diff --git "a/checkpoint-6000/trainer_state.json" "b/checkpoint-6000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-6000/trainer_state.json"
@@ -0,0 +1,4234 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.4538405621516841,
+  "eval_steps": 500,
+  "global_step": 6000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00242306760358614,
+      "grad_norm": 0.22300943732261658,
+      "learning_rate": 1.8e-06,
+      "loss": 1.2877,
+      "step": 10
+    },
+    {
+      "epoch": 0.00484613520717228,
+      "grad_norm": 0.13678084313869476,
+      "learning_rate": 3.8e-06,
+      "loss": 1.3473,
+      "step": 20
+    },
+    {
+      "epoch": 0.00726920281075842,
+      "grad_norm": 0.22025036811828613,
+      "learning_rate": 5.8e-06,
+      "loss": 1.4399,
+      "step": 30
+    },
+    {
+      "epoch": 0.00969227041434456,
+      "grad_norm": 0.1782158464193344,
+      "learning_rate": 7.8e-06,
+      "loss": 1.2627,
+      "step": 40
+    },
+    {
+      "epoch": 0.0121153380179307,
+      "grad_norm": 0.2191566377878189,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 1.3235,
+      "step": 50
+    },
+    {
+      "epoch": 0.01453840562151684,
+      "grad_norm": 0.2430618554353714,
+      "learning_rate": 1.18e-05,
+      "loss": 1.4062,
+      "step": 60
+    },
+    {
+      "epoch": 0.01696147322510298,
+      "grad_norm": 0.2611038088798523,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 1.1462,
+      "step": 70
+    },
+    {
+      "epoch": 0.01938454082868912,
+      "grad_norm": 0.18182355165481567,
+      "learning_rate": 1.58e-05,
+      "loss": 1.1858,
+      "step": 80
+    },
+    {
+      "epoch": 0.02180760843227526,
+      "grad_norm": 0.23076005280017853,
+      "learning_rate": 1.78e-05,
+      "loss": 1.0799,
+      "step": 90
+    },
+    {
+      "epoch": 0.0242306760358614,
+      "grad_norm": 0.4523642957210541,
+      "learning_rate": 1.9800000000000004e-05,
+      "loss": 1.1662,
+      "step": 100
+    },
+    {
+      "epoch": 0.02665374363944754,
+      "grad_norm": 0.23707886040210724,
+      "learning_rate": 2.18e-05,
+      "loss": 1.0855,
+      "step": 110
+    },
+    {
+      "epoch": 0.02907681124303368,
+      "grad_norm": 0.2294093519449234,
+      "learning_rate": 2.38e-05,
+      "loss": 0.9007,
+      "step": 120
+    },
+    {
+      "epoch": 0.03149987884661982,
+      "grad_norm": 0.23308613896369934,
+      "learning_rate": 2.58e-05,
+      "loss": 0.9462,
+      "step": 130
+    },
+    {
+      "epoch": 0.03392294645020596,
+      "grad_norm": 0.21758387982845306,
+      "learning_rate": 2.7800000000000005e-05,
+      "loss": 0.8586,
+      "step": 140
+    },
+    {
+      "epoch": 0.0363460140537921,
+      "grad_norm": 0.2084963023662567,
+      "learning_rate": 2.98e-05,
+      "loss": 0.6916,
+      "step": 150
+    },
+    {
+      "epoch": 0.03876908165737824,
+      "grad_norm": 0.2152082622051239,
+      "learning_rate": 3.18e-05,
+      "loss": 0.7014,
+      "step": 160
+    },
+    {
+      "epoch": 0.04119214926096438,
+      "grad_norm": 0.22858214378356934,
+      "learning_rate": 3.38e-05,
+      "loss": 0.6022,
+      "step": 170
+    },
+    {
+      "epoch": 0.04361521686455052,
+      "grad_norm": 0.1915079951286316,
+      "learning_rate": 3.58e-05,
+      "loss": 0.6187,
+      "step": 180
+    },
+    {
+      "epoch": 0.04603828446813666,
+      "grad_norm": 0.12963800132274628,
+      "learning_rate": 3.7800000000000004e-05,
+      "loss": 0.4887,
+      "step": 190
+    },
+    {
+      "epoch": 0.0484613520717228,
+      "grad_norm": 0.13217966258525848,
+      "learning_rate": 3.9800000000000005e-05,
+      "loss": 0.4853,
+      "step": 200
+    },
+    {
+      "epoch": 0.05088441967530894,
+      "grad_norm": 0.17833656072616577,
+      "learning_rate": 4.18e-05,
+      "loss": 0.4678,
+      "step": 210
+    },
+    {
+      "epoch": 0.05330748727889508,
+      "grad_norm": 0.1171719878911972,
+      "learning_rate": 4.38e-05,
+      "loss": 0.4291,
+      "step": 220
+    },
+    {
+      "epoch": 0.05573055488248122,
+      "grad_norm": 0.1501215249300003,
+      "learning_rate": 4.58e-05,
+      "loss": 0.4234,
+      "step": 230
+    },
+    {
+      "epoch": 0.05815362248606736,
+      "grad_norm": 0.10823249816894531,
+      "learning_rate": 4.78e-05,
+      "loss": 0.3753,
+      "step": 240
+    },
+    {
+      "epoch": 0.0605766900896535,
+      "grad_norm": 0.12995333969593048,
+      "learning_rate": 4.9800000000000004e-05,
+      "loss": 0.3648,
+      "step": 250
+    },
+    {
+      "epoch": 0.06299975769323964,
+      "grad_norm": 0.14510154724121094,
+      "learning_rate": 5.1800000000000005e-05,
+      "loss": 0.3304,
+      "step": 260
+    },
+    {
+      "epoch": 0.06542282529682578,
+      "grad_norm": 0.15389852225780487,
+      "learning_rate": 5.380000000000001e-05,
+      "loss": 0.3509,
+      "step": 270
+    },
+    {
+      "epoch": 0.06784589290041192,
+      "grad_norm": 0.13117581605911255,
+      "learning_rate": 5.580000000000001e-05,
+      "loss": 0.3177,
+      "step": 280
+    },
+    {
+      "epoch": 0.07026896050399806,
+      "grad_norm": 0.14186632633209229,
+      "learning_rate": 5.7799999999999995e-05,
+      "loss": 0.3259,
+      "step": 290
+    },
+    {
+      "epoch": 0.0726920281075842,
+      "grad_norm": 0.17051789164543152,
+      "learning_rate": 5.9800000000000003e-05,
+      "loss": 0.292,
+      "step": 300
+    },
+    {
+      "epoch": 0.07511509571117034,
+      "grad_norm": 0.14579974114894867,
+      "learning_rate": 6.18e-05,
+      "loss": 0.2928,
+      "step": 310
+    },
+    {
+      "epoch": 0.07753816331475648,
+      "grad_norm": 0.14740227162837982,
+      "learning_rate": 6.38e-05,
+      "loss": 0.2593,
+      "step": 320
+    },
+    {
+      "epoch": 0.07996123091834262,
+      "grad_norm": 0.16401788592338562,
+      "learning_rate": 6.58e-05,
+      "loss": 0.2481,
+      "step": 330
+    },
+    {
+      "epoch": 0.08238429852192876,
+      "grad_norm": 0.1546965092420578,
+      "learning_rate": 6.780000000000001e-05,
+      "loss": 0.2263,
+      "step": 340
+    },
+    {
+      "epoch": 0.0848073661255149,
+      "grad_norm": 0.22539666295051575,
+      "learning_rate": 6.98e-05,
+      "loss": 0.2216,
+      "step": 350
+    },
+    {
+      "epoch": 0.08723043372910104,
+      "grad_norm": 0.25954148173332214,
+      "learning_rate": 7.18e-05,
+      "loss": 0.2152,
+      "step": 360
+    },
+    {
+      "epoch": 0.08965350133268718,
+      "grad_norm": 0.18402907252311707,
+      "learning_rate": 7.38e-05,
+      "loss": 0.2177,
+      "step": 370
+    },
+    {
+      "epoch": 0.09207656893627332,
+      "grad_norm": 0.1356307864189148,
+      "learning_rate": 7.58e-05,
+      "loss": 0.1911,
+      "step": 380
+    },
+    {
+      "epoch": 0.09449963653985946,
