "loss": 0.8798,
+      "step": 6670
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.724125862121582,
+      "learning_rate": 8.007144168538198e-05,
+      "loss": 0.9017,
+      "step": 6675
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7241107821464539,
+      "learning_rate": 7.994142983361843e-05,
+      "loss": 0.8341,
+      "step": 6680
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7320983409881592,
+      "learning_rate": 7.981145329878867e-05,
+      "loss": 0.9171,
+      "step": 6685
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7066484689712524,
+      "learning_rate": 7.96815123097411e-05,
+      "loss": 0.789,
+      "step": 6690
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.6992103457450867,
+      "learning_rate": 7.955160709526167e-05,
+      "loss": 0.9138,
+      "step": 6695
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7294855713844299,
+      "learning_rate": 7.942173788407318e-05,
+      "loss": 0.8713,
+      "step": 6700
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.9073478579521179,
+      "learning_rate": 7.929190490483517e-05,
+      "loss": 0.9354,
+      "step": 6705
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7209857702255249,
+      "learning_rate": 7.916210838614331e-05,
+      "loss": 0.9218,
+      "step": 6710
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7639252543449402,
+      "learning_rate": 7.903234855652907e-05,
+      "loss": 0.9264,
+      "step": 6715
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.8197429776191711,
+      "learning_rate": 7.890262564445939e-05,
+      "loss": 0.8179,
+      "step": 6720
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.6600883603096008,
+      "learning_rate": 7.877293987833617e-05,
+      "loss": 0.8916,
+      "step": 6725
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7015143632888794,
+      "learning_rate": 7.864329148649584e-05,
+      "loss": 0.8594,
+      "step": 6730
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7198832035064697,
+      "learning_rate": 7.851368069720917e-05,
+      "loss": 0.8914,
+      "step": 6735
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7876847386360168,
+      "learning_rate": 7.838410773868061e-05,
+      "loss": 0.7863,
+      "step": 6740
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.8481395840644836,
+      "learning_rate": 7.825457283904802e-05,
+      "loss": 0.9198,
+      "step": 6745
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7551649212837219,
+      "learning_rate": 7.81250762263823e-05,
+      "loss": 0.944,
+      "step": 6750
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.6790037751197815,
+      "learning_rate": 7.799561812868691e-05,
+      "loss": 0.8491,
+      "step": 6755
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.789820671081543,
+      "learning_rate": 7.786619877389742e-05,
+      "loss": 0.9247,
+      "step": 6760
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.6964678168296814,
+      "learning_rate": 7.773681838988136e-05,
+      "loss": 0.7134,
+      "step": 6765
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.8050416707992554,
+      "learning_rate": 7.760747720443744e-05,
+      "loss": 0.8758,
+      "step": 6770
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.764661431312561,
+      "learning_rate": 7.747817544529555e-05,
+      "loss": 0.8431,
+      "step": 6775
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7200865745544434,
+      "learning_rate": 7.7348913340116e-05,
+      "loss": 0.7752,
+      "step": 6780
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.8310842514038086,
+      "learning_rate": 7.721969111648936e-05,
+      "loss": 0.9438,
+      "step": 6785
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.8838992714881897,
+      "learning_rate": 7.709050900193601e-05,
+      "loss": 0.9612,
+      "step": 6790
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.6813116073608398,
+      "learning_rate": 7.696136722390566e-05,
+      "loss": 0.7167,
+      "step": 6795
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.8536310791969299,
+      "learning_rate": 7.683226600977695e-05,
+      "loss": 1.0073,
+      "step": 6800
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.5965607762336731,
+      "learning_rate": 7.670320558685724e-05,
+      "loss": 0.7397,
+      "step": 6805
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7263707518577576,
+      "learning_rate": 7.657418618238196e-05,
+      "loss": 0.8697,
+      "step": 6810
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7662990689277649,
+      "learning_rate": 7.644520802351431e-05,
+      "loss": 0.9073,
+      "step": 6815
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.659342885017395,
+      "learning_rate": 7.631627133734497e-05,
+      "loss": 0.8674,
+      "step": 6820
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.737581729888916,
+      "learning_rate": 7.61873763508915e-05,
+      "loss": 0.9284,
+      "step": 6825
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7625646591186523,
+      "learning_rate": 7.605852329109808e-05,
+      "loss": 0.9188,
+      "step": 6830
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7271213531494141,
+      "learning_rate": 7.592971238483508e-05,
+      "loss": 0.7946,
+      "step": 6835
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7164754867553711,
+      "learning_rate": 7.580094385889862e-05,
+      "loss": 0.9425,
+      "step": 6840
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.6694819331169128,
+      "learning_rate": 7.567221794001025e-05,
+      "loss": 0.8806,
+      "step": 6845
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7427958846092224,
+      "learning_rate": 7.554353485481646e-05,
+      "loss": 0.8411,
+      "step": 6850
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7842068076133728,
+      "learning_rate": 7.54148948298883e-05,
+      "loss": 0.9323,
+      "step": 6855
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7708226442337036,
+      "learning_rate": 7.528629809172109e-05,
+      "loss": 0.8336,
+      "step": 6860
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.6797932982444763,
+      "learning_rate": 7.515774486673386e-05,
+      "loss": 0.9175,
+      "step": 6865
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.772397518157959,
+      "learning_rate": 7.502923538126903e-05,
+      "loss": 1.0511,
+      "step": 6870
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7092522382736206,
+      "learning_rate": 7.490076986159207e-05,
+      "loss": 0.9162,
+      "step": 6875
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.841741144657135,
+      "learning_rate": 7.477234853389099e-05,
+      "loss": 0.9273,
+      "step": 6880
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.682362973690033,
+      "learning_rate": 7.464397162427595e-05,
+      "loss": 0.8971,
+      "step": 6885
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7055883407592773,
+      "learning_rate": 7.451563935877901e-05,
+      "loss": 0.9621,
+      "step": 6890
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.6486619114875793,
+      "learning_rate": 7.438735196335361e-05,
+      "loss": 0.9299,
+      "step": 6895
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.6966800093650818,
+      "learning_rate": 7.425910966387399e-05,
+      "loss": 0.8397,
+      "step": 6900
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.6951152086257935,
+      "learning_rate": 7.413091268613535e-05,
+      "loss": 0.8891,
+      "step": 6905
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7122839689254761,
+      "learning_rate": 7.400276125585275e-05,
+      "loss": 0.8917,
+      "step": 6910
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.6868377923965454,
+      "learning_rate": 7.387465559866118e-05,
+      "loss": 0.9443,
+      "step": 6915
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.8608435988426208,
+      "learning_rate": 7.374659594011519e-05,
+      "loss": 0.744,
+      "step": 6920
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7597634196281433,
+      "learning_rate": 7.361858250568805e-05,
+      "loss": 0.9243,
+      "step": 6925
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6974212527275085,
+      "learning_rate": 7.34906155207719e-05,
+      "loss": 1.0219,
+      "step": 6930
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7687626481056213,
+      "learning_rate": 7.3362695210677e-05,
+      "loss": 0.8733,
+      "step": 6935
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7128692865371704,
+      "learning_rate": 7.32348218006313e-05,
+      "loss": 0.8652,
+      "step": 6940
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6069349646568298,
+      "learning_rate": 7.310699551578045e-05,
+      "loss": 0.8382,
+      "step": 6945
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.985689103603363,
+      "learning_rate": 7.29792165811869e-05,
+      "loss": 0.8298,
+      "step": 6950
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.8701884746551514,
+      "learning_rate": 7.285148522182975e-05,
+      "loss": 0.9955,
+      "step": 6955
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.834453821182251,
+      "learning_rate": 7.272380166260453e-05,
+      "loss": 0.8658,
+      "step": 6960
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.664952278137207,
+      "learning_rate": 7.259616612832237e-05,
+      "loss": 0.8893,
+      "step": 6965
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.8810997009277344,
+      "learning_rate": 7.24685788437099e-05,
+      "loss": 0.9426,
+      "step": 6970
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6937991380691528,
+      "learning_rate": 7.234104003340898e-05,
+      "loss": 0.9102,
+      "step": 6975
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6798984408378601,
+      "learning_rate": 7.221354992197587e-05,
+      "loss": 0.7986,
+      "step": 6980
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.811714768409729,
+      "learning_rate": 7.208610873388122e-05,
+      "loss": 0.9457,
+      "step": 6985
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7090261578559875,
+      "learning_rate": 7.195871669350953e-05,
+      "loss": 0.9537,
+      "step": 6990
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6901644468307495,
+      "learning_rate": 7.183137402515872e-05,
+      "loss": 0.8441,
+      "step": 6995
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.9163976907730103,
+      "learning_rate": 7.170408095303992e-05,
+      "loss": 0.9279,
+      "step": 7000
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6931922435760498,
+      "learning_rate": 7.157683770127671e-05,
+      "loss": 0.8883,
+      "step": 7005
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6846193075180054,
+      "learning_rate": 7.14496444939051e-05,
+      "loss": 0.9294,
+      "step": 7010
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7656509876251221,
+      "learning_rate": 7.132250155487304e-05,
+      "loss": 0.833,
+      "step": 7015
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7121906280517578,
+      "learning_rate": 7.119540910803982e-05,
+      "loss": 0.7782,
+      "step": 7020
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6763817667961121,
+      "learning_rate": 7.106836737717589e-05,
+      "loss": 0.8981,
+      "step": 7025
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7830772995948792,
+      "learning_rate": 7.094137658596247e-05,
+      "loss": 0.8412,
+      "step": 7030
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6828163862228394,
+      "learning_rate": 7.081443695799102e-05,
+      "loss": 0.7903,
+      "step": 7035
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7466477155685425,
+      "learning_rate": 7.068754871676291e-05,
+      "loss": 0.9161,
+      "step": 7040
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7557064294815063,
+      "learning_rate": 7.056071208568911e-05,
+      "loss": 0.9285,
+      "step": 7045
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6851228475570679,
+      "learning_rate": 7.043392728808962e-05,
+      "loss": 0.7728,
+      "step": 7050
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6479991674423218,
+      "learning_rate": 7.030719454719325e-05,
+      "loss": 0.7594,
+      "step": 7055
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.8061099648475647,
+      "learning_rate": 7.018051408613715e-05,
+      "loss": 0.921,
+      "step": 7060
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7075207233428955,
+      "learning_rate": 7.005388612796635e-05,
+      "loss": 0.8041,
+      "step": 7065
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7294245362281799,
+      "learning_rate": 6.992731089563356e-05,
+      "loss": 0.9568,
+      "step": 7070
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7740755081176758,
+      "learning_rate": 6.980078861199854e-05,
+      "loss": 0.8527,
+      "step": 7075
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7101545929908752,
+      "learning_rate": 6.967431949982789e-05,
+      "loss": 0.9794,
+      "step": 7080
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.728982150554657,
+      "learning_rate": 6.954790378179459e-05,
+      "loss": 0.8613,
+      "step": 7085
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6854989528656006,
+      "learning_rate": 6.942154168047756e-05,
+      "loss": 0.8637,
+      "step": 7090
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7399042248725891,
+      "learning_rate": 6.929523341836133e-05,
+      "loss": 0.9954,
+      "step": 7095
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.9394546151161194,
+      "learning_rate": 6.916897921783574e-05,
+      "loss": 0.784,
+      "step": 7100
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6976795792579651,
+      "learning_rate": 6.904277930119529e-05,
+      "loss": 0.896,
+      "step": 7105
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.905849277973175,
+      "learning_rate": 6.891663389063898e-05,
+      "loss": 0.8973,
+      "step": 7110
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.742719292640686,
+      "learning_rate": 6.879054320826988e-05,
+      "loss": 0.8795,
+      "step": 7115
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6740474104881287,
+      "learning_rate": 6.866450747609461e-05,
+      "loss": 0.8453,
+      "step": 7120
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7123729586601257,
+      "learning_rate": 6.853852691602309e-05,
+      "loss": 0.867,
+      "step": 7125
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.8164270520210266,
+      "learning_rate": 6.841260174986811e-05,
+      "loss": 0.9143,
+      "step": 7130
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7992966175079346,
+      "learning_rate": 6.828673219934491e-05,
+      "loss": 0.9265,
+      "step": 7135
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7531195878982544,
+      "learning_rate": 6.816091848607081e-05,
+      "loss": 0.8229,
+      "step": 7140
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.8514467477798462,
+      "learning_rate": 6.80351608315648e-05,
+      "loss": 0.697,
+      "step": 7145
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6383096575737,
+      "learning_rate": 6.790945945724721e-05,
+      "loss": 0.9276,
+      "step": 7150
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.8250300288200378,
+      "learning_rate": 6.778381458443925e-05,
+      "loss": 0.8385,
+      "step": 7155
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.692227303981781,
+      "learning_rate": 6.765822643436267e-05,
+      "loss": 0.8599,
+      "step": 7160
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.6989490985870361,
+      "learning_rate": 6.753269522813929e-05,
+      "loss": 0.7171,
+      "step": 7165
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7054385542869568,
+      "learning_rate": 6.740722118679075e-05,
+      "loss": 0.7917,
+      "step": 7170
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7425069212913513,
+      "learning_rate": 6.728180453123798e-05,
+      "loss": 0.8909,
+      "step": 7175
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7305125594139099,
+      "learning_rate": 6.715644548230086e-05,
+      "loss": 0.8232,
+      "step": 7180
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7323471307754517,
+      "learning_rate": 6.703114426069797e-05,
+      "loss": 0.8548,
+      "step": 7185
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8324387669563293,
+      "learning_rate": 6.69059010870459e-05,
+      "loss": 0.878,
+      "step": 7190
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.678524374961853,
+      "learning_rate": 6.678071618185913e-05,
+      "loss": 0.8011,
+      "step": 7195
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.6823922395706177,
+      "learning_rate": 6.665558976554957e-05,
+      "loss": 0.8471,
+      "step": 7200
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7461297512054443,
+      "learning_rate": 6.653052205842609e-05,
+      "loss": 0.895,
+      "step": 7205
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.6897065043449402,
+      "learning_rate": 6.640551328069414e-05,
+      "loss": 0.9243,
+      "step": 7210
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7159562706947327,
+      "learning_rate": 6.628056365245561e-05,
+      "loss": 0.8941,
+      "step": 7215
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8460803627967834,
+      "learning_rate": 6.615567339370803e-05,
+      "loss": 0.8553,
+      "step": 7220
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7752892971038818,
+      "learning_rate": 6.603084272434455e-05,
+      "loss": 0.836,
+      "step": 7225
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.9097653031349182,
+      "learning_rate": 6.59060718641533e-05,
+      "loss": 0.9746,
+      "step": 7230
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8083062767982483,
+      "learning_rate": 6.578136103281717e-05,
+      "loss": 0.9358,
+      "step": 7235
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8664425015449524,
+      "learning_rate": 6.565671044991335e-05,
+      "loss": 0.998,
+      "step": 7240
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7137695550918579,
+      "learning_rate": 6.553212033491291e-05,
+      "loss": 0.8507,
+      "step": 7245
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.6330139636993408,
+      "learning_rate": 6.540759090718047e-05,
+      "loss": 0.8012,
+      "step": 7250
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8543661832809448,
+      "learning_rate": 6.528312238597382e-05,
+      "loss": 0.8562,
+      "step": 7255
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7509561777114868,
+      "learning_rate": 6.515871499044358e-05,
+      "loss": 0.7816,
+      "step": 7260
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7578710317611694,
+      "learning_rate": 6.50343689396325e-05,
+      "loss": 0.8554,
+      "step": 7265
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7192575335502625,
+      "learning_rate": 6.491008445247563e-05,
+      "loss": 0.8258,
+      "step": 7270
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7355043292045593,
+      "learning_rate": 6.478586174779947e-05,
+      "loss": 0.7551,
+      "step": 7275
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8419028520584106,
+      "learning_rate": 6.466170104432166e-05,
+      "loss": 0.9343,
+      "step": 7280
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7209092378616333,
+      "learning_rate": 6.453760256065091e-05,
+      "loss": 0.9588,
+      "step": 7285
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.8162868618965149,
+      "learning_rate": 6.441356651528609e-05,
+      "loss": 0.9452,
+      "step": 7290
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7459548711776733,
+      "learning_rate": 6.428959312661642e-05,
+      "loss": 0.8994,
+      "step": 7295
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7295629382133484,
+      "learning_rate": 6.416568261292062e-05,
+      "loss": 0.8829,
+      "step": 7300
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7880662083625793,
+      "learning_rate": 6.404183519236669e-05,
+      "loss": 0.8597,
+      "step": 7305
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.6742045879364014,
+      "learning_rate": 6.391805108301167e-05,
+      "loss": 0.9528,
+      "step": 7310
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7547880411148071,
+      "learning_rate": 6.37943305028011e-05,
+      "loss": 0.9334,
+      "step": 7315
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7758142352104187,
+      "learning_rate": 6.367067366956854e-05,
+      "loss": 0.9598,
+      "step": 7320
