diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4509 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.999764280663157,
+  "eval_steps": 500,
+  "global_step": 3181,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0003142924491239098,
+      "grad_norm": 19.593584060668945,
+      "learning_rate": 3.134796238244514e-06,
+      "loss": 7.2211,
+      "step": 1
+    },
+    {
+      "epoch": 0.001571462245619549,
+      "grad_norm": 17.929481506347656,
+      "learning_rate": 1.567398119122257e-05,
+      "loss": 7.2713,
+      "step": 5
+    },
+    {
+      "epoch": 0.003142924491239098,
+      "grad_norm": 17.97577476501465,
+      "learning_rate": 3.134796238244514e-05,
+      "loss": 7.2643,
+      "step": 10
+    },
+    {
+      "epoch": 0.004714386736858647,
+      "grad_norm": 9.52695369720459,
+      "learning_rate": 4.7021943573667716e-05,
+      "loss": 6.9744,
+      "step": 15
+    },
+    {
+      "epoch": 0.006285848982478196,
+      "grad_norm": 4.145598888397217,
+      "learning_rate": 6.269592476489028e-05,
+      "loss": 6.5934,
+      "step": 20
+    },
+    {
+      "epoch": 0.007857311228097744,
+      "grad_norm": 3.3066627979278564,
+      "learning_rate": 7.836990595611286e-05,
+      "loss": 6.1739,
+      "step": 25
+    },
+    {
+      "epoch": 0.009428773473717294,
+      "grad_norm": 3.240945339202881,
+      "learning_rate": 9.404388714733543e-05,
+      "loss": 5.9143,
+      "step": 30
+    },
+    {
+      "epoch": 0.011000235719336842,
+      "grad_norm": 1.6531422138214111,
+      "learning_rate": 0.000109717868338558,
+      "loss": 5.6534,
+      "step": 35
+    },
+    {
+      "epoch": 0.012571697964956392,
+      "grad_norm": 0.9349701404571533,
+      "learning_rate": 0.00012539184952978057,
+      "loss": 5.471,
+      "step": 40
+    },
+    {
+      "epoch": 0.01414316021057594,
+      "grad_norm": 0.565790057182312,
+      "learning_rate": 0.00014106583072100311,
+      "loss": 5.3034,
+      "step": 45
+    },
+    {
+      "epoch": 0.01571462245619549,
+      "grad_norm": 0.4517938494682312,
+      "learning_rate": 0.00015673981191222572,
+      "loss": 5.1125,
+      "step": 50
+    },
+    {
+      "epoch": 0.01728608470181504,
+      "grad_norm": 0.4741256833076477,
+      "learning_rate": 0.0001724137931034483,
+      "loss": 5.009,
+      "step": 55
+    },
+    {
+      "epoch": 0.01885754694743459,
+      "grad_norm": 0.4407913386821747,
+      "learning_rate": 0.00018808777429467086,
+      "loss": 4.9626,
+      "step": 60
+    },
+    {
+      "epoch": 0.020429009193054137,
+      "grad_norm": 0.6502472758293152,
+      "learning_rate": 0.00020376175548589344,
+      "loss": 4.837,
+      "step": 65
+    },
+    {
+      "epoch": 0.022000471438673685,
+      "grad_norm": 0.6139042377471924,
+      "learning_rate": 0.000219435736677116,
+      "loss": 4.7188,
+      "step": 70
+    },
+    {
+      "epoch": 0.023571933684293236,
+      "grad_norm": 0.5190150141716003,
+      "learning_rate": 0.00023510971786833856,
+      "loss": 4.5818,
+      "step": 75
+    },
+    {
+      "epoch": 0.025143395929912785,
+      "grad_norm": 0.4606360197067261,
+      "learning_rate": 0.00025078369905956113,
+      "loss": 4.4429,
+      "step": 80
+    },
+    {
+      "epoch": 0.026714858175532333,
+      "grad_norm": 0.38270899653434753,
+      "learning_rate": 0.0002664576802507837,
+      "loss": 4.4131,
+      "step": 85
+    },
+    {
+      "epoch": 0.02828632042115188,
+      "grad_norm": 0.4721100330352783,
+      "learning_rate": 0.00028213166144200623,
+      "loss": 4.248,
+      "step": 90
+    },
+    {
+      "epoch": 0.02985778266677143,
+      "grad_norm": 0.4452402889728546,
+      "learning_rate": 0.00029780564263322886,
+      "loss": 4.1921,
+      "step": 95
+    },
+    {
+      "epoch": 0.03142924491239098,
+      "grad_norm": 0.4978654980659485,
+      "learning_rate": 0.00031347962382445143,
+      "loss": 4.0356,
+      "step": 100
+    },
+    {
+      "epoch": 0.033000707158010525,
+      "grad_norm": 0.6152021288871765,
+      "learning_rate": 0.000329153605015674,
+      "loss": 3.8516,
+      "step": 105
+    },
+    {
+      "epoch": 0.03457216940363008,
+      "grad_norm": 0.4162922203540802,
+      "learning_rate": 0.0003448275862068966,
+      "loss": 3.8018,
+      "step": 110
+    },
+    {
+      "epoch": 0.03614363164924963,
+      "grad_norm": 0.44047296047210693,
+      "learning_rate": 0.0003605015673981191,
+      "loss": 3.5919,
+      "step": 115
+    },
+    {
+      "epoch": 0.03771509389486918,
+      "grad_norm": 0.4196017384529114,
+      "learning_rate": 0.00037617554858934173,
+      "loss": 3.5899,
+      "step": 120
+    },
+    {
+      "epoch": 0.039286556140488725,
+      "grad_norm": 0.4823077917098999,
+      "learning_rate": 0.00039184952978056425,
+      "loss": 3.4321,
+      "step": 125
+    },
+    {
+      "epoch": 0.04085801838610827,
+      "grad_norm": 0.3624018132686615,
+      "learning_rate": 0.0004075235109717869,
+      "loss": 3.3484,
+      "step": 130
+    },
+    {
+      "epoch": 0.04242948063172782,
+      "grad_norm": 0.3260553181171417,
+      "learning_rate": 0.0004231974921630094,
+      "loss": 3.2639,
+      "step": 135
+    },
+    {
+      "epoch": 0.04400094287734737,
+      "grad_norm": 0.38659989833831787,
+      "learning_rate": 0.000438871473354232,
+      "loss": 3.1999,
+      "step": 140
+    },
+    {
+      "epoch": 0.04557240512296692,
+      "grad_norm": 0.4590894281864166,
+      "learning_rate": 0.00045454545454545455,
+      "loss": 3.1469,
+      "step": 145
+    },
+    {
+      "epoch": 0.04714386736858647,
+      "grad_norm": 0.39775681495666504,
+      "learning_rate": 0.0004702194357366771,
+      "loss": 3.0631,
+      "step": 150
+    },
+    {
+      "epoch": 0.04871532961420602,
+      "grad_norm": 0.43203604221343994,
+      "learning_rate": 0.0004858934169278997,
+      "loss": 3.0505,
+      "step": 155
+    },
+    {
+      "epoch": 0.05028679185982557,
+      "grad_norm": 0.3682110607624054,
+      "learning_rate": 0.0005015673981191223,
+      "loss": 3.0316,
+      "step": 160
+    },
+    {
+      "epoch": 0.05185825410544512,
+      "grad_norm": 0.32574963569641113,
+      "learning_rate": 0.0005172413793103448,
+      "loss": 3.0132,
+      "step": 165
+    },
+    {
+      "epoch": 0.053429716351064666,
+      "grad_norm": 0.4408009946346283,
+      "learning_rate": 0.0005329153605015674,
+      "loss": 3.0166,
+      "step": 170
+    },
+    {
+      "epoch": 0.055001178596684214,
+      "grad_norm": 0.3008413016796112,
+      "learning_rate": 0.00054858934169279,
+      "loss": 2.8596,
+      "step": 175
+    },
+    {
+      "epoch": 0.05657264084230376,
+      "grad_norm": 0.3133067786693573,
+      "learning_rate": 0.0005642633228840125,
+      "loss": 2.9631,
+      "step": 180
+    },
+    {
+      "epoch": 0.05814410308792331,
+      "grad_norm": 0.4327049255371094,
+      "learning_rate": 0.0005799373040752351,
+      "loss": 2.8666,
+      "step": 185
+    },
+    {
+      "epoch": 0.05971556533354286,
+      "grad_norm": 0.44304072856903076,
+      "learning_rate": 0.0005956112852664577,
+      "loss": 2.7938,
+      "step": 190
+    },
+    {
+      "epoch": 0.06128702757916241,
+      "grad_norm": 0.4220844805240631,
+      "learning_rate": 0.0006112852664576803,
+      "loss": 2.8245,
+      "step": 195
+    },
+    {
+      "epoch": 0.06285848982478195,
+      "grad_norm": 0.36100244522094727,
+      "learning_rate": 0.0006269592476489029,
+      "loss": 2.7925,
+      "step": 200
+    },
+    {
+      "epoch": 0.0644299520704015,
+      "grad_norm": 0.4607883095741272,
+      "learning_rate": 0.0006426332288401254,
+      "loss": 2.7257,
+      "step": 205
+    },
+    {
+      "epoch": 0.06600141431602105,
+      "grad_norm": 0.3919202387332916,
+      "learning_rate": 0.000658307210031348,
+      "loss": 2.6557,
+      "step": 210
+    },
+    {
+      "epoch": 0.06757287656164061,
+      "grad_norm": 0.3630659282207489,
+      "learning_rate": 0.0006739811912225705,
+      "loss": 2.7476,
+      "step": 215
+    },
+    {
+      "epoch": 0.06914433880726016,
+      "grad_norm": 0.3189098834991455,
+      "learning_rate": 0.0006896551724137932,
+      "loss": 2.6834,
+      "step": 220
+    },
+    {
+      "epoch": 0.07071580105287971,
+      "grad_norm": 0.3652380108833313,
+      "learning_rate": 0.0007053291536050157,
+      "loss": 2.6362,
+      "step": 225
+    },
+    {
+      "epoch": 0.07228726329849926,
+      "grad_norm": 0.35851606726646423,
+      "learning_rate": 0.0007210031347962382,
+      "loss": 2.637,
+      "step": 230
+    },
+    {
+      "epoch": 0.0738587255441188,
+      "grad_norm": 0.7696281671524048,
+      "learning_rate": 0.0007366771159874608,
+      "loss": 2.5676,
+      "step": 235
+    },
+    {
+      "epoch": 0.07543018778973835,
+      "grad_norm": 0.6321612000465393,
+      "learning_rate": 0.0007523510971786835,
+      "loss": 2.6147,
+      "step": 240
+    },
+    {
+      "epoch": 0.0770016500353579,
+      "grad_norm": 0.386197030544281,
+      "learning_rate": 0.000768025078369906,
+      "loss": 2.5907,
+      "step": 245
+    },
+    {
+      "epoch": 0.07857311228097745,
+      "grad_norm": 0.4827702045440674,
+      "learning_rate": 0.0007836990595611285,
+      "loss": 2.5142,
+      "step": 250
+    },
+    {
+      "epoch": 0.080144574526597,
+      "grad_norm": 0.292161762714386,
+      "learning_rate": 0.0007993730407523511,
+      "loss": 2.4859,
+      "step": 255
+    },
+    {
+      "epoch": 0.08171603677221655,
+      "grad_norm": 0.33461418747901917,
+      "learning_rate": 0.0008150470219435738,
+      "loss": 2.5654,
+      "step": 260
+    },
+    {
+      "epoch": 0.0832874990178361,
+      "grad_norm": 0.5843150019645691,
+      "learning_rate": 0.0008307210031347962,
+      "loss": 2.5168,
+      "step": 265
+    },
+    {
+      "epoch": 0.08485896126345564,
+      "grad_norm": 0.5514886975288391,
+      "learning_rate": 0.0008463949843260188,
+      "loss": 2.4514,
+      "step": 270
+    },
+    {
+      "epoch": 0.08643042350907519,
+      "grad_norm": 0.32894566655158997,
+      "learning_rate": 0.0008620689655172414,
+      "loss": 2.4504,
+      "step": 275
+    },
+    {
+      "epoch": 0.08800188575469474,
+      "grad_norm": 0.30089420080184937,
+      "learning_rate": 0.000877742946708464,
+      "loss": 2.5003,
+      "step": 280
+    },
+    {
+      "epoch": 0.08957334800031429,
+      "grad_norm": 0.41669008135795593,
+      "learning_rate": 0.0008934169278996865,
+      "loss": 2.4293,
+      "step": 285
+    },
+    {
+      "epoch": 0.09114481024593384,
+      "grad_norm": 0.25633400678634644,
+      "learning_rate": 0.0009090909090909091,
+      "loss": 2.4447,
+      "step": 290
+    },
+    {
+      "epoch": 0.09271627249155338,
+      "grad_norm": 0.27528685331344604,
+      "learning_rate": 0.0009247648902821318,
+      "loss": 2.4672,
+      "step": 295
+    },
+    {
+      "epoch": 0.09428773473717295,
+      "grad_norm": 0.3092777132987976,
+      "learning_rate": 0.0009404388714733542,
+      "loss": 2.3757,
+      "step": 300
+    },
+    {
+      "epoch": 0.0958591969827925,
+      "grad_norm": 0.29513272643089294,
+      "learning_rate": 0.0009561128526645768,
+      "loss": 2.4152,
+      "step": 305
+    },
+    {
+      "epoch": 0.09743065922841204,
+      "grad_norm": 0.43028154969215393,
+      "learning_rate": 0.0009717868338557994,
+      "loss": 2.4507,
+      "step": 310
+    },
+    {
+      "epoch": 0.09900212147403159,
+      "grad_norm": 0.33603623509407043,
+      "learning_rate": 0.000987460815047022,
+      "loss": 2.3948,
+      "step": 315
+    },
+    {
+      "epoch": 0.10057358371965114,
+      "grad_norm": 0.36047083139419556,
+      "learning_rate": 0.0009999996987684534,
+      "loss": 2.3747,
+      "step": 320
+    },
+    {
+      "epoch": 0.10214504596527069,
+      "grad_norm": 0.47083067893981934,
+      "learning_rate": 0.0009999891557024327,
+      "loss": 2.4144,
+      "step": 325
+    },
+    {
+      "epoch": 0.10371650821089023,
+      "grad_norm": 0.3539304733276367,
+      "learning_rate": 0.0009999635514220413,
+      "loss": 2.365,
+      "step": 330
+    },
+    {
+      "epoch": 0.10528797045650978,
+      "grad_norm": 0.3079536557197571,
+      "learning_rate": 0.0009999228866985584,
+      "loss": 2.3087,
+      "step": 335
+    },
+    {
+      "epoch": 0.10685943270212933,
+      "grad_norm": 0.3304389715194702,
+      "learning_rate": 0.0009998671627569314,
+      "loss": 2.2729,
+      "step": 340
+    },
+    {
+      "epoch": 0.10843089494774888,
+      "grad_norm": 0.3577703535556793,
+      "learning_rate": 0.0009997963812757367,
+      "loss": 2.362,
+      "step": 345
+    },
+    {
+      "epoch": 0.11000235719336843,
+      "grad_norm": 0.3626112937927246,
+      "learning_rate": 0.000999710544387131,
+      "loss": 2.3232,
+      "step": 350
+    },
+    {
+      "epoch": 0.11157381943898798,
+      "grad_norm": 0.33722984790802,
+      "learning_rate": 0.000999609654676786,
+      "loss": 2.3837,
+      "step": 355
+    },
+    {
+      "epoch": 0.11314528168460752,
+      "grad_norm": 0.6007051467895508,
+      "learning_rate": 0.0009994937151838103,
+      "loss": 2.2621,
+      "step": 360
+    },
+    {
+      "epoch": 0.11471674393022707,
+      "grad_norm": 0.24785129725933075,
+      "learning_rate": 0.0009993627294006592,
+      "loss": 2.3242,
+      "step": 365
+    },
+    {
+      "epoch": 0.11628820617584662,
+      "grad_norm": 0.3804435431957245,
+      "learning_rate": 0.000999216701273028,
+      "loss": 2.2901,
+      "step": 370
+    },
+    {
+      "epoch": 0.11785966842146617,
+      "grad_norm": 0.2868124544620514,
+      "learning_rate": 0.000999055635199734,
+      "loss": 2.2492,
+      "step": 375
+    },
+    {
+      "epoch": 0.11943113066708572,
+      "grad_norm": 0.45313236117362976,
+      "learning_rate": 0.0009988795360325836,
+      "loss": 2.3535,
+      "step": 380
+    },
+    {
+      "epoch": 0.12100259291270528,
+      "grad_norm": 0.37607526779174805,
+      "learning_rate": 0.0009986884090762266,
+      "loss": 2.2953,
+      "step": 385
+    },
+    {
+      "epoch": 0.12257405515832483,
+      "grad_norm": 0.27437832951545715,
+      "learning_rate": 0.000998482260087996,
+      "loss": 2.2992,
+      "step": 390
+    },
+    {
+      "epoch": 0.12414551740394437,
+      "grad_norm": 1.2427669763565063,