+      "grad_norm": 0.13355639576911926,
+      "learning_rate": 7.780000000000001e-05,
+      "loss": 0.1817,
+      "step": 390
+    },
+    {
+      "epoch": 0.0969227041434456,
+      "grad_norm": 0.14798177778720856,
+      "learning_rate": 7.98e-05,
+      "loss": 0.1782,
+      "step": 400
+    },
+    {
+      "epoch": 0.09934577174703174,
+      "grad_norm": 0.2079785168170929,
+      "learning_rate": 8.18e-05,
+      "loss": 0.1817,
+      "step": 410
+    },
+    {
+      "epoch": 0.10176883935061788,
+      "grad_norm": 0.1562364548444748,
+      "learning_rate": 8.38e-05,
+      "loss": 0.1699,
+      "step": 420
+    },
+    {
+      "epoch": 0.10419190695420402,
+      "grad_norm": 0.14666175842285156,
+      "learning_rate": 8.58e-05,
+      "loss": 0.1691,
+      "step": 430
+    },
+    {
+      "epoch": 0.10661497455779016,
+      "grad_norm": 0.28095802664756775,
+      "learning_rate": 8.78e-05,
+      "loss": 0.1613,
+      "step": 440
+    },
+    {
+      "epoch": 0.1090380421613763,
+      "grad_norm": 0.16671571135520935,
+      "learning_rate": 8.98e-05,
+      "loss": 0.1575,
+      "step": 450
+    },
+    {
+      "epoch": 0.11146110976496244,
+      "grad_norm": 0.17770862579345703,
+      "learning_rate": 9.180000000000001e-05,
+      "loss": 0.1646,
+      "step": 460
+    },
+    {
+      "epoch": 0.11388417736854858,
+      "grad_norm": 0.2173616737127304,
+      "learning_rate": 9.38e-05,
+      "loss": 0.1568,
+      "step": 470
+    },
+    {
+      "epoch": 0.11630724497213472,
+      "grad_norm": 0.18545781075954437,
+      "learning_rate": 9.58e-05,
+      "loss": 0.1509,
+      "step": 480
+    },
+    {
+      "epoch": 0.11873031257572086,
+      "grad_norm": 0.1835421770811081,
+      "learning_rate": 9.78e-05,
+      "loss": 0.1424,
+      "step": 490
+    },
+    {
+      "epoch": 0.121153380179307,
+      "grad_norm": 0.14738881587982178,
+      "learning_rate": 9.98e-05,
+      "loss": 0.1352,
+      "step": 500
+    },
+    {
+      "epoch": 0.12357644778289315,
+      "grad_norm": 0.1982613503932953,
+      "learning_rate": 9.9999778549206e-05,
+      "loss": 0.1336,
+      "step": 510
+    },
+    {
+      "epoch": 0.12599951538647927,
+      "grad_norm": 0.23628918826580048,
+      "learning_rate": 9.999901304280685e-05,
+      "loss": 0.131,
+      "step": 520
+    },
+    {
+      "epoch": 0.12842258299006543,
+      "grad_norm": 0.3614969253540039,
+      "learning_rate": 9.999770075521164e-05,
+      "loss": 0.1377,
+      "step": 530
+    },
+    {
+      "epoch": 0.13084565059365155,
+      "grad_norm": 0.15150132775306702,
+      "learning_rate": 9.99958417007713e-05,
+      "loss": 0.1254,
+      "step": 540
+    },
+    {
+      "epoch": 0.1332687181972377,
+      "grad_norm": 0.3162827789783478,
+      "learning_rate": 9.999343589981615e-05,
+      "loss": 0.1204,
+      "step": 550
+    },
+    {
+      "epoch": 0.13569178580082383,
+      "grad_norm": 0.18979772925376892,
+      "learning_rate": 9.999048337865568e-05,
+      "loss": 0.1176,
+      "step": 560
+    },
+    {
+      "epoch": 0.13811485340441,
+      "grad_norm": 0.22147120535373688,
+      "learning_rate": 9.998698416957815e-05,
+      "loss": 0.1146,
+      "step": 570
+    },
+    {
+      "epoch": 0.1405379210079961,
+      "grad_norm": 0.222473606467247,
+      "learning_rate": 9.998293831085037e-05,
+      "loss": 0.1212,
+      "step": 580
+    },
+    {
+      "epoch": 0.14296098861158227,
+      "grad_norm": 0.14630824327468872,
+      "learning_rate": 9.997834584671719e-05,
+      "loss": 0.1241,
+      "step": 590
+    },
+    {
+      "epoch": 0.1453840562151684,
+      "grad_norm": 0.15462113916873932,
+      "learning_rate": 9.997320682740107e-05,
+      "loss": 0.12,
+      "step": 600
+    },
+    {
+      "epoch": 0.14780712381875455,
+      "grad_norm": 0.19773511588573456,
+      "learning_rate": 9.996752130910149e-05,
+      "loss": 0.1246,
+      "step": 610
+    },
+    {
+      "epoch": 0.15023019142234068,
+      "grad_norm": 0.22675354778766632,
+      "learning_rate": 9.99612893539944e-05,
+      "loss": 0.1145,
+      "step": 620
+    },
+    {
+      "epoch": 0.15265325902592683,
+      "grad_norm": 0.15683938562870026,
+      "learning_rate": 9.995451103023144e-05,
+      "loss": 0.0963,
+      "step": 630
+    },
+    {
+      "epoch": 0.15507632662951296,
+      "grad_norm": 0.23586426675319672,
+      "learning_rate": 9.994718641193928e-05,
+      "loss": 0.1043,
+      "step": 640
+    },
+    {
+      "epoch": 0.1574993942330991,
+      "grad_norm": 0.28452053666114807,
+      "learning_rate": 9.993931557921874e-05,
+      "loss": 0.1126,
+      "step": 650
+    },
+    {
+      "epoch": 0.15992246183668524,
+      "grad_norm": 0.21146593987941742,
+      "learning_rate": 9.993089861814402e-05,
+      "loss": 0.1118,
+      "step": 660
+    },
+    {
+      "epoch": 0.1623455294402714,
+      "grad_norm": 0.21904832124710083,
+      "learning_rate": 9.992193562076166e-05,
+      "loss": 0.125,
+      "step": 670
+    },
+    {
+      "epoch": 0.16476859704385752,
+      "grad_norm": 0.1765352338552475,
+      "learning_rate": 9.991242668508954e-05,
+      "loss": 0.1084,
+      "step": 680
+    },
+    {
+      "epoch": 0.16719166464744367,
+      "grad_norm": 0.2080189287662506,
+      "learning_rate": 9.990237191511587e-05,
+      "loss": 0.0988,
+      "step": 690
+    },
+    {
+      "epoch": 0.1696147322510298,
+      "grad_norm": 0.27506083250045776,
+      "learning_rate": 9.989177142079802e-05,
+      "loss": 0.0961,
+      "step": 700
+    },
+    {
+      "epoch": 0.17203779985461595,
+      "grad_norm": 0.1665004938840866,
+      "learning_rate": 9.988062531806126e-05,
+      "loss": 0.1093,
+      "step": 710
+    },
+    {
+      "epoch": 0.17446086745820208,
+      "grad_norm": 0.1868591457605362,
+      "learning_rate": 9.986893372879762e-05,
+      "loss": 0.0962,
+      "step": 720
+    },
+    {
+      "epoch": 0.17688393506178823,
+      "grad_norm": 0.14843980967998505,
+      "learning_rate": 9.985669678086443e-05,
+      "loss": 0.096,
+      "step": 730
+    },
+    {
+      "epoch": 0.17930700266537436,
+      "grad_norm": 0.12363710254430771,
+      "learning_rate": 9.984391460808298e-05,
+      "loss": 0.0822,
+      "step": 740
+    },
+    {
+      "epoch": 0.18173007026896051,
+      "grad_norm": 0.1875678449869156,
+      "learning_rate": 9.983058735023709e-05,
+      "loss": 0.0921,
+      "step": 750
+    },
+    {
+      "epoch": 0.18415313787254664,
+      "grad_norm": 0.15753141045570374,
+      "learning_rate": 9.98167151530715e-05,
+      "loss": 0.0852,
+      "step": 760
+    },
+    {
+      "epoch": 0.1865762054761328,
+      "grad_norm": 0.14286907017230988,
+      "learning_rate": 9.980229816829034e-05,
+      "loss": 0.0879,
+      "step": 770
+    },
+    {
+      "epoch": 0.18899927307971892,