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7337788939476013,
+      "learning_rate": 6.354708080103548e-05,
+      "loss": 1.0353,
+      "step": 7325
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7313909530639648,
+      "learning_rate": 6.342355211481065e-05,
+      "loss": 0.9179,
+      "step": 7330
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.735283613204956,
+      "learning_rate": 6.33000878283898e-05,
+      "loss": 0.9905,
+      "step": 7335
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.714606761932373,
+      "learning_rate": 6.317668815915547e-05,
+      "loss": 0.8085,
+      "step": 7340
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7330592274665833,
+      "learning_rate": 6.305335332437617e-05,
+      "loss": 0.7654,
+      "step": 7345
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.6161234974861145,
+      "learning_rate": 6.293008354120635e-05,
+      "loss": 0.9045,
+      "step": 7350
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7620159387588501,
+      "learning_rate": 6.280687902668604e-05,
+      "loss": 0.9067,
+      "step": 7355
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.6839978694915771,
+      "learning_rate": 6.26837399977402e-05,
+      "loss": 0.8462,
+      "step": 7360
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.6819284558296204,
+      "learning_rate": 6.256066667117855e-05,
+      "loss": 0.824,
+      "step": 7365
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7531672716140747,
+      "learning_rate": 6.243765926369513e-05,
+      "loss": 1.0016,
+      "step": 7370
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.6551452875137329,
+      "learning_rate": 6.231471799186788e-05,
+      "loss": 0.8784,
+      "step": 7375
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7985334396362305,
+      "learning_rate": 6.219184307215843e-05,
+      "loss": 0.8138,
+      "step": 7380
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.8459438681602478,
+      "learning_rate": 6.206903472091139e-05,
+      "loss": 0.9275,
+      "step": 7385
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.6439704298973083,
+      "learning_rate": 6.194629315435426e-05,
+      "loss": 0.9211,
+      "step": 7390
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.6462088227272034,
+      "learning_rate": 6.182361858859699e-05,
+      "loss": 0.9553,
+      "step": 7395
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7741279006004333,
+      "learning_rate": 6.170101123963152e-05,
+      "loss": 0.8159,
+      "step": 7400
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.810332179069519,
+      "learning_rate": 6.157847132333138e-05,
+      "loss": 0.9315,
+      "step": 7405
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6735566854476929,
+      "learning_rate": 6.145599905545151e-05,
+      "loss": 0.9615,
+      "step": 7410
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7178578972816467,
+      "learning_rate": 6.133359465162767e-05,
+      "loss": 0.8496,
+      "step": 7415
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.8918485045433044,
+      "learning_rate": 6.121125832737605e-05,
+      "loss": 0.8955,
+      "step": 7420
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6969971060752869,
+      "learning_rate": 6.108899029809313e-05,
+      "loss": 0.8493,
+      "step": 7425
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7077623605728149,
+      "learning_rate": 6.0966790779055036e-05,
+      "loss": 0.7798,
+      "step": 7430
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6919378042221069,
+      "learning_rate": 6.0844659985417285e-05,
+      "loss": 0.829,
+      "step": 7435
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.2355577945709229,
+      "learning_rate": 6.0722598132214445e-05,
+      "loss": 0.8555,
+      "step": 7440
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6976816654205322,
+      "learning_rate": 6.060060543435961e-05,
+      "loss": 0.897,
+      "step": 7445
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6664767861366272,
+      "learning_rate": 6.0478682106644225e-05,
+      "loss": 0.815,
+      "step": 7450
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6910175085067749,
+      "learning_rate": 6.0356828363737484e-05,
+      "loss": 0.7017,
+      "step": 7455
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7369092702865601,
+      "learning_rate": 6.0235044420186125e-05,
+      "loss": 0.8366,
+      "step": 7460
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7320170998573303,
+      "learning_rate": 6.0113330490413985e-05,
+      "loss": 0.8964,
+      "step": 7465
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6991029977798462,
+      "learning_rate": 5.9991686788721646e-05,
+      "loss": 0.7967,
+      "step": 7470
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.5753344297409058,
+      "learning_rate": 5.9870113529285956e-05,
+      "loss": 0.9125,
+      "step": 7475
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7163155674934387,
+      "learning_rate": 5.974861092615985e-05,
+      "loss": 0.8387,
+      "step": 7480
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7662732601165771,
+      "learning_rate": 5.96271791932718e-05,
+      "loss": 0.9932,
+      "step": 7485
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.8402357697486877,
+      "learning_rate": 5.950581854442547e-05,
+      "loss": 0.8119,
+      "step": 7490
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7967143654823303,
+      "learning_rate": 5.9384529193299444e-05,
+      "loss": 0.945,
+      "step": 7495
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7467031478881836,
+      "learning_rate": 5.926331135344671e-05,
+      "loss": 0.8427,
+      "step": 7500
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6836127042770386,
+      "learning_rate": 5.9142165238294344e-05,
+      "loss": 0.885,
+      "step": 7505
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.8548532128334045,
+      "learning_rate": 5.9021091061143194e-05,
+      "loss": 0.9242,
+      "step": 7510
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6110352873802185,
+      "learning_rate": 5.89000890351674e-05,
+      "loss": 0.9173,
+      "step": 7515
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7760581970214844,
+      "learning_rate": 5.877915937341407e-05,
+      "loss": 0.9793,
+      "step": 7520
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.6202065944671631,
+      "learning_rate": 5.865830228880294e-05,
+      "loss": 0.8601,
+      "step": 7525
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.9943656921386719,
+      "learning_rate": 5.8537517994125876e-05,
+      "loss": 0.7854,
+      "step": 7530
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7132951617240906,
+      "learning_rate": 5.84168067020467e-05,
+      "loss": 0.9711,
+      "step": 7535
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.8035358786582947,
+      "learning_rate": 5.829616862510059e-05,
+      "loss": 0.8294,
+      "step": 7540
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7671014070510864,
+      "learning_rate": 5.817560397569385e-05,
+      "loss": 0.8559,
+      "step": 7545
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.6550133228302002,
+      "learning_rate": 5.805511296610362e-05,
+      "loss": 0.8136,
+      "step": 7550
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.594433069229126,
+      "learning_rate": 5.793469580847714e-05,
+      "loss": 0.8446,
+      "step": 7555
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7266082167625427,
+      "learning_rate": 5.7814352714831774e-05,
+      "loss": 0.9847,
+      "step": 7560
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.8068563342094421,
+      "learning_rate": 5.769408389705453e-05,
+      "loss": 0.8966,
+      "step": 7565
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7309244871139526,
+      "learning_rate": 5.757388956690155e-05,
+      "loss": 0.926,
+      "step": 7570
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.6696233749389648,
+      "learning_rate": 5.7453769935997825e-05,
+      "loss": 0.9283,
+      "step": 7575
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7572994828224182,
+      "learning_rate": 5.733372521583686e-05,
+      "loss": 1.004,
+      "step": 7580
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7896938920021057,
+      "learning_rate": 5.721375561778026e-05,
+      "loss": 0.881,
+      "step": 7585
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.6995894908905029,
+      "learning_rate": 5.70938613530573e-05,
+      "loss": 0.8758,
+      "step": 7590
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.8638085722923279,
+      "learning_rate": 5.697404263276476e-05,
+      "loss": 0.8162,
+      "step": 7595
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7239465117454529,
+      "learning_rate": 5.685429966786628e-05,
+      "loss": 0.8784,
+      "step": 7600
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7965220808982849,
+      "learning_rate": 5.673463266919216e-05,
+      "loss": 0.7788,
+      "step": 7605
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7969363927841187,
+      "learning_rate": 5.661504184743895e-05,
+      "loss": 0.9163,
+      "step": 7610
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7120541930198669,
+      "learning_rate": 5.6495527413169026e-05,
+      "loss": 0.9791,
+      "step": 7615
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7569546103477478,
+      "learning_rate": 5.6376089576810396e-05,
+      "loss": 0.8799,
+      "step": 7620
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.6639888286590576,
+      "learning_rate": 5.625672854865609e-05,
+      "loss": 0.8349,
+      "step": 7625
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.8096963763237,
+      "learning_rate": 5.613744453886394e-05,
+      "loss": 0.8983,
+      "step": 7630
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7809593677520752,
+      "learning_rate": 5.6018237757456163e-05,
+      "loss": 0.8812,
+      "step": 7635
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.8204253315925598,
+      "learning_rate": 5.5899108414318994e-05,
+      "loss": 0.862,
+      "step": 7640
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7801651358604431,
+      "learning_rate": 5.5780056719202304e-05,
+      "loss": 0.873,
+      "step": 7645
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.9668874144554138,
+      "learning_rate": 5.566108288171936e-05,
+      "loss": 0.8918,
+      "step": 7650
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7693642377853394,
+      "learning_rate": 5.5542187111346224e-05,
+      "loss": 0.8444,
+      "step": 7655
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.6818474531173706,
+      "learning_rate": 5.5423369617421564e-05,
+      "loss": 0.7371,
+      "step": 7660
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.748066246509552,
+      "learning_rate": 5.530463060914619e-05,
+      "loss": 0.8924,
+      "step": 7665
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.6938193440437317,
+      "learning_rate": 5.5185970295582726e-05,
+      "loss": 0.762,
+      "step": 7670
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7946302890777588,
+      "learning_rate": 5.50673888856553e-05,
+      "loss": 0.8746,
+      "step": 7675
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.6880958080291748,
+      "learning_rate": 5.494888658814907e-05,
+      "loss": 0.8095,
+      "step": 7680
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.9050551652908325,
+      "learning_rate": 5.483046361170992e-05,
+      "loss": 0.9121,
+      "step": 7685
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7461682558059692,
+      "learning_rate": 5.471212016484399e-05,
+      "loss": 0.9817,
+      "step": 7690
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.819770872592926,
+      "learning_rate": 5.4593856455917536e-05,
+      "loss": 0.8736,
+      "step": 7695
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.8093503713607788,
+      "learning_rate": 5.447567269315627e-05,
+      "loss": 0.9012,
+      "step": 7700
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.6116926074028015,
+      "learning_rate": 5.435756908464529e-05,
+      "loss": 0.8433,
+      "step": 7705
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.8973492383956909,
+      "learning_rate": 5.4239545838328475e-05,
+      "loss": 0.8598,
+      "step": 7710
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7746146321296692,
+      "learning_rate": 5.4121603162008226e-05,
+      "loss": 0.904,
+      "step": 7715
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.867224395275116,
+      "learning_rate": 5.400374126334511e-05,
+      "loss": 0.8575,
+      "step": 7720
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.6685112714767456,
+      "learning_rate": 5.388596034985742e-05,
+      "loss": 0.9585,
+      "step": 7725
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7715603113174438,
+      "learning_rate": 5.376826062892086e-05,
+      "loss": 0.8547,
+      "step": 7730
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.8174254894256592,
+      "learning_rate": 5.365064230776831e-05,
+      "loss": 0.8832,
+      "step": 7735
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7603541016578674,
+      "learning_rate": 5.3533105593489163e-05,
+      "loss": 0.931,
+      "step": 7740
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.6725593209266663,
+      "learning_rate": 5.3415650693029205e-05,
+      "loss": 0.8013,
+      "step": 7745
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.9110146164894104,
+      "learning_rate": 5.329827781319018e-05,
+      "loss": 0.9516,
+      "step": 7750
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.8404948711395264,
+      "learning_rate": 5.318098716062934e-05,
+      "loss": 0.9361,
+      "step": 7755
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7202340364456177,
+      "learning_rate": 5.30637789418593e-05,
+      "loss": 0.9126,
+      "step": 7760
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7240476608276367,
+      "learning_rate": 5.294665336324742e-05,
+      "loss": 0.9588,
+      "step": 7765
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7280794382095337,
+      "learning_rate": 5.2829610631015606e-05,
+      "loss": 0.9005,
+      "step": 7770
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.8151552677154541,
+      "learning_rate": 5.271265095123987e-05,
+      "loss": 0.9236,
+      "step": 7775
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7294365763664246,
+      "learning_rate": 5.2595774529850006e-05,
+      "loss": 0.8351,
+      "step": 7780
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7628800868988037,
+      "learning_rate": 5.24789815726292e-05,
+      "loss": 0.8084,
+      "step": 7785
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7836165428161621,
+      "learning_rate": 5.2362272285213756e-05,
+      "loss": 0.8823,
+      "step": 7790
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7417895793914795,
+      "learning_rate": 5.224564687309261e-05,
+      "loss": 1.0214,
+      "step": 7795
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7525092363357544,
+      "learning_rate": 5.2129105541606916e-05,
+      "loss": 0.945,
+      "step": 7800
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7636805176734924,
+      "learning_rate": 5.2012648495949976e-05,
+      "loss": 0.8494,
+      "step": 7805
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.8036842942237854,
+      "learning_rate": 5.189627594116657e-05,
+      "loss": 0.872,
+      "step": 7810
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.8050177693367004,
+      "learning_rate": 5.1779988082152786e-05,
+      "loss": 0.896,
+      "step": 7815
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7123312950134277,
+      "learning_rate": 5.166378512365552e-05,
+      "loss": 0.8844,
+      "step": 7820
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.6821147799491882,
+      "learning_rate": 5.1547667270272226e-05,
+      "loss": 0.8998,
+      "step": 7825
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7495633959770203,
+      "learning_rate": 5.143163472645049e-05,
+      "loss": 0.8506,
+      "step": 7830
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7350967526435852,
+      "learning_rate": 5.131568769648775e-05,
+      "loss": 0.7543,
+      "step": 7835
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7220240235328674,
+      "learning_rate": 5.119982638453075e-05,
+      "loss": 0.8647,
+      "step": 7840
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.6074078679084778,
+      "learning_rate": 5.108405099457549e-05,
+      "loss": 0.7925,
+      "step": 7845
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7537201046943665,
+      "learning_rate": 5.096836173046663e-05,
+      "loss": 0.953,
+      "step": 7850
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.8651964068412781,
+      "learning_rate": 5.0852758795897006e-05,
+      "loss": 0.8693,
+      "step": 7855
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.6949556469917297,
+      "learning_rate": 5.073724239440773e-05,
+      "loss": 0.8184,
+      "step": 7860
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.8523896932601929,
+      "learning_rate": 5.06218127293874e-05,
+      "loss": 0.9198,
+      "step": 7865
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7108052968978882,
+      "learning_rate": 5.050647000407189e-05,
+      "loss": 0.841,
+      "step": 7870
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.8096588850021362,
+      "learning_rate": 5.039121442154415e-05,
+      "loss": 0.9523,
+      "step": 7875
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7509374618530273,
+      "learning_rate": 5.027604618473347e-05,
+      "loss": 0.8155,
+      "step": 7880
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6036114692687988,
+      "learning_rate": 5.016096549641549e-05,
+      "loss": 0.7785,
+      "step": 7885
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7784599661827087,
+      "learning_rate": 5.004597255921174e-05,
+      "loss": 0.9645,
+      "step": 7890
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7571439146995544,
+      "learning_rate": 4.993106757558912e-05,
+      "loss": 0.9329,
+      "step": 7895
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.5869072079658508,
+      "learning_rate": 4.981625074785986e-05,
+      "loss": 0.809,
+      "step": 7900
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.8537865877151489,
+      "learning_rate": 4.9701522278180736e-05,
+      "loss": 0.9381,
+      "step": 7905
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.2609446048736572,
+      "learning_rate": 4.958688236855308e-05,
+      "loss": 0.899,
+      "step": 7910
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.8565067052841187,
+      "learning_rate": 4.9472331220822366e-05,
+      "loss": 1.046,
+      "step": 7915
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.9478633403778076,
+      "learning_rate": 4.935786903667767e-05,
+      "loss": 0.8329,
+      "step": 7920
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7316033244132996,
+      "learning_rate": 4.9243496017651434e-05,
+      "loss": 0.8807,
+      "step": 7925
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.9690425395965576,
+      "learning_rate": 4.912921236511927e-05,
+      "loss": 0.8134,
+      "step": 7930
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7974583506584167,
+      "learning_rate": 4.901501828029919e-05,
+      "loss": 0.991,
+      "step": 7935
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.8434284925460815,
+      "learning_rate": 4.890091396425163e-05,
+      "loss": 0.8801,
+      "step": 7940
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7458634972572327,
+      "learning_rate": 4.878689961787907e-05,
+      "loss": 0.8981,
+      "step": 7945
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6761431097984314,
+      "learning_rate": 4.8672975441925425e-05,
+      "loss": 0.9697,
+      "step": 7950
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7856532335281372,
+      "learning_rate": 4.8559141636975925e-05,
+      "loss": 0.7743,
+      "step": 7955
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7205869555473328,
+      "learning_rate": 4.844539840345666e-05,
+      "loss": 0.9266,
+      "step": 7960
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7787695527076721,
+      "learning_rate": 4.8331745941634235e-05,
+      "loss": 0.8912,
+      "step": 7965
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.800104558467865,
+      "learning_rate": 4.821818445161551e-05,
+      "loss": 0.9171,
+      "step": 7970