+      "learning_rate": 0.0009982610952777348,
+      "loss": 2.23,
+      "step": 395
+    },
+    {
+      "epoch": 0.1257169796495639,
+      "grad_norm": 0.2712255120277405,
+      "learning_rate": 0.0009980249213076085,
+      "loss": 2.2312,
+      "step": 400
+    },
+    {
+      "epoch": 0.12728844189518346,
+      "grad_norm": 0.3325655162334442,
+      "learning_rate": 0.0009977737452919052,
+      "loss": 2.24,
+      "step": 405
+    },
+    {
+      "epoch": 0.128859904140803,
+      "grad_norm": 0.3058350384235382,
+      "learning_rate": 0.0009975075747968203,
+      "loss": 2.2233,
+      "step": 410
+    },
+    {
+      "epoch": 0.13043136638642255,
+      "grad_norm": 0.46767985820770264,
+      "learning_rate": 0.00099722641784023,
+      "loss": 2.2153,
+      "step": 415
+    },
+    {
+      "epoch": 0.1320028286320421,
+      "grad_norm": 0.27317899465560913,
+      "learning_rate": 0.0009969302828914477,
+      "loss": 2.2048,
+      "step": 420
+    },
+    {
+      "epoch": 0.13357429087766168,
+      "grad_norm": 0.49403005838394165,
+      "learning_rate": 0.0009966191788709714,
+      "loss": 2.1963,
+      "step": 425
+    },
+    {
+      "epoch": 0.13514575312328123,
+      "grad_norm": 0.28752318024635315,
+      "learning_rate": 0.0009962931151502128,
+      "loss": 2.2472,
+      "step": 430
+    },
+    {
+      "epoch": 0.13671721536890077,
+      "grad_norm": 0.2463676631450653,
+      "learning_rate": 0.000995952101551216,
+      "loss": 2.1985,
+      "step": 435
+    },
+    {
+      "epoch": 0.13828867761452032,
+      "grad_norm": 0.24134370684623718,
+      "learning_rate": 0.000995596148346362,
+      "loss": 2.2209,
+      "step": 440
+    },
+    {
+      "epoch": 0.13986013986013987,
+      "grad_norm": 0.28822633624076843,
+      "learning_rate": 0.000995225266258058,
+      "loss": 2.1771,
+      "step": 445
+    },
+    {
+      "epoch": 0.14143160210575942,
+      "grad_norm": 0.35365989804267883,
+      "learning_rate": 0.0009948394664584155,
+      "loss": 2.1384,
+      "step": 450
+    },
+    {
+      "epoch": 0.14300306435137897,
+      "grad_norm": 0.25629669427871704,
+      "learning_rate": 0.0009944387605689139,
+      "loss": 2.1782,
+      "step": 455
+    },
+    {
+      "epoch": 0.14457452659699851,
+      "grad_norm": 0.3197285532951355,
+      "learning_rate": 0.0009940231606600494,
+      "loss": 2.1927,
+      "step": 460
+    },
+    {
+      "epoch": 0.14614598884261806,
+      "grad_norm": 0.2365492731332779,
+      "learning_rate": 0.0009935926792509723,
+      "loss": 2.1824,
+      "step": 465
+    },
+    {
+      "epoch": 0.1477174510882376,
+      "grad_norm": 0.3567672073841095,
+      "learning_rate": 0.000993147329309109,
+      "loss": 2.1744,
+      "step": 470
+    },
+    {
+      "epoch": 0.14928891333385716,
+      "grad_norm": 0.28331291675567627,
+      "learning_rate": 0.0009926871242497731,
+      "loss": 2.1513,
+      "step": 475
+    },
+    {
+      "epoch": 0.1508603755794767,
+      "grad_norm": 0.2945978343486786,
+      "learning_rate": 0.0009922120779357587,
+      "loss": 2.1584,
+      "step": 480
+    },
+    {
+      "epoch": 0.15243183782509626,
+      "grad_norm": 0.2360486388206482,
+      "learning_rate": 0.0009917222046769252,
+      "loss": 2.1771,
+      "step": 485
+    },
+    {
+      "epoch": 0.1540033000707158,
+      "grad_norm": 0.26957565546035767,
+      "learning_rate": 0.0009912175192297648,
+      "loss": 2.1623,
+      "step": 490
+    },
+    {
+      "epoch": 0.15557476231633535,
+      "grad_norm": 0.2816142141819,
+      "learning_rate": 0.0009906980367969589,
+      "loss": 2.1792,
+      "step": 495
+    },
+    {
+      "epoch": 0.1571462245619549,
+      "grad_norm": 0.23650062084197998,
+      "learning_rate": 0.0009901637730269192,
+      "loss": 2.1513,
+      "step": 500
+    },
+    {
+      "epoch": 0.15871768680757445,
+      "grad_norm": 0.20437929034233093,
+      "learning_rate": 0.0009896147440133173,
+      "loss": 2.1128,
+      "step": 505
+    },
+    {
+      "epoch": 0.160289149053194,
+      "grad_norm": 0.4117949903011322,
+      "learning_rate": 0.0009890509662945992,
+      "loss": 2.1282,
+      "step": 510
+    },
+    {
+      "epoch": 0.16186061129881354,
+      "grad_norm": 0.2904440462589264,
+      "learning_rate": 0.0009884724568534873,
+      "loss": 2.1036,
+      "step": 515
+    },
+    {
+      "epoch": 0.1634320735444331,
+      "grad_norm": 0.30711230635643005,
+      "learning_rate": 0.000987879233116469,
+      "loss": 2.1463,
+      "step": 520
+    },
+    {
+      "epoch": 0.16500353579005264,
+      "grad_norm": 0.28438258171081543,
+      "learning_rate": 0.0009872713129532717,
+      "loss": 2.0971,
+      "step": 525
+    },
+    {
+      "epoch": 0.1665749980356722,
+      "grad_norm": 0.2756672203540802,
+      "learning_rate": 0.000986648714676324,
+      "loss": 2.1291,
+      "step": 530
+    },
+    {
+      "epoch": 0.16814646028129174,
+      "grad_norm": 0.23554356396198273,
+      "learning_rate": 0.0009860114570402056,
+      "loss": 2.0797,
+      "step": 535
+    },
+    {
+      "epoch": 0.16971792252691129,
+      "grad_norm": 0.3189121186733246,
+      "learning_rate": 0.0009853595592410798,
+      "loss": 2.1122,
+      "step": 540
+    },
+    {
+      "epoch": 0.17128938477253083,
+      "grad_norm": 0.2167348712682724,
+      "learning_rate": 0.000984693040916118,
+      "loss": 2.1103,
+      "step": 545
+    },
+    {
+      "epoch": 0.17286084701815038,
+      "grad_norm": 0.3020518124103546,
+      "learning_rate": 0.0009840119221429062,
+      "loss": 2.1314,
+      "step": 550
+    },
+    {
+      "epoch": 0.17443230926376993,
+      "grad_norm": 0.2934524118900299,
+      "learning_rate": 0.0009833162234388414,
+      "loss": 2.1118,
+      "step": 555
+    },
+    {
+      "epoch": 0.17600377150938948,
+      "grad_norm": 0.4206439256668091,
+      "learning_rate": 0.0009826059657605123,
+      "loss": 2.0729,
+      "step": 560
+    },
+    {
+      "epoch": 0.17757523375500903,
+      "grad_norm": 0.2568371593952179,
+      "learning_rate": 0.0009818811705030695,
+      "loss": 2.0941,
+      "step": 565
+    },
+    {
+      "epoch": 0.17914669600062857,
+      "grad_norm": 0.3140527009963989,
+      "learning_rate": 0.00098114185949958,
+      "loss": 2.0994,
+      "step": 570
+    },
+    {
+      "epoch": 0.18071815824624812,
+      "grad_norm": 0.2604500651359558,
+      "learning_rate": 0.0009803880550203698,
+      "loss": 2.1462,
+      "step": 575
+    },
+    {
+      "epoch": 0.18228962049186767,
+      "grad_norm": 0.3531113266944885,
+      "learning_rate": 0.0009796197797723532,
+      "loss": 2.0743,
+      "step": 580
+    },
+    {
+      "epoch": 0.18386108273748722,
+      "grad_norm": 0.24229241907596588,
+      "learning_rate": 0.0009788370568983484,
+      "loss": 2.0614,
+      "step": 585
+    },
+    {
+      "epoch": 0.18543254498310677,
+      "grad_norm": 0.43031787872314453,
+      "learning_rate": 0.000978039909976381,
+      "loss": 2.1287,
+      "step": 590
+    },
+    {
+      "epoch": 0.18700400722872634,
+      "grad_norm": 0.210410937666893,
+      "learning_rate": 0.0009772283630189726,
+      "loss": 2.0234,
+      "step": 595
+    },
+    {
+      "epoch": 0.1885754694743459,
+      "grad_norm": 0.28944680094718933,
+      "learning_rate": 0.0009764024404724194,
+      "loss": 2.0951,
+      "step": 600
+    },
+    {
+      "epoch": 0.19014693171996544,
+      "grad_norm": 0.2910195589065552,
+      "learning_rate": 0.0009755621672160539,
+      "loss": 2.077,
+      "step": 605
+    },
+    {
+      "epoch": 0.191718393965585,
+      "grad_norm": 0.27437564730644226,
+      "learning_rate": 0.0009747075685614961,
+      "loss": 2.0639,
+      "step": 610
+    },
+    {
+      "epoch": 0.19328985621120454,
+      "grad_norm": 0.26965805888175964,
+      "learning_rate": 0.0009738386702518917,
+      "loss": 2.0205,
+      "step": 615
+    },
+    {
+      "epoch": 0.19486131845682408,
+      "grad_norm": 0.22133469581604004,
+      "learning_rate": 0.0009729554984611357,
+      "loss": 2.0871,
+      "step": 620
+    },
+    {
+      "epoch": 0.19643278070244363,
+      "grad_norm": 0.26669520139694214,
+      "learning_rate": 0.0009720580797930845,
+      "loss": 2.1095,
+      "step": 625
+    },
+    {
+      "epoch": 0.19800424294806318,
+      "grad_norm": 0.22317343950271606,
+      "learning_rate": 0.0009711464412807542,
+      "loss": 2.0449,
+      "step": 630
+    },
+    {
+      "epoch": 0.19957570519368273,
+      "grad_norm": 0.4366435706615448,
+      "learning_rate": 0.0009702206103855065,
+      "loss": 2.0923,
+      "step": 635
+    },
+    {
+      "epoch": 0.20114716743930228,
+      "grad_norm": 0.29953792691230774,
+      "learning_rate": 0.0009692806149962215,
+      "loss": 2.0278,
+      "step": 640
+    },
+    {
+      "epoch": 0.20271862968492182,
+      "grad_norm": 0.3391275405883789,
+      "learning_rate": 0.0009683264834284575,
+      "loss": 2.0462,
+      "step": 645
+    },
+    {
+      "epoch": 0.20429009193054137,
+      "grad_norm": 0.3684813976287842,
+      "learning_rate": 0.0009673582444235978,
+      "loss": 2.0406,
+      "step": 650
+    },
+    {
+      "epoch": 0.20586155417616092,
+      "grad_norm": 0.3173034191131592,
+      "learning_rate": 0.0009663759271479858,
+      "loss": 1.9821,
+      "step": 655
+    },
+    {
+      "epoch": 0.20743301642178047,
+      "grad_norm": 0.28240126371383667,
+      "learning_rate": 0.0009653795611920448,
+      "loss": 2.0112,
+      "step": 660
+    },
+    {
+      "epoch": 0.20900447866740002,
+      "grad_norm": 0.31004390120506287,
+      "learning_rate": 0.000964369176569389,
+      "loss": 2.0155,
+      "step": 665
+    },
+    {
+      "epoch": 0.21057594091301957,
+      "grad_norm": 0.29802826046943665,
+      "learning_rate": 0.0009633448037159167,
+      "loss": 2.0641,
+      "step": 670
+    },
+    {
+      "epoch": 0.21214740315863911,
+      "grad_norm": 0.2916626036167145,
+      "learning_rate": 0.0009623064734888958,
+      "loss": 1.9948,
+      "step": 675
+    },
+    {
+      "epoch": 0.21371886540425866,
+      "grad_norm": 0.24406270682811737,
+      "learning_rate": 0.0009612542171660328,
+      "loss": 1.9714,
+      "step": 680
+    },
+    {
+      "epoch": 0.2152903276498782,
+      "grad_norm": 0.31989896297454834,
+      "learning_rate": 0.0009601880664445312,
+      "loss": 2.0075,
+      "step": 685
+    },
+    {
+      "epoch": 0.21686178989549776,
+      "grad_norm": 0.16441141068935394,
+      "learning_rate": 0.000959108053440137,
+      "loss": 1.9964,
+      "step": 690
+    },
+    {
+      "epoch": 0.2184332521411173,
+      "grad_norm": 0.27429190278053284,
+      "learning_rate": 0.0009580142106861702,
+      "loss": 2.013,
+      "step": 695
+    },
+    {
+      "epoch": 0.22000471438673685,
+      "grad_norm": 0.21225886046886444,
+      "learning_rate": 0.0009569065711325461,
+      "loss": 2.0092,
+      "step": 700
+    },
+    {
+      "epoch": 0.2215761766323564,
+      "grad_norm": 0.24281561374664307,
+      "learning_rate": 0.0009557851681447816,
+      "loss": 2.0172,
+      "step": 705
+    },
+    {
+      "epoch": 0.22314763887797595,
+      "grad_norm": 0.20721524953842163,
+      "learning_rate": 0.0009546500355029912,
+      "loss": 2.0265,
+      "step": 710
+    },
+    {
+      "epoch": 0.2247191011235955,
+      "grad_norm": 0.268388032913208,
+      "learning_rate": 0.0009535012074008687,
+      "loss": 2.0329,
+      "step": 715
+    },
+    {
+      "epoch": 0.22629056336921505,
+      "grad_norm": 0.25884491205215454,
+      "learning_rate": 0.0009523387184446569,
+      "loss": 1.9673,
+      "step": 720
+    },
+    {
+      "epoch": 0.2278620256148346,
+      "grad_norm": 0.26054859161376953,
+      "learning_rate": 0.0009511626036521064,
+      "loss": 1.9775,
+      "step": 725
+    },
+    {
+      "epoch": 0.22943348786045414,
+      "grad_norm": 0.23918680846691132,
+      "learning_rate": 0.0009499728984514194,
+      "loss": 1.9753,
+      "step": 730
+    },
+    {
+      "epoch": 0.2310049501060737,
+      "grad_norm": 0.18698784708976746,
+      "learning_rate": 0.0009487696386801834,
+      "loss": 2.0547,
+      "step": 735
+    },
+    {
+      "epoch": 0.23257641235169324,
+      "grad_norm": 0.23608291149139404,
+      "learning_rate": 0.0009475528605842913,
+      "loss": 1.9701,
+      "step": 740
+    },
+    {
+      "epoch": 0.2341478745973128,
+      "grad_norm": 0.20181405544281006,
+      "learning_rate": 0.0009463226008168493,
+      "loss": 1.9994,
+      "step": 745
+    },
+    {
+      "epoch": 0.23571933684293234,
+      "grad_norm": 0.421236515045166,
+      "learning_rate": 0.0009450788964370737,
+      "loss": 1.9943,
+      "step": 750
+    },
+    {
+      "epoch": 0.23729079908855188,
+      "grad_norm": 0.2977412939071655,
+      "learning_rate": 0.0009438217849091732,
+      "loss": 1.998,
+      "step": 755
+    },
+    {
+      "epoch": 0.23886226133417143,
+      "grad_norm": 0.3429923355579376,
+      "learning_rate": 0.0009425513041012219,
+      "loss": 2.0098,
+      "step": 760
+    },
+    {
+      "epoch": 0.240433723579791,
+      "grad_norm": 0.28545793890953064,
+      "learning_rate": 0.0009412674922840173,
+      "loss": 1.9684,
+      "step": 765
+    },
+    {
+      "epoch": 0.24200518582541056,
+      "grad_norm": 0.1793077141046524,
+      "learning_rate": 0.0009399703881299281,
+      "loss": 2.0074,
+      "step": 770
+    },
+    {
+      "epoch": 0.2435766480710301,
+      "grad_norm": 0.280935674905777,
+      "learning_rate": 0.0009386600307117292,
+      "loss": 2.0081,
+      "step": 775
+    },
+    {
+      "epoch": 0.24514811031664965,
+      "grad_norm": 0.30971866846084595,
+      "learning_rate": 0.0009373364595014245,
+      "loss": 1.9568,
+      "step": 780
+    },
+    {
+      "epoch": 0.2467195725622692,
+      "grad_norm": 0.26543349027633667,
+      "learning_rate": 0.0009359997143690587,
+      "loss": 1.9926,
+      "step": 785
+    },
+    {
+      "epoch": 0.24829103480788875,
+      "grad_norm": 0.27692025899887085,
+      "learning_rate": 0.0009346498355815143,
+      "loss": 1.9465,
+      "step": 790
+    },
+    {
+      "epoch": 0.2498624970535083,
+      "grad_norm": 0.2559816539287567,
+      "learning_rate": 0.0009332868638013016,
+      "loss": 1.9473,