+      "grad_norm": 0.18874263763427734,
+      "learning_rate": 9.978733655355544e-05,
+      "loss": 0.0848,
+      "step": 780
+    },
+    {
+      "epoch": 0.19142234068330508,
+      "grad_norm": 0.21680274605751038,
+      "learning_rate": 9.977183047248464e-05,
+      "loss": 0.0849,
+      "step": 790
+    },
+    {
+      "epoch": 0.1938454082868912,
+      "grad_norm": 0.24747930467128754,
+      "learning_rate": 9.975578009464992e-05,
+      "loss": 0.0892,
+      "step": 800
+    },
+    {
+      "epoch": 0.19626847589047736,
+      "grad_norm": 0.14206859469413757,
+      "learning_rate": 9.97391855955757e-05,
+      "loss": 0.0928,
+      "step": 810
+    },
+    {
+      "epoch": 0.19869154349406348,
+      "grad_norm": 0.22649221122264862,
+      "learning_rate": 9.972204715673669e-05,
+      "loss": 0.0873,
+      "step": 820
+    },
+    {
+      "epoch": 0.20111461109764964,
+      "grad_norm": 0.19492876529693604,
+      "learning_rate": 9.970436496555617e-05,
+      "loss": 0.0894,
+      "step": 830
+    },
+    {
+      "epoch": 0.20353767870123576,
+      "grad_norm": 0.20269566774368286,
+      "learning_rate": 9.968613921540373e-05,
+      "loss": 0.081,
+      "step": 840
+    },
+    {
+      "epoch": 0.20596074630482192,
+      "grad_norm": 0.24085013568401337,
+      "learning_rate": 9.966737010559326e-05,
+      "loss": 0.0989,
+      "step": 850
+    },
+    {
+      "epoch": 0.20838381390840804,
+      "grad_norm": 0.15436244010925293,
+      "learning_rate": 9.964805784138072e-05,
+      "loss": 0.091,
+      "step": 860
+    },
+    {
+      "epoch": 0.21080688151199417,
+      "grad_norm": 0.17235654592514038,
+      "learning_rate": 9.962820263396195e-05,
+      "loss": 0.0862,
+      "step": 870
+    },
+    {
+      "epoch": 0.21322994911558033,
+      "grad_norm": 0.20409919321537018,
+      "learning_rate": 9.960780470047033e-05,
+      "loss": 0.0797,
+      "step": 880
+    },
+    {
+      "epoch": 0.21565301671916645,
+      "grad_norm": 0.20658552646636963,
+      "learning_rate": 9.958686426397437e-05,
+      "loss": 0.0778,
+      "step": 890
+    },
+    {
+      "epoch": 0.2180760843227526,
+      "grad_norm": 0.1812807023525238,
+      "learning_rate": 9.956538155347534e-05,
+      "loss": 0.0903,
+      "step": 900
+    },
+    {
+      "epoch": 0.22049915192633873,
+      "grad_norm": 0.21331733465194702,
+      "learning_rate": 9.95433568039047e-05,
+      "loss": 0.0681,
+      "step": 910
+    },
+    {
+      "epoch": 0.2229222195299249,
+      "grad_norm": 0.17142294347286224,
+      "learning_rate": 9.952079025612162e-05,
+      "loss": 0.0863,
+      "step": 920
+    },
+    {
+      "epoch": 0.225345287133511,
+      "grad_norm": 0.3264574706554413,
+      "learning_rate": 9.949768215691022e-05,
+      "loss": 0.0782,
+      "step": 930
+    },
+    {
+      "epoch": 0.22776835473709717,
+      "grad_norm": 0.18573996424674988,
+      "learning_rate": 9.9474032758977e-05,
+      "loss": 0.075,
+      "step": 940
+    },
+    {
+      "epoch": 0.2301914223406833,
+      "grad_norm": 0.2647434175014496,
+      "learning_rate": 9.944984232094794e-05,
+      "loss": 0.089,
+      "step": 950
+    },
+    {
+      "epoch": 0.23261448994426945,
+      "grad_norm": 0.1720058023929596,
+      "learning_rate": 9.942511110736584e-05,
+      "loss": 0.0709,
+      "step": 960
+    },
+    {
+      "epoch": 0.23503755754785557,
+      "grad_norm": 0.23745214939117432,
+      "learning_rate": 9.939983938868726e-05,
+      "loss": 0.0727,
+      "step": 970
+    },
+    {
+      "epoch": 0.23746062515144173,
+      "grad_norm": 0.20860709249973297,
+      "learning_rate": 9.93740274412797e-05,
+      "loss": 0.0834,
+      "step": 980
+    },
+    {
+      "epoch": 0.23988369275502786,
+      "grad_norm": 0.19187010824680328,
+      "learning_rate": 9.934767554741846e-05,
+      "loss": 0.0789,
+      "step": 990
+    },
+    {
+      "epoch": 0.242306760358614,
+      "grad_norm": 0.1633918285369873,
+      "learning_rate": 9.932078399528361e-05,
+      "loss": 0.0798,
+      "step": 1000
+    },
+    {
+      "epoch": 0.24472982796220014,
+      "grad_norm": 0.15381617844104767,
+      "learning_rate": 9.929335307895689e-05,
+      "loss": 0.0701,
+      "step": 1010
+    },
+    {
+      "epoch": 0.2471528955657863,
+      "grad_norm": 0.17352166771888733,
+      "learning_rate": 9.926538309841839e-05,
+      "loss": 0.0792,
+      "step": 1020
+    },
+    {
+      "epoch": 0.24957596316937242,
+      "grad_norm": 0.40674543380737305,
+      "learning_rate": 9.923687435954334e-05,
+      "loss": 0.0754,
+      "step": 1030
+    },
+    {
+      "epoch": 0.25199903077295854,
+      "grad_norm": 0.2632424831390381,
+      "learning_rate": 9.920782717409873e-05,
+      "loss": 0.0773,
+      "step": 1040
+    },
+    {
+      "epoch": 0.2544220983765447,
+      "grad_norm": 0.19672192633152008,
+      "learning_rate": 9.917824185973994e-05,
+      "loss": 0.0688,
+      "step": 1050
+    },
+    {
+      "epoch": 0.25684516598013085,
+      "grad_norm": 0.22081975638866425,
+      "learning_rate": 9.914811874000723e-05,
+      "loss": 0.0875,
+      "step": 1060
+    },
+    {
+      "epoch": 0.259268233583717,
+      "grad_norm": 0.35084763169288635,
+      "learning_rate": 9.911745814432218e-05,
+      "loss": 0.08,
+      "step": 1070
+    },
+    {
+      "epoch": 0.2616913011873031,
+      "grad_norm": 0.25521281361579895,
+      "learning_rate": 9.90862604079842e-05,
+      "loss": 0.0842,
+      "step": 1080
+    },
+    {
+      "epoch": 0.2641143687908893,
+      "grad_norm": 0.18135966360569,
+      "learning_rate": 9.90545258721667e-05,
+      "loss": 0.0632,
+      "step": 1090
+    },
+    {
+      "epoch": 0.2665374363944754,
+      "grad_norm": 0.21649938821792603,
+      "learning_rate": 9.90222548839135e-05,
+      "loss": 0.0737,
+      "step": 1100
+    },
+    {
+      "epoch": 0.26896050399806154,
+      "grad_norm": 0.22486425936222076,
+      "learning_rate": 9.898944779613495e-05,
+      "loss": 0.0781,
+      "step": 1110
+    },
+    {
+      "epoch": 0.27138357160164767,
+      "grad_norm": 0.1296594738960266,
+      "learning_rate": 9.89561049676041e-05,
+      "loss": 0.0712,
+      "step": 1120
+    },
+    {
+      "epoch": 0.27380663920523385,
+      "grad_norm": 0.20417578518390656,
+      "learning_rate": 9.89222267629528e-05,
+      "loss": 0.0754,
+      "step": 1130
+    },
+    {
+      "epoch": 0.27622970680882,
+      "grad_norm": 0.36884164810180664,
+      "learning_rate": 9.888781355266763e-05,
+      "loss": 0.0788,
+      "step": 1140
+    },
+    {
+      "epoch": 0.2786527744124061,
+      "grad_norm": 0.1669790893793106,
+      "learning_rate": 9.885286571308598e-05,
+      "loss": 0.0752,
+      "step": 1150
+    },
+    {
+      "epoch": 0.2810758420159922,
+      "grad_norm": 0.22811958193778992,