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6946626305580139,
+      "learning_rate": 4.810471413334711e-05,
+      "loss": 0.9681,
+      "step": 7975
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.8018020391464233,
+      "learning_rate": 4.7991335186615126e-05,
+      "loss": 0.9078,
+      "step": 7980
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.8994207978248596,
+      "learning_rate": 4.78780478110448e-05,
+      "loss": 0.9104,
+      "step": 7985
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6811192631721497,
+      "learning_rate": 4.776485220610014e-05,
+      "loss": 0.8681,
+      "step": 7990
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7817435264587402,
+      "learning_rate": 4.765174857108352e-05,
+      "loss": 0.8882,
+      "step": 7995
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.6679404377937317,
+      "learning_rate": 4.7538737105135526e-05,
+      "loss": 0.887,
+      "step": 8000
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.6360458135604858,
+      "learning_rate": 4.7425818007234324e-05,
+      "loss": 0.8876,
+      "step": 8005
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8776546716690063,
+      "learning_rate": 4.73129914761955e-05,
+      "loss": 0.7742,
+      "step": 8010
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.6663452982902527,
+      "learning_rate": 4.720025771067166e-05,
+      "loss": 0.7151,
+      "step": 8015
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.6508984565734863,
+      "learning_rate": 4.708761690915206e-05,
+      "loss": 0.9013,
+      "step": 8020
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8622150421142578,
+      "learning_rate": 4.697506926996226e-05,
+      "loss": 0.9069,
+      "step": 8025
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.759907603263855,
+      "learning_rate": 4.686261499126389e-05,
+      "loss": 0.9119,
+      "step": 8030
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.6916012763977051,
+      "learning_rate": 4.6750254271054087e-05,
+      "loss": 0.7963,
+      "step": 8035
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8521272540092468,
+      "learning_rate": 4.663798730716532e-05,
+      "loss": 0.9725,
+      "step": 8040
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7710113525390625,
+      "learning_rate": 4.6525814297264945e-05,
+      "loss": 0.9471,
+      "step": 8045
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7958170175552368,
+      "learning_rate": 4.641373543885489e-05,
+      "loss": 0.8105,
+      "step": 8050
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.823060154914856,
+      "learning_rate": 4.6301750929271404e-05,
+      "loss": 0.8821,
+      "step": 8055
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.6921396851539612,
+      "learning_rate": 4.61898609656845e-05,
+      "loss": 0.8424,
+      "step": 8060
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7996007800102234,
+      "learning_rate": 4.607806574509781e-05,
+      "loss": 0.8629,
+      "step": 8065
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7803316116333008,
+      "learning_rate": 4.596636546434807e-05,
+      "loss": 0.9515,
+      "step": 8070
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.855900228023529,
+      "learning_rate": 4.585476032010494e-05,
+      "loss": 0.9281,
+      "step": 8075
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.9303337335586548,
+      "learning_rate": 4.5743250508870475e-05,
+      "loss": 0.9563,
+      "step": 8080
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8133184313774109,
+      "learning_rate": 4.5631836226979017e-05,
+      "loss": 0.9454,
+      "step": 8085
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7453633546829224,
+      "learning_rate": 4.5520517670596607e-05,
+      "loss": 0.8819,
+      "step": 8090
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8156158924102783,
+      "learning_rate": 4.540929503572077e-05,
+      "loss": 0.874,
+      "step": 8095
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7186123132705688,
+      "learning_rate": 4.5298168518180115e-05,
+      "loss": 0.8392,
+      "step": 8100
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.6981141567230225,
+      "learning_rate": 4.518713831363408e-05,
+      "loss": 0.887,
+      "step": 8105
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.9109984040260315,
+      "learning_rate": 4.5076204617572425e-05,
+      "loss": 0.9101,
+      "step": 8110
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7632050514221191,
+      "learning_rate": 4.4965367625315146e-05,
+      "loss": 0.8918,
+      "step": 8115
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.9145953059196472,
+      "learning_rate": 4.4854627532011836e-05,
+      "loss": 0.9091,
+      "step": 8120
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.748193621635437,
+      "learning_rate": 4.474398453264154e-05,
+      "loss": 0.8552,
+      "step": 8125
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7607966661453247,
+      "learning_rate": 4.463343882201231e-05,
+      "loss": 0.9048,
+      "step": 8130
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.8794971704483032,
+      "learning_rate": 4.452299059476091e-05,
+      "loss": 0.9144,
+      "step": 8135
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.8333990573883057,
+      "learning_rate": 4.441264004535254e-05,
+      "loss": 0.9613,
+      "step": 8140
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.8173981308937073,
+      "learning_rate": 4.430238736808033e-05,
+      "loss": 0.9441,
+      "step": 8145
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.8113160729408264,
+      "learning_rate": 4.419223275706515e-05,
+      "loss": 0.8207,
+      "step": 8150
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.6924260854721069,
+      "learning_rate": 4.408217640625514e-05,
+      "loss": 0.8045,
+      "step": 8155
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.8391642570495605,
+      "learning_rate": 4.397221850942549e-05,
+      "loss": 0.8219,
+      "step": 8160
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.8425463438034058,
+      "learning_rate": 4.386235926017798e-05,
+      "loss": 0.9439,
+      "step": 8165
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.745590090751648,
+      "learning_rate": 4.3752598851940805e-05,
+      "loss": 1.0404,
+      "step": 8170
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.6746026277542114,
+      "learning_rate": 4.3642937477968105e-05,
+      "loss": 0.8602,
+      "step": 8175
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.6796015501022339,
+      "learning_rate": 4.3533375331339486e-05,
+      "loss": 0.7994,
+      "step": 8180
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.6459152102470398,
+      "learning_rate": 4.3423912604960095e-05,
+      "loss": 0.7885,
+      "step": 8185
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7240301966667175,
+      "learning_rate": 4.331454949155983e-05,
+      "loss": 0.914,
+      "step": 8190
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7516999244689941,
+      "learning_rate": 4.320528618369337e-05,
+      "loss": 0.8743,
+      "step": 8195
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7468701601028442,
+      "learning_rate": 4.309612287373957e-05,
+      "loss": 0.9048,
+      "step": 8200
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7464987635612488,
+      "learning_rate": 4.298705975390115e-05,
+      "loss": 0.8997,
+      "step": 8205
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7245876789093018,
+      "learning_rate": 4.287809701620459e-05,
+      "loss": 0.8149,
+      "step": 8210
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7378844022750854,
+      "learning_rate": 4.2769234852499505e-05,
+      "loss": 0.9746,
+      "step": 8215
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.8340911865234375,
+      "learning_rate": 4.266047345445846e-05,
+      "loss": 0.8128,
+      "step": 8220
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7118589878082275,
+      "learning_rate": 4.255181301357668e-05,
+      "loss": 0.8448,
+      "step": 8225
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7692140340805054,
+      "learning_rate": 4.244325372117156e-05,
+      "loss": 0.8732,
+      "step": 8230
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.6922775506973267,
+      "learning_rate": 4.2334795768382306e-05,
+      "loss": 0.749,
+      "step": 8235
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.8570601344108582,
+      "learning_rate": 4.2226439346169924e-05,
+      "loss": 0.8071,
+      "step": 8240
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7879267334938049,
+      "learning_rate": 4.211818464531649e-05,
+      "loss": 0.8962,
+      "step": 8245
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.8359821438789368,
+      "learning_rate": 4.2010031856425e-05,
+      "loss": 0.8961,
+      "step": 8250
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.8399404287338257,
+      "learning_rate": 4.190198116991915e-05,
+      "loss": 0.9703,
+      "step": 8255
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7860932946205139,
+      "learning_rate": 4.179403277604259e-05,
+      "loss": 0.9162,
+      "step": 8260
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7792479991912842,
+      "learning_rate": 4.168618686485916e-05,
+      "loss": 0.8027,
+      "step": 8265
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7603457570075989,
+      "learning_rate": 4.1578443626252094e-05,
+      "loss": 0.7886,
+      "step": 8270
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7225006818771362,
+      "learning_rate": 4.147080324992384e-05,
+      "loss": 0.8427,
+      "step": 8275
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.8202854990959167,
+      "learning_rate": 4.136326592539591e-05,
+      "loss": 0.716,
+      "step": 8280
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7011755108833313,
+      "learning_rate": 4.125583184200812e-05,
+      "loss": 0.8174,
+      "step": 8285
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.8102407455444336,
+      "learning_rate": 4.114850118891866e-05,
+      "loss": 0.9644,
+      "step": 8290
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7152637839317322,
+      "learning_rate": 4.104127415510365e-05,
+      "loss": 0.9355,
+      "step": 8295
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.766459584236145,
+      "learning_rate": 4.093415092935667e-05,
+      "loss": 0.8724,
+      "step": 8300
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.6977735757827759,
+      "learning_rate": 4.082713170028858e-05,
+      "loss": 0.8373,
+      "step": 8305
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.6987833380699158,
+      "learning_rate": 4.0720216656327105e-05,
+      "loss": 0.9344,
+      "step": 8310
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7315753698348999,
+      "learning_rate": 4.0613405985716554e-05,
+      "loss": 0.8926,
+      "step": 8315
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.684941828250885,
+      "learning_rate": 4.050669987651742e-05,
+      "loss": 0.9252,
+      "step": 8320
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7240857481956482,
+      "learning_rate": 4.04000985166062e-05,
+      "loss": 0.8779,
+      "step": 8325
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7166629433631897,
+      "learning_rate": 4.029360209367487e-05,
+      "loss": 0.8747,
+      "step": 8330
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7198646664619446,
+      "learning_rate": 4.0187210795230677e-05,
+      "loss": 0.8418,
+      "step": 8335
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.8439791202545166,
+      "learning_rate": 4.008092480859574e-05,
+      "loss": 0.8697,
+      "step": 8340
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7171686291694641,
+      "learning_rate": 3.997474432090679e-05,
+      "loss": 0.8472,
+      "step": 8345
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7596126198768616,
+      "learning_rate": 3.986866951911483e-05,
+      "loss": 0.8471,
+      "step": 8350
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.8424491882324219,
+      "learning_rate": 3.9762700589984744e-05,
+      "loss": 0.9175,
+      "step": 8355
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.9210354685783386,
+      "learning_rate": 3.965683772009502e-05,
+      "loss": 0.9507,
+      "step": 8360
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7487869262695312,
+      "learning_rate": 3.95510810958374e-05,
+      "loss": 0.9011,
+      "step": 8365
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7913604974746704,
+      "learning_rate": 3.944543090341656e-05,
+      "loss": 0.9093,
+      "step": 8370
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.6716275811195374,
+      "learning_rate": 3.933988732884976e-05,
+      "loss": 0.8307,
+      "step": 8375
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.8774741888046265,
+      "learning_rate": 3.923445055796664e-05,
+      "loss": 0.946,
+      "step": 8380
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7749642729759216,
+      "learning_rate": 3.912912077640869e-05,
+      "loss": 1.0164,
+      "step": 8385
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.0464688539505005,
+      "learning_rate": 3.9023898169629046e-05,
+      "loss": 0.9048,
+      "step": 8390
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.715448796749115,
+      "learning_rate": 3.891878292289216e-05,
+      "loss": 0.9282,
+      "step": 8395
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7132737040519714,
+      "learning_rate": 3.881377522127343e-05,
+      "loss": 0.7596,
+      "step": 8400
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.8384308218955994,
+      "learning_rate": 3.8708875249658905e-05,
+      "loss": 0.9182,
+      "step": 8405
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.0072834491729736,
+      "learning_rate": 3.8604083192745036e-05,
+      "loss": 0.818,
+      "step": 8410
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.8128643035888672,
+      "learning_rate": 3.849939923503815e-05,
+      "loss": 0.9118,
+      "step": 8415
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7836415767669678,
+      "learning_rate": 3.83948235608543e-05,
+      "loss": 0.9227,
+      "step": 8420
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.8463979363441467,
+      "learning_rate": 3.829035635431889e-05,
+      "loss": 0.8403,
+      "step": 8425
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7599585056304932,
+      "learning_rate": 3.818599779936629e-05,
+      "loss": 0.9183,
+      "step": 8430
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7512611150741577,
+      "learning_rate": 3.80817480797397e-05,
+      "loss": 0.8014,
+      "step": 8435
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7594391703605652,
+      "learning_rate": 3.7977607378990574e-05,
+      "loss": 0.8771,
+      "step": 8440
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.6978799104690552,
+      "learning_rate": 3.787357588047844e-05,
+      "loss": 0.7696,
+      "step": 8445
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.8497031927108765,
+      "learning_rate": 3.7769653767370586e-05,
+      "loss": 0.9166,
+      "step": 8450
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.8748128414154053,
+      "learning_rate": 3.766584122264166e-05,
+      "loss": 0.9172,
+      "step": 8455
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.797017514705658,
+      "learning_rate": 3.7562138429073424e-05,
+      "loss": 0.7935,
+      "step": 8460
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7100562453269958,
+      "learning_rate": 3.7458545569254445e-05,
+      "loss": 0.8367,
+      "step": 8465
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.6721343994140625,
+      "learning_rate": 3.735506282557967e-05,
+      "loss": 0.7493,
+      "step": 8470
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.8016105890274048,
+      "learning_rate": 3.725169038025016e-05,
+      "loss": 0.8932,
+      "step": 8475
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.787413477897644,
+      "learning_rate": 3.714842841527282e-05,
+      "loss": 0.8338,
+      "step": 8480
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7610762715339661,
+      "learning_rate": 3.7045277112459954e-05,
+      "loss": 0.9025,
+      "step": 8485
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7663581371307373,
+      "learning_rate": 3.694223665342915e-05,
+      "loss": 0.8799,
+      "step": 8490
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.8082067966461182,
+      "learning_rate": 3.683930721960276e-05,
+      "loss": 0.8652,
+      "step": 8495
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.9462847113609314,
+      "learning_rate": 3.6736488992207615e-05,
+      "loss": 0.8302,
+      "step": 8500
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7060618996620178,
+      "learning_rate": 3.663378215227483e-05,
+      "loss": 0.7578,
+      "step": 8505
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.8916406035423279,
+      "learning_rate": 3.653118688063935e-05,
+      "loss": 0.9445,
+      "step": 8510
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7780551910400391,
+      "learning_rate": 3.6428703357939644e-05,
+      "loss": 0.9411,
+      "step": 8515
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7502923607826233,
+      "learning_rate": 3.632633176461755e-05,
+      "loss": 0.8078,
+      "step": 8520
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.146773338317871,
+      "learning_rate": 3.622407228091774e-05,
+      "loss": 0.9264,
+      "step": 8525
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7078317403793335,
+      "learning_rate": 3.612192508688751e-05,
+      "loss": 0.8356,
+      "step": 8530
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7490567564964294,
+      "learning_rate": 3.601989036237644e-05,
+      "loss": 0.9093,
+      "step": 8535
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7482819557189941,
+      "learning_rate": 3.5917968287036104e-05,
+      "loss": 0.7811,
+      "step": 8540
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.8336611390113831,
+      "learning_rate": 3.5816159040319716e-05,
+      "loss": 0.9027,
+      "step": 8545
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.9120607972145081,
+      "learning_rate": 3.5714462801481895e-05,
+      "loss": 0.8714,
+      "step": 8550
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.6962396502494812,
+      "learning_rate": 3.5612879749578244e-05,
+      "loss": 0.7858,
+      "step": 8555
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7170752882957458,
+      "learning_rate": 3.551141006346499e-05,
+      "loss": 0.9845,
+      "step": 8560
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.821403443813324,
+      "learning_rate": 3.5410053921798926e-05,
+      "loss": 0.7884,
+      "step": 8565
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.825884997844696,
+      "learning_rate": 3.530881150303679e-05,
+      "loss": 0.8375,
+      "step": 8570
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7249986529350281,
+      "learning_rate": 3.5207682985435206e-05,
+      "loss": 0.9092,
+      "step": 8575
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.6975976228713989,
+      "learning_rate": 3.510666854705021e-05,
+      "loss": 0.9401,
+      "step": 8580
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7805627584457397,
+      "learning_rate": 3.5005768365736855e-05,
+      "loss": 0.8207,
+      "step": 8585
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7622156143188477,
+      "learning_rate": 3.490498261914923e-05,
+      "loss": 0.8758,
+      "step": 8590
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.8169152736663818,
+      "learning_rate": 3.48043114847398e-05,
+      "loss": 0.9381,
+      "step": 8595
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7095611691474915,
+      "learning_rate": 3.470375513975925e-05,
+      "loss": 0.8201,
+      "step": 8600
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7938575744628906,
+      "learning_rate": 3.460331376125624e-05,
+      "loss": 0.8418,
+      "step": 8605
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.9179478287696838,
+      "learning_rate": 3.450298752607696e-05,
+      "loss": 0.9845,
+      "step": 8610
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7515020966529846,
+      "learning_rate": 3.440277661086475e-05,
+      "loss": 0.8611,
+      "step": 8615
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7615182399749756,
+      "learning_rate": 3.4302681192060114e-05,