+      "step": 795
+    },
+    {
+      "epoch": 0.2514339592991278,
+      "grad_norm": 0.330941766500473,
+      "learning_rate": 0.0009319108400853309,
+      "loss": 1.9892,
+      "step": 800
+    },
+    {
+      "epoch": 0.25300542154474737,
+      "grad_norm": 0.33171382546424866,
+      "learning_rate": 0.0009305218058836777,
+      "loss": 1.9819,
+      "step": 805
+    },
+    {
+      "epoch": 0.2545768837903669,
+      "grad_norm": 0.49775874614715576,
+      "learning_rate": 0.0009291198030383335,
+      "loss": 1.9979,
+      "step": 810
+    },
+    {
+      "epoch": 0.25614834603598646,
+      "grad_norm": 0.4397469758987427,
+      "learning_rate": 0.0009277048737819445,
+      "loss": 1.9923,
+      "step": 815
+    },
+    {
+      "epoch": 0.257719808281606,
+      "grad_norm": 0.38636040687561035,
+      "learning_rate": 0.0009262770607365412,
+      "loss": 1.989,
+      "step": 820
+    },
+    {
+      "epoch": 0.25929127052722556,
+      "grad_norm": 0.19002190232276917,
+      "learning_rate": 0.0009248364069122531,
+      "loss": 1.9449,
+      "step": 825
+    },
+    {
+      "epoch": 0.2608627327728451,
+      "grad_norm": 0.21801802515983582,
+      "learning_rate": 0.0009233829557060136,
+      "loss": 1.9449,
+      "step": 830
+    },
+    {
+      "epoch": 0.26243419501846466,
+      "grad_norm": 0.2232774794101715,
+      "learning_rate": 0.0009219167509002526,
+      "loss": 2.0174,
+      "step": 835
+    },
+    {
+      "epoch": 0.2640056572640842,
+      "grad_norm": 0.21366550028324127,
+      "learning_rate": 0.0009204378366615778,
+      "loss": 1.9458,
+      "step": 840
+    },
+    {
+      "epoch": 0.26557711950970375,
+      "grad_norm": 0.22979532182216644,
+      "learning_rate": 0.0009189462575394443,
+      "loss": 1.968,
+      "step": 845
+    },
+    {
+      "epoch": 0.26714858175532336,
+      "grad_norm": 0.3232949376106262,
+      "learning_rate": 0.0009174420584648122,
+      "loss": 1.9344,
+      "step": 850
+    },
+    {
+      "epoch": 0.2687200440009429,
+      "grad_norm": 0.3545362055301666,
+      "learning_rate": 0.0009159252847487938,
+      "loss": 1.9558,
+      "step": 855
+    },
+    {
+      "epoch": 0.27029150624656245,
+      "grad_norm": 0.25105616450309753,
+      "learning_rate": 0.0009143959820812882,
+      "loss": 1.9726,
+      "step": 860
+    },
+    {
+      "epoch": 0.271862968492182,
+      "grad_norm": 0.37967053055763245,
+      "learning_rate": 0.0009128541965296051,
+      "loss": 1.9593,
+      "step": 865
+    },
+    {
+      "epoch": 0.27343443073780155,
+      "grad_norm": 0.47272658348083496,
+      "learning_rate": 0.0009112999745370774,
+      "loss": 1.9306,
+      "step": 870
+    },
+    {
+      "epoch": 0.2750058929834211,
+      "grad_norm": 0.40520063042640686,
+      "learning_rate": 0.0009097333629216616,
+      "loss": 2.0052,
+      "step": 875
+    },
+    {
+      "epoch": 0.27657735522904064,
+      "grad_norm": 0.16213001310825348,
+      "learning_rate": 0.000908154408874528,
+      "loss": 1.9685,
+      "step": 880
+    },
+    {
+      "epoch": 0.2781488174746602,
+      "grad_norm": 0.2853317856788635,
+      "learning_rate": 0.000906563159958639,
+      "loss": 1.972,
+      "step": 885
+    },
+    {
+      "epoch": 0.27972027972027974,
+      "grad_norm": 0.3446250259876251,
+      "learning_rate": 0.0009049596641073161,
+      "loss": 1.9114,
+      "step": 890
+    },
+    {
+      "epoch": 0.2812917419658993,
+      "grad_norm": 0.19377753138542175,
+      "learning_rate": 0.0009033439696227966,
+      "loss": 1.9119,
+      "step": 895
+    },
+    {
+      "epoch": 0.28286320421151884,
+      "grad_norm": 0.249973326921463,
+      "learning_rate": 0.0009017161251747779,
+      "loss": 1.9657,
+      "step": 900
+    },
+    {
+      "epoch": 0.2844346664571384,
+      "grad_norm": 0.26847654581069946,
+      "learning_rate": 0.0009000761797989521,
+      "loss": 1.9543,
+      "step": 905
+    },
+    {
+      "epoch": 0.28600612870275793,
+      "grad_norm": 0.23439465463161469,
+      "learning_rate": 0.0008984241828955281,
+      "loss": 1.9104,
+      "step": 910
+    },
+    {
+      "epoch": 0.2875775909483775,
+      "grad_norm": 0.37913578748703003,
+      "learning_rate": 0.0008967601842277444,
+      "loss": 1.9149,
+      "step": 915
+    },
+    {
+      "epoch": 0.28914905319399703,
+      "grad_norm": 0.2542668581008911,
+      "learning_rate": 0.0008950842339203695,
+      "loss": 1.9682,
+      "step": 920
+    },
+    {
+      "epoch": 0.2907205154396166,
+      "grad_norm": 0.3627631962299347,
+      "learning_rate": 0.0008933963824581919,
+      "loss": 1.9878,
+      "step": 925
+    },
+    {
+      "epoch": 0.2922919776852361,
+      "grad_norm": 0.25258249044418335,
+      "learning_rate": 0.0008916966806844996,
+      "loss": 1.9425,
+      "step": 930
+    },
+    {
+      "epoch": 0.2938634399308557,
+      "grad_norm": 0.502772867679596,
+      "learning_rate": 0.0008899851797995489,
+      "loss": 1.956,
+      "step": 935
+    },
+    {
+      "epoch": 0.2954349021764752,
+      "grad_norm": 0.3263178765773773,
+      "learning_rate": 0.0008882619313590213,
+      "loss": 1.9896,
+      "step": 940
+    },
+    {
+      "epoch": 0.29700636442209477,
+      "grad_norm": 0.24707098305225372,
+      "learning_rate": 0.0008865269872724708,
+      "loss": 1.901,
+      "step": 945
+    },
+    {
+      "epoch": 0.2985778266677143,
+      "grad_norm": 0.26257702708244324,
+      "learning_rate": 0.000884780399801761,
+      "loss": 1.957,
+      "step": 950
+    },
+    {
+      "epoch": 0.30014928891333387,
+      "grad_norm": 0.2532920241355896,
+      "learning_rate": 0.000883022221559489,
+      "loss": 1.9059,
+      "step": 955
+    },
+    {
+      "epoch": 0.3017207511589534,
+      "grad_norm": 0.26241347193717957,
+      "learning_rate": 0.0008812525055074028,
+      "loss": 1.938,
+      "step": 960
+    },
+    {
+      "epoch": 0.30329221340457296,
+      "grad_norm": 0.28568172454833984,
+      "learning_rate": 0.000879471304954804,
+      "loss": 1.8915,
+      "step": 965
+    },
+    {
+      "epoch": 0.3048636756501925,
+      "grad_norm": 0.23444630205631256,
+      "learning_rate": 0.0008776786735569431,
+      "loss": 1.9095,
+      "step": 970
+    },
+    {
+      "epoch": 0.30643513789581206,
+      "grad_norm": 0.22640594840049744,
+      "learning_rate": 0.0008758746653134029,
+      "loss": 1.8879,
+      "step": 975
+    },
+    {
+      "epoch": 0.3080066001414316,
+      "grad_norm": 0.5037649869918823,
+      "learning_rate": 0.0008740593345664716,
+      "loss": 1.889,
+      "step": 980
+    },
+    {
+      "epoch": 0.30957806238705116,
+      "grad_norm": 0.18424461781978607,
+      "learning_rate": 0.0008722327359995064,
+      "loss": 1.8806,
+      "step": 985
+    },
+    {
+      "epoch": 0.3111495246326707,
+      "grad_norm": 0.2199254333972931,
+      "learning_rate": 0.0008703949246352858,
+      "loss": 1.8767,
+      "step": 990
+    },
+    {
+      "epoch": 0.31272098687829025,
+      "grad_norm": 0.26489025354385376,
+      "learning_rate": 0.0008685459558343523,
+      "loss": 1.8609,
+      "step": 995
+    },
+    {
+      "epoch": 0.3142924491239098,
+      "grad_norm": 0.164434015750885,
+      "learning_rate": 0.000866685885293345,
+      "loss": 1.8941,
+      "step": 1000
+    },
+    {
+      "epoch": 0.31586391136952935,
+      "grad_norm": 0.21282632648944855,
+      "learning_rate": 0.0008648147690433212,
+      "loss": 1.8818,
+      "step": 1005
+    },
+    {
+      "epoch": 0.3174353736151489,
+      "grad_norm": 0.21168777346611023,
+      "learning_rate": 0.0008629326634480697,
+      "loss": 1.917,
+      "step": 1010
+    },
+    {
+      "epoch": 0.31900683586076845,
+      "grad_norm": 0.2421833872795105,
+      "learning_rate": 0.0008610396252024113,
+      "loss": 1.9548,
+      "step": 1015
+    },
+    {
+      "epoch": 0.320578298106388,
+      "grad_norm": 0.17949452996253967,
+      "learning_rate": 0.0008591357113304927,
+      "loss": 1.907,
+      "step": 1020
+    },
+    {
+      "epoch": 0.32214976035200754,
+      "grad_norm": 0.25564640760421753,
+      "learning_rate": 0.0008572209791840677,
+      "loss": 1.9133,
+      "step": 1025
+    },
+    {
+      "epoch": 0.3237212225976271,
+      "grad_norm": 0.21499499678611755,
+      "learning_rate": 0.0008552954864407698,
+      "loss": 1.8724,
+      "step": 1030
+    },
+    {
+      "epoch": 0.32529268484324664,
+      "grad_norm": 0.286145955324173,
+      "learning_rate": 0.000853359291102375,
+      "loss": 1.9174,
+      "step": 1035
+    },
+    {
+      "epoch": 0.3268641470888662,
+      "grad_norm": 0.216496080160141,
+      "learning_rate": 0.0008514124514930544,
+      "loss": 1.9212,
+      "step": 1040
+    },
+    {
+      "epoch": 0.32843560933448573,
+      "grad_norm": 0.20780375599861145,
+      "learning_rate": 0.0008494550262576173,
+      "loss": 1.9042,
+      "step": 1045
+    },
+    {
+      "epoch": 0.3300070715801053,
+      "grad_norm": 0.19447444379329681,
+      "learning_rate": 0.0008474870743597448,
+      "loss": 1.9154,
+      "step": 1050
+    },
+    {
+      "epoch": 0.33157853382572483,
+      "grad_norm": 0.2113378793001175,
+      "learning_rate": 0.0008455086550802132,
+      "loss": 1.9357,
+      "step": 1055
+    },
+    {
+      "epoch": 0.3331499960713444,
+      "grad_norm": 0.2649160921573639,
+      "learning_rate": 0.0008435198280151091,
+      "loss": 1.9016,
+      "step": 1060
+    },
+    {
+      "epoch": 0.3347214583169639,
+      "grad_norm": 0.37894386053085327,
+      "learning_rate": 0.0008415206530740331,
+      "loss": 1.9078,
+      "step": 1065
+    },
+    {
+      "epoch": 0.3362929205625835,
+      "grad_norm": 0.19303838908672333,
+      "learning_rate": 0.0008395111904782964,
+      "loss": 1.9018,
+      "step": 1070
+    },
+    {
+      "epoch": 0.337864382808203,
+      "grad_norm": 0.25691303610801697,
+      "learning_rate": 0.0008374915007591053,
+      "loss": 1.9319,
+      "step": 1075
+    },
+    {
+      "epoch": 0.33943584505382257,
+      "grad_norm": 0.3956037759780884,
+      "learning_rate": 0.0008354616447557392,
+      "loss": 1.9003,
+      "step": 1080
+    },
+    {
+      "epoch": 0.3410073072994421,
+      "grad_norm": 0.29999446868896484,
+      "learning_rate": 0.000833421683613717,
+      "loss": 1.9377,
+      "step": 1085
+    },
+    {
+      "epoch": 0.34257876954506167,
+      "grad_norm": 0.23759868741035461,
+      "learning_rate": 0.0008313716787829558,
+      "loss": 1.8833,
+      "step": 1090
+    },
+    {
+      "epoch": 0.3441502317906812,
+      "grad_norm": 0.33085566759109497,
+      "learning_rate": 0.000829311692015919,
+      "loss": 1.936,
+      "step": 1095
+    },
+    {
+      "epoch": 0.34572169403630076,
+      "grad_norm": 0.32732662558555603,
+      "learning_rate": 0.0008272417853657571,
+      "loss": 1.8785,
+      "step": 1100
+    },
+    {
+      "epoch": 0.3472931562819203,
+      "grad_norm": 0.23105382919311523,
+      "learning_rate": 0.0008251620211844383,
+      "loss": 1.9553,
+      "step": 1105
+    },
+    {
+      "epoch": 0.34886461852753986,
+      "grad_norm": 0.34907397627830505,
+      "learning_rate": 0.0008230724621208693,
+      "loss": 1.8749,
+      "step": 1110
+    },
+    {
+      "epoch": 0.3504360807731594,
+      "grad_norm": 0.2972997725009918,
+      "learning_rate": 0.0008209731711190099,
+      "loss": 1.8532,
+      "step": 1115
+    },
+    {
+      "epoch": 0.35200754301877896,
+      "grad_norm": 0.21005867421627045,
+      "learning_rate": 0.0008188642114159747,
+      "loss": 1.8804,
+      "step": 1120
+    },
+    {
+      "epoch": 0.3535790052643985,
+      "grad_norm": 0.1943141371011734,
+      "learning_rate": 0.0008167456465401299,
+      "loss": 1.8385,
+      "step": 1125
+    },
+    {
+      "epoch": 0.35515046751001805,
+      "grad_norm": 0.2459149956703186,
+      "learning_rate": 0.0008146175403091795,
+      "loss": 1.8629,
+      "step": 1130
+    },
+    {
+      "epoch": 0.3567219297556376,
+      "grad_norm": 0.2223600149154663,
+      "learning_rate": 0.0008124799568282418,
+      "loss": 1.8939,
+      "step": 1135
+    },
+    {
+      "epoch": 0.35829339200125715,
+      "grad_norm": 0.3591344654560089,
+      "learning_rate": 0.0008103329604879195,
+      "loss": 1.835,
+      "step": 1140
+    },
+    {
+      "epoch": 0.3598648542468767,
+      "grad_norm": 0.33568888902664185,
+      "learning_rate": 0.0008081766159623596,
+      "loss": 1.9111,
+      "step": 1145
+    },
+    {
+      "epoch": 0.36143631649249625,
+      "grad_norm": 0.19357386231422424,
+      "learning_rate": 0.0008060109882073055,
+      "loss": 1.8748,
+      "step": 1150
+    },
+    {
+      "epoch": 0.3630077787381158,
+      "grad_norm": 0.20973102748394012,
+      "learning_rate": 0.0008038361424581392,
+      "loss": 1.8755,
+      "step": 1155
+    },
+    {
+      "epoch": 0.36457924098373534,
+      "grad_norm": 0.19260567426681519,
+      "learning_rate": 0.000801652144227918,
+      "loss": 1.8516,
+      "step": 1160
+    },
+    {
+      "epoch": 0.3661507032293549,
+      "grad_norm": 0.21391819417476654,
+      "learning_rate": 0.0007994590593054001,
+      "loss": 1.8751,
+      "step": 1165
+    },
+    {
+      "epoch": 0.36772216547497444,
+      "grad_norm": 0.23929333686828613,
+      "learning_rate": 0.0007972569537530623,
+      "loss": 1.8371,
+      "step": 1170
+    },
+    {
+      "epoch": 0.369293627720594,
+      "grad_norm": 0.2769240438938141,
+      "learning_rate": 0.0007950458939051108,
+      "loss": 1.8872,
+      "step": 1175
+    },
+    {
+      "epoch": 0.37086508996621353,
+      "grad_norm": 0.22298012673854828,
+      "learning_rate": 0.000792825946365483,
+      "loss": 1.8764,
+      "step": 1180
+    },
+    {
+      "epoch": 0.37243655221183314,
+      "grad_norm": 0.1722257286310196,
+      "learning_rate": 0.000790597178005841,
+      "loss": 1.802,
+      "step": 1185
+    },
+    {
+      "epoch": 0.3740080144574527,
+      "grad_norm": 0.29332205653190613,
+      "learning_rate": 0.0007883596559635567,
+      "loss": 1.8884,
+      "step": 1190
+    },
+    {
+      "epoch": 0.37557947670307223,
+      "grad_norm": 0.2480962872505188,
+      "learning_rate": 0.0007861134476396901,
+      "loss": 1.8287,
+      "step": 1195
+    },
+    {
+      "epoch": 0.3771509389486918,