+      "learning_rate": 9.881738362639182e-05,
+      "loss": 0.0694,
+      "step": 1160
+    },
+    {
+      "epoch": 0.2834989096195784,
+      "grad_norm": 0.17611163854599,
+      "learning_rate": 9.878136768061154e-05,
+      "loss": 0.0747,
+      "step": 1170
+    },
+    {
+      "epoch": 0.28592197722316454,
+      "grad_norm": 0.21130016446113586,
+      "learning_rate": 9.874481826960979e-05,
+      "loss": 0.0723,
+      "step": 1180
+    },
+    {
+      "epoch": 0.28834504482675066,
+      "grad_norm": 0.18295472860336304,
+      "learning_rate": 9.870773579308503e-05,
+      "loss": 0.0717,
+      "step": 1190
+    },
+    {
+      "epoch": 0.2907681124303368,
+      "grad_norm": 0.23597301542758942,
+      "learning_rate": 9.867012065656533e-05,
+      "loss": 0.0757,
+      "step": 1200
+    },
+    {
+      "epoch": 0.29319118003392297,
+      "grad_norm": 0.26741477847099304,
+      "learning_rate": 9.863197327140376e-05,
+      "loss": 0.0731,
+      "step": 1210
+    },
+    {
+      "epoch": 0.2956142476375091,
+      "grad_norm": 0.20920027792453766,
+      "learning_rate": 9.859329405477403e-05,
+      "loss": 0.0636,
+      "step": 1220
+    },
+    {
+      "epoch": 0.2980373152410952,
+      "grad_norm": 0.224365234375,
+      "learning_rate": 9.855408342966585e-05,
+      "loss": 0.0748,
+      "step": 1230
+    },
+    {
+      "epoch": 0.30046038284468135,
+      "grad_norm": 0.1771516501903534,
+      "learning_rate": 9.851434182488033e-05,
+      "loss": 0.0687,
+      "step": 1240
+    },
+    {
+      "epoch": 0.30288345044826753,
+      "grad_norm": 0.21521975100040436,
+      "learning_rate": 9.84740696750253e-05,
+      "loss": 0.0688,
+      "step": 1250
+    },
+    {
+      "epoch": 0.30530651805185366,
+      "grad_norm": 0.21606016159057617,
+      "learning_rate": 9.843326742051055e-05,
+      "loss": 0.0679,
+      "step": 1260
+    },
+    {
+      "epoch": 0.3077295856554398,
+      "grad_norm": 0.38219350576400757,
+      "learning_rate": 9.839193550754297e-05,
+      "loss": 0.0731,
+      "step": 1270
+    },
+    {
+      "epoch": 0.3101526532590259,
+      "grad_norm": 0.35934513807296753,
+      "learning_rate": 9.835007438812177e-05,
+      "loss": 0.0754,
+      "step": 1280
+    },
+    {
+      "epoch": 0.3125757208626121,
+      "grad_norm": 0.27372294664382935,
+      "learning_rate": 9.830768452003341e-05,
+      "loss": 0.0703,
+      "step": 1290
+    },
+    {
+      "epoch": 0.3149987884661982,
+      "grad_norm": 0.19996874034404755,
+      "learning_rate": 9.826476636684671e-05,
+      "loss": 0.0646,
+      "step": 1300
+    },
+    {
+      "epoch": 0.31742185606978435,
+      "grad_norm": 0.22204864025115967,
+      "learning_rate": 9.822132039790773e-05,
+      "loss": 0.069,
+      "step": 1310
+    },
+    {
+      "epoch": 0.3198449236733705,
+      "grad_norm": 0.23537422716617584,
+      "learning_rate": 9.817734708833461e-05,
+      "loss": 0.0722,
+      "step": 1320
+    },
+    {
+      "epoch": 0.3222679912769566,
+      "grad_norm": 0.1366153061389923,
+      "learning_rate": 9.813284691901243e-05,
+      "loss": 0.0602,
+      "step": 1330
+    },
+    {
+      "epoch": 0.3246910588805428,
+      "grad_norm": 0.2823709547519684,
+      "learning_rate": 9.808782037658792e-05,
+      "loss": 0.0669,
+      "step": 1340
+    },
+    {
+      "epoch": 0.3271141264841289,
+      "grad_norm": 0.23131944239139557,
+      "learning_rate": 9.804226795346411e-05,
+      "loss": 0.0625,
+      "step": 1350
+    },
+    {
+      "epoch": 0.32953719408771504,
+      "grad_norm": 0.18494287133216858,
+      "learning_rate": 9.799619014779503e-05,
+      "loss": 0.0602,
+      "step": 1360
+    },
+    {
+      "epoch": 0.33196026169130116,
+      "grad_norm": 0.26136401295661926,
+      "learning_rate": 9.794958746348013e-05,
+      "loss": 0.0593,
+      "step": 1370
+    },
+    {
+      "epoch": 0.33438332929488734,
+      "grad_norm": 0.16153356432914734,
+      "learning_rate": 9.790246041015896e-05,
+      "loss": 0.0652,
+      "step": 1380
+    },
+    {
+      "epoch": 0.33680639689847347,
+      "grad_norm": 0.14857329428195953,
+      "learning_rate": 9.785480950320538e-05,
+      "loss": 0.0556,
+      "step": 1390
+    },
+    {
+      "epoch": 0.3392294645020596,
+      "grad_norm": 0.18751142919063568,
+      "learning_rate": 9.78066352637221e-05,
+      "loss": 0.0669,
+      "step": 1400
+    },
+    {
+      "epoch": 0.3416525321056457,
+      "grad_norm": 0.18614909052848816,
+      "learning_rate": 9.775793821853488e-05,
+      "loss": 0.0582,
+      "step": 1410
+    },
+    {
+      "epoch": 0.3440755997092319,
+      "grad_norm": 0.1765611320734024,
+      "learning_rate": 9.77087189001868e-05,
+      "loss": 0.0661,
+      "step": 1420
+    },
+    {
+      "epoch": 0.34649866731281803,
+      "grad_norm": 0.1687687337398529,
+      "learning_rate": 9.765897784693243e-05,
+      "loss": 0.0656,
+      "step": 1430
+    },
+    {
+      "epoch": 0.34892173491640416,
+      "grad_norm": 0.1597665697336197,
+      "learning_rate": 9.760871560273197e-05,
+      "loss": 0.0619,
+      "step": 1440
+    },
+    {
+      "epoch": 0.3513448025199903,
+      "grad_norm": 0.29115739464759827,
+      "learning_rate": 9.755793271724526e-05,
+      "loss": 0.0703,
+      "step": 1450
+    },
+    {
+      "epoch": 0.35376787012357647,
+      "grad_norm": 0.36171144247055054,
+      "learning_rate": 9.750662974582584e-05,
+      "loss": 0.0638,
+      "step": 1460
+    },
+    {
+      "epoch": 0.3561909377271626,
+      "grad_norm": 0.24505296349525452,
+      "learning_rate": 9.745480724951473e-05,
+      "loss": 0.07,
+      "step": 1470
+    },
+    {
+      "epoch": 0.3586140053307487,
+      "grad_norm": 0.26854684948921204,
+      "learning_rate": 9.740246579503447e-05,
+      "loss": 0.072,
+      "step": 1480
+    },
+    {
+      "epoch": 0.36103707293433485,
+      "grad_norm": 0.23773345351219177,
+      "learning_rate": 9.734960595478284e-05,
+      "loss": 0.0567,
+      "step": 1490
+    },
+    {
+      "epoch": 0.36346014053792103,
+      "grad_norm": 0.2043919414281845,
+      "learning_rate": 9.729622830682657e-05,
+      "loss": 0.0778,
+      "step": 1500
+    },
+    {
+      "epoch": 0.36588320814150715,
+      "grad_norm": 0.17598839104175568,
+      "learning_rate": 9.724233343489504e-05,
+      "loss": 0.0615,
+      "step": 1510
+    },
+    {
+      "epoch": 0.3683062757450933,
+      "grad_norm": 0.17013928294181824,
+      "learning_rate": 9.718792192837396e-05,
+      "loss": 0.0599,
+      "step": 1520
+    },
+    {
+      "epoch": 0.3707293433486794,
+      "grad_norm": 0.19057349860668182,
+      "learning_rate": 9.713299438229886e-05,
+      "loss": 0.0671,
+      "step": 1530
+    },
+    {
+      "epoch": 0.3731524109522656,
+      "grad_norm": 0.21346819400787354,