+      "loss": 0.9925,
+      "step": 8620
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.1498469114303589,
+      "learning_rate": 3.4202701445900085e-05,
+      "loss": 0.8428,
+      "step": 8625
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.6935446858406067,
+      "learning_rate": 3.410283754841801e-05,
+      "loss": 0.8424,
+      "step": 8630
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7446281313896179,
+      "learning_rate": 3.40030896754434e-05,
+      "loss": 0.9002,
+      "step": 8635
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7671748399734497,
+      "learning_rate": 3.390345800260125e-05,
+      "loss": 0.8108,
+      "step": 8640
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.9478299021720886,
+      "learning_rate": 3.380394270531221e-05,
+      "loss": 0.9014,
+      "step": 8645
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7975957989692688,
+      "learning_rate": 3.370454395879188e-05,
+      "loss": 0.7999,
+      "step": 8650
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7455906271934509,
+      "learning_rate": 3.360526193805065e-05,
+      "loss": 0.903,
+      "step": 8655
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.8040593266487122,
+      "learning_rate": 3.3506096817893526e-05,
+      "loss": 0.9848,
+      "step": 8660
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.8673847913742065,
+      "learning_rate": 3.3407048772919514e-05,
+      "loss": 0.9344,
+      "step": 8665
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.8976979851722717,
+      "learning_rate": 3.3308117977521544e-05,
+      "loss": 0.8125,
+      "step": 8670
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.6757309436798096,
+      "learning_rate": 3.32093046058862e-05,
+      "loss": 0.8485,
+      "step": 8675
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7074803709983826,
+      "learning_rate": 3.311060883199323e-05,
+      "loss": 0.9082,
+      "step": 8680
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.79212486743927,
+      "learning_rate": 3.301203082961532e-05,
+      "loss": 0.8276,
+      "step": 8685
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.8590596318244934,
+      "learning_rate": 3.291357077231781e-05,
+      "loss": 0.8158,
+      "step": 8690
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.8269075155258179,
+      "learning_rate": 3.281522883345843e-05,
+      "loss": 0.7806,
+      "step": 8695
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.8562278151512146,
+      "learning_rate": 3.271700518618683e-05,
+      "loss": 0.8634,
+      "step": 8700
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.8395061492919922,
+      "learning_rate": 3.261890000344453e-05,
+      "loss": 0.9034,
+      "step": 8705
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7919216156005859,
+      "learning_rate": 3.252091345796432e-05,
+      "loss": 0.8768,
+      "step": 8710
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.706246554851532,
+      "learning_rate": 3.2423045722270294e-05,
+      "loss": 0.8709,
+      "step": 8715
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.8488343954086304,
+      "learning_rate": 3.232529696867712e-05,
+      "loss": 0.842,
+      "step": 8720
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7069640159606934,
+      "learning_rate": 3.222766736929013e-05,
+      "loss": 0.7587,
+      "step": 8725
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7644367218017578,
+      "learning_rate": 3.2130157096004864e-05,
+      "loss": 0.8415,
+      "step": 8730
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.6687289476394653,
+      "learning_rate": 3.203276632050671e-05,
+      "loss": 0.7619,
+      "step": 8735
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.864510178565979,
+      "learning_rate": 3.1935495214270705e-05,
+      "loss": 0.9298,
+      "step": 8740
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.8863157629966736,
+      "learning_rate": 3.1838343948561136e-05,
+      "loss": 0.851,
+      "step": 8745
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7367367744445801,
+      "learning_rate": 3.1741312694431315e-05,
+      "loss": 0.8596,
+      "step": 8750
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.6925660371780396,
+      "learning_rate": 3.164440162272322e-05,
+      "loss": 0.9713,
+      "step": 8755
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7945080399513245,
+      "learning_rate": 3.1547610904067325e-05,
+      "loss": 0.7486,
+      "step": 8760
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.9216223955154419,
+      "learning_rate": 3.145094070888208e-05,
+      "loss": 0.9921,
+      "step": 8765
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.9635785818099976,
+      "learning_rate": 3.13543912073738e-05,
+      "loss": 0.9232,
+      "step": 8770
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.8723745942115784,
+      "learning_rate": 3.125796256953625e-05,
+      "loss": 1.0093,
+      "step": 8775
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.713442325592041,
+      "learning_rate": 3.1161654965150436e-05,
+      "loss": 0.7681,
+      "step": 8780
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.8069811463356018,
+      "learning_rate": 3.1065468563784196e-05,
+      "loss": 0.9341,
+      "step": 8785
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.8023787140846252,
+      "learning_rate": 3.096940353479208e-05,
+      "loss": 0.8578,
+      "step": 8790
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.8353567719459534,
+      "learning_rate": 3.087346004731485e-05,
+      "loss": 0.9737,
+      "step": 8795
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7994593977928162,
+      "learning_rate": 3.077763827027929e-05,
+      "loss": 0.845,
+      "step": 8800
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.6210119128227234,
+      "learning_rate": 3.0681938372397865e-05,
+      "loss": 0.9005,
+      "step": 8805
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7582188844680786,
+      "learning_rate": 3.0586360522168476e-05,
+      "loss": 1.0126,
+      "step": 8810
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.8581645488739014,
+      "learning_rate": 3.0490904887874183e-05,
+      "loss": 0.929,
+      "step": 8815
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.752768874168396,
+      "learning_rate": 3.039557163758279e-05,
+      "loss": 0.9874,
+      "step": 8820
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.9083414673805237,
+      "learning_rate": 3.030036093914663e-05,
+      "loss": 0.923,
+      "step": 8825
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.6985150575637817,
+      "learning_rate": 3.0205272960202292e-05,
+      "loss": 0.8371,
+      "step": 8830
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8764124512672424,
+      "learning_rate": 3.0110307868170263e-05,
+      "loss": 1.0147,
+      "step": 8835
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8891944885253906,
+      "learning_rate": 3.0015465830254663e-05,
+      "loss": 0.9144,
+      "step": 8840
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7759283185005188,
+      "learning_rate": 2.9920747013443007e-05,
+      "loss": 0.926,
+      "step": 8845
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8171085715293884,
+      "learning_rate": 2.98261515845058e-05,
+      "loss": 0.8523,
+      "step": 8850
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8419032692909241,
+      "learning_rate": 2.9731679709996306e-05,
+      "loss": 0.8821,
+      "step": 8855
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8328216075897217,
+      "learning_rate": 2.963733155625026e-05,
+      "loss": 0.7794,
+      "step": 8860
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8300039768218994,
+      "learning_rate": 2.954310728938553e-05,
+      "loss": 0.7377,
+      "step": 8865
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8087813854217529,
+      "learning_rate": 2.944900707530195e-05,
+      "loss": 0.739,
+      "step": 8870
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8231684565544128,
+      "learning_rate": 2.9355031079680827e-05,
+      "loss": 0.9466,
+      "step": 8875
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7937616109848022,
+      "learning_rate": 2.9261179467984822e-05,
+      "loss": 0.9411,
+      "step": 8880
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8774697780609131,
+      "learning_rate": 2.9167452405457562e-05,
+      "loss": 0.8134,
+      "step": 8885
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7183670401573181,
+      "learning_rate": 2.907385005712341e-05,
+      "loss": 0.8746,
+      "step": 8890
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.921839714050293,
+      "learning_rate": 2.8980372587787087e-05,
+      "loss": 0.8537,
+      "step": 8895
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.6738794445991516,
+      "learning_rate": 2.888702016203354e-05,
+      "loss": 0.9334,
+      "step": 8900
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7114542722702026,
+      "learning_rate": 2.879379294422748e-05,
+      "loss": 0.7629,
+      "step": 8905
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.9150660634040833,
+      "learning_rate": 2.8700691098513188e-05,
+      "loss": 0.8519,
+      "step": 8910
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7252589464187622,
+      "learning_rate": 2.8607714788814176e-05,
+      "loss": 0.8837,
+      "step": 8915
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7726362943649292,
+      "learning_rate": 2.8514864178832967e-05,
+      "loss": 0.8531,
+      "step": 8920
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7412396669387817,
+      "learning_rate": 2.842213943205072e-05,
+      "loss": 0.8902,
+      "step": 8925
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7849807739257812,
+      "learning_rate": 2.8329540711727054e-05,
+      "loss": 0.827,
+      "step": 8930
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.9399614930152893,
+      "learning_rate": 2.823706818089965e-05,
+      "loss": 0.8699,
+      "step": 8935
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7663363814353943,
+      "learning_rate": 2.8144722002383993e-05,
+      "loss": 0.8914,
+      "step": 8940
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7527604699134827,
+      "learning_rate": 2.8052502338773146e-05,
+      "loss": 0.81,
+      "step": 8945
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.8573604226112366,
+      "learning_rate": 2.7960409352437333e-05,
+      "loss": 0.9121,
+      "step": 8950
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.8559805750846863,
+      "learning_rate": 2.7868443205523888e-05,
+      "loss": 0.8892,
+      "step": 8955
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.8466255068778992,
+      "learning_rate": 2.777660405995671e-05,
+      "loss": 0.7399,
+      "step": 8960
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6989338397979736,
+      "learning_rate": 2.768489207743603e-05,
+      "loss": 0.8253,
+      "step": 8965
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7643797993659973,
+      "learning_rate": 2.7593307419438354e-05,
+      "loss": 0.8435,
+      "step": 8970
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.8357900381088257,
+      "learning_rate": 2.7501850247215878e-05,
+      "loss": 0.856,
+      "step": 8975
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.8707523345947266,
+      "learning_rate": 2.741052072179636e-05,
+      "loss": 0.8189,
+      "step": 8980
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6612287759780884,
+      "learning_rate": 2.7319319003982925e-05,
+      "loss": 0.9007,
+      "step": 8985
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.8610745668411255,
+      "learning_rate": 2.7228245254353444e-05,
+      "loss": 0.8348,
+      "step": 8990
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7897176742553711,
+      "learning_rate": 2.7137299633260638e-05,
+      "loss": 0.8131,
+      "step": 8995
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.8576532602310181,
+      "learning_rate": 2.7046482300831642e-05,
+      "loss": 1.0017,
+      "step": 9000
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7004991769790649,
+      "learning_rate": 2.6955793416967646e-05,
+      "loss": 0.7315,
+      "step": 9005
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.9065749645233154,
+      "learning_rate": 2.686523314134367e-05,
+      "loss": 0.85,
+      "step": 9010
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.8746364116668701,
+      "learning_rate": 2.6774801633408418e-05,
+      "loss": 0.8827,
+      "step": 9015
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5582964420318604,
+      "learning_rate": 2.668449905238367e-05,
+      "loss": 0.8187,
+      "step": 9020
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7435264587402344,
+      "learning_rate": 2.659432555726441e-05,
+      "loss": 0.8926,
+      "step": 9025
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7577239871025085,
+      "learning_rate": 2.6504281306818225e-05,
+      "loss": 0.7974,
+      "step": 9030
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.8755051493644714,
+      "learning_rate": 2.641436645958515e-05,
+      "loss": 0.8204,
+      "step": 9035
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7866493463516235,
+      "learning_rate": 2.6324581173877473e-05,
+      "loss": 0.9189,
+      "step": 9040
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6970306634902954,
+      "learning_rate": 2.6234925607779215e-05,
+      "loss": 0.8448,
+      "step": 9045
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.775654673576355,
+      "learning_rate": 2.6145399919146086e-05,
+      "loss": 0.8325,
+      "step": 9050
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.796474277973175,
+      "learning_rate": 2.6056004265605148e-05,
+      "loss": 0.9613,
+      "step": 9055
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7175946235656738,
+      "learning_rate": 2.596673880455448e-05,
+      "loss": 0.8525,
+      "step": 9060
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.8016606569290161,
+      "learning_rate": 2.587760369316291e-05,
+      "loss": 0.8111,
+      "step": 9065
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.8967776894569397,
+      "learning_rate": 2.578859908836979e-05,
+      "loss": 0.8669,
+      "step": 9070
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.8003263473510742,
+      "learning_rate": 2.569972514688468e-05,
+      "loss": 0.8308,
+      "step": 9075
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7419909834861755,
+      "learning_rate": 2.5610982025187046e-05,
+      "loss": 0.8653,
+      "step": 9080
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7726119756698608,
+      "learning_rate": 2.552236987952612e-05,
+      "loss": 0.8031,
+      "step": 9085
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7673059701919556,
+      "learning_rate": 2.543388886592045e-05,
+      "loss": 1.0341,
+      "step": 9090
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7712087035179138,
+      "learning_rate": 2.5345539140157705e-05,
+      "loss": 0.811,
+      "step": 9095
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7213789224624634,
+      "learning_rate": 2.5257320857794397e-05,
+      "loss": 0.7562,
+      "step": 9100
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7315651178359985,
+      "learning_rate": 2.5169234174155608e-05,
+      "loss": 0.7916,
+      "step": 9105
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.807244598865509,
+      "learning_rate": 2.5081279244334764e-05,
+      "loss": 0.865,
+      "step": 9110
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7596588134765625,
+      "learning_rate": 2.4993456223193266e-05,
+      "loss": 0.907,
+      "step": 9115
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.9872735142707825,
+      "learning_rate": 2.490576526536025e-05,
+      "loss": 0.9589,
+      "step": 9120
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.8341203927993774,
+      "learning_rate": 2.4818206525232356e-05,
+      "loss": 0.873,
+      "step": 9125
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.8057723045349121,
+      "learning_rate": 2.4730780156973442e-05,
+      "loss": 0.9904,
+      "step": 9130
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.8961916565895081,
+      "learning_rate": 2.464348631451424e-05,
+      "loss": 0.8134,
+      "step": 9135
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.030664086341858,
+      "learning_rate": 2.455632515155224e-05,
+      "loss": 0.8357,
+      "step": 9140
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.6567131280899048,
+      "learning_rate": 2.4469296821551257e-05,
+      "loss": 0.7706,
+      "step": 9145
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.74416583776474,
+      "learning_rate": 2.4382401477741244e-05,
+      "loss": 0.9082,
+      "step": 9150
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.9838562607765198,
+      "learning_rate": 2.429563927311801e-05,
+      "loss": 1.0202,
+      "step": 9155
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7462790608406067,
+      "learning_rate": 2.4209010360442896e-05,
+      "loss": 0.9368,
+      "step": 9160
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7337406277656555,
+      "learning_rate": 2.4122514892242677e-05,
+      "loss": 0.8163,
+      "step": 9165
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.6417105197906494,
+      "learning_rate": 2.4036153020809072e-05,
+      "loss": 0.833,
+      "step": 9170
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.9543890357017517,
+      "learning_rate": 2.3949924898198604e-05,
+      "loss": 0.8944,
+      "step": 9175
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.763455867767334,
+      "learning_rate": 2.3863830676232313e-05,
+      "loss": 0.8355,
+      "step": 9180
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7808805108070374,
+      "learning_rate": 2.377787050649547e-05,
+      "loss": 0.9126,
+      "step": 9185
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7466232180595398,
+      "learning_rate": 2.36920445403373e-05,
+      "loss": 0.9283,
+      "step": 9190
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7460727095603943,
+      "learning_rate": 2.3606352928870835e-05,
+      "loss": 0.8184,
+      "step": 9195
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.8474850654602051,
+      "learning_rate": 2.352079582297244e-05,
+      "loss": 0.941,
+      "step": 9200
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.8749337792396545,
+      "learning_rate": 2.34353733732817e-05,
+      "loss": 0.873,
+      "step": 9205
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.8273049592971802,
+      "learning_rate": 2.335008573020111e-05,
+      "loss": 0.9853,
+      "step": 9210
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7353382706642151,
+      "learning_rate": 2.326493304389582e-05,
+      "loss": 0.9153,
+      "step": 9215
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.8182071447372437,
+      "learning_rate": 2.3179915464293323e-05,
+      "loss": 0.8271,
+      "step": 9220
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.952183187007904,
+      "learning_rate": 2.309503314108331e-05,
+      "loss": 0.8598,
+      "step": 9225
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.8443077206611633,
+      "learning_rate": 2.301028622371726e-05,
+      "loss": 0.8799,
+      "step": 9230
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7569529414176941,
+      "learning_rate": 2.2925674861408264e-05,
+      "loss": 0.8918,
+      "step": 9235
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.8164322376251221,
+      "learning_rate": 2.2841199203130747e-05,
+      "loss": 0.9618,
+      "step": 9240
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.883344829082489,
+      "learning_rate": 2.2756859397620156e-05,
+      "loss": 0.9899,
+      "step": 9245
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.713097095489502,
+      "learning_rate": 2.267265559337286e-05,
+      "loss": 0.8803,
+      "step": 9250
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6419798135757446,
+      "learning_rate": 2.2588587938645656e-05,
+      "loss": 0.8278,
+      "step": 9255
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.75221186876297,
+      "learning_rate": 2.2504656581455665e-05,
+      "loss": 0.9868,
+      "step": 9260
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.8278515934944153,
+      "learning_rate": 2.242086166958004e-05,
+      "loss": 0.8268,
+      "step": 9265
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.9154884219169617,