+      "grad_norm": 0.22237135469913483,
+      "learning_rate": 0.0007838586206969593,
+      "loss": 1.8414,
+      "step": 1200
+    },
+    {
+      "epoch": 0.37872240119431133,
+      "grad_norm": 0.2007599025964737,
+      "learning_rate": 0.0007815952430577014,
+      "loss": 1.9258,
+      "step": 1205
+    },
+    {
+      "epoch": 0.3802938634399309,
+      "grad_norm": 0.2816416323184967,
+      "learning_rate": 0.0007793233829018263,
+      "loss": 1.8544,
+      "step": 1210
+    },
+    {
+      "epoch": 0.3818653256855504,
+      "grad_norm": 0.1743992269039154,
+      "learning_rate": 0.0007770431086647642,
+      "loss": 1.8735,
+      "step": 1215
+    },
+    {
+      "epoch": 0.38343678793117,
+      "grad_norm": 0.20147810876369476,
+      "learning_rate": 0.000774754489035403,
+      "loss": 1.8541,
+      "step": 1220
+    },
+    {
+      "epoch": 0.3850082501767895,
+      "grad_norm": 0.19584012031555176,
+      "learning_rate": 0.0007724575929540197,
+      "loss": 1.821,
+      "step": 1225
+    },
+    {
+      "epoch": 0.38657971242240907,
+      "grad_norm": 0.32170602679252625,
+      "learning_rate": 0.0007701524896102037,
+      "loss": 1.8277,
+      "step": 1230
+    },
+    {
+      "epoch": 0.3881511746680286,
+      "grad_norm": 0.16368649899959564,
+      "learning_rate": 0.0007678392484407717,
+      "loss": 1.8317,
+      "step": 1235
+    },
+    {
+      "epoch": 0.38972263691364817,
+      "grad_norm": 0.17887534201145172,
+      "learning_rate": 0.0007655179391276775,
+      "loss": 1.8234,
+      "step": 1240
+    },
+    {
+      "epoch": 0.3912940991592677,
+      "grad_norm": 0.2335747331380844,
+      "learning_rate": 0.0007631886315959121,
+      "loss": 1.8789,
+      "step": 1245
+    },
+    {
+      "epoch": 0.39286556140488726,
+      "grad_norm": 0.16466860473155975,
+      "learning_rate": 0.0007608513960113975,
+      "loss": 1.8777,
+      "step": 1250
+    },
+    {
+      "epoch": 0.3944370236505068,
+      "grad_norm": 0.1995929628610611,
+      "learning_rate": 0.000758506302778873,
+      "loss": 1.8159,
+      "step": 1255
+    },
+    {
+      "epoch": 0.39600848589612636,
+      "grad_norm": 0.28415942192077637,
+      "learning_rate": 0.0007561534225397744,
+      "loss": 1.8151,
+      "step": 1260
+    },
+    {
+      "epoch": 0.3975799481417459,
+      "grad_norm": 0.15427584946155548,
+      "learning_rate": 0.0007537928261701064,
+      "loss": 1.8402,
+      "step": 1265
+    },
+    {
+      "epoch": 0.39915141038736546,
+      "grad_norm": 0.211939737200737,
+      "learning_rate": 0.0007514245847783069,
+      "loss": 1.8554,
+      "step": 1270
+    },
+    {
+      "epoch": 0.400722872632985,
+      "grad_norm": 0.1703938990831375,
+      "learning_rate": 0.0007490487697031061,
+      "loss": 1.8369,
+      "step": 1275
+    },
+    {
+      "epoch": 0.40229433487860455,
+      "grad_norm": 0.19989508390426636,
+      "learning_rate": 0.0007466654525113761,
+      "loss": 1.8075,
+      "step": 1280
+    },
+    {
+      "epoch": 0.4038657971242241,
+      "grad_norm": 0.3463574945926666,
+      "learning_rate": 0.0007442747049959765,
+      "loss": 1.8606,
+      "step": 1285
+    },
+    {
+      "epoch": 0.40543725936984365,
+      "grad_norm": 0.2496725171804428,
+      "learning_rate": 0.0007418765991735908,
+      "loss": 1.8269,
+      "step": 1290
+    },
+    {
+      "epoch": 0.4070087216154632,
+      "grad_norm": 0.29229408502578735,
+      "learning_rate": 0.0007394712072825576,
+      "loss": 1.8034,
+      "step": 1295
+    },
+    {
+      "epoch": 0.40858018386108275,
+      "grad_norm": 0.2583361566066742,
+      "learning_rate": 0.0007370586017806941,
+      "loss": 1.8845,
+      "step": 1300
+    },
+    {
+      "epoch": 0.4101516461067023,
+      "grad_norm": 0.168927401304245,
+      "learning_rate": 0.0007346388553431141,
+      "loss": 1.8509,
+      "step": 1305
+    },
+    {
+      "epoch": 0.41172310835232184,
+      "grad_norm": 0.24876132607460022,
+      "learning_rate": 0.0007322120408600379,
+      "loss": 1.9044,
+      "step": 1310
+    },
+    {
+      "epoch": 0.4132945705979414,
+      "grad_norm": 0.16044846177101135,
+      "learning_rate": 0.0007297782314345972,
+      "loss": 1.8406,
+      "step": 1315
+    },
+    {
+      "epoch": 0.41486603284356094,
+      "grad_norm": 0.18333998322486877,
+      "learning_rate": 0.0007273375003806335,
+      "loss": 1.8755,
+      "step": 1320
+    },
+    {
+      "epoch": 0.4164374950891805,
+      "grad_norm": 0.2609647512435913,
+      "learning_rate": 0.0007248899212204883,
+      "loss": 1.8146,
+      "step": 1325
+    },
+    {
+      "epoch": 0.41800895733480004,
+      "grad_norm": 0.20326748490333557,
+      "learning_rate": 0.0007224355676827897,
+      "loss": 1.8173,
+      "step": 1330
+    },
+    {
+      "epoch": 0.4195804195804196,
+      "grad_norm": 0.32762306928634644,
+      "learning_rate": 0.0007199745137002305,
+      "loss": 1.892,
+      "step": 1335
+    },
+    {
+      "epoch": 0.42115188182603913,
+      "grad_norm": 0.25635045766830444,
+      "learning_rate": 0.000717506833407342,
+      "loss": 1.8104,
+      "step": 1340
+    },
+    {
+      "epoch": 0.4227233440716587,
+      "grad_norm": 0.29682666063308716,
+      "learning_rate": 0.0007150326011382603,
+      "loss": 1.8018,
+      "step": 1345
+    },
+    {
+      "epoch": 0.42429480631727823,
+      "grad_norm": 0.22666814923286438,
+      "learning_rate": 0.0007125518914244868,
+      "loss": 1.8358,
+      "step": 1350
+    },
+    {
+      "epoch": 0.4258662685628978,
+      "grad_norm": 0.26551052927970886,
+      "learning_rate": 0.000710064778992644,
+      "loss": 1.8332,
+      "step": 1355
+    },
+    {
+      "epoch": 0.4274377308085173,
+      "grad_norm": 0.16310109198093414,
+      "learning_rate": 0.0007075713387622236,
+      "loss": 1.7835,
+      "step": 1360
+    },
+    {
+      "epoch": 0.4290091930541369,
+      "grad_norm": 0.19358539581298828,
+      "learning_rate": 0.0007050716458433305,
+      "loss": 1.7956,
+      "step": 1365
+    },
+    {
+      "epoch": 0.4305806552997564,
+      "grad_norm": 0.16724836826324463,
+      "learning_rate": 0.0007025657755344197,
+      "loss": 1.874,
+      "step": 1370
+    },
+    {
+      "epoch": 0.43215211754537597,
+      "grad_norm": 0.2080153524875641,
+      "learning_rate": 0.0007000538033200279,
+      "loss": 1.8706,
+      "step": 1375
+    },
+    {
+      "epoch": 0.4337235797909955,
+      "grad_norm": 0.1837550699710846,
+      "learning_rate": 0.0006975358048685004,
+      "loss": 1.8576,
+      "step": 1380
+    },
+    {
+      "epoch": 0.43529504203661507,
+      "grad_norm": 0.26517677307128906,
+      "learning_rate": 0.0006950118560297112,
+      "loss": 1.8121,
+      "step": 1385
+    },
+    {
+      "epoch": 0.4368665042822346,
+      "grad_norm": 0.2400379180908203,
+      "learning_rate": 0.0006924820328327785,
+      "loss": 1.7908,
+      "step": 1390
+    },
+    {
+      "epoch": 0.43843796652785416,
+      "grad_norm": 0.1548314243555069,
+      "learning_rate": 0.0006899464114837739,
+      "loss": 1.8439,
+      "step": 1395
+    },
+    {
+      "epoch": 0.4400094287734737,
+      "grad_norm": 0.1790788173675537,
+      "learning_rate": 0.0006874050683634273,
+      "loss": 1.8511,
+      "step": 1400
+    },
+    {
+      "epoch": 0.44158089101909326,
+      "grad_norm": 0.2156343162059784,
+      "learning_rate": 0.0006848580800248262,
+      "loss": 1.8442,
+      "step": 1405
+    },
+    {
+      "epoch": 0.4431523532647128,
+      "grad_norm": 0.24280230700969696,
+      "learning_rate": 0.0006823055231911093,
+      "loss": 1.8091,
+      "step": 1410
+    },
+    {
+      "epoch": 0.44472381551033235,
+      "grad_norm": 0.18132703006267548,
+      "learning_rate": 0.0006797474747531558,
+      "loss": 1.8596,
+      "step": 1415
+    },
+    {
+      "epoch": 0.4462952777559519,
+      "grad_norm": 0.2499558925628662,
+      "learning_rate": 0.0006771840117672684,
+      "loss": 1.7982,
+      "step": 1420
+    },
+    {
+      "epoch": 0.44786674000157145,
+      "grad_norm": 0.14542344212532043,
+      "learning_rate": 0.0006746152114528531,
+      "loss": 1.8436,
+      "step": 1425
+    },
+    {
+      "epoch": 0.449438202247191,
+      "grad_norm": 0.23537123203277588,
+      "learning_rate": 0.0006720411511900927,
+      "loss": 1.8342,
+      "step": 1430
+    },
+    {
+      "epoch": 0.45100966449281055,
+      "grad_norm": 0.2291416972875595,
+      "learning_rate": 0.000669461908517616,
+      "loss": 1.815,
+      "step": 1435
+    },
+    {
+      "epoch": 0.4525811267384301,
+      "grad_norm": 0.17745240032672882,
+      "learning_rate": 0.0006668775611301611,
+      "loss": 1.807,
+      "step": 1440
+    },
+    {
+      "epoch": 0.45415258898404964,
+      "grad_norm": 0.2866009771823883,
+      "learning_rate": 0.0006642881868762368,
+      "loss": 1.8313,
+      "step": 1445
+    },
+    {
+      "epoch": 0.4557240512296692,
+      "grad_norm": 0.23607899248600006,
+      "learning_rate": 0.0006616938637557761,
+      "loss": 1.8308,
+      "step": 1450
+    },
+    {
+      "epoch": 0.45729551347528874,
+      "grad_norm": 0.25843170285224915,
+      "learning_rate": 0.0006590946699177875,
+      "loss": 1.8076,
+      "step": 1455
+    },
+    {
+      "epoch": 0.4588669757209083,
+      "grad_norm": 0.2201550155878067,
+      "learning_rate": 0.0006564906836580004,
+      "loss": 1.7999,
+      "step": 1460
+    },
+    {
+      "epoch": 0.46043843796652784,
+      "grad_norm": 0.2689286172389984,
+      "learning_rate": 0.0006538819834165061,
+      "loss": 1.8498,
+      "step": 1465
+    },
+    {
+      "epoch": 0.4620099002121474,
+      "grad_norm": 0.19224600493907928,
+      "learning_rate": 0.0006512686477753966,
+      "loss": 1.8525,
+      "step": 1470
+    },
+    {
+      "epoch": 0.46358136245776693,
+      "grad_norm": 0.18337461352348328,
+      "learning_rate": 0.0006486507554563953,
+      "loss": 1.8378,
+      "step": 1475
+    },
+    {
+      "epoch": 0.4651528247033865,
+      "grad_norm": 0.22880521416664124,
+      "learning_rate": 0.0006460283853184879,
+      "loss": 1.8124,
+      "step": 1480
+    },
+    {
+      "epoch": 0.46672428694900603,
+      "grad_norm": 0.3217616081237793,
+      "learning_rate": 0.0006434016163555452,
+      "loss": 1.8509,
+      "step": 1485
+    },
+    {
+      "epoch": 0.4682957491946256,
+      "grad_norm": 0.1924976408481598,
+      "learning_rate": 0.000640770527693944,
+      "loss": 1.8106,
+      "step": 1490
+    },
+    {
+      "epoch": 0.4698672114402451,
+      "grad_norm": 0.24654747545719147,
+      "learning_rate": 0.0006381351985901842,
+      "loss": 1.79,
+      "step": 1495
+    },
+    {
+      "epoch": 0.4714386736858647,
+      "grad_norm": 0.18487554788589478,
+      "learning_rate": 0.0006354957084285007,
+      "loss": 1.7589,
+      "step": 1500
+    },
+    {
+      "epoch": 0.4730101359314842,
+      "grad_norm": 0.19454938173294067,
+      "learning_rate": 0.0006328521367184721,
+      "loss": 1.7756,
+      "step": 1505
+    },
+    {
+      "epoch": 0.47458159817710377,
+      "grad_norm": 0.2697226107120514,
+      "learning_rate": 0.000630204563092626,
+      "loss": 1.8154,
+      "step": 1510
+    },
+    {
+      "epoch": 0.4761530604227233,
+      "grad_norm": 0.2261214405298233,
+      "learning_rate": 0.0006275530673040401,
+      "loss": 1.7989,
+      "step": 1515
+    },
+    {
+      "epoch": 0.47772452266834287,
+      "grad_norm": 0.29914212226867676,
+      "learning_rate": 0.0006248977292239395,
+      "loss": 1.7528,
+      "step": 1520
+    },
+    {
+      "epoch": 0.47929598491396247,
+      "grad_norm": 0.21949878334999084,
+      "learning_rate": 0.0006222386288392914,
+      "loss": 1.7694,
+      "step": 1525
+    },
+    {
+      "epoch": 0.480867447159582,
+      "grad_norm": 0.22492919862270355,
+      "learning_rate": 0.0006195758462503947,
+      "loss": 1.8911,
+      "step": 1530
+    },
+    {
+      "epoch": 0.48243890940520157,
+      "grad_norm": 0.1728420853614807,
+      "learning_rate": 0.0006169094616684678,
+      "loss": 1.7795,
+      "step": 1535
+    },
+    {
+      "epoch": 0.4840103716508211,
+      "grad_norm": 0.1900889128446579,
+      "learning_rate": 0.0006142395554132324,
+      "loss": 1.8095,
+      "step": 1540
+    },
+    {
+      "epoch": 0.48558183389644066,
+      "grad_norm": 0.2331051081418991,
+      "learning_rate": 0.0006115662079104937,
+      "loss": 1.8101,
+      "step": 1545
+    },
+    {
+      "epoch": 0.4871532961420602,
+      "grad_norm": 0.28242024779319763,
+      "learning_rate": 0.000608889499689718,
+      "loss": 1.8349,
+      "step": 1550
+    },
+    {
+      "epoch": 0.48872475838767976,
+      "grad_norm": 0.2434227615594864,
+      "learning_rate": 0.0006062095113816069,
+      "loss": 1.7964,
+      "step": 1555
+    },
+    {
+      "epoch": 0.4902962206332993,
+      "grad_norm": 0.2203354686498642,
+      "learning_rate": 0.0006035263237156676,
+      "loss": 1.7928,
+      "step": 1560
+    },
+    {
+      "epoch": 0.49186768287891885,
+      "grad_norm": 0.2670794427394867,
+      "learning_rate": 0.0006008400175177827,
+      "loss": 1.8127,
+      "step": 1565
+    },
+    {
+      "epoch": 0.4934391451245384,
+      "grad_norm": 0.3301917016506195,
+      "learning_rate": 0.0005981506737077743,
+      "loss": 1.8027,
+      "step": 1570
+    },
+    {
+      "epoch": 0.49501060737015795,
+      "grad_norm": 0.20628106594085693,
+      "learning_rate": 0.0005954583732969666,
+      "loss": 1.7938,
+      "step": 1575
+    },
+    {
+      "epoch": 0.4965820696157775,
+      "grad_norm": 0.16303130984306335,
+      "learning_rate": 0.000592763197385746,
+      "loss": 1.8486,
+      "step": 1580
+    },
+    {
+      "epoch": 0.49815353186139705,
+      "grad_norm": 0.15457695722579956,
+      "learning_rate": 0.0005900652271611175,
+      "loss": 1.773,
+      "step": 1585
+    },
+    {
+      "epoch": 0.4997249941070166,
+      "grad_norm": 0.1589762419462204,
+      "learning_rate": 0.0005873645438942595,
+      "loss": 1.7507,
+      "step": 1590
+    },
+    {
+      "epoch": 0.5012964563526361,
+      "grad_norm": 0.19667518138885498,
+      "learning_rate": 0.000584661228938076,
+      "loss": 1.8197,
+      "step": 1595
+    },
+    {
+      "epoch": 0.5028679185982556,
+      "grad_norm": 0.40122613310813904,