+      "learning_rate": 9.707755139734855e-05,
+      "loss": 0.0726,
+      "step": 1540
+    },
+    {
+      "epoch": 0.3755754785558517,
+      "grad_norm": 0.18674880266189575,
+      "learning_rate": 9.702159357983866e-05,
+      "loss": 0.0655,
+      "step": 1550
+    },
+    {
+      "epoch": 0.37799854615943784,
+      "grad_norm": 0.1875901222229004,
+      "learning_rate": 9.696512154171492e-05,
+      "loss": 0.0617,
+      "step": 1560
+    },
+    {
+      "epoch": 0.38042161376302397,
+      "grad_norm": 0.2749476432800293,
+      "learning_rate": 9.690813590054645e-05,
+      "loss": 0.0571,
+      "step": 1570
+    },
+    {
+      "epoch": 0.38284468136661015,
+      "grad_norm": 0.29953062534332275,
+      "learning_rate": 9.685063727951914e-05,
+      "loss": 0.0521,
+      "step": 1580
+    },
+    {
+      "epoch": 0.3852677489701963,
+      "grad_norm": 0.181138813495636,
+      "learning_rate": 9.679262630742865e-05,
+      "loss": 0.0632,
+      "step": 1590
+    },
+    {
+      "epoch": 0.3876908165737824,
+      "grad_norm": 0.16172991693019867,
+      "learning_rate": 9.673410361867373e-05,
+      "loss": 0.0576,
+      "step": 1600
+    },
+    {
+      "epoch": 0.39011388417736853,
+      "grad_norm": 0.20111921429634094,
+      "learning_rate": 9.667506985324909e-05,
+      "loss": 0.0582,
+      "step": 1610
+    },
+    {
+      "epoch": 0.3925369517809547,
+      "grad_norm": 0.29993516206741333,
+      "learning_rate": 9.661552565673855e-05,
+      "loss": 0.0576,
+      "step": 1620
+    },
+    {
+      "epoch": 0.39496001938454084,
+      "grad_norm": 0.2669925093650818,
+      "learning_rate": 9.655547168030789e-05,
+      "loss": 0.0541,
+      "step": 1630
+    },
+    {
+      "epoch": 0.39738308698812697,
+      "grad_norm": 0.15487892925739288,
+      "learning_rate": 9.649490858069777e-05,
+      "loss": 0.0606,
+      "step": 1640
+    },
+    {
+      "epoch": 0.3998061545917131,
+      "grad_norm": 0.2327035367488861,
+      "learning_rate": 9.643383702021658e-05,
+      "loss": 0.0676,
+      "step": 1650
+    },
+    {
+      "epoch": 0.4022292221952993,
+      "grad_norm": 0.20327667891979218,
+      "learning_rate": 9.637225766673307e-05,
+      "loss": 0.0483,
+      "step": 1660
+    },
+    {
+      "epoch": 0.4046522897988854,
+      "grad_norm": 0.2192804366350174,
+      "learning_rate": 9.631017119366922e-05,
+      "loss": 0.0588,
+      "step": 1670
+    },
+    {
+      "epoch": 0.4070753574024715,
+      "grad_norm": 0.18243132531642914,
+      "learning_rate": 9.624757827999273e-05,
+      "loss": 0.0646,
+      "step": 1680
+    },
+    {
+      "epoch": 0.40949842500605765,
+      "grad_norm": 0.21957585215568542,
+      "learning_rate": 9.618447961020971e-05,
+      "loss": 0.0597,
+      "step": 1690
+    },
+    {
+      "epoch": 0.41192149260964384,
+      "grad_norm": 0.2594870924949646,
+      "learning_rate": 9.612087587435707e-05,
+      "loss": 0.0624,
+      "step": 1700
+    },
+    {
+      "epoch": 0.41434456021322996,
+      "grad_norm": 0.15151965618133545,
+      "learning_rate": 9.605676776799508e-05,
+      "loss": 0.0571,
+      "step": 1710
+    },
+    {
+      "epoch": 0.4167676278168161,
+      "grad_norm": 0.2595440447330475,
+      "learning_rate": 9.599215599219973e-05,
+      "loss": 0.0645,
+      "step": 1720
+    },
+    {
+      "epoch": 0.4191906954204022,
+      "grad_norm": 0.17983701825141907,
+      "learning_rate": 9.592704125355505e-05,
+      "loss": 0.062,
+      "step": 1730
+    },
+    {
+      "epoch": 0.42161376302398834,
+      "grad_norm": 0.21827834844589233,
+      "learning_rate": 9.586142426414538e-05,
+      "loss": 0.0531,
+      "step": 1740
+    },
+    {
+      "epoch": 0.4240368306275745,
+      "grad_norm": 0.2171270102262497,
+      "learning_rate": 9.57953057415476e-05,
+      "loss": 0.0604,
+      "step": 1750
+    },
+    {
+      "epoch": 0.42645989823116065,
+      "grad_norm": 0.20583873987197876,
+      "learning_rate": 9.572868640882328e-05,
+      "loss": 0.0661,
+      "step": 1760
+    },
+    {
+      "epoch": 0.4288829658347468,
+      "grad_norm": 0.19157806038856506,
+      "learning_rate": 9.56615669945108e-05,
+      "loss": 0.0612,
+      "step": 1770
+    },
+    {
+      "epoch": 0.4313060334383329,
+      "grad_norm": 0.12651243805885315,
+      "learning_rate": 9.55939482326173e-05,
+      "loss": 0.0588,
+      "step": 1780
+    },
+    {
+      "epoch": 0.4337291010419191,
+      "grad_norm": 0.17638131976127625,
+      "learning_rate": 9.552583086261069e-05,
+      "loss": 0.0618,
+      "step": 1790
+    },
+    {
+      "epoch": 0.4361521686455052,
+      "grad_norm": 0.213279590010643,
+      "learning_rate": 9.545721562941168e-05,
+      "loss": 0.0591,
+      "step": 1800
+    },
+    {
+      "epoch": 0.43857523624909134,
+      "grad_norm": 0.23938247561454773,
+      "learning_rate": 9.538810328338543e-05,
+      "loss": 0.057,
+      "step": 1810
+    },
+    {
+      "epoch": 0.44099830385267746,
+      "grad_norm": 0.1525634527206421,
+      "learning_rate": 9.531849458033349e-05,
+      "loss": 0.0486,
+      "step": 1820
+    },
+    {
+      "epoch": 0.44342137145626365,
+      "grad_norm": 0.3473944067955017,
+      "learning_rate": 9.524839028148547e-05,
+      "loss": 0.0552,
+      "step": 1830
+    },
+    {
+      "epoch": 0.4458444390598498,
+      "grad_norm": 0.18791596591472626,
+      "learning_rate": 9.517779115349077e-05,
+      "loss": 0.053,
+      "step": 1840
+    },
+    {
+      "epoch": 0.4482675066634359,
+      "grad_norm": 0.28670239448547363,
+      "learning_rate": 9.510669796841014e-05,
+      "loss": 0.0517,
+      "step": 1850
+    },
+    {
+      "epoch": 0.450690574267022,
+      "grad_norm": 0.3638632297515869,
+      "learning_rate": 9.503511150370727e-05,
+      "loss": 0.0651,
+      "step": 1860
+    },
+    {
+      "epoch": 0.4531136418706082,
+      "grad_norm": 0.22422659397125244,
+      "learning_rate": 9.496303254224024e-05,
+      "loss": 0.0556,
+      "step": 1870
+    },
+    {
+      "epoch": 0.45553670947419433,
+      "grad_norm": 0.2053101509809494,
+      "learning_rate": 9.489046187225306e-05,
+      "loss": 0.0499,
+      "step": 1880
+    },
+    {
+      "epoch": 0.45795977707778046,
+      "grad_norm": 0.2781410217285156,
+      "learning_rate": 9.481740028736692e-05,
+      "loss": 0.0491,
+      "step": 1890
+    },
+    {
+      "epoch": 0.4603828446813666,
+      "grad_norm": 0.18476931750774384,
+      "learning_rate": 9.474384858657164e-05,
+      "loss": 0.0628,
+      "step": 1900
+    },
+    {
+      "epoch": 0.46280591228495277,
+      "grad_norm": 0.19517628848552704,
+      "learning_rate": 9.466980757421679e-05,
+      "loss": 0.0603,
+      "step": 1910
+    },
+    {
+      "epoch": 0.4652289798885389,
+      "grad_norm": 0.1381346881389618,