+      "learning_rate": 2.233720335055567e-05,
+      "loss": 0.8536,
+      "step": 9270
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7821744680404663,
+      "learning_rate": 2.2253681771678946e-05,
+      "loss": 0.9334,
+      "step": 9275
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7095476984977722,
+      "learning_rate": 2.2170297080005564e-05,
+      "loss": 0.8938,
+      "step": 9280
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7716023921966553,
+      "learning_rate": 2.208704942235017e-05,
+      "loss": 0.7645,
+      "step": 9285
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.8259437680244446,
+      "learning_rate": 2.200393894528603e-05,
+      "loss": 0.7961,
+      "step": 9290
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7871156334877014,
+      "learning_rate": 2.1920965795145054e-05,
+      "loss": 0.9552,
+      "step": 9295
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7442185878753662,
+      "learning_rate": 2.1838130118017252e-05,
+      "loss": 0.871,
+      "step": 9300
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7882496118545532,
+      "learning_rate": 2.175543205975059e-05,
+      "loss": 0.883,
+      "step": 9305
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7419693470001221,
+      "learning_rate": 2.1672871765950808e-05,
+      "loss": 0.8593,
+      "step": 9310
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.9180772304534912,
+      "learning_rate": 2.1590449381980993e-05,
+      "loss": 0.9116,
+      "step": 9315
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7324300408363342,
+      "learning_rate": 2.150816505296147e-05,
+      "loss": 0.8584,
+      "step": 9320
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.8154258728027344,
+      "learning_rate": 2.1426018923769464e-05,
+      "loss": 0.869,
+      "step": 9325
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7851940989494324,
+      "learning_rate": 2.1344011139038843e-05,
+      "loss": 0.8994,
+      "step": 9330
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.8780238628387451,
+      "learning_rate": 2.126214184316002e-05,
+      "loss": 0.8458,
+      "step": 9335
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.75881427526474,
+      "learning_rate": 2.1180411180279458e-05,
+      "loss": 0.9026,
+      "step": 9340
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7425323128700256,
+      "learning_rate": 2.1098819294299498e-05,
+      "loss": 0.9413,
+      "step": 9345
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7396215796470642,
+      "learning_rate": 2.101736632887825e-05,
+      "loss": 0.7693,
+      "step": 9350
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7948014140129089,
+      "learning_rate": 2.0936052427429186e-05,
+      "loss": 0.8503,
+      "step": 9355
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7535755634307861,
+      "learning_rate": 2.085487773312086e-05,
+      "loss": 0.9074,
+      "step": 9360
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7665767073631287,
+      "learning_rate": 2.0773842388876884e-05,
+      "loss": 0.8667,
+      "step": 9365
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7539588809013367,
+      "learning_rate": 2.0692946537375336e-05,
+      "loss": 0.8238,
+      "step": 9370
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.6851677894592285,
+      "learning_rate": 2.0612190321048762e-05,
+      "loss": 0.8384,
+      "step": 9375
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.9282670617103577,
+      "learning_rate": 2.053157388208393e-05,
+      "loss": 0.8211,
+      "step": 9380
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.897520899772644,
+      "learning_rate": 2.0451097362421366e-05,
+      "loss": 0.9642,
+      "step": 9385
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.8395950794219971,
+      "learning_rate": 2.037076090375539e-05,
+      "loss": 0.8843,
+      "step": 9390
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.8132792115211487,
+      "learning_rate": 2.029056464753363e-05,
+      "loss": 0.9268,
+      "step": 9395
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.6976551413536072,
+      "learning_rate": 2.021050873495679e-05,
+      "loss": 0.7874,
+      "step": 9400
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.8054299354553223,
+      "learning_rate": 2.013059330697864e-05,
+      "loss": 0.9546,
+      "step": 9405
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.8405972719192505,
+      "learning_rate": 2.005081850430548e-05,
+      "loss": 0.8365,
+      "step": 9410
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.6910131573677063,
+      "learning_rate": 1.9971184467396022e-05,
+      "loss": 0.9169,
+      "step": 9415
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7588268518447876,
+      "learning_rate": 1.989169133646124e-05,
+      "loss": 0.8404,
+      "step": 9420
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7150195240974426,
+      "learning_rate": 1.981233925146385e-05,
+      "loss": 0.8692,
+      "step": 9425
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.8912619948387146,
+      "learning_rate": 1.9733128352118324e-05,
+      "loss": 0.9376,
+      "step": 9430
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.8724376559257507,
+      "learning_rate": 1.9654058777890573e-05,
+      "loss": 0.9965,
+      "step": 9435
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7318885326385498,
+      "learning_rate": 1.9575130667997643e-05,
+      "loss": 0.9009,
+      "step": 9440
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.8611711859703064,
+      "learning_rate": 1.9496344161407487e-05,
+      "loss": 1.0043,
+      "step": 9445
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.753633439540863,
+      "learning_rate": 1.9417699396838764e-05,
+      "loss": 0.8654,
+      "step": 9450
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.8039273619651794,
+      "learning_rate": 1.9339196512760538e-05,
+      "loss": 0.8386,
+      "step": 9455
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.760884165763855,
+      "learning_rate": 1.926083564739215e-05,
+      "loss": 0.8392,
+      "step": 9460
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.805150032043457,
+      "learning_rate": 1.9182616938702792e-05,
+      "loss": 0.8525,
+      "step": 9465
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7307751774787903,
+      "learning_rate": 1.910454052441141e-05,
+      "loss": 0.8812,
+      "step": 9470
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.6862212419509888,
+      "learning_rate": 1.9026606541986393e-05,
+      "loss": 0.8585,
+      "step": 9475
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7243665456771851,
+      "learning_rate": 1.894881512864537e-05,
+      "loss": 0.8165,
+      "step": 9480
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.6453014612197876,
+      "learning_rate": 1.8871166421354924e-05,
+      "loss": 0.8449,
+      "step": 9485
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7892723679542542,
+      "learning_rate": 1.879366055683044e-05,
+      "loss": 0.9335,
+      "step": 9490
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.70809006690979,
+      "learning_rate": 1.871629767153573e-05,
+      "loss": 0.9325,
+      "step": 9495
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.816694438457489,
+      "learning_rate": 1.863907790168289e-05,
+      "loss": 0.7737,
+      "step": 9500
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.8121705055236816,
+      "learning_rate": 1.8562001383232043e-05,
+      "loss": 0.7726,
+      "step": 9505
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.8642139434814453,
+      "learning_rate": 1.848506825189107e-05,
+      "loss": 0.8563,
+      "step": 9510
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7992377877235413,
+      "learning_rate": 1.8408278643115384e-05,
+      "loss": 0.9831,
+      "step": 9515
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.8327411413192749,
+      "learning_rate": 1.833163269210777e-05,
+      "loss": 0.8809,
+      "step": 9520
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.6830605268478394,
+      "learning_rate": 1.825513053381801e-05,
+      "loss": 0.7633,
+      "step": 9525
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.8190145492553711,
+      "learning_rate": 1.8178772302942705e-05,
+      "loss": 0.9256,
+      "step": 9530
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.857457160949707,
+      "learning_rate": 1.8102558133925084e-05,
+      "loss": 0.8816,
+      "step": 9535
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7766903638839722,
+      "learning_rate": 1.802648816095468e-05,
+      "loss": 0.8836,
+      "step": 9540
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7608227133750916,
+      "learning_rate": 1.7950562517967217e-05,
+      "loss": 0.9066,
+      "step": 9545
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.67827969789505,
+      "learning_rate": 1.787478133864423e-05,
+      "loss": 0.7088,
+      "step": 9550
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.8599781394004822,
+      "learning_rate": 1.779914475641292e-05,
+      "loss": 0.9545,
+      "step": 9555
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7540460824966431,
+      "learning_rate": 1.7723652904445907e-05,
+      "loss": 0.8754,
+      "step": 9560
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7129548192024231,
+      "learning_rate": 1.7648305915660968e-05,
+      "loss": 0.8777,
+      "step": 9565
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7389048337936401,
+      "learning_rate": 1.75731039227208e-05,
+      "loss": 0.802,
+      "step": 9570
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7623948454856873,
+      "learning_rate": 1.7498047058032896e-05,
+      "loss": 0.9349,
+      "step": 9575
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.8034424781799316,
+      "learning_rate": 1.742313545374914e-05,
+      "loss": 0.936,
+      "step": 9580
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.8239856958389282,
+      "learning_rate": 1.7348369241765683e-05,
+      "loss": 0.7999,
+      "step": 9585
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.8029961585998535,
+      "learning_rate": 1.7273748553722668e-05,
+      "loss": 0.8171,
+      "step": 9590
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7298925518989563,
+      "learning_rate": 1.7199273521004046e-05,
+      "loss": 0.8307,
+      "step": 9595
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7932624220848083,
+      "learning_rate": 1.7124944274737274e-05,
+      "loss": 1.0165,
+      "step": 9600
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.8447069525718689,
+      "learning_rate": 1.7050760945793187e-05,
+      "loss": 0.963,
+      "step": 9605
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7499577403068542,
+      "learning_rate": 1.6976723664785653e-05,
+      "loss": 0.7964,
+      "step": 9610
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7524833679199219,
+      "learning_rate": 1.6902832562071404e-05,
+      "loss": 0.7665,
+      "step": 9615
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.8401513695716858,
+      "learning_rate": 1.682908776774981e-05,
+      "loss": 0.8132,
+      "step": 9620
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.8104890584945679,
+      "learning_rate": 1.6755489411662595e-05,
+      "loss": 0.9056,
+      "step": 9625
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7588112354278564,
+      "learning_rate": 1.668203762339373e-05,
+      "loss": 0.7794,
+      "step": 9630
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.8475006222724915,
+      "learning_rate": 1.6608732532269077e-05,
+      "loss": 0.7724,
+      "step": 9635
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.8192122578620911,
+      "learning_rate": 1.6535574267356192e-05,
+      "loss": 0.9124,
+      "step": 9640
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7818514704704285,
+      "learning_rate": 1.6462562957464132e-05,
+      "loss": 0.8726,
+      "step": 9645
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8724757432937622,
+      "learning_rate": 1.6389698731143242e-05,
+      "loss": 0.8193,
+      "step": 9650
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8011074066162109,
+      "learning_rate": 1.631698171668483e-05,
+      "loss": 0.8256,
+      "step": 9655
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.0343480110168457,
+      "learning_rate": 1.6244412042121105e-05,
+      "loss": 0.9657,
+      "step": 9660
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8504678010940552,
+      "learning_rate": 1.61719898352248e-05,
+      "loss": 0.9206,
+      "step": 9665
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8102021813392639,
+      "learning_rate": 1.6099715223508937e-05,
+      "loss": 0.9319,
+      "step": 9670
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8873084783554077,
+      "learning_rate": 1.6027588334226807e-05,
+      "loss": 0.927,
+      "step": 9675
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7379308938980103,
+      "learning_rate": 1.59556092943715e-05,
+      "loss": 0.7433,
+      "step": 9680
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8241307139396667,
+      "learning_rate": 1.5883778230675862e-05,
+      "loss": 0.8641,
+      "step": 9685
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7821060419082642,
+      "learning_rate": 1.5812095269612136e-05,
+      "loss": 0.9231,
+      "step": 9690
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7354114651679993,
+      "learning_rate": 1.5740560537391858e-05,
+      "loss": 0.8959,
+      "step": 9695
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8831083178520203,
+      "learning_rate": 1.5669174159965517e-05,
+      "loss": 1.0867,
+      "step": 9700
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7737128138542175,
+      "learning_rate": 1.559793626302245e-05,
+      "loss": 0.7872,
+      "step": 9705
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.9330393075942993,
+      "learning_rate": 1.5526846971990505e-05,
+      "loss": 0.7577,
+      "step": 9710
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8273800611495972,
+      "learning_rate": 1.545590641203599e-05,
+      "loss": 0.867,
+      "step": 9715
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8835793137550354,
+      "learning_rate": 1.5385114708063265e-05,
+      "loss": 0.8406,
+      "step": 9720
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7108529806137085,
+      "learning_rate": 1.531447198471453e-05,
+      "loss": 0.8639,
+      "step": 9725
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8037097454071045,
+      "learning_rate": 1.5243978366369837e-05,
+      "loss": 0.8974,
+      "step": 9730
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8605689406394958,
+      "learning_rate": 1.5173633977146595e-05,
+      "loss": 0.8906,
+      "step": 9735
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7327417731285095,
+      "learning_rate": 1.5103438940899494e-05,
+      "loss": 0.8373,
+      "step": 9740
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8748462200164795,
+      "learning_rate": 1.5033393381220329e-05,
+      "loss": 0.8786,
+      "step": 9745
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8213547468185425,
+      "learning_rate": 1.4963497421437577e-05,
+      "loss": 1.0147,
+      "step": 9750
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8919892311096191,
+      "learning_rate": 1.48937511846164e-05,
+      "loss": 0.7929,
+      "step": 9755
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8843607306480408,
+      "learning_rate": 1.4824154793558375e-05,
+      "loss": 0.8334,
+      "step": 9760
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8356348872184753,
+      "learning_rate": 1.4754708370801151e-05,
+      "loss": 0.9379,
+      "step": 9765
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7609606385231018,
+      "learning_rate": 1.4685412038618473e-05,
+      "loss": 0.7203,
+      "step": 9770
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8738039135932922,
+      "learning_rate": 1.4616265919019645e-05,
+      "loss": 0.793,
+      "step": 9775
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7854296565055847,
+      "learning_rate": 1.454727013374959e-05,
+      "loss": 0.8969,
+      "step": 9780
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8933621048927307,
+      "learning_rate": 1.4478424804288582e-05,
+      "loss": 0.9128,
+      "step": 9785
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.9402585625648499,
+      "learning_rate": 1.440973005185191e-05,
+      "loss": 0.899,
+      "step": 9790
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7768909931182861,
+      "learning_rate": 1.434118599738975e-05,
+      "loss": 0.8688,
+      "step": 9795
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8614407777786255,
+      "learning_rate": 1.427279276158704e-05,
+      "loss": 1.0387,
+      "step": 9800
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7587859034538269,
+      "learning_rate": 1.4204550464863021e-05,
+      "loss": 0.9036,
+      "step": 9805
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8442284464836121,
+      "learning_rate": 1.4136459227371269e-05,
+      "loss": 0.9267,
+      "step": 9810
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8466005921363831,
+      "learning_rate": 1.4068519168999405e-05,
+      "loss": 0.8621,
+      "step": 9815
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8399500250816345,
+      "learning_rate": 1.4000730409368845e-05,
+      "loss": 0.8269,
+      "step": 9820
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.0434014797210693,
+      "learning_rate": 1.3933093067834601e-05,
+      "loss": 0.9427,
+      "step": 9825
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8561471104621887,
+      "learning_rate": 1.3865607263485091e-05,
+      "loss": 0.8998,
+      "step": 9830
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.097008228302002,
+      "learning_rate": 1.3798273115141912e-05,
+      "loss": 0.8711,
+      "step": 9835
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7780779600143433,
+      "learning_rate": 1.373109074135972e-05,
+      "loss": 0.838,
+      "step": 9840
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.9400659799575806,
+      "learning_rate": 1.3664060260425827e-05,
+      "loss": 0.8199,
+      "step": 9845
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8247981071472168,
+      "learning_rate": 1.359718179036019e-05,
+      "loss": 0.8271,
+      "step": 9850
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7556101679801941,
+      "learning_rate": 1.353045544891508e-05,
+      "loss": 0.9536,
+      "step": 9855
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7069379687309265,
+      "learning_rate": 1.3463881353574947e-05,
+      "loss": 0.7715,
+      "step": 9860
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7760348916053772,
+      "learning_rate": 1.339745962155613e-05,
+      "loss": 0.7853,
+      "step": 9865
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8063364028930664,
+      "learning_rate": 1.33311903698068e-05,
+      "loss": 0.7978,
+      "step": 9870
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7314766049385071,
+      "learning_rate": 1.326507371500656e-05,
+      "loss": 0.9842,
+      "step": 9875
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7383533120155334,
+      "learning_rate": 1.3199109773566387e-05,
+      "loss": 0.9943,
+      "step": 9880
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7715831995010376,
+      "learning_rate": 1.3133298661628368e-05,
+      "loss": 0.8209,
+      "step": 9885
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7701368927955627,
+      "learning_rate": 1.3067640495065492e-05,
+      "loss": 0.8278,
+      "step": 9890
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.9878485798835754,
+      "learning_rate": 1.3002135389481451e-05,
+      "loss": 0.9727,
+      "step": 9895
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.8185439109802246,
+      "learning_rate": 1.29367834602105e-05,
+      "loss": 0.8967,
+      "step": 9900
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7079369425773621,
+      "learning_rate": 1.2871584822317151e-05,
+      "loss": 0.8743,
+      "step": 9905
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.9138286113739014,
+      "learning_rate": 1.2806539590596023e-05,
+      "loss": 0.8651,
+      "step": 9910
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7398014068603516,
+      "learning_rate": 1.2741647879571627e-05,
+      "loss": 0.9221,
+      "step": 9915