+      "learning_rate": 0.000581955363724745,
+      "loss": 1.8094,
+      "step": 1600
+    },
+    {
+      "epoch": 0.5044393808438752,
+      "grad_norm": 0.19897930324077606,
+      "learning_rate": 0.0005792470297632666,
+      "loss": 1.7865,
+      "step": 1605
+    },
+    {
+      "epoch": 0.5060108430894947,
+      "grad_norm": 0.3172074854373932,
+      "learning_rate": 0.0005765363086370068,
+      "loss": 1.7962,
+      "step": 1610
+    },
+    {
+      "epoch": 0.5075823053351143,
+      "grad_norm": 0.20447920262813568,
+      "learning_rate": 0.0005738232820012407,
+      "loss": 1.7691,
+      "step": 1615
+    },
+    {
+      "epoch": 0.5091537675807338,
+      "grad_norm": 0.16863569617271423,
+      "learning_rate": 0.0005711080315806921,
+      "loss": 1.8213,
+      "step": 1620
+    },
+    {
+      "epoch": 0.5107252298263534,
+      "grad_norm": 0.29164841771125793,
+      "learning_rate": 0.0005683906391670727,
+      "loss": 1.7405,
+      "step": 1625
+    },
+    {
+      "epoch": 0.5122966920719729,
+      "grad_norm": 0.1419832408428192,
+      "learning_rate": 0.0005656711866166167,
+      "loss": 1.8176,
+      "step": 1630
+    },
+    {
+      "epoch": 0.5138681543175925,
+      "grad_norm": 0.1718250811100006,
+      "learning_rate": 0.0005629497558476167,
+      "loss": 1.8431,
+      "step": 1635
+    },
+    {
+      "epoch": 0.515439616563212,
+      "grad_norm": 0.14950774610042572,
+      "learning_rate": 0.0005602264288379551,
+      "loss": 1.8238,
+      "step": 1640
+    },
+    {
+      "epoch": 0.5170110788088316,
+      "grad_norm": 0.204507514834404,
+      "learning_rate": 0.0005575012876226347,
+      "loss": 1.831,
+      "step": 1645
+    },
+    {
+      "epoch": 0.5185825410544511,
+      "grad_norm": 0.16311819851398468,
+      "learning_rate": 0.0005547744142913084,
+      "loss": 1.7805,
+      "step": 1650
+    },
+    {
+      "epoch": 0.5201540033000707,
+      "grad_norm": 0.17227725684642792,
+      "learning_rate": 0.0005520458909858048,
+      "loss": 1.8119,
+      "step": 1655
+    },
+    {
+      "epoch": 0.5217254655456902,
+      "grad_norm": 0.22001691162586212,
+      "learning_rate": 0.0005493157998976559,
+      "loss": 1.7803,
+      "step": 1660
+    },
+    {
+      "epoch": 0.5232969277913098,
+      "grad_norm": 0.3537323474884033,
+      "learning_rate": 0.0005465842232656194,
+      "loss": 1.8142,
+      "step": 1665
+    },
+    {
+      "epoch": 0.5248683900369293,
+      "grad_norm": 0.7048746347427368,
+      "learning_rate": 0.0005438512433732023,
+      "loss": 1.7881,
+      "step": 1670
+    },
+    {
+      "epoch": 0.5264398522825489,
+      "grad_norm": 0.24534624814987183,
+      "learning_rate": 0.0005411169425461822,
+      "loss": 1.8228,
+      "step": 1675
+    },
+    {
+      "epoch": 0.5280113145281684,
+      "grad_norm": 0.16271162033081055,
+      "learning_rate": 0.0005383814031501272,
+      "loss": 1.8155,
+      "step": 1680
+    },
+    {
+      "epoch": 0.529582776773788,
+      "grad_norm": 0.3233349323272705,
+      "learning_rate": 0.0005356447075879153,
+      "loss": 1.8067,
+      "step": 1685
+    },
+    {
+      "epoch": 0.5311542390194075,
+      "grad_norm": 0.20079372823238373,
+      "learning_rate": 0.0005329069382972513,
+      "loss": 1.8035,
+      "step": 1690
+    },
+    {
+      "epoch": 0.5327257012650272,
+      "grad_norm": 0.17062976956367493,
+      "learning_rate": 0.0005301681777481846,
+      "loss": 1.7636,
+      "step": 1695
+    },
+    {
+      "epoch": 0.5342971635106467,
+      "grad_norm": 0.24763086438179016,
+      "learning_rate": 0.0005274285084406234,
+      "loss": 1.8238,
+      "step": 1700
+    },
+    {
+      "epoch": 0.5358686257562663,
+      "grad_norm": 0.18531553447246552,
+      "learning_rate": 0.0005246880129018515,
+      "loss": 1.7826,
+      "step": 1705
+    },
+    {
+      "epoch": 0.5374400880018858,
+      "grad_norm": 0.13745392858982086,
+      "learning_rate": 0.0005219467736840409,
+      "loss": 1.7463,
+      "step": 1710
+    },
+    {
+      "epoch": 0.5390115502475054,
+      "grad_norm": 0.2211742103099823,
+      "learning_rate": 0.0005192048733617654,
+      "loss": 1.775,
+      "step": 1715
+    },
+    {
+      "epoch": 0.5405830124931249,
+      "grad_norm": 0.1478564590215683,
+      "learning_rate": 0.0005164623945295136,
+      "loss": 1.7752,
+      "step": 1720
+    },
+    {
+      "epoch": 0.5421544747387445,
+      "grad_norm": 0.19443662464618683,
+      "learning_rate": 0.0005137194197992001,
+      "loss": 1.7751,
+      "step": 1725
+    },
+    {
+      "epoch": 0.543725936984364,
+      "grad_norm": 0.27985528111457825,
+      "learning_rate": 0.0005109760317976782,
+      "loss": 1.7892,
+      "step": 1730
+    },
+    {
+      "epoch": 0.5452973992299835,
+      "grad_norm": 0.3474476635456085,
+      "learning_rate": 0.0005082323131642496,
+      "loss": 1.7856,
+      "step": 1735
+    },
+    {
+      "epoch": 0.5468688614756031,
+      "grad_norm": 0.19026850163936615,
+      "learning_rate": 0.0005054883465481761,
+      "loss": 1.7514,
+      "step": 1740
+    },
+    {
+      "epoch": 0.5484403237212226,
+      "grad_norm": 0.15917906165122986,
+      "learning_rate": 0.0005027442146061889,
+      "loss": 1.8218,
+      "step": 1745
+    },
+    {
+      "epoch": 0.5500117859668422,
+      "grad_norm": 0.1620151400566101,
+      "learning_rate": 0.0005,
+      "loss": 1.7814,
+      "step": 1750
+    },
+    {
+      "epoch": 0.5515832482124617,
+      "grad_norm": 0.19481198489665985,
+      "learning_rate": 0.0004972557853938111,
+      "loss": 1.8251,
+      "step": 1755
+    },
+    {
+      "epoch": 0.5531547104580813,
+      "grad_norm": 0.20308136940002441,
+      "learning_rate": 0.000494511653451824,
+      "loss": 1.8222,
+      "step": 1760
+    },
+    {
+      "epoch": 0.5547261727037008,
+      "grad_norm": 0.1648699939250946,
+      "learning_rate": 0.0004917676868357503,
+      "loss": 1.81,
+      "step": 1765
+    },
+    {
+      "epoch": 0.5562976349493204,
+      "grad_norm": 0.2805931568145752,
+      "learning_rate": 0.0004890239682023217,
+      "loss": 1.7743,
+      "step": 1770
+    },
+    {
+      "epoch": 0.5578690971949399,
+      "grad_norm": 0.22301295399665833,
+      "learning_rate": 0.00048628058020080007,
+      "loss": 1.7481,
+      "step": 1775
+    },
+    {
+      "epoch": 0.5594405594405595,
+      "grad_norm": 0.18746043741703033,
+      "learning_rate": 0.0004835376054704866,
+      "loss": 1.7545,
+      "step": 1780
+    },
+    {
+      "epoch": 0.561012021686179,
+      "grad_norm": 0.22121259570121765,
+      "learning_rate": 0.00048079512663823474,
+      "loss": 1.8134,
+      "step": 1785
+    },
+    {
+      "epoch": 0.5625834839317986,
+      "grad_norm": 0.22781746089458466,
+      "learning_rate": 0.0004780532263159592,
+      "loss": 1.8077,
+      "step": 1790
+    },
+    {
+      "epoch": 0.5641549461774181,
+      "grad_norm": 0.26571694016456604,
+      "learning_rate": 0.00047531198709814857,
+      "loss": 1.8487,
+      "step": 1795
+    },
+    {
+      "epoch": 0.5657264084230377,
+      "grad_norm": 0.1296350359916687,
+      "learning_rate": 0.00047257149155937667,
+      "loss": 1.7883,
+      "step": 1800
+    },
+    {
+      "epoch": 0.5672978706686572,
+      "grad_norm": 0.18556547164916992,
+      "learning_rate": 0.00046983182225181555,
+      "loss": 1.7306,
+      "step": 1805
+    },
+    {
+      "epoch": 0.5688693329142768,
+      "grad_norm": 0.1794668436050415,
+      "learning_rate": 0.00046709306170274867,
+      "loss": 1.8121,
+      "step": 1810
+    },
+    {
+      "epoch": 0.5704407951598963,
+      "grad_norm": 0.14622707664966583,
+      "learning_rate": 0.0004643552924120847,
+      "loss": 1.7723,
+      "step": 1815
+    },
+    {
+      "epoch": 0.5720122574055159,
+      "grad_norm": 0.2089068442583084,
+      "learning_rate": 0.00046161859684987303,
+      "loss": 1.792,
+      "step": 1820
+    },
+    {
+      "epoch": 0.5735837196511354,
+      "grad_norm": 0.18128232657909393,
+      "learning_rate": 0.000458883057453818,
+      "loss": 1.7802,
+      "step": 1825
+    },
+    {
+      "epoch": 0.575155181896755,
+      "grad_norm": 0.3529801666736603,
+      "learning_rate": 0.00045614875662679797,
+      "loss": 1.7848,
+      "step": 1830
+    },
+    {
+      "epoch": 0.5767266441423745,
+      "grad_norm": 0.1691349595785141,
+      "learning_rate": 0.00045341577673438073,
+      "loss": 1.7563,
+      "step": 1835
+    },
+    {
+      "epoch": 0.5782981063879941,
+      "grad_norm": 0.2139570415019989,
+      "learning_rate": 0.00045068420010234417,
+      "loss": 1.7557,
+      "step": 1840
+    },
+    {
+      "epoch": 0.5798695686336136,
+      "grad_norm": 0.3797873556613922,
+      "learning_rate": 0.00044795410901419527,
+      "loss": 1.8131,
+      "step": 1845
+    },
+    {
+      "epoch": 0.5814410308792332,
+      "grad_norm": 0.20959897339344025,
+      "learning_rate": 0.00044522558570869177,
+      "loss": 1.7783,
+      "step": 1850
+    },
+    {
+      "epoch": 0.5830124931248527,
+      "grad_norm": 0.21909403800964355,
+      "learning_rate": 0.0004424987123773653,
+      "loss": 1.7801,
+      "step": 1855
+    },
+    {
+      "epoch": 0.5845839553704723,
+      "grad_norm": 0.18797878921031952,
+      "learning_rate": 0.0004397735711620451,
+      "loss": 1.7918,
+      "step": 1860
+    },
+    {
+      "epoch": 0.5861554176160918,
+      "grad_norm": 0.15942728519439697,
+      "learning_rate": 0.0004370502441523834,
+      "loss": 1.7746,
+      "step": 1865
+    },
+    {
+      "epoch": 0.5877268798617113,
+      "grad_norm": 0.14537079632282257,
+      "learning_rate": 0.0004343288133833835,
+      "loss": 1.7475,
+      "step": 1870
+    },
+    {
+      "epoch": 0.5892983421073309,
+      "grad_norm": 0.2374078631401062,
+      "learning_rate": 0.0004316093608329275,
+      "loss": 1.802,
+      "step": 1875
+    },
+    {
+      "epoch": 0.5908698043529504,
+      "grad_norm": 0.16053801774978638,
+      "learning_rate": 0.000428891968419308,
+      "loss": 1.7559,
+      "step": 1880
+    },
+    {
+      "epoch": 0.59244126659857,
+      "grad_norm": 0.15984462201595306,
+      "learning_rate": 0.00042617671799875947,
+      "loss": 1.7853,
+      "step": 1885
+    },
+    {
+      "epoch": 0.5940127288441895,
+      "grad_norm": 0.1509033590555191,
+      "learning_rate": 0.00042346369136299334,
+      "loss": 1.78,
+      "step": 1890
+    },
+    {
+      "epoch": 0.5955841910898091,
+      "grad_norm": 0.21747443079948425,
+      "learning_rate": 0.0004207529702367335,
+      "loss": 1.7661,
+      "step": 1895
+    },
+    {
+      "epoch": 0.5971556533354286,
+      "grad_norm": 0.23470966517925262,
+      "learning_rate": 0.00041804463627525504,
+      "loss": 1.7534,
+      "step": 1900
+    },
+    {
+      "epoch": 0.5987271155810482,
+      "grad_norm": 0.1689888834953308,
+      "learning_rate": 0.00041533877106192407,
+      "loss": 1.7705,
+      "step": 1905
+    },
+    {
+      "epoch": 0.6002985778266677,
+      "grad_norm": 0.1531875878572464,
+      "learning_rate": 0.0004126354561057404,
+      "loss": 1.7506,
+      "step": 1910
+    },
+    {
+      "epoch": 0.6018700400722873,
+      "grad_norm": 0.22668935358524323,
+      "learning_rate": 0.00040993477283888266,
+      "loss": 1.781,
+      "step": 1915
+    },
+    {
+      "epoch": 0.6034415023179068,
+      "grad_norm": 0.15343786776065826,
+      "learning_rate": 0.0004072368026142541,
+      "loss": 1.7676,
+      "step": 1920
+    },
+    {
+      "epoch": 0.6050129645635264,
+      "grad_norm": 0.19431781768798828,
+      "learning_rate": 0.0004045416267030335,
+      "loss": 1.7892,
+      "step": 1925
+    },
+    {
+      "epoch": 0.6065844268091459,
+      "grad_norm": 0.1378578096628189,
+      "learning_rate": 0.00040184932629222574,
+      "loss": 1.7712,
+      "step": 1930
+    },
+    {
+      "epoch": 0.6081558890547655,
+      "grad_norm": 0.19014595448970795,
+      "learning_rate": 0.0003991599824822174,
+      "loss": 1.8178,
+      "step": 1935
+    },
+    {
+      "epoch": 0.609727351300385,
+      "grad_norm": 0.21545611321926117,
+      "learning_rate": 0.00039647367628433246,
+      "loss": 1.8122,
+      "step": 1940
+    },
+    {
+      "epoch": 0.6112988135460046,
+      "grad_norm": 0.12300246208906174,
+      "learning_rate": 0.0003937904886183933,
+      "loss": 1.7544,
+      "step": 1945
+    },
+    {
+      "epoch": 0.6128702757916241,
+      "grad_norm": 0.18749244511127472,
+      "learning_rate": 0.00039111050031028193,
+      "loss": 1.7679,
+      "step": 1950
+    },
+    {
+      "epoch": 0.6144417380372437,
+      "grad_norm": 0.20615063607692719,
+      "learning_rate": 0.00038843379208950617,
+      "loss": 1.7784,
+      "step": 1955
+    },
+    {
+      "epoch": 0.6160132002828632,
+      "grad_norm": 0.1621728390455246,
+      "learning_rate": 0.0003857604445867677,
+      "loss": 1.7688,
+      "step": 1960
+    },
+    {
+      "epoch": 0.6175846625284828,
+      "grad_norm": 0.2304105907678604,
+      "learning_rate": 0.00038309053833153234,
+      "loss": 1.8177,
+      "step": 1965
+    },
+    {
+      "epoch": 0.6191561247741023,
+      "grad_norm": 0.19086576998233795,
+      "learning_rate": 0.0003804241537496055,
+      "loss": 1.7566,
+      "step": 1970
+    },
+    {
+      "epoch": 0.6207275870197219,
+      "grad_norm": 0.21728602051734924,
+      "learning_rate": 0.00037776137116070867,
+      "loss": 1.7514,
+      "step": 1975
+    },
+    {
+      "epoch": 0.6222990492653414,
+      "grad_norm": 0.1872587502002716,
+      "learning_rate": 0.0003751022707760605,
+      "loss": 1.8045,
+      "step": 1980
+    },
+    {
+      "epoch": 0.623870511510961,
+      "grad_norm": 0.21256718039512634,
+      "learning_rate": 0.00037244693269596,
+      "loss": 1.7916,
+      "step": 1985
+    },
+    {
+      "epoch": 0.6254419737565805,
+      "grad_norm": 0.2095334529876709,
+      "learning_rate": 0.00036979543690737407,
+      "loss": 1.7581,
+      "step": 1990
+    },
+    {
+      "epoch": 0.6270134360022,
+      "grad_norm": 0.22193801403045654,
+      "learning_rate": 0.00036714786328152804,
+      "loss": 1.779,
+      "step": 1995
+    },
+    {
+      "epoch": 0.6285848982478196,
+      "grad_norm": 0.16215133666992188,
+      "learning_rate": 0.00036450429157149934,