+      "learning_rate": 9.459527806000305e-05,
+      "loss": 0.0563,
+      "step": 1920
+    },
+    {
+      "epoch": 0.467652047492125,
+      "grad_norm": 0.16588693857192993,
+      "learning_rate": 9.452026085897325e-05,
+      "loss": 0.0571,
+      "step": 1930
+    },
+    {
+      "epoch": 0.47007511509571115,
+      "grad_norm": 0.30310162901878357,
+      "learning_rate": 9.444475679150348e-05,
+      "loss": 0.0593,
+      "step": 1940
+    },
+    {
+      "epoch": 0.47249818269929733,
+      "grad_norm": 0.18577975034713745,
+      "learning_rate": 9.436876668329411e-05,
+      "loss": 0.0587,
+      "step": 1950
+    },
+    {
+      "epoch": 0.47492125030288346,
+      "grad_norm": 0.17943626642227173,
+      "learning_rate": 9.429229136536079e-05,
+      "loss": 0.0503,
+      "step": 1960
+    },
+    {
+      "epoch": 0.4773443179064696,
+      "grad_norm": 0.25463947653770447,
+      "learning_rate": 9.421533167402534e-05,
+      "loss": 0.0644,
+      "step": 1970
+    },
+    {
+      "epoch": 0.4797673855100557,
+      "grad_norm": 0.2563469111919403,
+      "learning_rate": 9.413788845090666e-05,
+      "loss": 0.066,
+      "step": 1980
+    },
+    {
+      "epoch": 0.4821904531136419,
+      "grad_norm": 0.19609716534614563,
+      "learning_rate": 9.405996254291136e-05,
+      "loss": 0.0564,
+      "step": 1990
+    },
+    {
+      "epoch": 0.484613520717228,
+      "grad_norm": 0.23780830204486847,
+      "learning_rate": 9.398155480222474e-05,
+      "loss": 0.051,
+      "step": 2000
+    },
+    {
+      "epoch": 0.48703658832081415,
+      "grad_norm": 0.16558371484279633,
+      "learning_rate": 9.390266608630128e-05,
+      "loss": 0.0544,
+      "step": 2010
+    },
+    {
+      "epoch": 0.48945965592440027,
+      "grad_norm": 0.22054706513881683,
+      "learning_rate": 9.38232972578553e-05,
+      "loss": 0.056,
+      "step": 2020
+    },
+    {
+      "epoch": 0.49188272352798645,
+      "grad_norm": 0.1278141885995865,
+      "learning_rate": 9.374344918485164e-05,
+      "loss": 0.0573,
+      "step": 2030
+    },
+    {
+      "epoch": 0.4943057911315726,
+      "grad_norm": 0.168584942817688,
+      "learning_rate": 9.366312274049602e-05,
+      "loss": 0.06,
+      "step": 2040
+    },
+    {
+      "epoch": 0.4967288587351587,
+      "grad_norm": 0.25215259194374084,
+      "learning_rate": 9.358231880322554e-05,
+      "loss": 0.0544,
+      "step": 2050
+    },
+    {
+      "epoch": 0.49915192633874483,
+      "grad_norm": 0.13378572463989258,
+      "learning_rate": 9.350103825669916e-05,
+      "loss": 0.0461,
+      "step": 2060
+    },
+    {
+      "epoch": 0.501574993942331,
+      "grad_norm": 0.2712815999984741,
+      "learning_rate": 9.341928198978787e-05,
+      "loss": 0.0492,
+      "step": 2070
+    },
+    {
+      "epoch": 0.5039980615459171,
+      "grad_norm": 0.1256439983844757,
+      "learning_rate": 9.333705089656512e-05,
+      "loss": 0.0467,
+      "step": 2080
+    },
+    {
+      "epoch": 0.5064211291495033,
+      "grad_norm": 0.19592291116714478,
+      "learning_rate": 9.325434587629698e-05,
+      "loss": 0.0524,
+      "step": 2090
+    },
+    {
+      "epoch": 0.5088441967530895,
+      "grad_norm": 0.13289900124073029,
+      "learning_rate": 9.31711678334323e-05,
+      "loss": 0.047,
+      "step": 2100
+    },
+    {
+      "epoch": 0.5112672643566756,
+      "grad_norm": 0.21564555168151855,
+      "learning_rate": 9.308751767759282e-05,
+      "loss": 0.0498,
+      "step": 2110
+    },
+    {
+      "epoch": 0.5136903319602617,
+      "grad_norm": 0.21520939469337463,
+      "learning_rate": 9.300339632356325e-05,
+      "loss": 0.0466,
+      "step": 2120
+    },
+    {
+      "epoch": 0.5161133995638478,
+      "grad_norm": 0.18444831669330597,
+      "learning_rate": 9.291880469128124e-05,
+      "loss": 0.0514,
+      "step": 2130
+    },
+    {
+      "epoch": 0.518536467167434,
+      "grad_norm": 0.19666188955307007,
+      "learning_rate": 9.283374370582732e-05,
+      "loss": 0.0535,
+      "step": 2140
+    },
+    {
+      "epoch": 0.5209595347710201,
+      "grad_norm": 0.20858798921108246,
+      "learning_rate": 9.274821429741482e-05,
+      "loss": 0.0596,
+      "step": 2150
+    },
+    {
+      "epoch": 0.5233826023746062,
+      "grad_norm": 0.2382081151008606,
+      "learning_rate": 9.266221740137961e-05,
+      "loss": 0.0501,
+      "step": 2160
+    },
+    {
+      "epoch": 0.5258056699781924,
+      "grad_norm": 0.16628485918045044,
+      "learning_rate": 9.257575395817001e-05,
+      "loss": 0.047,
+      "step": 2170
+    },
+    {
+      "epoch": 0.5282287375817786,
+      "grad_norm": 0.18956221640110016,
+      "learning_rate": 9.248882491333637e-05,
+      "loss": 0.0554,
+      "step": 2180
+    },
+    {
+      "epoch": 0.5306518051853647,
+      "grad_norm": 0.2803729772567749,
+      "learning_rate": 9.240143121752076e-05,
+      "loss": 0.0471,
+      "step": 2190
+    },
+    {
+      "epoch": 0.5330748727889508,
+      "grad_norm": 0.13001462817192078,
+      "learning_rate": 9.23135738264467e-05,
+      "loss": 0.0486,
+      "step": 2200
+    },
+    {
+      "epoch": 0.535497940392537,
+      "grad_norm": 0.15444600582122803,
+      "learning_rate": 9.222525370090849e-05,
+      "loss": 0.0441,
+      "step": 2210
+    },
+    {
+      "epoch": 0.5379210079961231,
+      "grad_norm": 0.35646572709083557,
+      "learning_rate": 9.213647180676088e-05,
+      "loss": 0.0494,
+      "step": 2220
+    },
+    {
+      "epoch": 0.5403440755997092,
+      "grad_norm": 0.24044056236743927,
+      "learning_rate": 9.204722911490846e-05,
+      "loss": 0.0515,
+      "step": 2230
+    },
+    {
+      "epoch": 0.5427671432032953,
+      "grad_norm": 0.2854156494140625,
+      "learning_rate": 9.1957526601295e-05,
+      "loss": 0.046,
+      "step": 2240
+    },
+    {
+      "epoch": 0.5451902108068815,
+      "grad_norm": 0.2488689422607422,
+      "learning_rate": 9.186736524689281e-05,
+      "loss": 0.0545,
+      "step": 2250
+    },
+    {
+      "epoch": 0.5476132784104677,
+      "grad_norm": 0.15807735919952393,
+      "learning_rate": 9.177674603769204e-05,
+      "loss": 0.0558,
+      "step": 2260
+    },
+    {
+      "epoch": 0.5500363460140538,
+      "grad_norm": 0.20337934792041779,
+      "learning_rate": 9.168566996468983e-05,
+      "loss": 0.0496,
+      "step": 2270
+    },
+    {
+      "epoch": 0.55245941361764,
+      "grad_norm": 0.1580781489610672,
+      "learning_rate": 9.159413802387951e-05,
+      "loss": 0.0568,
+      "step": 2280
+    },
+    {
+      "epoch": 0.5548824812212261,
+      "grad_norm": 0.1751731038093567,
+      "learning_rate": 9.150215121623974e-05,
+      "loss": 0.051,
+      "step": 2290
+    },
+    {
+      "epoch": 0.5573055488248122,
+      "grad_norm": 0.1994052231311798,
+      "learning_rate": 9.140971054772349e-05,