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7842152118682861,
+      "learning_rate": 1.2676909803498161e-05,
+      "loss": 0.9457,
+      "step": 9920
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.0707106590270996,
+      "learning_rate": 1.2612325476359388e-05,
+      "loss": 1.025,
+      "step": 9925
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.802452564239502,
+      "learning_rate": 1.2547895011868304e-05,
+      "loss": 0.7072,
+      "step": 9930
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7343943119049072,
+      "learning_rate": 1.2483618523467e-05,
+      "loss": 0.8953,
+      "step": 9935
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7321301698684692,
+      "learning_rate": 1.241949612432649e-05,
+      "loss": 0.8973,
+      "step": 9940
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7284245491027832,
+      "learning_rate": 1.2355527927346478e-05,
+      "loss": 0.8672,
+      "step": 9945
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.00664222240448,
+      "learning_rate": 1.229171404515511e-05,
+      "loss": 0.9601,
+      "step": 9950
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7685894966125488,
+      "learning_rate": 1.2228054590108962e-05,
+      "loss": 0.9258,
+      "step": 9955
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.870555579662323,
+      "learning_rate": 1.2164549674292581e-05,
+      "loss": 0.9566,
+      "step": 9960
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.8318189382553101,
+      "learning_rate": 1.2101199409518483e-05,
+      "loss": 0.8342,
+      "step": 9965
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.9991118907928467,
+      "learning_rate": 1.2038003907326867e-05,
+      "loss": 0.8669,
+      "step": 9970
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.6647642254829407,
+      "learning_rate": 1.1974963278985463e-05,
+      "loss": 0.8771,
+      "step": 9975
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.8214685916900635,
+      "learning_rate": 1.1912077635489282e-05,
+      "loss": 0.9524,
+      "step": 9980
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.8654179573059082,
+      "learning_rate": 1.1849347087560525e-05,
+      "loss": 0.92,
+      "step": 9985
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.765496551990509,
+      "learning_rate": 1.1786771745648229e-05,
+      "loss": 0.8282,
+      "step": 9990
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.944416880607605,
+      "learning_rate": 1.1724351719928228e-05,
+      "loss": 0.8714,
+      "step": 9995
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7935224771499634,
+      "learning_rate": 1.1662087120302867e-05,
+      "loss": 0.8278,
+      "step": 10000
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.806790292263031,
+      "learning_rate": 1.1599978056400796e-05,
+      "loss": 0.888,
+      "step": 10005
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.8951229453086853,
+      "learning_rate": 1.1538024637576905e-05,
+      "loss": 0.914,
+      "step": 10010
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.7594240307807922,
+      "learning_rate": 1.1476226972911974e-05,
+      "loss": 0.889,
+      "step": 10015
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.8180289268493652,
+      "learning_rate": 1.1414585171212555e-05,
+      "loss": 0.8932,
+      "step": 10020
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.7589259743690491,
+      "learning_rate": 1.1353099341010786e-05,
+      "loss": 0.8497,
+      "step": 10025
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.6496753096580505,
+      "learning_rate": 1.1291769590564182e-05,
+      "loss": 0.8868,
+      "step": 10030
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.7347110509872437,
+      "learning_rate": 1.1230596027855434e-05,
+      "loss": 0.8546,
+      "step": 10035
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.9040740132331848,
+      "learning_rate": 1.1169578760592292e-05,
+      "loss": 0.9046,
+      "step": 10040
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.1045706272125244,
+      "learning_rate": 1.1108717896207276e-05,
+      "loss": 0.958,
+      "step": 10045
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.7006523013114929,
+      "learning_rate": 1.1048013541857472e-05,
+      "loss": 0.8458,
+      "step": 10050
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.8821123838424683,
+      "learning_rate": 1.0987465804424512e-05,
+      "loss": 0.8466,
+      "step": 10055
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.8056004047393799,
+      "learning_rate": 1.0927074790514203e-05,
+      "loss": 0.8797,
+      "step": 10060
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.7699093222618103,
+      "learning_rate": 1.0866840606456452e-05,
+      "loss": 1.0301,
+      "step": 10065
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.7734396457672119,
+      "learning_rate": 1.0806763358305005e-05,
+      "loss": 0.7968,
+      "step": 10070
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.7000718116760254,
+      "learning_rate": 1.074684315183727e-05,
+      "loss": 0.8832,
+      "step": 10075
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.6902223825454712,
+      "learning_rate": 1.0687080092554225e-05,
+      "loss": 0.8448,
+      "step": 10080
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.0849089622497559,
+      "learning_rate": 1.0627474285680105e-05,
+      "loss": 0.8238,
+      "step": 10085
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.7399893999099731,
+      "learning_rate": 1.0568025836162265e-05,
+      "loss": 0.8334,
+      "step": 10090
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.6186020374298096,
+      "learning_rate": 1.0508734848671064e-05,
+      "loss": 0.7938,
+      "step": 10095
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.6724085807800293,
+      "learning_rate": 1.0449601427599588e-05,
+      "loss": 0.8982,
+      "step": 10100
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.7217341661453247,
+      "learning_rate": 1.0390625677063415e-05,
+      "loss": 0.8907,
+      "step": 10105
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.8844561576843262,
+      "learning_rate": 1.0331807700900664e-05,
+      "loss": 0.9729,
+      "step": 10110
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.9020631909370422,
+      "learning_rate": 1.0273147602671562e-05,
+      "loss": 0.9127,
+      "step": 10115
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.6652135252952576,
+      "learning_rate": 1.0214645485658358e-05,
+      "loss": 0.7979,
+      "step": 10120
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.8186726570129395,
+      "learning_rate": 1.0156301452865246e-05,
+      "loss": 0.795,
+      "step": 10125
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.9490936994552612,
+      "learning_rate": 1.0098115607017922e-05,
+      "loss": 0.7368,
+      "step": 10130
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.8582313060760498,
+      "learning_rate": 1.00400880505637e-05,
+      "loss": 0.7767,
+      "step": 10135
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7754263877868652,
+      "learning_rate": 9.982218885671158e-06,
+      "loss": 0.8886,
+      "step": 10140
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7403327822685242,
+      "learning_rate": 9.924508214229933e-06,
+      "loss": 0.8946,
+      "step": 10145
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.787521481513977,
+      "learning_rate": 9.866956137850736e-06,
+      "loss": 0.7958,
+      "step": 10150
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.8559316992759705,
+      "learning_rate": 9.809562757864887e-06,
+      "loss": 0.9273,
+      "step": 10155
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.6812841296195984,
+      "learning_rate": 9.752328175324366e-06,
+      "loss": 0.7358,
+      "step": 10160
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.9591403007507324,
+      "learning_rate": 9.695252491001617e-06,
+      "loss": 0.9783,
+      "step": 10165
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.6938751935958862,
+      "learning_rate": 9.638335805389209e-06,
+      "loss": 0.9006,
+      "step": 10170
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7790308594703674,
+      "learning_rate": 9.581578218699805e-06,
+      "loss": 0.9647,
+      "step": 10175
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7733349204063416,
+      "learning_rate": 9.524979830865999e-06,
+      "loss": 0.8754,
+      "step": 10180
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7591155171394348,
+      "learning_rate": 9.468540741539988e-06,
+      "loss": 0.843,
+      "step": 10185
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7738540768623352,
+      "learning_rate": 9.41226105009353e-06,
+      "loss": 0.8567,
+      "step": 10190
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.9674339294433594,
+      "learning_rate": 9.356140855617778e-06,
+      "loss": 0.8106,
+      "step": 10195
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.8703681826591492,
+      "learning_rate": 9.30018025692302e-06,
+      "loss": 0.8847,
+      "step": 10200
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7185772657394409,
+      "learning_rate": 9.244379352538535e-06,
+      "loss": 0.8332,
+      "step": 10205
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7870998382568359,
+      "learning_rate": 9.188738240712447e-06,
+      "loss": 0.8033,
+      "step": 10210
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.8352511525154114,
+      "learning_rate": 9.133257019411524e-06,
+      "loss": 0.7773,
+      "step": 10215
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.8999930620193481,
+      "learning_rate": 9.077935786321045e-06,
+      "loss": 0.8611,
+      "step": 10220
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.8633418083190918,
+      "learning_rate": 9.022774638844588e-06,
+      "loss": 0.9518,
+      "step": 10225
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.8065072298049927,
+      "learning_rate": 8.96777367410383e-06,
+      "loss": 0.9339,
+      "step": 10230
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7075096964836121,
+      "learning_rate": 8.912932988938472e-06,
+      "loss": 0.7301,
+      "step": 10235
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7145190834999084,
+      "learning_rate": 8.858252679905966e-06,
+      "loss": 0.7512,
+      "step": 10240
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.8400455713272095,
+      "learning_rate": 8.803732843281409e-06,
+      "loss": 0.8201,
+      "step": 10245
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.839592695236206,
+      "learning_rate": 8.749373575057384e-06,
+      "loss": 1.0019,
+      "step": 10250
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.8692545890808105,
+      "learning_rate": 8.695174970943732e-06,
+      "loss": 0.9288,
+      "step": 10255
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.8020569086074829,
+      "learning_rate": 8.641137126367416e-06,
+      "loss": 0.8789,
+      "step": 10260
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.8302674293518066,
+      "learning_rate": 8.587260136472353e-06,
+      "loss": 0.8592,
+      "step": 10265
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.8927515745162964,
+      "learning_rate": 8.53354409611924e-06,
+      "loss": 0.7836,
+      "step": 10270
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.8325286507606506,
+      "learning_rate": 8.479989099885388e-06,
+      "loss": 0.8751,
+      "step": 10275
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.808384895324707,
+      "learning_rate": 8.426595242064606e-06,
+      "loss": 0.9749,
+      "step": 10280
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.8181533813476562,
+      "learning_rate": 8.373362616666936e-06,
+      "loss": 1.0068,
+      "step": 10285
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7378880381584167,
+      "learning_rate": 8.320291317418549e-06,
+      "loss": 0.9654,
+      "step": 10290
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7099074721336365,
+      "learning_rate": 8.26738143776159e-06,
+      "loss": 0.9094,
+      "step": 10295
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.762239396572113,
+      "learning_rate": 8.214633070853938e-06,
+      "loss": 0.8217,
+      "step": 10300
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7620306015014648,
+      "learning_rate": 8.162046309569205e-06,
+      "loss": 0.7677,
+      "step": 10305
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.8328052163124084,
+      "learning_rate": 8.109621246496368e-06,
+      "loss": 0.9343,
+      "step": 10310
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.8162596821784973,
+      "learning_rate": 8.057357973939727e-06,
+      "loss": 0.9812,
+      "step": 10315
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.8823369145393372,
+      "learning_rate": 8.005256583918763e-06,
+      "loss": 1.0342,
+      "step": 10320
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7189355492591858,
+      "learning_rate": 7.953317168167862e-06,
+      "loss": 0.8651,
+      "step": 10325
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7958583235740662,
+      "learning_rate": 7.901539818136261e-06,
+      "loss": 0.8575,
+      "step": 10330
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.8764750957489014,
+      "learning_rate": 7.849924624987881e-06,
+      "loss": 1.0525,
+      "step": 10335
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.8272556662559509,
+      "learning_rate": 7.798471679601082e-06,
+      "loss": 0.8466,
+      "step": 10340
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.8034465909004211,
+      "learning_rate": 7.747181072568576e-06,
+      "loss": 0.8984,
+      "step": 10345
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.8183860182762146,
+      "learning_rate": 7.696052894197247e-06,
+      "loss": 0.7763,
+      "step": 10350
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7499911189079285,
+      "learning_rate": 7.645087234507975e-06,
+      "loss": 0.9159,
+      "step": 10355
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7751929759979248,
+      "learning_rate": 7.594284183235556e-06,
+      "loss": 0.89,
+      "step": 10360
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.9942512512207031,
+      "learning_rate": 7.543643829828406e-06,
+      "loss": 0.9464,
+      "step": 10365
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.915164589881897,
+      "learning_rate": 7.493166263448515e-06,
+      "loss": 0.9092,
+      "step": 10370
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.8597160577774048,
+      "learning_rate": 7.442851572971265e-06,
+      "loss": 0.923,
+      "step": 10375
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7946441173553467,
+      "learning_rate": 7.392699846985263e-06,
+      "loss": 0.8659,
+      "step": 10380
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7779476642608643,
+      "learning_rate": 7.342711173792127e-06,
+      "loss": 0.8092,
+      "step": 10385
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7479021549224854,
+      "learning_rate": 7.2928856414064996e-06,
+      "loss": 0.8293,
+      "step": 10390
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.8338636755943298,
+      "learning_rate": 7.243223337555693e-06,
+      "loss": 0.9277,
+      "step": 10395
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7314404845237732,
+      "learning_rate": 7.193724349679654e-06,
+      "loss": 0.8412,
+      "step": 10400
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.8216047286987305,
+      "learning_rate": 7.144388764930788e-06,
+      "loss": 0.9765,
+      "step": 10405
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7718465328216553,
+      "learning_rate": 7.095216670173776e-06,
+      "loss": 0.9366,
+      "step": 10410
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7446076273918152,
+      "learning_rate": 7.046208151985456e-06,
+      "loss": 0.7936,
+      "step": 10415
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7243603467941284,
+      "learning_rate": 6.997363296654691e-06,
+      "loss": 0.8252,
+      "step": 10420
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.8296288847923279,
+      "learning_rate": 6.9486821901821435e-06,
+      "loss": 0.7981,
+      "step": 10425
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.8079470992088318,
+      "learning_rate": 6.900164918280128e-06,
+      "loss": 0.9989,
+      "step": 10430
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.8123701214790344,
+      "learning_rate": 6.851811566372601e-06,
+      "loss": 0.8505,
+      "step": 10435
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.870762825012207,
+      "learning_rate": 6.8036222195948075e-06,
+      "loss": 0.8892,
+      "step": 10440
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.9291538596153259,
+      "learning_rate": 6.755596962793309e-06,
+      "loss": 0.8043,
+      "step": 10445
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.8943066596984863,
+      "learning_rate": 6.707735880525723e-06,
+      "loss": 0.8753,
+      "step": 10450
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.6992756724357605,
+      "learning_rate": 6.660039057060552e-06,
+      "loss": 0.8899,
+      "step": 10455
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.6595776677131653,
+      "learning_rate": 6.612506576377175e-06,
+      "loss": 0.957,
+      "step": 10460
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.871720016002655,
+      "learning_rate": 6.565138522165581e-06,
+      "loss": 0.8798,
+      "step": 10465
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.8589921593666077,
+      "learning_rate": 6.517934977826223e-06,
+      "loss": 0.7956,
+      "step": 10470
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.9183969497680664,
+      "learning_rate": 6.4708960264699745e-06,
+      "loss": 0.9662,
+      "step": 10475
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7627586722373962,
+      "learning_rate": 6.424021750917864e-06,
+      "loss": 0.922,
+      "step": 10480
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.8349488973617554,
+      "learning_rate": 6.377312233700938e-06,
+      "loss": 0.8646,
+      "step": 10485
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7143064141273499,
+      "learning_rate": 6.3307675570602354e-06,
+      "loss": 0.8173,
+      "step": 10490
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.8462641835212708,
+      "learning_rate": 6.284387802946534e-06,
+      "loss": 0.8991,
+      "step": 10495
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.8399547934532166,
+      "learning_rate": 6.238173053020191e-06,
+      "loss": 0.8205,
+      "step": 10500
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.8168681859970093,
+      "learning_rate": 6.192123388651128e-06,
+      "loss": 0.8276,
+      "step": 10505
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.8551720380783081,
+      "learning_rate": 6.146238890918488e-06,
+      "loss": 0.8583,
+      "step": 10510
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7600584030151367,
+      "learning_rate": 6.100519640610725e-06,
+      "loss": 0.8404,
+      "step": 10515
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7038664817810059,
+      "learning_rate": 6.054965718225258e-06,
+      "loss": 0.7928,
+      "step": 10520
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7905198335647583,
+      "learning_rate": 6.009577203968453e-06,
+      "loss": 0.8677,
+      "step": 10525
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.8545170426368713,
+      "learning_rate": 5.964354177755449e-06,
+      "loss": 0.864,
+      "step": 10530
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.8659030795097351,
+      "learning_rate": 5.919296719209988e-06,
+      "loss": 0.8788,
+      "step": 10535
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.6199814677238464,
+      "learning_rate": 5.874404907664277e-06,
+      "loss": 0.8115,
+      "step": 10540
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7567442655563354,
+      "learning_rate": 5.8296788221589575e-06,
+      "loss": 0.8215,
+      "step": 10545
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.8200069665908813,
+      "learning_rate": 5.785118541442791e-06,
+      "loss": 0.889,
+      "step": 10550
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.8444991707801819,
+      "learning_rate": 5.740724143972642e-06,
+      "loss": 0.8148,
+      "step": 10555
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.8217705488204956,
+      "learning_rate": 5.6964957079133186e-06,
+      "loss": 0.8814,
+      "step": 10560