+      "loss": 1.7565,
+      "step": 2000
+    },
+    {
+      "epoch": 0.6301563604934391,
+      "grad_norm": 0.17526470124721527,
+      "learning_rate": 0.00036186480140981583,
+      "loss": 1.789,
+      "step": 2005
+    },
+    {
+      "epoch": 0.6317278227390587,
+      "grad_norm": 0.3180091679096222,
+      "learning_rate": 0.00035922947230605605,
+      "loss": 1.7617,
+      "step": 2010
+    },
+    {
+      "epoch": 0.6332992849846782,
+      "grad_norm": 0.11836399137973785,
+      "learning_rate": 0.00035659838364445503,
+      "loss": 1.8111,
+      "step": 2015
+    },
+    {
+      "epoch": 0.6348707472302978,
+      "grad_norm": 0.15318524837493896,
+      "learning_rate": 0.0003539716146815122,
+      "loss": 1.7409,
+      "step": 2020
+    },
+    {
+      "epoch": 0.6364422094759173,
+      "grad_norm": 0.13768509030342102,
+      "learning_rate": 0.0003513492445436048,
+      "loss": 1.7733,
+      "step": 2025
+    },
+    {
+      "epoch": 0.6380136717215369,
+      "grad_norm": 0.16969747841358185,
+      "learning_rate": 0.0003487313522246036,
+      "loss": 1.8085,
+      "step": 2030
+    },
+    {
+      "epoch": 0.6395851339671564,
+      "grad_norm": 0.15154893696308136,
+      "learning_rate": 0.00034611801658349393,
+      "loss": 1.7053,
+      "step": 2035
+    },
+    {
+      "epoch": 0.641156596212776,
+      "grad_norm": 0.16899384558200836,
+      "learning_rate": 0.0003435093163419998,
+      "loss": 1.8229,
+      "step": 2040
+    },
+    {
+      "epoch": 0.6427280584583955,
+      "grad_norm": 0.23929907381534576,
+      "learning_rate": 0.00034090533008221234,
+      "loss": 1.7719,
+      "step": 2045
+    },
+    {
+      "epoch": 0.6442995207040151,
+      "grad_norm": 0.12470386922359467,
+      "learning_rate": 0.00033830613624422377,
+      "loss": 1.8131,
+      "step": 2050
+    },
+    {
+      "epoch": 0.6458709829496346,
+      "grad_norm": 0.15960881114006042,
+      "learning_rate": 0.00033571181312376335,
+      "loss": 1.7428,
+      "step": 2055
+    },
+    {
+      "epoch": 0.6474424451952542,
+      "grad_norm": 0.2688332796096802,
+      "learning_rate": 0.00033312243886983906,
+      "loss": 1.7652,
+      "step": 2060
+    },
+    {
+      "epoch": 0.6490139074408737,
+      "grad_norm": 0.20620514452457428,
+      "learning_rate": 0.00033053809148238423,
+      "loss": 1.7579,
+      "step": 2065
+    },
+    {
+      "epoch": 0.6505853696864933,
+      "grad_norm": 0.18862473964691162,
+      "learning_rate": 0.0003279588488099073,
+      "loss": 1.7364,
+      "step": 2070
+    },
+    {
+      "epoch": 0.6521568319321128,
+      "grad_norm": 0.155025452375412,
+      "learning_rate": 0.0003253847885471469,
+      "loss": 1.7118,
+      "step": 2075
+    },
+    {
+      "epoch": 0.6537282941777324,
+      "grad_norm": 0.27940723299980164,
+      "learning_rate": 0.0003228159882327317,
+      "loss": 1.6794,
+      "step": 2080
+    },
+    {
+      "epoch": 0.6552997564233519,
+      "grad_norm": 0.19810332357883453,
+      "learning_rate": 0.0003202525252468443,
+      "loss": 1.7879,
+      "step": 2085
+    },
+    {
+      "epoch": 0.6568712186689715,
+      "grad_norm": 0.1640649139881134,
+      "learning_rate": 0.00031769447680889064,
+      "loss": 1.7904,
+      "step": 2090
+    },
+    {
+      "epoch": 0.658442680914591,
+      "grad_norm": 0.2099301815032959,
+      "learning_rate": 0.00031514191997517385,
+      "loss": 1.7922,
+      "step": 2095
+    },
+    {
+      "epoch": 0.6600141431602106,
+      "grad_norm": 0.19281212985515594,
+      "learning_rate": 0.0003125949316365728,
+      "loss": 1.7957,
+      "step": 2100
+    },
+    {
+      "epoch": 0.6615856054058301,
+      "grad_norm": 0.20853348076343536,
+      "learning_rate": 0.00031005358851622633,
+      "loss": 1.7531,
+      "step": 2105
+    },
+    {
+      "epoch": 0.6631570676514497,
+      "grad_norm": 0.16896933317184448,
+      "learning_rate": 0.00030751796716722157,
+      "loss": 1.7632,
+      "step": 2110
+    },
+    {
+      "epoch": 0.6647285298970692,
+      "grad_norm": 0.6465707421302795,
+      "learning_rate": 0.0003049881439702888,
+      "loss": 1.7804,
+      "step": 2115
+    },
+    {
+      "epoch": 0.6662999921426888,
+      "grad_norm": 0.1943897157907486,
+      "learning_rate": 0.00030246419513149967,
+      "loss": 1.7897,
+      "step": 2120
+    },
+    {
+      "epoch": 0.6678714543883083,
+      "grad_norm": 0.20024192333221436,
+      "learning_rate": 0.00029994619667997216,
+      "loss": 1.734,
+      "step": 2125
+    },
+    {
+      "epoch": 0.6694429166339279,
+      "grad_norm": 0.18751543760299683,
+      "learning_rate": 0.0002974342244655804,
+      "loss": 1.7113,
+      "step": 2130
+    },
+    {
+      "epoch": 0.6710143788795474,
+      "grad_norm": 0.1718306988477707,
+      "learning_rate": 0.0002949283541566694,
+      "loss": 1.6794,
+      "step": 2135
+    },
+    {
+      "epoch": 0.672585841125167,
+      "grad_norm": 0.18411104381084442,
+      "learning_rate": 0.0002924286612377764,
+      "loss": 1.7223,
+      "step": 2140
+    },
+    {
+      "epoch": 0.6741573033707865,
+      "grad_norm": 0.14647985994815826,
+      "learning_rate": 0.0002899352210073562,
+      "loss": 1.7483,
+      "step": 2145
+    },
+    {
+      "epoch": 0.675728765616406,
+      "grad_norm": 0.27714547514915466,
+      "learning_rate": 0.0002874481085755133,
+      "loss": 1.7302,
+      "step": 2150
+    },
+    {
+      "epoch": 0.6773002278620256,
+      "grad_norm": 0.32130590081214905,
+      "learning_rate": 0.0002849673988617399,
+      "loss": 1.812,
+      "step": 2155
+    },
+    {
+      "epoch": 0.6788716901076451,
+      "grad_norm": 0.17637619376182556,
+      "learning_rate": 0.000282493166592658,
+      "loss": 1.729,
+      "step": 2160
+    },
+    {
+      "epoch": 0.6804431523532647,
+      "grad_norm": 0.19782552123069763,
+      "learning_rate": 0.0002800254862997695,
+      "loss": 1.7661,
+      "step": 2165
+    },
+    {
+      "epoch": 0.6820146145988842,
+      "grad_norm": 0.21681202948093414,
+      "learning_rate": 0.0002775644323172105,
+      "loss": 1.7431,
+      "step": 2170
+    },
+    {
+      "epoch": 0.6835860768445038,
+      "grad_norm": 0.21317927539348602,
+      "learning_rate": 0.0002751100787795118,
+      "loss": 1.7573,
+      "step": 2175
+    },
+    {
+      "epoch": 0.6851575390901233,
+      "grad_norm": 0.2029709368944168,
+      "learning_rate": 0.0002726624996193665,
+      "loss": 1.776,
+      "step": 2180
+    },
+    {
+      "epoch": 0.6867290013357429,
+      "grad_norm": 0.15296722948551178,
+      "learning_rate": 0.0002702217685654028,
+      "loss": 1.7741,
+      "step": 2185
+    },
+    {
+      "epoch": 0.6883004635813624,
+      "grad_norm": 0.14518578350543976,
+      "learning_rate": 0.00026778795913996224,
+      "loss": 1.7665,
+      "step": 2190
+    },
+    {
+      "epoch": 0.689871925826982,
+      "grad_norm": 0.17168502509593964,
+      "learning_rate": 0.0002653611446568861,
+      "loss": 1.7154,
+      "step": 2195
+    },
+    {
+      "epoch": 0.6914433880726015,
+      "grad_norm": 0.1342601329088211,
+      "learning_rate": 0.00026294139821930593,
+      "loss": 1.7174,
+      "step": 2200
+    },
+    {
+      "epoch": 0.6930148503182211,
+      "grad_norm": 0.1369861364364624,
+      "learning_rate": 0.00026052879271744263,
+      "loss": 1.753,
+      "step": 2205
+    },
+    {
+      "epoch": 0.6945863125638406,
+      "grad_norm": 0.1410820037126541,
+      "learning_rate": 0.00025812340082640936,
+      "loss": 1.6835,
+      "step": 2210
+    },
+    {
+      "epoch": 0.6961577748094602,
+      "grad_norm": 0.16122332215309143,
+      "learning_rate": 0.00025572529500402365,
+      "loss": 1.7404,
+      "step": 2215
+    },
+    {
+      "epoch": 0.6977292370550797,
+      "grad_norm": 0.19913320243358612,
+      "learning_rate": 0.00025333454748862396,
+      "loss": 1.7498,
+      "step": 2220
+    },
+    {
+      "epoch": 0.6993006993006993,
+      "grad_norm": 0.14058250188827515,
+      "learning_rate": 0.0002509512302968941,
+      "loss": 1.7571,
+      "step": 2225
+    },
+    {
+      "epoch": 0.7008721615463188,
+      "grad_norm": 0.2978239059448242,
+      "learning_rate": 0.0002485754152216931,
+      "loss": 1.7602,
+      "step": 2230
+    },
+    {
+      "epoch": 0.7024436237919384,
+      "grad_norm": 0.13628768920898438,
+      "learning_rate": 0.0002462071738298936,
+      "loss": 1.7331,
+      "step": 2235
+    },
+    {
+      "epoch": 0.7040150860375579,
+      "grad_norm": 0.16833730041980743,
+      "learning_rate": 0.00024384657746022564,
+      "loss": 1.7697,
+      "step": 2240
+    },
+    {
+      "epoch": 0.7055865482831775,
+      "grad_norm": 0.19926880300045013,
+      "learning_rate": 0.00024149369722112717,
+      "loss": 1.7079,
+      "step": 2245
+    },
+    {
+      "epoch": 0.707158010528797,
+      "grad_norm": 0.13357147574424744,
+      "learning_rate": 0.00023914860398860255,
+      "loss": 1.702,
+      "step": 2250
+    },
+    {
+      "epoch": 0.7087294727744166,
+      "grad_norm": 0.1692400723695755,
+      "learning_rate": 0.00023681136840408786,
+      "loss": 1.7342,
+      "step": 2255
+    },
+    {
+      "epoch": 0.7103009350200361,
+      "grad_norm": 0.1352614313364029,
+      "learning_rate": 0.00023448206087232267,
+      "loss": 1.7437,
+      "step": 2260
+    },
+    {
+      "epoch": 0.7118723972656557,
+      "grad_norm": 0.17008154094219208,
+      "learning_rate": 0.00023216075155922845,
+      "loss": 1.6892,
+      "step": 2265
+    },
+    {
+      "epoch": 0.7134438595112752,
+      "grad_norm": 0.17110054194927216,
+      "learning_rate": 0.0002298475103897964,
+      "loss": 1.7326,
+      "step": 2270
+    },
+    {
+      "epoch": 0.7150153217568948,
+      "grad_norm": 0.16124136745929718,
+      "learning_rate": 0.0002275424070459803,
+      "loss": 1.766,
+      "step": 2275
+    },
+    {
+      "epoch": 0.7165867840025143,
+      "grad_norm": 0.14922770857810974,
+      "learning_rate": 0.000225245510964597,
+      "loss": 1.7667,
+      "step": 2280
+    },
+    {
+      "epoch": 0.7181582462481338,
+      "grad_norm": 0.17472444474697113,
+      "learning_rate": 0.000222956891335236,
+      "loss": 1.7224,
+      "step": 2285
+    },
+    {
+      "epoch": 0.7197297084937534,
+      "grad_norm": 0.14927974343299866,
+      "learning_rate": 0.00022067661709817383,
+      "loss": 1.7444,
+      "step": 2290
+    },
+    {
+      "epoch": 0.7213011707393729,
+      "grad_norm": 0.1595926582813263,
+      "learning_rate": 0.00021840475694229888,
+      "loss": 1.7501,
+      "step": 2295
+    },
+    {
+      "epoch": 0.7228726329849925,
+      "grad_norm": 0.1755470633506775,
+      "learning_rate": 0.00021614137930304068,
+      "loss": 1.7742,
+      "step": 2300
+    },
+    {
+      "epoch": 0.724444095230612,
+      "grad_norm": 0.15455584228038788,
+      "learning_rate": 0.00021388655236030985,
+      "loss": 1.7152,
+      "step": 2305
+    },
+    {
+      "epoch": 0.7260155574762316,
+      "grad_norm": 0.13549718260765076,
+      "learning_rate": 0.00021164034403644338,
+      "loss": 1.7603,
+      "step": 2310
+    },
+    {
+      "epoch": 0.7275870197218511,
+      "grad_norm": 0.20018717646598816,
+      "learning_rate": 0.00020940282199415915,
+      "loss": 1.7403,
+      "step": 2315
+    },
+    {
+      "epoch": 0.7291584819674707,
+      "grad_norm": 0.14150027930736542,
+      "learning_rate": 0.00020717405363451696,
+      "loss": 1.7578,
+      "step": 2320
+    },
+    {
+      "epoch": 0.7307299442130902,
+      "grad_norm": 0.17683018743991852,
+      "learning_rate": 0.00020495410609488912,
+      "loss": 1.7105,
+      "step": 2325
+    },
+    {
+      "epoch": 0.7323014064587098,
+      "grad_norm": 0.1376308798789978,
+      "learning_rate": 0.00020274304624693778,
+      "loss": 1.6991,
+      "step": 2330
+    },
+    {
+      "epoch": 0.7338728687043293,
+      "grad_norm": 0.2307780683040619,
+      "learning_rate": 0.0002005409406946,
+      "loss": 1.7478,
+      "step": 2335
+    },
+    {
+      "epoch": 0.7354443309499489,
+      "grad_norm": 0.22559094429016113,
+      "learning_rate": 0.00019834785577208192,
+      "loss": 1.7321,
+      "step": 2340
+    },
+    {
+      "epoch": 0.7370157931955684,
+      "grad_norm": 0.2081470936536789,
+      "learning_rate": 0.00019616385754186078,
+      "loss": 1.7659,
+      "step": 2345
+    },
+    {
+      "epoch": 0.738587255441188,
+      "grad_norm": 0.17917020618915558,
+      "learning_rate": 0.00019398901179269474,
+      "loss": 1.7489,
+      "step": 2350
+    },
+    {
+      "epoch": 0.7401587176868075,
+      "grad_norm": 0.1390395164489746,
+      "learning_rate": 0.00019182338403764038,
+      "loss": 1.7142,
+      "step": 2355
+    },
+    {
+      "epoch": 0.7417301799324271,
+      "grad_norm": 0.13578888773918152,
+      "learning_rate": 0.00018966703951208048,
+      "loss": 1.7468,
+      "step": 2360
+    },
+    {
+      "epoch": 0.7433016421780466,
+      "grad_norm": 0.1478307545185089,
+      "learning_rate": 0.00018752004317175832,
+      "loss": 1.7042,
+      "step": 2365
+    },
+    {
+      "epoch": 0.7448731044236663,
+      "grad_norm": 0.1503387987613678,
+      "learning_rate": 0.00018538245969082056,
+      "loss": 1.7176,
+      "step": 2370
+    },
+    {
+      "epoch": 0.7464445666692858,
+      "grad_norm": 0.15558657050132751,
+      "learning_rate": 0.00018325435345986995,
+      "loss": 1.7821,
+      "step": 2375
+    },
+    {
+      "epoch": 0.7480160289149054,
+      "grad_norm": 0.14257632195949554,
+      "learning_rate": 0.0001811357885840254,
+      "loss": 1.7148,
+      "step": 2380
+    },
+    {
+      "epoch": 0.7495874911605249,
+      "grad_norm": 0.13766352832317352,
+      "learning_rate": 0.00017902682888099026,
+      "loss": 1.7506,
+      "step": 2385
+    },
+    {
+      "epoch": 0.7511589534061445,
+      "grad_norm": 0.16231706738471985,
+      "learning_rate": 0.00017692753787913057,
+      "loss": 1.7785,
+      "step": 2390
+    },
+    {
+      "epoch": 0.752730415651764,
+      "grad_norm": 0.15460623800754547,
+      "learning_rate": 0.00017483797881556173,
+      "loss": 1.7757,
+      "step": 2395
+    },
+    {
+      "epoch": 0.7543018778973836,
+      "grad_norm": 0.15915700793266296,
+      "learning_rate": 0.000172758214634243,
+      "loss": 1.6892,