+      "loss": 0.0441,
+      "step": 2300
+    },
+    {
+      "epoch": 0.5597286164283983,
+      "grad_norm": 0.24094025790691376,
+      "learning_rate": 9.131681702924713e-05,
+      "loss": 0.0577,
+      "step": 2310
+    },
+    {
+      "epoch": 0.5621516840319845,
+      "grad_norm": 0.17011752724647522,
+      "learning_rate": 9.122347167667926e-05,
+      "loss": 0.0428,
+      "step": 2320
+    },
+    {
+      "epoch": 0.5645747516355706,
+      "grad_norm": 0.16502590477466583,
+      "learning_rate": 9.112967551082973e-05,
+      "loss": 0.0583,
+      "step": 2330
+    },
+    {
+      "epoch": 0.5669978192391568,
+      "grad_norm": 0.2089923918247223,
+      "learning_rate": 9.103542955743835e-05,
+      "loss": 0.0491,
+      "step": 2340
+    },
+    {
+      "epoch": 0.569420886842743,
+      "grad_norm": 0.1554768830537796,
+      "learning_rate": 9.094073484716381e-05,
+      "loss": 0.054,
+      "step": 2350
+    },
+    {
+      "epoch": 0.5718439544463291,
+      "grad_norm": 0.2179958075284958,
+      "learning_rate": 9.084559241557226e-05,
+      "loss": 0.0445,
+      "step": 2360
+    },
+    {
+      "epoch": 0.5742670220499152,
+      "grad_norm": 0.16002824902534485,
+      "learning_rate": 9.075000330312608e-05,
+      "loss": 0.0506,
+      "step": 2370
+    },
+    {
+      "epoch": 0.5766900896535013,
+      "grad_norm": 0.28972455859184265,
+      "learning_rate": 9.065396855517253e-05,
+      "loss": 0.0424,
+      "step": 2380
+    },
+    {
+      "epoch": 0.5791131572570875,
+      "grad_norm": 0.23938970267772675,
+      "learning_rate": 9.055748922193219e-05,
+      "loss": 0.0551,
+      "step": 2390
+    },
+    {
+      "epoch": 0.5815362248606736,
+      "grad_norm": 0.15034063160419464,
+      "learning_rate": 9.046056635848761e-05,
+      "loss": 0.051,
+      "step": 2400
+    },
+    {
+      "epoch": 0.5839592924642597,
+      "grad_norm": 0.18481327593326569,
+      "learning_rate": 9.036320102477169e-05,
+      "loss": 0.047,
+      "step": 2410
+    },
+    {
+      "epoch": 0.5863823600678459,
+      "grad_norm": 0.23522192239761353,
+      "learning_rate": 9.02653942855561e-05,
+      "loss": 0.0478,
+      "step": 2420
+    },
+    {
+      "epoch": 0.5888054276714321,
+      "grad_norm": 0.25239166617393494,
+      "learning_rate": 9.016714721043971e-05,
+      "loss": 0.0458,
+      "step": 2430
+    },
+    {
+      "epoch": 0.5912284952750182,
+      "grad_norm": 0.256253182888031,
+      "learning_rate": 9.006846087383675e-05,
+      "loss": 0.053,
+      "step": 2440
+    },
+    {
+      "epoch": 0.5936515628786043,
+      "grad_norm": 0.20433920621871948,
+      "learning_rate": 8.996933635496523e-05,
+      "loss": 0.0485,
+      "step": 2450
+    },
+    {
+      "epoch": 0.5960746304821904,
+      "grad_norm": 0.18060754239559174,
+      "learning_rate": 8.986977473783498e-05,
+      "loss": 0.0515,
+      "step": 2460
+    },
+    {
+      "epoch": 0.5984976980857766,
+      "grad_norm": 0.1614294946193695,
+      "learning_rate": 8.97697771112359e-05,
+      "loss": 0.0481,
+      "step": 2470
+    },
+    {
+      "epoch": 0.6009207656893627,
+      "grad_norm": 0.181077778339386,
+      "learning_rate": 8.966934456872602e-05,
+      "loss": 0.0483,
+      "step": 2480
+    },
+    {
+      "epoch": 0.6033438332929488,
+      "grad_norm": 0.1722615659236908,
+      "learning_rate": 8.95684782086195e-05,
+      "loss": 0.041,
+      "step": 2490
+    },
+    {
+      "epoch": 0.6057669008965351,
+      "grad_norm": 0.14602269232273102,
+      "learning_rate": 8.946717913397476e-05,
+      "loss": 0.0493,
+      "step": 2500
+    },
+    {
+      "epoch": 0.6081899685001212,
+      "grad_norm": 0.1512426882982254,
+      "learning_rate": 8.93654484525822e-05,
+      "loss": 0.0497,
+      "step": 2510
+    },
+    {
+      "epoch": 0.6106130361037073,
+      "grad_norm": 0.13178949058055878,
+      "learning_rate": 8.926328727695226e-05,
+      "loss": 0.046,
+      "step": 2520
+    },
+    {
+      "epoch": 0.6130361037072934,
+      "grad_norm": 0.16389816999435425,
+      "learning_rate": 8.916069672430319e-05,
+      "loss": 0.0441,
+      "step": 2530
+    },
+    {
+      "epoch": 0.6154591713108796,
+      "grad_norm": 0.210100919008255,
+      "learning_rate": 8.905767791654884e-05,
+      "loss": 0.0492,
+      "step": 2540
+    },
+    {
+      "epoch": 0.6178822389144657,
+      "grad_norm": 0.2015194296836853,
+      "learning_rate": 8.895423198028638e-05,
+      "loss": 0.057,
+      "step": 2550
+    },
+    {
+      "epoch": 0.6203053065180518,
+      "grad_norm": 0.24084636569023132,
+      "learning_rate": 8.885036004678402e-05,
+      "loss": 0.0506,
+      "step": 2560
+    },
+    {
+      "epoch": 0.622728374121638,
+      "grad_norm": 0.1615552008152008,
+      "learning_rate": 8.874606325196857e-05,
+      "loss": 0.0474,
+      "step": 2570
+    },
+    {
+      "epoch": 0.6251514417252242,
+      "grad_norm": 0.17708715796470642,
+      "learning_rate": 8.864134273641304e-05,
+      "loss": 0.0469,
+      "step": 2580
+    },
+    {
+      "epoch": 0.6275745093288103,
+      "grad_norm": 0.17544691264629364,
+      "learning_rate": 8.853619964532427e-05,
+      "loss": 0.0492,
+      "step": 2590
+    },
+    {
+      "epoch": 0.6299975769323964,
+      "grad_norm": 0.1414787769317627,
+      "learning_rate": 8.843063512853019e-05,
+      "loss": 0.0457,
+      "step": 2600
+    },
+    {
+      "epoch": 0.6324206445359826,
+      "grad_norm": 0.14779818058013916,
+      "learning_rate": 8.832465034046749e-05,
+      "loss": 0.0416,
+      "step": 2610
+    },
+    {
+      "epoch": 0.6348437121395687,
+      "grad_norm": 0.19263465702533722,
+      "learning_rate": 8.821824644016882e-05,
+      "loss": 0.0465,
+      "step": 2620
+    },
+    {
+      "epoch": 0.6372667797431548,
+      "grad_norm": 0.3165013790130615,
+      "learning_rate": 8.811142459125019e-05,
+      "loss": 0.0431,
+      "step": 2630
+    },
+    {
+      "epoch": 0.639689847346741,
+      "grad_norm": 0.17729122936725616,
+      "learning_rate": 8.800418596189822e-05,
+      "loss": 0.0405,
+      "step": 2640
+    },
+    {
+      "epoch": 0.6421129149503271,
+      "grad_norm": 0.17129915952682495,
+      "learning_rate": 8.789653172485737e-05,
+      "loss": 0.045,
+      "step": 2650
+    },
+    {
+      "epoch": 0.6445359825539132,
+      "grad_norm": 0.17606666684150696,
+      "learning_rate": 8.778846305741715e-05,
+      "loss": 0.056,
+      "step": 2660
+    },
+    {
+      "epoch": 0.6469590501574994,
+      "grad_norm": 0.10136760026216507,
+      "learning_rate": 8.767998114139918e-05,
+      "loss": 0.0491,
+      "step": 2670
+    },
+    {
+      "epoch": 0.6493821177610856,
+      "grad_norm": 0.1798381209373474,
+      "learning_rate": 8.757108716314429e-05,
+      "loss": 0.0505,