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.8477857112884521,
+      "learning_rate": 5.652433311137384e-06,
+      "loss": 0.8157,
+      "step": 10565
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.8516268730163574,
+      "learning_rate": 5.608537031225092e-06,
+      "loss": 0.8576,
+      "step": 10570
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7571528553962708,
+      "learning_rate": 5.564806945464218e-06,
+      "loss": 0.7577,
+      "step": 10575
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7651041150093079,
+      "learning_rate": 5.521243130849873e-06,
+      "loss": 0.9706,
+      "step": 10580
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.9462747573852539,
+      "learning_rate": 5.4778456640845135e-06,
+      "loss": 0.8431,
+      "step": 10585
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.8326951861381531,
+      "learning_rate": 5.434614621577594e-06,
+      "loss": 0.7995,
+      "step": 10590
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.9692691564559937,
+      "learning_rate": 5.391550079445606e-06,
+      "loss": 0.9605,
+      "step": 10595
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.8991673588752747,
+      "learning_rate": 5.348652113511898e-06,
+      "loss": 0.866,
+      "step": 10600
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.8957405686378479,
+      "learning_rate": 5.305920799306496e-06,
+      "loss": 0.7906,
+      "step": 10605
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7040393352508545,
+      "learning_rate": 5.263356212066028e-06,
+      "loss": 0.8215,
+      "step": 10610
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.8569772243499756,
+      "learning_rate": 5.220958426733558e-06,
+      "loss": 0.9202,
+      "step": 10615
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.0359724760055542,
+      "learning_rate": 5.178727517958459e-06,
+      "loss": 0.9172,
+      "step": 10620
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7228214144706726,
+      "learning_rate": 5.136663560096277e-06,
+      "loss": 0.7619,
+      "step": 10625
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7785528302192688,
+      "learning_rate": 5.094766627208647e-06,
+      "loss": 0.8318,
+      "step": 10630
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.9546996355056763,
+      "learning_rate": 5.053036793063093e-06,
+      "loss": 0.9362,
+      "step": 10635
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.841090738773346,
+      "learning_rate": 5.011474131132931e-06,
+      "loss": 0.8019,
+      "step": 10640
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.6441927552223206,
+      "learning_rate": 4.970078714597149e-06,
+      "loss": 0.8103,
+      "step": 10645
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.8087906241416931,
+      "learning_rate": 4.928850616340252e-06,
+      "loss": 0.9268,
+      "step": 10650
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.6910356283187866,
+      "learning_rate": 4.887789908952178e-06,
+      "loss": 0.952,
+      "step": 10655
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7780956029891968,
+      "learning_rate": 4.846896664728118e-06,
+      "loss": 0.8688,
+      "step": 10660
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7802953720092773,
+      "learning_rate": 4.806170955668421e-06,
+      "loss": 0.7813,
+      "step": 10665
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.8613384366035461,
+      "learning_rate": 4.765612853478451e-06,
+      "loss": 0.8596,
+      "step": 10670
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.990271806716919,
+      "learning_rate": 4.725222429568477e-06,
+      "loss": 0.8501,
+      "step": 10675
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.8030913472175598,
+      "learning_rate": 4.68499975505351e-06,
+      "loss": 0.9165,
+      "step": 10680
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.9801160097122192,
+      "learning_rate": 4.644944900753278e-06,
+      "loss": 0.9632,
+      "step": 10685
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7507076263427734,
+      "learning_rate": 4.605057937191947e-06,
+      "loss": 1.0431,
+      "step": 10690
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7139765620231628,
+      "learning_rate": 4.565338934598129e-06,
+      "loss": 0.9047,
+      "step": 10695
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.8364464044570923,
+      "learning_rate": 4.525787962904682e-06,
+      "loss": 0.738,
+      "step": 10700
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.9670020341873169,
+      "learning_rate": 4.4864050917486355e-06,
+      "loss": 0.9677,
+      "step": 10705
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.9008302092552185,
+      "learning_rate": 4.447190390471024e-06,
+      "loss": 0.9228,
+      "step": 10710
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.8564368486404419,
+      "learning_rate": 4.408143928116815e-06,
+      "loss": 0.9183,
+      "step": 10715
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.8975895047187805,
+      "learning_rate": 4.369265773434739e-06,
+      "loss": 0.7976,
+      "step": 10720
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7680640816688538,
+      "learning_rate": 4.330555994877195e-06,
+      "loss": 0.8273,
+      "step": 10725
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.8345970511436462,
+      "learning_rate": 4.292014660600119e-06,
+      "loss": 0.8742,
+      "step": 10730
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.8377253413200378,
+      "learning_rate": 4.253641838462852e-06,
+      "loss": 0.9009,
+      "step": 10735
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7228429317474365,
+      "learning_rate": 4.2154375960280935e-06,
+      "loss": 0.8395,
+      "step": 10740
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.8424662947654724,
+      "learning_rate": 4.17740200056167e-06,
+      "loss": 0.9902,
+      "step": 10745
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.72088623046875,
+      "learning_rate": 4.139535119032501e-06,
+      "loss": 0.9015,
+      "step": 10750
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7103576064109802,
+      "learning_rate": 4.1018370181124424e-06,
+      "loss": 0.9158,
+      "step": 10755
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7628330588340759,
+      "learning_rate": 4.064307764176168e-06,
+      "loss": 0.8489,
+      "step": 10760
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7091038823127747,
+      "learning_rate": 4.0269474233010865e-06,
+      "loss": 0.8728,
+      "step": 10765
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.8292171359062195,
+      "learning_rate": 3.9897560612672136e-06,
+      "loss": 0.8284,
+      "step": 10770
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.6915936470031738,
+      "learning_rate": 3.9527337435570025e-06,
+      "loss": 0.811,
+      "step": 10775
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.9190654754638672,
+      "learning_rate": 3.915880535355298e-06,
+      "loss": 0.9775,
+      "step": 10780
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.9156875610351562,
+      "learning_rate": 3.879196501549209e-06,
+      "loss": 0.9032,
+      "step": 10785
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7918872833251953,
+      "learning_rate": 3.842681706727957e-06,
+      "loss": 0.9325,
+      "step": 10790
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7252178192138672,
+      "learning_rate": 3.806336215182782e-06,
+      "loss": 0.7474,
+      "step": 10795
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.8236010074615479,
+      "learning_rate": 3.7701600909068714e-06,
+      "loss": 0.9378,
+      "step": 10800
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.8983059525489807,
+      "learning_rate": 3.734153397595164e-06,
+      "loss": 0.8263,
+      "step": 10805
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.8560591340065002,
+      "learning_rate": 3.6983161986443027e-06,
+      "loss": 0.8704,
+      "step": 10810
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.9709807634353638,
+      "learning_rate": 3.662648557152515e-06,
+      "loss": 0.8396,
+      "step": 10815
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7081771492958069,
+      "learning_rate": 3.6271505359194547e-06,
+      "loss": 0.8082,
+      "step": 10820
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.8575358390808105,
+      "learning_rate": 3.591822197446182e-06,
+      "loss": 0.8739,
+      "step": 10825
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.9410017728805542,
+      "learning_rate": 3.556663603934951e-06,
+      "loss": 0.8653,
+      "step": 10830
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.8749052882194519,
+      "learning_rate": 3.5216748172891446e-06,
+      "loss": 0.9879,
+      "step": 10835
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.7416231036186218,
+      "learning_rate": 3.486855899113217e-06,
+      "loss": 0.8421,
+      "step": 10840
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.9364011287689209,
+      "learning_rate": 3.4522069107124966e-06,
+      "loss": 0.866,
+      "step": 10845
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.774756669998169,
+      "learning_rate": 3.4177279130931163e-06,
+      "loss": 0.8056,
+      "step": 10850
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.0904358625411987,
+      "learning_rate": 3.3834189669619377e-06,
+      "loss": 0.8952,
+      "step": 10855
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.8134381771087646,
+      "learning_rate": 3.3492801327263843e-06,
+      "loss": 0.8085,
+      "step": 10860
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.7840288877487183,
+      "learning_rate": 3.3153114704943756e-06,
+      "loss": 1.0089,
+      "step": 10865
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.1242091655731201,
+      "learning_rate": 3.2815130400742133e-06,
+      "loss": 0.805,
+      "step": 10870
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.88893061876297,
+      "learning_rate": 3.247884900974474e-06,
+      "loss": 0.7861,
+      "step": 10875
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.7033008933067322,
+      "learning_rate": 3.214427112403906e-06,
+      "loss": 0.893,
+      "step": 10880
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.9423139691352844,
+      "learning_rate": 3.181139733271332e-06,
+      "loss": 0.8106,
+      "step": 10885
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.831916332244873,
+      "learning_rate": 3.1480228221854923e-06,
+      "loss": 0.7484,
+      "step": 10890
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.6366778612136841,
+      "learning_rate": 3.1150764374550443e-06,
+      "loss": 0.8293,
+      "step": 10895
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.787534773349762,
+      "learning_rate": 3.0823006370883534e-06,
+      "loss": 0.859,
+      "step": 10900
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.747705340385437,
+      "learning_rate": 3.0496954787934684e-06,
+      "loss": 0.8347,
+      "step": 10905
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.802186906337738,
+      "learning_rate": 3.0172610199780017e-06,
+      "loss": 0.7809,
+      "step": 10910
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.92281574010849,
+      "learning_rate": 2.984997317748972e-06,
+      "loss": 0.9261,
+      "step": 10915
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.7391403913497925,
+      "learning_rate": 2.9529044289127726e-06,
+      "loss": 0.6701,
+      "step": 10920
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.921058177947998,
+      "learning_rate": 2.9209824099750595e-06,
+      "loss": 0.8643,
+      "step": 10925
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.8696638345718384,
+      "learning_rate": 2.889231317140617e-06,
+      "loss": 0.9082,
+      "step": 10930
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.0707058906555176,
+      "learning_rate": 2.857651206313305e-06,
+      "loss": 0.8325,
+      "step": 10935
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.7705253958702087,
+      "learning_rate": 2.8262421330959244e-06,
+      "loss": 0.9603,
+      "step": 10940
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.8354006409645081,
+      "learning_rate": 2.795004152790115e-06,
+      "loss": 0.8568,
+      "step": 10945
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.9094474911689758,
+      "learning_rate": 2.7639373203963036e-06,
+      "loss": 0.9253,
+      "step": 10950
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7950231432914734,
+      "learning_rate": 2.7330416906135582e-06,
+      "loss": 0.8863,
+      "step": 10955
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.8168395161628723,
+      "learning_rate": 2.702317317839531e-06,
+      "loss": 0.769,
+      "step": 10960
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.8256956934928894,
+      "learning_rate": 2.6717642561703505e-06,
+      "loss": 0.8708,
+      "step": 10965
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7272620797157288,
+      "learning_rate": 2.6413825594004625e-06,
+      "loss": 0.847,
+      "step": 10970
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.8060124516487122,
+      "learning_rate": 2.611172281022645e-06,
+      "loss": 0.804,
+      "step": 10975
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.8608812689781189,
+      "learning_rate": 2.5811334742278593e-06,
+      "loss": 0.9412,
+      "step": 10980
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7618398666381836,
+      "learning_rate": 2.551266191905133e-06,
+      "loss": 0.8519,
+      "step": 10985
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7475120425224304,
+      "learning_rate": 2.5215704866415224e-06,
+      "loss": 0.708,
+      "step": 10990
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7691943049430847,
+      "learning_rate": 2.492046410721971e-06,
+      "loss": 0.8495,
+      "step": 10995
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.8780038952827454,
+      "learning_rate": 2.4626940161292187e-06,
+      "loss": 1.1176,
+      "step": 11000
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7955008745193481,
+      "learning_rate": 2.4335133545437596e-06,
+      "loss": 0.8303,
+      "step": 11005
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7584604620933533,
+      "learning_rate": 2.4045044773437163e-06,
+      "loss": 0.8121,
+      "step": 11010
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7853428721427917,
+      "learning_rate": 2.3756674356047338e-06,
+      "loss": 0.8703,
+      "step": 11015
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.8115850687026978,
+      "learning_rate": 2.3470022800999193e-06,
+      "loss": 0.8613,
+      "step": 11020
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7144448161125183,
+      "learning_rate": 2.318509061299745e-06,
+      "loss": 0.8079,
+      "step": 11025
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.923337459564209,
+      "learning_rate": 2.2901878293719257e-06,
+      "loss": 0.9307,
+      "step": 11030
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7687123417854309,
+      "learning_rate": 2.2620386341814182e-06,
+      "loss": 0.8111,
+      "step": 11035
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.949176549911499,
+      "learning_rate": 2.234061525290232e-06,
+      "loss": 0.8639,
+      "step": 11040
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.8756553530693054,
+      "learning_rate": 2.2062565519573865e-06,
+      "loss": 0.9837,
+      "step": 11045
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.8877837657928467,
+      "learning_rate": 2.1786237631388428e-06,
+      "loss": 0.9086,
+      "step": 11050
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7283702492713928,
+      "learning_rate": 2.1511632074873835e-06,
+      "loss": 0.8531,
+      "step": 11055
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.1859776973724365,
+      "learning_rate": 2.1238749333525543e-06,
+      "loss": 0.8264,
+      "step": 11060
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7941679358482361,
+      "learning_rate": 2.096758988780556e-06,
+      "loss": 0.9445,
+      "step": 11065
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.724950909614563,
+      "learning_rate": 2.069815421514176e-06,
+      "loss": 0.7968,
+      "step": 11070
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.7822914123535156,
+      "learning_rate": 2.0430442789927007e-06,
+      "loss": 0.8489,
+      "step": 11075
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.7305450439453125,
+      "learning_rate": 2.0164456083518246e-06,
+      "loss": 0.8252,
+      "step": 11080
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.9346940517425537,
+      "learning_rate": 1.990019456423564e-06,
+      "loss": 0.8682,
+      "step": 11085
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.9352343678474426,
+      "learning_rate": 1.9637658697362003e-06,
+      "loss": 0.9635,
+      "step": 11090
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6705776453018188,
+      "learning_rate": 1.93768489451418e-06,
+      "loss": 0.7376,
+      "step": 11095
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6965312361717224,
+      "learning_rate": 1.911776576678015e-06,
+      "loss": 0.802,
+      "step": 11100
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.8443041443824768,
+      "learning_rate": 1.8860409618442488e-06,
+      "loss": 0.8767,
+      "step": 11105
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.8731304407119751,
+      "learning_rate": 1.8604780953253353e-06,
+      "loss": 0.8648,
+      "step": 11110
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.8138821721076965,
+      "learning_rate": 1.8350880221295496e-06,
+      "loss": 0.7863,
+      "step": 11115
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.8005531430244446,
+      "learning_rate": 1.8098707869609654e-06,
+      "loss": 0.8392,
+      "step": 11120
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.7327057719230652,
+      "learning_rate": 1.7848264342193333e-06,
+      "loss": 1.0973,
+      "step": 11125
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.8505914807319641,
+      "learning_rate": 1.7599550080000027e-06,
+      "loss": 0.8634,
+      "step": 11130
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.8741824626922607,
+      "learning_rate": 1.7352565520938558e-06,
+      "loss": 0.8809,
+      "step": 11135
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6897190809249878,
+      "learning_rate": 1.7107311099872403e-06,
+      "loss": 0.8575,
+      "step": 11140
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.772314190864563,
+      "learning_rate": 1.6863787248618367e-06,
+      "loss": 1.0116,
+      "step": 11145
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.7905736565589905,
+      "learning_rate": 1.6621994395946916e-06,
+      "loss": 0.8715,
+      "step": 11150
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.7269569039344788,
+      "learning_rate": 1.6381932967580505e-06,
+      "loss": 0.9212,
+      "step": 11155
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.7961433529853821,
+      "learning_rate": 1.6143603386192474e-06,
+      "loss": 0.8239,
+      "step": 11160
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.811042845249176,
+      "learning_rate": 1.5907006071408049e-06,
+      "loss": 0.9451,
+      "step": 11165
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.7868227958679199,
+      "learning_rate": 1.5672141439801446e-06,
+      "loss": 0.8004,
+      "step": 11170
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.7977198362350464,
+      "learning_rate": 1.5439009904896773e-06,
+      "loss": 0.9105,
+      "step": 11175
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.7236701250076294,
+      "learning_rate": 1.5207611877166573e-06,
+      "loss": 0.8506,
+      "step": 11180
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6770230531692505,
+      "learning_rate": 1.4977947764031053e-06,
+      "loss": 0.7893,
+      "step": 11185
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7709506154060364,
+      "learning_rate": 1.4750017969857643e-06,
+      "loss": 0.8009,
+      "step": 11190
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.8483154773712158,
+      "learning_rate": 1.4523822895960216e-06,
+      "loss": 0.7705,
+      "step": 11195
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.8976624608039856,
+      "learning_rate": 1.4299362940598194e-06,
+      "loss": 0.9007,
+      "step": 11200
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.9494867324829102,
+      "learning_rate": 1.4076638498976113e-06,
+      "loss": 0.8144,