+      "step": 2400
+    },
+    {
+      "epoch": 0.7558733401430031,
+      "grad_norm": 0.14042919874191284,
+      "learning_rate": 0.0001706883079840812,
+      "loss": 1.7892,
+      "step": 2405
+    },
+    {
+      "epoch": 0.7574448023886227,
+      "grad_norm": 0.23500895500183105,
+      "learning_rate": 0.00016862832121704435,
+      "loss": 1.7211,
+      "step": 2410
+    },
+    {
+      "epoch": 0.7590162646342422,
+      "grad_norm": 0.20954306423664093,
+      "learning_rate": 0.00016657831638628297,
+      "loss": 1.7364,
+      "step": 2415
+    },
+    {
+      "epoch": 0.7605877268798618,
+      "grad_norm": 0.18037594854831696,
+      "learning_rate": 0.00016453835524426086,
+      "loss": 1.7445,
+      "step": 2420
+    },
+    {
+      "epoch": 0.7621591891254813,
+      "grad_norm": 0.1708739697933197,
+      "learning_rate": 0.00016250849924089484,
+      "loss": 1.7493,
+      "step": 2425
+    },
+    {
+      "epoch": 0.7637306513711009,
+      "grad_norm": 0.16356390714645386,
+      "learning_rate": 0.00016048880952170374,
+      "loss": 1.7218,
+      "step": 2430
+    },
+    {
+      "epoch": 0.7653021136167204,
+      "grad_norm": 0.1173071339726448,
+      "learning_rate": 0.00015847934692596688,
+      "loss": 1.7069,
+      "step": 2435
+    },
+    {
+      "epoch": 0.76687357586234,
+      "grad_norm": 0.15243308246135712,
+      "learning_rate": 0.00015648017198489106,
+      "loss": 1.7909,
+      "step": 2440
+    },
+    {
+      "epoch": 0.7684450381079595,
+      "grad_norm": 0.12692369520664215,
+      "learning_rate": 0.00015449134491978683,
+      "loss": 1.7751,
+      "step": 2445
+    },
+    {
+      "epoch": 0.770016500353579,
+      "grad_norm": 0.13145235180854797,
+      "learning_rate": 0.00015251292564025527,
+      "loss": 1.76,
+      "step": 2450
+    },
+    {
+      "epoch": 0.7715879625991986,
+      "grad_norm": 0.12512874603271484,
+      "learning_rate": 0.00015054497374238275,
+      "loss": 1.7219,
+      "step": 2455
+    },
+    {
+      "epoch": 0.7731594248448181,
+      "grad_norm": 0.1528131067752838,
+      "learning_rate": 0.0001485875485069456,
+      "loss": 1.7519,
+      "step": 2460
+    },
+    {
+      "epoch": 0.7747308870904377,
+      "grad_norm": 0.213288813829422,
+      "learning_rate": 0.00014664070889762492,
+      "loss": 1.7176,
+      "step": 2465
+    },
+    {
+      "epoch": 0.7763023493360572,
+      "grad_norm": 0.12732981145381927,
+      "learning_rate": 0.00014470451355923025,
+      "loss": 1.7407,
+      "step": 2470
+    },
+    {
+      "epoch": 0.7778738115816768,
+      "grad_norm": 0.13688194751739502,
+      "learning_rate": 0.00014277902081593252,
+      "loss": 1.7018,
+      "step": 2475
+    },
+    {
+      "epoch": 0.7794452738272963,
+      "grad_norm": 0.1379719078540802,
+      "learning_rate": 0.00014086428866950744,
+      "loss": 1.7401,
+      "step": 2480
+    },
+    {
+      "epoch": 0.7810167360729159,
+      "grad_norm": 0.15917198359966278,
+      "learning_rate": 0.00013896037479758878,
+      "loss": 1.7188,
+      "step": 2485
+    },
+    {
+      "epoch": 0.7825881983185354,
+      "grad_norm": 0.16691961884498596,
+      "learning_rate": 0.00013706733655193055,
+      "loss": 1.6855,
+      "step": 2490
+    },
+    {
+      "epoch": 0.784159660564155,
+      "grad_norm": 0.11223277449607849,
+      "learning_rate": 0.0001351852309566788,
+      "loss": 1.7897,
+      "step": 2495
+    },
+    {
+      "epoch": 0.7857311228097745,
+      "grad_norm": 0.13406723737716675,
+      "learning_rate": 0.00013331411470665505,
+      "loss": 1.7386,
+      "step": 2500
+    },
+    {
+      "epoch": 0.7873025850553941,
+      "grad_norm": 0.2705506980419159,
+      "learning_rate": 0.0001314540441656476,
+      "loss": 1.6985,
+      "step": 2505
+    },
+    {
+      "epoch": 0.7888740473010136,
+      "grad_norm": 0.22507880628108978,
+      "learning_rate": 0.00012960507536471428,
+      "loss": 1.721,
+      "step": 2510
+    },
+    {
+      "epoch": 0.7904455095466332,
+      "grad_norm": 0.1900377720594406,
+      "learning_rate": 0.0001277672640004936,
+      "loss": 1.7351,
+      "step": 2515
+    },
+    {
+      "epoch": 0.7920169717922527,
+      "grad_norm": 0.17875802516937256,
+      "learning_rate": 0.0001259406654335285,
+      "loss": 1.7385,
+      "step": 2520
+    },
+    {
+      "epoch": 0.7935884340378723,
+      "grad_norm": 0.23769760131835938,
+      "learning_rate": 0.0001241253346865972,
+      "loss": 1.7105,
+      "step": 2525
+    },
+    {
+      "epoch": 0.7951598962834918,
+      "grad_norm": 0.15305249392986298,
+      "learning_rate": 0.000122321326443057,
+      "loss": 1.7535,
+      "step": 2530
+    },
+    {
+      "epoch": 0.7967313585291114,
+      "grad_norm": 0.13712617754936218,
+      "learning_rate": 0.00012052869504519603,
+      "loss": 1.6869,
+      "step": 2535
+    },
+    {
+      "epoch": 0.7983028207747309,
+      "grad_norm": 0.14712879061698914,
+      "learning_rate": 0.0001187474944925972,
+      "loss": 1.6889,
+      "step": 2540
+    },
+    {
+      "epoch": 0.7998742830203505,
+      "grad_norm": 0.20566821098327637,
+      "learning_rate": 0.00011697777844051105,
+      "loss": 1.73,
+      "step": 2545
+    },
+    {
+      "epoch": 0.80144574526597,
+      "grad_norm": 0.16050195693969727,
+      "learning_rate": 0.00011521960019823913,
+      "loss": 1.705,
+      "step": 2550
+    },
+    {
+      "epoch": 0.8030172075115896,
+      "grad_norm": 0.2171618640422821,
+      "learning_rate": 0.00011347301272752913,
+      "loss": 1.7078,
+      "step": 2555
+    },
+    {
+      "epoch": 0.8045886697572091,
+      "grad_norm": 0.19696617126464844,
+      "learning_rate": 0.00011173806864097885,
+      "loss": 1.7577,
+      "step": 2560
+    },
+    {
+      "epoch": 0.8061601320028287,
+      "grad_norm": 0.15522879362106323,
+      "learning_rate": 0.00011001482020045128,
+      "loss": 1.7271,
+      "step": 2565
+    },
+    {
+      "epoch": 0.8077315942484482,
+      "grad_norm": 0.13360817730426788,
+      "learning_rate": 0.00010830331931550047,
+      "loss": 1.7681,
+      "step": 2570
+    },
+    {
+      "epoch": 0.8093030564940678,
+      "grad_norm": 0.17120471596717834,
+      "learning_rate": 0.0001066036175418082,
+      "loss": 1.7188,
+      "step": 2575
+    },
+    {
+      "epoch": 0.8108745187396873,
+      "grad_norm": 0.13265547156333923,
+      "learning_rate": 0.00010491576607963066,
+      "loss": 1.7485,
+      "step": 2580
+    },
+    {
+      "epoch": 0.8124459809853068,
+      "grad_norm": 0.14022652804851532,
+      "learning_rate": 0.0001032398157722556,
+      "loss": 1.6629,
+      "step": 2585
+    },
+    {
+      "epoch": 0.8140174432309264,
+      "grad_norm": 0.14860029518604279,
+      "learning_rate": 0.0001015758171044719,
+      "loss": 1.6937,
+      "step": 2590
+    },
+    {
+      "epoch": 0.815588905476546,
+      "grad_norm": 0.1722240149974823,
+      "learning_rate": 9.992382020104807e-05,
+      "loss": 1.7502,
+      "step": 2595
+    },
+    {
+      "epoch": 0.8171603677221655,
+      "grad_norm": 0.1788044422864914,
+      "learning_rate": 9.828387482522216e-05,
+      "loss": 1.6794,
+      "step": 2600
+    },
+    {
+      "epoch": 0.818731829967785,
+      "grad_norm": 0.17370399832725525,
+      "learning_rate": 9.66560303772035e-05,
+      "loss": 1.6838,
+      "step": 2605
+    },
+    {
+      "epoch": 0.8203032922134046,
+      "grad_norm": 0.12996020913124084,
+      "learning_rate": 9.504033589268401e-05,
+      "loss": 1.7152,
+      "step": 2610
+    },
+    {
+      "epoch": 0.8218747544590241,
+      "grad_norm": 0.14151506125926971,
+      "learning_rate": 9.343684004136121e-05,
+      "loss": 1.7185,
+      "step": 2615
+    },
+    {
+      "epoch": 0.8234462167046437,
+      "grad_norm": 0.15836787223815918,
+      "learning_rate": 9.184559112547208e-05,
+      "loss": 1.7237,
+      "step": 2620
+    },
+    {
+      "epoch": 0.8250176789502632,
+      "grad_norm": 0.1398027092218399,
+      "learning_rate": 9.026663707833843e-05,
+      "loss": 1.7814,
+      "step": 2625
+    },
+    {
+      "epoch": 0.8265891411958828,
+      "grad_norm": 0.14515486359596252,
+      "learning_rate": 8.870002546292256e-05,
+      "loss": 1.6791,
+      "step": 2630
+    },
+    {
+      "epoch": 0.8281606034415023,
+      "grad_norm": 0.16488778591156006,
+      "learning_rate": 8.714580347039492e-05,
+      "loss": 1.75,
+      "step": 2635
+    },
+    {
+      "epoch": 0.8297320656871219,
+      "grad_norm": 0.1568734496831894,
+      "learning_rate": 8.560401791871186e-05,
+      "loss": 1.7421,
+      "step": 2640
+    },
+    {
+      "epoch": 0.8313035279327414,
+      "grad_norm": 0.15567830204963684,
+      "learning_rate": 8.407471525120625e-05,
+      "loss": 1.7411,
+      "step": 2645
+    },
+    {
+      "epoch": 0.832874990178361,
+      "grad_norm": 0.16733458638191223,
+      "learning_rate": 8.255794153518798e-05,
+      "loss": 1.7286,
+      "step": 2650
+    },
+    {
+      "epoch": 0.8344464524239805,
+      "grad_norm": 0.1331174075603485,
+      "learning_rate": 8.10537424605558e-05,
+      "loss": 1.6844,
+      "step": 2655
+    },
+    {
+      "epoch": 0.8360179146696001,
+      "grad_norm": 0.18669481575489044,
+      "learning_rate": 7.95621633384223e-05,
+      "loss": 1.7329,
+      "step": 2660
+    },
+    {
+      "epoch": 0.8375893769152196,
+      "grad_norm": 0.1392640769481659,
+      "learning_rate": 7.808324909974745e-05,
+      "loss": 1.7276,
+      "step": 2665
+    },
+    {
+      "epoch": 0.8391608391608392,
+      "grad_norm": 0.15594840049743652,
+      "learning_rate": 7.661704429398653e-05,
+      "loss": 1.6907,
+      "step": 2670
+    },
+    {
+      "epoch": 0.8407323014064587,
+      "grad_norm": 0.15163709223270416,
+      "learning_rate": 7.516359308774695e-05,
+      "loss": 1.7359,
+      "step": 2675
+    },
+    {
+      "epoch": 0.8423037636520783,
+      "grad_norm": 0.14341649413108826,
+      "learning_rate": 7.37229392634588e-05,
+      "loss": 1.7746,
+      "step": 2680
+    },
+    {
+      "epoch": 0.8438752258976978,
+      "grad_norm": 0.1520870178937912,
+      "learning_rate": 7.229512621805562e-05,
+      "loss": 1.7143,
+      "step": 2685
+    },
+    {
+      "epoch": 0.8454466881433174,
+      "grad_norm": 0.1629820466041565,
+      "learning_rate": 7.08801969616667e-05,
+      "loss": 1.818,
+      "step": 2690
+    },
+    {
+      "epoch": 0.8470181503889369,
+      "grad_norm": 0.14012764394283295,
+      "learning_rate": 6.947819411632222e-05,
+      "loss": 1.7606,
+      "step": 2695
+    },
+    {
+      "epoch": 0.8485896126345565,
+      "grad_norm": 0.18674196302890778,
+      "learning_rate": 6.808915991466902e-05,
+      "loss": 1.7707,
+      "step": 2700
+    },
+    {
+      "epoch": 0.850161074880176,
+      "grad_norm": 0.15248626470565796,
+      "learning_rate": 6.671313619869857e-05,
+      "loss": 1.7617,
+      "step": 2705
+    },
+    {
+      "epoch": 0.8517325371257956,
+      "grad_norm": 0.185978963971138,
+      "learning_rate": 6.535016441848573e-05,
+      "loss": 1.6929,
+      "step": 2710
+    },
+    {
+      "epoch": 0.8533039993714151,
+      "grad_norm": 0.14573198556900024,
+      "learning_rate": 6.400028563094152e-05,
+      "loss": 1.725,
+      "step": 2715
+    },
+    {
+      "epoch": 0.8548754616170346,
+      "grad_norm": 0.18840889632701874,
+      "learning_rate": 6.266354049857543e-05,
+      "loss": 1.7846,
+      "step": 2720
+    },
+    {
+      "epoch": 0.8564469238626542,
+      "grad_norm": 0.13628707826137543,
+      "learning_rate": 6.13399692882709e-05,
+      "loss": 1.7298,
+      "step": 2725
+    },
+    {
+      "epoch": 0.8580183861082737,
+      "grad_norm": 0.14333242177963257,
+      "learning_rate": 6.002961187007194e-05,
+      "loss": 1.7341,
+      "step": 2730
+    },
+    {
+      "epoch": 0.8595898483538933,
+      "grad_norm": 0.13763877749443054,
+      "learning_rate": 5.873250771598265e-05,
+      "loss": 1.7192,
+      "step": 2735
+    },
+    {
+      "epoch": 0.8611613105995128,
+      "grad_norm": 0.14965565502643585,
+      "learning_rate": 5.7448695898778106e-05,
+      "loss": 1.7346,
+      "step": 2740
+    },
+    {
+      "epoch": 0.8627327728451324,
+      "grad_norm": 0.11032088100910187,
+      "learning_rate": 5.617821509082671e-05,
+      "loss": 1.6652,
+      "step": 2745
+    },
+    {
+      "epoch": 0.8643042350907519,
+      "grad_norm": 0.17685818672180176,
+      "learning_rate": 5.49211035629264e-05,
+      "loss": 1.8149,
+      "step": 2750
+    },
+    {
+      "epoch": 0.8658756973363715,
+      "grad_norm": 0.14297960698604584,
+      "learning_rate": 5.3677399183150674e-05,
+      "loss": 1.7199,
+      "step": 2755
+    },
+    {
+      "epoch": 0.867447159581991,
+      "grad_norm": 0.1297394186258316,
+      "learning_rate": 5.244713941570889e-05,
+      "loss": 1.7095,
+      "step": 2760
+    },
+    {
+      "epoch": 0.8690186218276106,
+      "grad_norm": 0.13748817145824432,
+      "learning_rate": 5.123036131981668e-05,
+      "loss": 1.7151,
+      "step": 2765
+    },
+    {
+      "epoch": 0.8705900840732301,
+      "grad_norm": 0.11178262531757355,
+      "learning_rate": 5.002710154858065e-05,
+      "loss": 1.7202,
+      "step": 2770
+    },
+    {
+      "epoch": 0.8721615463188497,
+      "grad_norm": 0.12300541251897812,
+      "learning_rate": 4.883739634789375e-05,
+      "loss": 1.7699,
+      "step": 2775
+    },
+    {
+      "epoch": 0.8737330085644692,
+      "grad_norm": 0.12739528715610504,
+      "learning_rate": 4.7661281555343164e-05,
+      "loss": 1.7716,
+      "step": 2780
+    },
+    {
+      "epoch": 0.8753044708100888,
+      "grad_norm": 0.12120873481035233,
+      "learning_rate": 4.649879259913137e-05,
+      "loss": 1.7218,
+      "step": 2785
+    },
+    {
+      "epoch": 0.8768759330557083,
+      "grad_norm": 0.1548471450805664,
+      "learning_rate": 4.534996449700879e-05,
+      "loss": 1.7433,
+      "step": 2790
+    },
+    {
+      "epoch": 0.8784473953013279,
+      "grad_norm": 0.16176049411296844,
+      "learning_rate": 4.421483185521835e-05,
+      "loss": 1.728,
+      "step": 2795
+    },
+    {
+      "epoch": 0.8800188575469474,
+      "grad_norm": 0.19101834297180176,
+      "learning_rate": 4.309342886745399e-05,
+      "loss": 1.7275,
+      "step": 2800
+    },
+    {
+      "epoch": 0.881590319792567,
+      "grad_norm": 0.14591899514198303,
+      "learning_rate": 4.198578931382979e-05,
+      "loss": 1.7068,
+      "step": 2805
+    },
+    {
+      "epoch": 0.8831617820381865,
+      "grad_norm": 0.16475893557071686,
+      "learning_rate": 4.0891946559863055e-05,
+      "loss": 1.7479,
+      "step": 2810
+    },
+    {
+      "epoch": 0.8847332442838061,
+      "grad_norm": 0.18267378211021423,
+      "learning_rate": 3.981193355546869e-05,
+      "loss": 1.7445,
+      "step": 2815
+    },
+    {
+      "epoch": 0.8863047065294256,
+      "grad_norm": 0.10409973561763763,
+      "learning_rate": 3.874578283396718e-05,
+      "loss": 1.7277,
+      "step": 2820
+    },
+    {
+      "epoch": 0.8878761687750452,
+      "grad_norm": 0.13992512226104736,
+      "learning_rate": 3.769352651110419e-05,
+      "loss": 1.7217,
+      "step": 2825
+    },
+    {
+      "epoch": 0.8894476310206647,
+      "grad_norm": 0.1338614523410797,
+      "learning_rate": 3.6655196284083314e-05,
+      "loss": 1.7598,
+      "step": 2830
+    },
+    {
+      "epoch": 0.8910190932662843,
+      "grad_norm": 0.15006397664546967,
+      "learning_rate": 3.563082343061108e-05,
+      "loss": 1.7017,
+      "step": 2835
+    },
+    {
+      "epoch": 0.8925905555119038,
+      "grad_norm": 0.14285215735435486,
+      "learning_rate": 3.4620438807955125e-05,
+      "loss": 1.7068,
+      "step": 2840
+    },
+    {
+      "epoch": 0.8941620177575234,
+      "grad_norm": 0.11642735451459885,
+      "learning_rate": 3.3624072852014354e-05,
+      "loss": 1.7363,
+      "step": 2845
+    },
+    {
+      "epoch": 0.8957334800031429,
+      "grad_norm": 0.1364676058292389,
+      "learning_rate": 3.2641755576402255e-05,
+      "loss": 1.79,
+      "step": 2850
+    },
+    {
+      "epoch": 0.8973049422487624,
+      "grad_norm": 0.13236261904239655,
+      "learning_rate": 3.16735165715426e-05,
+      "loss": 1.7463,
+      "step": 2855
+    },
+    {
+      "epoch": 0.898876404494382,
+      "grad_norm": 0.16500313580036163,
+      "learning_rate": 3.071938500377852e-05,
+      "loss": 1.7057,
+      "step": 2860
+    },
+    {
+      "epoch": 0.9004478667400015,
+      "grad_norm": 0.15041331946849823,
+      "learning_rate": 2.9779389614493558e-05,
+      "loss": 1.8108,
+      "step": 2865
+    },
+    {
+      "epoch": 0.9020193289856211,
+      "grad_norm": 0.1336473524570465,
+      "learning_rate": 2.8853558719245833e-05,
+      "loss": 1.7186,
+      "step": 2870
+    },
+    {
+      "epoch": 0.9035907912312406,
+      "grad_norm": 0.1339365541934967,
+      "learning_rate": 2.794192020691544e-05,
+      "loss": 1.7605,
+      "step": 2875
+    },
+    {
+      "epoch": 0.9051622534768602,
+      "grad_norm": 0.09974883496761322,
+      "learning_rate": 2.704450153886423e-05,
+      "loss": 1.7394,
+      "step": 2880
+    },
+    {
+      "epoch": 0.9067337157224797,
+      "grad_norm": 0.1279599815607071,
+      "learning_rate": 2.6161329748108253e-05,
+      "loss": 1.7415,
+      "step": 2885
+    },
+    {
+      "epoch": 0.9083051779680993,
+      "grad_norm": 0.1555188149213791,
+      "learning_rate": 2.5292431438503905e-05,
+      "loss": 1.8056,
+      "step": 2890
+    },
+    {
+      "epoch": 0.9098766402137188,
+      "grad_norm": 0.19093474745750427,
+      "learning_rate": 2.4437832783946234e-05,
+      "loss": 1.7738,
+      "step": 2895
+    },
+    {
+      "epoch": 0.9114481024593384,
+      "grad_norm": 0.09890997409820557,
+      "learning_rate": 2.3597559527580692e-05,
+      "loss": 1.6863,
+      "step": 2900
+    },
+    {
+      "epoch": 0.9130195647049579,
+      "grad_norm": 0.14242489635944366,
+      "learning_rate": 2.2771636981027467e-05,
+      "loss": 1.6858,
+      "step": 2905
+    },
+    {
+      "epoch": 0.9145910269505775,
+      "grad_norm": 0.15494489669799805,
+      "learning_rate": 2.1960090023619205e-05,
+      "loss": 1.7568,
+      "step": 2910
+    },
+    {
+      "epoch": 0.916162489196197,
+      "grad_norm": 0.1707945168018341,
+      "learning_rate": 2.1162943101651622e-05,
+      "loss": 1.7361,
+      "step": 2915
+    },
+    {
+      "epoch": 0.9177339514418166,
+      "grad_norm": 0.17627616226673126,
+      "learning_rate": 2.038022022764685e-05,
+      "loss": 1.791,
+      "step": 2920
+    },
+    {
+      "epoch": 0.9193054136874361,
+      "grad_norm": 0.15799571573734283,
+      "learning_rate": 1.9611944979630204e-05,
+      "loss": 1.7233,
+      "step": 2925
+    },
+    {
+      "epoch": 0.9208768759330557,
+      "grad_norm": 0.11753001809120178,
+      "learning_rate": 1.8858140500420005e-05,
+      "loss": 1.7659,
+      "step": 2930
+    },
+    {
+      "epoch": 0.9224483381786752,
+      "grad_norm": 0.11653709411621094,
+      "learning_rate": 1.8118829496930557e-05,
+      "loss": 1.7428,
+      "step": 2935
+    },
+    {
+      "epoch": 0.9240198004242948,
+      "grad_norm": 0.1365276426076889,
+      "learning_rate": 1.739403423948782e-05,
+      "loss": 1.728,
+      "step": 2940
+    },
+    {
+      "epoch": 0.9255912626699143,
+      "grad_norm": 0.1241711974143982,
+      "learning_rate": 1.668377656115877e-05,
+      "loss": 1.7144,
+      "step": 2945
+    },
+    {
+      "epoch": 0.9271627249155339,
+      "grad_norm": 0.15429584681987762,
+      "learning_rate": 1.5988077857093775e-05,
+      "loss": 1.6854,
+      "step": 2950
+    },
+    {
+      "epoch": 0.9287341871611534,
+      "grad_norm": 0.11500924080610275,
+      "learning_rate": 1.5306959083882078e-05,
+      "loss": 1.753,
+      "step": 2955
+    },
+    {
+      "epoch": 0.930305649406773,
+      "grad_norm": 0.12633784115314484,
+      "learning_rate": 1.4640440758920293e-05,
+      "loss": 1.7387,
+      "step": 2960
+    },
+    {
+      "epoch": 0.9318771116523925,
+      "grad_norm": 0.13106156885623932,
+      "learning_rate": 1.3988542959794625e-05,
+      "loss": 1.7124,
+      "step": 2965
+    },
+    {
+      "epoch": 0.9334485738980121,
+      "grad_norm": 0.10034507513046265,
+      "learning_rate": 1.3351285323676022e-05,
+      "loss": 1.7571,
+      "step": 2970
+    },
+    {
+      "epoch": 0.9350200361436316,
+      "grad_norm": 0.1519390344619751,
+      "learning_rate": 1.2728687046728526e-05,
+      "loss": 1.6967,
+      "step": 2975
+    },
+    {
+      "epoch": 0.9365914983892512,
+      "grad_norm": 0.15289808809757233,
+      "learning_rate": 1.2120766883531087e-05,
+      "loss": 1.7167,
+      "step": 2980
+    },
+    {
+      "epoch": 0.9381629606348707,
+      "grad_norm": 0.11695173382759094,
+      "learning_rate": 1.152754314651283e-05,
+      "loss": 1.7096,
+      "step": 2985
+    },
+    {
+      "epoch": 0.9397344228804902,
+      "grad_norm": 0.11231189966201782,
+      "learning_rate": 1.0949033705400902e-05,
+      "loss": 1.7092,
+      "step": 2990
+    },
+    {
+      "epoch": 0.9413058851261098,
+      "grad_norm": 0.12071159482002258,
+      "learning_rate": 1.0385255986682718e-05,
+      "loss": 1.7096,
+      "step": 2995
+    },
+    {
+      "epoch": 0.9428773473717293,
+      "grad_norm": 0.12497507780790329,
+      "learning_rate": 9.836226973080786e-06,
+      "loss": 1.7723,
+      "step": 3000
+    },
+    {
+      "epoch": 0.9444488096173489,
+      "grad_norm": 0.11592131853103638,
+      "learning_rate": 9.30196320304122e-06,
+      "loss": 1.7458,
+      "step": 3005
+    },
+    {
+      "epoch": 0.9460202718629684,
+      "grad_norm": 0.16098622977733612,
+      "learning_rate": 8.782480770235246e-06,
+      "loss": 1.7568,
+      "step": 3010
+    },
+    {
+      "epoch": 0.947591734108588,
+      "grad_norm": 0.1251290738582611,
+      "learning_rate": 8.277795323074933e-06,
+      "loss": 1.7218,
+      "step": 3015
+    },
+    {
+      "epoch": 0.9491631963542075,
+      "grad_norm": 0.15294887125492096,
+      "learning_rate": 7.787922064241393e-06,
+      "loss": 1.7038,
+      "step": 3020
+    },
+    {
+      "epoch": 0.9507346585998271,
+      "grad_norm": 0.12903992831707,
+      "learning_rate": 7.312875750227044e-06,
+      "loss": 1.7287,
+      "step": 3025
+    },
+    {
+      "epoch": 0.9523061208454466,
+      "grad_norm": 0.17125993967056274,
+      "learning_rate": 6.852670690890961e-06,
+      "loss": 1.7112,
+      "step": 3030
+    },
+    {
+      "epoch": 0.9538775830910662,
+      "grad_norm": 0.13478563725948334,
+      "learning_rate": 6.40732074902789e-06,
+      "loss": 1.7588,
+      "step": 3035
+    },
+    {
+      "epoch": 0.9554490453366857,
+      "grad_norm": 0.14833632111549377,
+      "learning_rate": 5.97683933995069e-06,
+      "loss": 1.7438,
+      "step": 3040
+    },
+    {
+      "epoch": 0.9570205075823053,
+      "grad_norm": 0.16726098954677582,
+      "learning_rate": 5.561239431086218e-06,
+      "loss": 1.7639,
+      "step": 3045
+    },
+    {
+      "epoch": 0.9585919698279249,
+      "grad_norm": 0.16357098519802094,
+      "learning_rate": 5.160533541584578e-06,
+      "loss": 1.6912,
+      "step": 3050
+    },
+    {
+      "epoch": 0.9601634320735445,
+      "grad_norm": 0.12722498178482056,
+      "learning_rate": 4.774733741942205e-06,
+      "loss": 1.7576,
+      "step": 3055
+    },
+    {
+      "epoch": 0.961734894319164,
+      "grad_norm": 0.16135632991790771,
+      "learning_rate": 4.403851653638158e-06,
+      "loss": 1.7702,
+      "step": 3060
+    },
+    {
+      "epoch": 0.9633063565647836,
+      "grad_norm": 0.156170055270195,
+      "learning_rate": 4.0478984487838935e-06,
+      "loss": 1.7429,
+      "step": 3065
+    },
+    {
+      "epoch": 0.9648778188104031,
+      "grad_norm": 0.1406138390302658,
+      "learning_rate": 3.706884849787151e-06,
+      "loss": 1.7209,
+      "step": 3070
+    },
+    {
+      "epoch": 0.9664492810560227,
+      "grad_norm": 0.1389617770910263,
+      "learning_rate": 3.3808211290284885e-06,
+      "loss": 1.7126,
+      "step": 3075
+    },
+    {
+      "epoch": 0.9680207433016422,
+      "grad_norm": 0.1216338723897934,
+      "learning_rate": 3.0697171085521946e-06,
+      "loss": 1.7186,
+      "step": 3080
+    },
+    {
+      "epoch": 0.9695922055472618,
+      "grad_norm": 0.13428504765033722,
+      "learning_rate": 2.7735821597701382e-06,
+      "loss": 1.7334,
+      "step": 3085
+    },
+    {
+      "epoch": 0.9711636677928813,
+      "grad_norm": 0.1258799433708191,
+      "learning_rate": 2.49242520317966e-06,
+      "loss": 1.7303,
+      "step": 3090
+    },
+    {
+      "epoch": 0.9727351300385009,
+      "grad_norm": 0.144920215010643,
+      "learning_rate": 2.2262547080948992e-06,
+      "loss": 1.7595,
+      "step": 3095
+    },
+    {
+      "epoch": 0.9743065922841204,
+      "grad_norm": 0.12858329713344574,
+      "learning_rate": 1.975078692391552e-06,
+      "loss": 1.7411,
+      "step": 3100
+    },
+    {
+      "epoch": 0.97587805452974,
+      "grad_norm": 0.1087680459022522,
+      "learning_rate": 1.7389047222652888e-06,
+      "loss": 1.7469,
+      "step": 3105
+    },
+    {
+      "epoch": 0.9774495167753595,
+      "grad_norm": 0.12177930027246475,
+      "learning_rate": 1.5177399120039904e-06,
+      "loss": 1.6998,
+      "step": 3110
+    },
+    {
+      "epoch": 0.9790209790209791,
+      "grad_norm": 0.11609877645969391,
+      "learning_rate": 1.3115909237734204e-06,
+      "loss": 1.7057,
+      "step": 3115
+    },
+    {
+      "epoch": 0.9805924412665986,
+      "grad_norm": 0.13257728517055511,
+      "learning_rate": 1.1204639674164962e-06,
+      "loss": 1.7442,
+      "step": 3120
+    },
+    {
+      "epoch": 0.9821639035122182,
+      "grad_norm": 0.12050619721412659,
+      "learning_rate": 9.44364800266162e-07,
+      "loss": 1.7151,
+      "step": 3125
+    },
+    {
+      "epoch": 0.9837353657578377,
+      "grad_norm": 0.12853524088859558,
+      "learning_rate": 7.832987269720815e-07,
+      "loss": 1.7062,
+      "step": 3130
+    },
+    {
+      "epoch": 0.9853068280034573,
+      "grad_norm": 0.11707114428281784,
+      "learning_rate": 6.372705993408223e-07,
+      "loss": 1.7023,
+      "step": 3135
+    },
+    {
+      "epoch": 0.9868782902490768,
+      "grad_norm": 0.12321787327528,
+      "learning_rate": 5.062848161896394e-07,
+      "loss": 1.7308,
+      "step": 3140
+    },
+    {
+      "epoch": 0.9884497524946964,
+      "grad_norm": 0.13473524153232574,
+      "learning_rate": 3.903453232140808e-07,
+      "loss": 1.7428,
+      "step": 3145
+    },
+    {
+      "epoch": 0.9900212147403159,
+      "grad_norm": 0.12741941213607788,
+      "learning_rate": 2.894556128689163e-07,
+      "loss": 1.6881,
+      "step": 3150
+    },
+    {
+      "epoch": 0.9915926769859355,
+      "grad_norm": 0.1763051599264145,
+      "learning_rate": 2.03618724263277e-07,
+      "loss": 1.7244,
+      "step": 3155
+    },
+    {
+      "epoch": 0.993164139231555,
+      "grad_norm": 0.11934591829776764,
+      "learning_rate": 1.3283724306867306e-07,
+      "loss": 1.7025,
+      "step": 3160
+    },
+    {
+      "epoch": 0.9947356014771745,
+      "grad_norm": 0.12859931588172913,
+      "learning_rate": 7.711330144161144e-08,
+      "loss": 1.7378,
+      "step": 3165
+    },
+    {
+      "epoch": 0.9963070637227941,
+      "grad_norm": 0.14746056497097015,
+      "learning_rate": 3.644857795886969e-08,
+      "loss": 1.7293,
+      "step": 3170
+    },
+    {
+      "epoch": 0.9978785259684136,
+      "grad_norm": 0.16102327406406403,
+      "learning_rate": 1.0844297567258466e-08,
+      "loss": 1.6806,
+      "step": 3175
+    },
+    {
+      "epoch": 0.9994499882140332,
+      "grad_norm": 0.15685699880123138,
+      "learning_rate": 3.012315465955595e-10,
+      "loss": 1.8048,
+      "step": 3180
+    },
+    {
+      "epoch": 0.999764280663157,
+      "eval_loss": 1.7209471464157104,
+      "eval_runtime": 333.4066,
+      "eval_samples_per_second": 31.946,
+      "eval_steps_per_second": 1.998,
+      "step": 3181
+    },
+    {
+      "epoch": 0.999764280663157,
+      "step": 3181,
+      "total_flos": 2.656972328528773e+17,
+      "train_loss": 2.0235598598475426,
+      "train_runtime": 14726.7001,
+      "train_samples_per_second": 13.827,
+      "train_steps_per_second": 0.216
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 3181,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": false,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.656972328528773e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}