+      "step": 11205
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7936916351318359,
+      "learning_rate": 1.3855649963242957e-06,
+      "loss": 0.8789,
+      "step": 11210
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7598190307617188,
+      "learning_rate": 1.3636397722490813e-06,
+      "loss": 0.8265,
+      "step": 11215
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7455407977104187,
+      "learning_rate": 1.3418882162755219e-06,
+      "loss": 0.8494,
+      "step": 11220
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.8700529336929321,
+      "learning_rate": 1.3203103667013827e-06,
+      "loss": 0.9347,
+      "step": 11225
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.8811212778091431,
+      "learning_rate": 1.298906261518551e-06,
+      "loss": 0.7147,
+      "step": 11230
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.8739584684371948,
+      "learning_rate": 1.2776759384130698e-06,
+      "loss": 0.8998,
+      "step": 11235
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.8095049858093262,
+      "learning_rate": 1.2566194347649385e-06,
+      "loss": 0.9268,
+      "step": 11240
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7221877574920654,
+      "learning_rate": 1.2357367876481452e-06,
+      "loss": 0.8186,
+      "step": 11245
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.8922355771064758,
+      "learning_rate": 1.2150280338305787e-06,
+      "loss": 0.8335,
+      "step": 11250
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7169233560562134,
+      "learning_rate": 1.194493209773928e-06,
+      "loss": 0.7704,
+      "step": 11255
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.8172361254692078,
+      "learning_rate": 1.1741323516336832e-06,
+      "loss": 0.8524,
+      "step": 11260
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7009252309799194,
+      "learning_rate": 1.1539454952590123e-06,
+      "loss": 0.8321,
+      "step": 11265
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7184674143791199,
+      "learning_rate": 1.133932676192695e-06,
+      "loss": 0.8192,
+      "step": 11270
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.8454670906066895,
+      "learning_rate": 1.114093929671145e-06,
+      "loss": 0.9189,
+      "step": 11275
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7921102046966553,
+      "learning_rate": 1.0944292906242326e-06,
+      "loss": 0.8797,
+      "step": 11280
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7686296701431274,
+      "learning_rate": 1.0749387936753064e-06,
+      "loss": 0.9228,
+      "step": 11285
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.78754061460495,
+      "learning_rate": 1.0556224731411157e-06,
+      "loss": 0.857,
+      "step": 11290
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.749705970287323,
+      "learning_rate": 1.0364803630316887e-06,
+      "loss": 0.84,
+      "step": 11295
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.751549482345581,
+      "learning_rate": 1.017512497050377e-06,
+      "loss": 0.7576,
+      "step": 11300
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.6023859977722168,
+      "learning_rate": 9.98718908593732e-07,
+      "loss": 0.8114,
+      "step": 11305
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.9303193092346191,
+      "learning_rate": 9.8009963075143e-07,
+      "loss": 0.8892,
+      "step": 11310
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.9002019166946411,
+      "learning_rate": 9.61654696306258e-07,
+      "loss": 0.8828,
+      "step": 11315
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.8884360790252686,
+      "learning_rate": 9.43384137734038e-07,
+      "loss": 0.8621,
+      "step": 11320
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.8538162708282471,
+      "learning_rate": 9.252879872035713e-07,
+      "loss": 0.7332,
+      "step": 11325
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.7713654041290283,
+      "learning_rate": 9.073662765765823e-07,
+      "loss": 0.754,
+      "step": 11330
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.7992544174194336,
+      "learning_rate": 8.896190374076518e-07,
+      "loss": 0.9635,
+      "step": 11335
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.85658198595047,
+      "learning_rate": 8.720463009441626e-07,
+      "loss": 0.8007,
+      "step": 11340
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.8936882615089417,
+      "learning_rate": 8.546480981262872e-07,
+      "loss": 0.9472,
+      "step": 11345
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.8276568651199341,
+      "learning_rate": 8.374244595868664e-07,
+      "loss": 0.978,
+      "step": 11350
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.8250508308410645,
+      "learning_rate": 8.203754156513865e-07,
+      "loss": 0.9266,
+      "step": 11355
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.7581374049186707,
+      "learning_rate": 8.03500996337958e-07,
+      "loss": 0.8211,
+      "step": 11360
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.0592025518417358,
+      "learning_rate": 7.868012313571927e-07,
+      "loss": 0.9002,
+      "step": 11365
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.9346907138824463,
+      "learning_rate": 7.702761501122147e-07,
+      "loss": 0.8058,
+      "step": 11370
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.866581916809082,
+      "learning_rate": 7.539257816985835e-07,
+      "loss": 0.8184,
+      "step": 11375
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.8073518872261047,
+      "learning_rate": 7.377501549042265e-07,
+      "loss": 0.8043,
+      "step": 11380
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.8920919895172119,
+      "learning_rate": 7.217492982094176e-07,
+      "loss": 0.8727,
+      "step": 11385
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.7557914853096008,
+      "learning_rate": 7.059232397867099e-07,
+      "loss": 0.8605,
+      "step": 11390
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.7152570486068726,
+      "learning_rate": 6.902720075009139e-07,
+      "loss": 0.8231,
+      "step": 11395
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.7431889772415161,
+      "learning_rate": 6.747956289089863e-07,
+      "loss": 0.9424,
+      "step": 11400
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.7541787624359131,
+      "learning_rate": 6.594941312600411e-07,
+      "loss": 0.9519,
+      "step": 11405
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.7901094555854797,
+      "learning_rate": 6.443675414952833e-07,
+      "loss": 0.9928,
+      "step": 11410
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.0786751508712769,
+      "learning_rate": 6.294158862479527e-07,
+      "loss": 0.8042,
+      "step": 11415
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.8062751293182373,
+      "learning_rate": 6.146391918433026e-07,
+      "loss": 0.8099,
+      "step": 11420
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.8185136318206787,
+      "learning_rate": 6.000374842984991e-07,
+      "loss": 1.0152,
+      "step": 11425
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.7455396056175232,
+      "learning_rate": 5.856107893226325e-07,
+      "loss": 0.8298,
+      "step": 11430
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.8589451909065247,
+      "learning_rate": 5.713591323166622e-07,
+      "loss": 0.8908,
+      "step": 11435
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.0530104637145996,
+      "learning_rate": 5.57282538373316e-07,
+      "loss": 0.9245,
+      "step": 11440
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.8141350746154785,
+      "learning_rate": 5.433810322771571e-07,
+      "loss": 0.861,
+      "step": 11445
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.7331737875938416,
+      "learning_rate": 5.296546385044065e-07,
+      "loss": 0.8259,
+      "step": 11450
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.8267128467559814,
+      "learning_rate": 5.161033812229987e-07,
+      "loss": 0.9983,
+      "step": 11455
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.7879713177680969,
+      "learning_rate": 5.027272842925146e-07,
+      "loss": 0.8164,
+      "step": 11460
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.83629310131073,
+      "learning_rate": 4.895263712641151e-07,
+      "loss": 0.947,
+      "step": 11465
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.8497070074081421,
+      "learning_rate": 4.7650066538051927e-07,
+      "loss": 0.8704,
+      "step": 11470
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.8327845931053162,
+      "learning_rate": 4.636501895759704e-07,
+      "loss": 0.8605,
+      "step": 11475
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.8554712533950806,
+      "learning_rate": 4.5097496647616977e-07,
+      "loss": 0.8408,
+      "step": 11480
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.9654614329338074,
+      "learning_rate": 4.3847501839827666e-07,
+      "loss": 0.894,
+      "step": 11485
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.794713020324707,
+      "learning_rate": 4.261503673508194e-07,
+      "loss": 0.8114,
+      "step": 11490
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.8144010305404663,
+      "learning_rate": 4.1400103503368425e-07,
+      "loss": 0.8534,
+      "step": 11495
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.1428261995315552,
+      "learning_rate": 4.0202704283810456e-07,
+      "loss": 0.8794,
+      "step": 11500
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.9131525158882141,
+      "learning_rate": 3.9022841184657155e-07,
+      "loss": 0.7327,
+      "step": 11505
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.6796863675117493,
+      "learning_rate": 3.7860516283282355e-07,
+      "loss": 0.8537,
+      "step": 11510
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.7078248262405396,
+      "learning_rate": 3.6715731626179027e-07,
+      "loss": 0.8454,
+      "step": 11515
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.745050847530365,
+      "learning_rate": 3.55884892289593e-07,
+      "loss": 0.7683,
+      "step": 11520
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.64480060338974,
+      "learning_rate": 3.447879107634888e-07,
+      "loss": 0.768,
+      "step": 11525
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.8762190341949463,
+      "learning_rate": 3.338663912218265e-07,
+      "loss": 0.794,
+      "step": 11530
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.8294702172279358,
+      "learning_rate": 3.23120352894013e-07,
+      "loss": 0.914,
+      "step": 11535
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.7077884078025818,
+      "learning_rate": 3.1254981470049126e-07,
+      "loss": 0.8384,
+      "step": 11540
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.8789534568786621,
+      "learning_rate": 3.021547952527293e-07,
+      "loss": 0.8172,
+      "step": 11545
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.9262539744377136,
+      "learning_rate": 2.9193531285311993e-07,
+      "loss": 0.9344,
+      "step": 11550
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.981154203414917,
+      "learning_rate": 2.818913854950256e-07,
+      "loss": 0.8827,
+      "step": 11555
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.6264580488204956,
+      "learning_rate": 2.720230308626781e-07,
+      "loss": 0.7482,
+      "step": 11560
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7378256320953369,
+      "learning_rate": 2.6233026633118994e-07,
+      "loss": 0.7769,
+      "step": 11565
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7126420736312866,
+      "learning_rate": 2.528131089665431e-07,
+      "loss": 0.7937,
+      "step": 11570
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.800485372543335,
+      "learning_rate": 2.4347157552548907e-07,
+      "loss": 0.9828,
+      "step": 11575
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.8482642769813538,
+      "learning_rate": 2.3430568245558227e-07,
+      "loss": 1.0101,
+      "step": 11580
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.9000346064567566,
+      "learning_rate": 2.2531544589512454e-07,
+      "loss": 0.8043,
+      "step": 11585
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.6278972029685974,
+      "learning_rate": 2.1650088167313177e-07,
+      "loss": 0.6926,
+      "step": 11590
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.8249235153198242,
+      "learning_rate": 2.0786200530933387e-07,
+      "loss": 0.9562,
+      "step": 11595
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7574777603149414,
+      "learning_rate": 1.9939883201410826e-07,
+      "loss": 0.8049,
+      "step": 11600
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.8108986020088196,
+      "learning_rate": 1.911113766884909e-07,
+      "loss": 0.8037,
+      "step": 11605
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.8171221613883972,
+      "learning_rate": 1.8299965392413187e-07,
+      "loss": 0.8173,
+      "step": 11610
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.8570646047592163,
+      "learning_rate": 1.7506367800325108e-07,
+      "loss": 0.8189,
+      "step": 11615
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7071337699890137,
+      "learning_rate": 1.6730346289864918e-07,
+      "loss": 0.9277,
+      "step": 11620
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.8421007394790649,
+      "learning_rate": 1.597190222736633e-07,
+      "loss": 0.8913,
+      "step": 11625
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.746640682220459,
+      "learning_rate": 1.5231036948215594e-07,
+      "loss": 0.8051,
+      "step": 11630
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.10215163230896,
+      "learning_rate": 1.4507751756845934e-07,
+      "loss": 0.9606,
+      "step": 11635
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7056140303611755,
+      "learning_rate": 1.380204792673867e-07,
+      "loss": 0.9655,
+      "step": 11640
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.896172046661377,
+      "learning_rate": 1.3113926700420998e-07,
+      "loss": 0.9775,
+      "step": 11645
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7562682032585144,
+      "learning_rate": 1.2443389289460427e-07,
+      "loss": 0.7869,
+      "step": 11650
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.8313950300216675,
+      "learning_rate": 1.1790436874465904e-07,
+      "loss": 0.8413,
+      "step": 11655
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.9028708338737488,
+      "learning_rate": 1.1155070605085583e-07,
+      "loss": 0.8534,
+      "step": 11660
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8687404990196228,
+      "learning_rate": 1.0537291600000165e-07,
+      "loss": 0.8209,
+      "step": 11665
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8653462529182434,
+      "learning_rate": 9.937100946930677e-08,
+      "loss": 0.7397,
+      "step": 11670
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7406777739524841,
+      "learning_rate": 9.354499702625141e-08,
+      "loss": 0.7934,
+      "step": 11675
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.6351541876792908,
+      "learning_rate": 8.789488892864129e-08,
+      "loss": 0.8001,
+      "step": 11680
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8214691877365112,
+      "learning_rate": 8.242069512456318e-08,
+      "loss": 0.8547,
+      "step": 11685
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8716254830360413,
+      "learning_rate": 7.71224252523961e-08,
+      "loss": 0.8056,
+      "step": 11690
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8704701662063599,
+      "learning_rate": 7.200008864073349e-08,
+      "loss": 0.9058,
+      "step": 11695
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8255096077919006,
+      "learning_rate": 6.705369430843878e-08,
+      "loss": 0.8133,
+      "step": 11700
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.6788018941879272,
+      "learning_rate": 6.228325096457876e-08,
+      "loss": 0.6706,
+      "step": 11705
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.2154595851898193,
+      "learning_rate": 5.7688767008423627e-08,
+      "loss": 0.9283,
+      "step": 11710
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.6604523062705994,
+      "learning_rate": 5.327025052943579e-08,
+      "loss": 0.7873,
+      "step": 11715
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8354642987251282,
+      "learning_rate": 4.902770930725886e-08,
+      "loss": 0.8338,
+      "step": 11720
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.9577195048332214,
+      "learning_rate": 4.4961150811695384e-08,
+      "loss": 0.9615,
+      "step": 11725
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7846125364303589,
+      "learning_rate": 4.107058220270687e-08,
+      "loss": 0.9433,
+      "step": 11730
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8085949420928955,
+      "learning_rate": 3.735601033035829e-08,
+      "loss": 0.87,
+      "step": 11735
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.022504210472107,
+      "learning_rate": 3.3817441734862455e-08,
+      "loss": 0.8262,
+      "step": 11740
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.686873197555542,
+      "learning_rate": 3.045488264656893e-08,
+      "loss": 0.799,
+      "step": 11745
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8503823280334473,
+      "learning_rate": 2.7268338985875218e-08,
+      "loss": 0.8444,
+      "step": 11750
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7915425300598145,
+      "learning_rate": 2.4257816363326692e-08,
+      "loss": 0.8874,
+      "step": 11755
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.6971069574356079,
+      "learning_rate": 2.1423320079494435e-08,
+      "loss": 0.8429,
+      "step": 11760
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8105065226554871,
+      "learning_rate": 1.8764855125052993e-08,
+      "loss": 0.9761,
+      "step": 11765
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8171855211257935,
+      "learning_rate": 1.6282426180758148e-08,
+      "loss": 0.9214,
+      "step": 11770
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.099576711654663,
+      "learning_rate": 1.3976037617380311e-08,
+      "loss": 1.024,
+      "step": 11775
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7364673614501953,
+      "learning_rate": 1.1845693495760035e-08,
+      "loss": 0.759,
+      "step": 11780
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.0711894035339355,
+      "learning_rate": 9.891397566774708e-09,
+      "loss": 0.7868,
+      "step": 11785
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7905126214027405,
+      "learning_rate": 8.113153271327446e-09,
+      "loss": 0.8227,
+      "step": 11790
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7753369808197021,
+      "learning_rate": 6.510963740369303e-09,
+      "loss": 0.7911,
+      "step": 11795
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6878575086593628,
+      "learning_rate": 5.08483179485486e-09,
+      "loss": 0.9471,
+      "step": 11800
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7836639285087585,
+      "learning_rate": 3.83475994575333e-09,
+      "loss": 0.8245,
+      "step": 11805
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7580632567405701,
+      "learning_rate": 2.7607503940707546e-09,
+      "loss": 0.8384,
+      "step": 11810
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7631229162216187,
+      "learning_rate": 1.862805030783399e-09,
+      "loss": 0.8937,
+      "step": 11815
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.764155387878418,
+      "learning_rate": 1.1409254369154632e-09,
+      "loss": 0.8793,
+      "step": 11820
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.8263822793960571,
+      "learning_rate": 5.951128834613684e-10,
+      "loss": 0.8754,
+      "step": 11825
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.8471561670303345,
+      "learning_rate": 2.2536833143016467e-10,
+      "loss": 0.8039,
+      "step": 11830
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7613112330436707,
+      "learning_rate": 3.169243183442916e-11,
+      "loss": 1.0197,
+      "step": 11835
+    },
+    {
+      "epoch": 1.0,
+      "step": 11838,
+      "total_flos": 6.323834158736998e+16,
+      "train_loss": 0.0,
+      "train_runtime": 0.0141,
+      "train_samples_per_second": 837064.766,
+      "train_steps_per_second": 837064.766
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 11838,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 6.323834158736998e+16,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}