+      "step": 2680
+    },
+    {
+      "epoch": 0.6518051853646717,
+      "grad_norm": 0.20055389404296875,
+      "learning_rate": 8.746178231349962e-05,
+      "loss": 0.0452,
+      "step": 2690
+    },
+    {
+      "epoch": 0.6542282529682578,
+      "grad_norm": 0.16758450865745544,
+      "learning_rate": 8.735206778780549e-05,
+      "loss": 0.0527,
+      "step": 2700
+    },
+    {
+      "epoch": 0.6566513205718439,
+      "grad_norm": 0.17579123377799988,
+      "learning_rate": 8.724194478588234e-05,
+      "loss": 0.0444,
+      "step": 2710
+    },
+    {
+      "epoch": 0.6590743881754301,
+      "grad_norm": 0.163347527384758,
+      "learning_rate": 8.713141451201772e-05,
+      "loss": 0.0481,
+      "step": 2720
+    },
+    {
+      "epoch": 0.6614974557790162,
+      "grad_norm": 0.19324535131454468,
+      "learning_rate": 8.702047817495295e-05,
+      "loss": 0.046,
+      "step": 2730
+    },
+    {
+      "epoch": 0.6639205233826023,
+      "grad_norm": 0.21531912684440613,
+      "learning_rate": 8.69091369878701e-05,
+      "loss": 0.0458,
+      "step": 2740
+    },
+    {
+      "epoch": 0.6663435909861886,
+      "grad_norm": 0.12496834993362427,
+      "learning_rate": 8.679739216837849e-05,
+      "loss": 0.0373,
+      "step": 2750
+    },
+    {
+      "epoch": 0.6687666585897747,
+      "grad_norm": 0.14187116920948029,
+      "learning_rate": 8.66852449385016e-05,
+      "loss": 0.0439,
+      "step": 2760
+    },
+    {
+      "epoch": 0.6711897261933608,
+      "grad_norm": 0.1545686572790146,
+      "learning_rate": 8.657269652466356e-05,
+      "loss": 0.0448,
+      "step": 2770
+    },
+    {
+      "epoch": 0.6736127937969469,
+      "grad_norm": 0.20154084265232086,
+      "learning_rate": 8.645974815767577e-05,
+      "loss": 0.0439,
+      "step": 2780
+    },
+    {
+      "epoch": 0.6760358614005331,
+      "grad_norm": 0.1700827032327652,
+      "learning_rate": 8.634640107272351e-05,
+      "loss": 0.0387,
+      "step": 2790
+    },
+    {
+      "epoch": 0.6784589290041192,
+      "grad_norm": 0.1399426907300949,
+      "learning_rate": 8.623265650935234e-05,
+      "loss": 0.0443,
+      "step": 2800
+    },
+    {
+      "epoch": 0.6808819966077053,
+      "grad_norm": 0.16368074715137482,
+      "learning_rate": 8.611851571145456e-05,
+      "loss": 0.0456,
+      "step": 2810
+    },
+    {
+      "epoch": 0.6833050642112914,
+      "grad_norm": 0.22185085713863373,
+      "learning_rate": 8.600397992725566e-05,
+      "loss": 0.0399,
+      "step": 2820
+    },
+    {
+      "epoch": 0.6857281318148777,
+      "grad_norm": 0.16658170521259308,
+      "learning_rate": 8.588905040930061e-05,
+      "loss": 0.0457,
+      "step": 2830
+    },
+    {
+      "epoch": 0.6881511994184638,
+      "grad_norm": 0.1847662329673767,
+      "learning_rate": 8.577372841444022e-05,
+      "loss": 0.0429,
+      "step": 2840
+    },
+    {
+      "epoch": 0.6905742670220499,
+      "grad_norm": 0.20223012566566467,
+      "learning_rate": 8.565801520381736e-05,
+      "loss": 0.0459,
+      "step": 2850
+    },
+    {
+      "epoch": 0.6929973346256361,
+      "grad_norm": 0.1493610143661499,
+      "learning_rate": 8.554191204285313e-05,
+      "loss": 0.0431,
+      "step": 2860
+    },
+    {
+      "epoch": 0.6954204022292222,
+      "grad_norm": 0.2627032399177551,
+      "learning_rate": 8.542542020123315e-05,
+      "loss": 0.0445,
+      "step": 2870
+    },
+    {
+      "epoch": 0.6978434698328083,
+      "grad_norm": 0.1458875983953476,
+      "learning_rate": 8.530854095289347e-05,
+      "loss": 0.0388,
+      "step": 2880
+    },
+    {
+      "epoch": 0.7002665374363944,
+      "grad_norm": 0.1582096666097641,
+      "learning_rate": 8.519127557600688e-05,
+      "loss": 0.0425,
+      "step": 2890
+    },
+    {
+      "epoch": 0.7026896050399806,
+      "grad_norm": 0.12833742797374725,
+      "learning_rate": 8.507362535296871e-05,
+      "loss": 0.0421,
+      "step": 2900
+    },
+    {
+      "epoch": 0.7051126726435668,
+      "grad_norm": 0.2219606339931488,
+      "learning_rate": 8.495559157038299e-05,
+      "loss": 0.0606,
+      "step": 2910
+    },
+    {
+      "epoch": 0.7075357402471529,
+      "grad_norm": 0.2175474762916565,
+      "learning_rate": 8.483717551904823e-05,
+      "loss": 0.04,
+      "step": 2920
+    },
+    {
+      "epoch": 0.7099588078507391,
+      "grad_norm": 0.2039986401796341,
+      "learning_rate": 8.47183784939434e-05,
+      "loss": 0.04,
+      "step": 2930
+    },
+    {
+      "epoch": 0.7123818754543252,
+      "grad_norm": 0.11814648658037186,
+      "learning_rate": 8.459920179421374e-05,
+      "loss": 0.0398,
+      "step": 2940
+    },
+    {
+      "epoch": 0.7148049430579113,
+      "grad_norm": 0.15184038877487183,
+      "learning_rate": 8.447964672315656e-05,
+      "loss": 0.0428,
+      "step": 2950
+    },
+    {
+      "epoch": 0.7172280106614974,
+      "grad_norm": 0.2321150004863739,
+      "learning_rate": 8.435971458820692e-05,
+      "loss": 0.052,
+      "step": 2960
+    },
+    {
+      "epoch": 0.7196510782650836,
+      "grad_norm": 0.19223789870738983,
+      "learning_rate": 8.423940670092345e-05,
+      "loss": 0.0493,
+      "step": 2970
+    },
+    {
+      "epoch": 0.7220741458686697,
+      "grad_norm": 0.21668773889541626,
+      "learning_rate": 8.411872437697394e-05,
+      "loss": 0.0426,
+      "step": 2980
+    },
+    {
+      "epoch": 0.7244972134722559,
+      "grad_norm": 0.16379350423812866,
+      "learning_rate": 8.399766893612096e-05,
+      "loss": 0.0386,
+      "step": 2990
+    },
+    {
+      "epoch": 0.7269202810758421,
+      "grad_norm": 0.19127467274665833,
+      "learning_rate": 8.38762417022074e-05,
+      "loss": 0.0525,
+      "step": 3000
+    },
+    {
+      "epoch": 0.7293433486794282,
+      "grad_norm": 0.1838960200548172,
+      "learning_rate": 8.375444400314204e-05,
+      "loss": 0.0434,
+      "step": 3010
+    },
+    {
+      "epoch": 0.7317664162830143,
+      "grad_norm": 0.18451040983200073,
+      "learning_rate": 8.3632277170885e-05,
+      "loss": 0.0532,
+      "step": 3020
+    },
+    {
+      "epoch": 0.7341894838866004,
+      "grad_norm": 0.1776730865240097,
+      "learning_rate": 8.350974254143318e-05,
+      "loss": 0.0372,
+      "step": 3030
+    },
+    {
+      "epoch": 0.7366125514901866,
+      "grad_norm": 0.15609405934810638,
+      "learning_rate": 8.338684145480566e-05,
+      "loss": 0.036,
+      "step": 3040
+    },
+    {
+      "epoch": 0.7390356190937727,
+      "grad_norm": 0.1353120058774948,
+      "learning_rate": 8.326357525502904e-05,
+      "loss": 0.0439,
+      "step": 3050
+    },
+    {
+      "epoch": 0.7414586866973588,
+      "grad_norm": 0.12224511802196503,
+      "learning_rate": 8.313994529012273e-05,
+      "loss": 0.0446,
+      "step": 3060
+    },
+    {
+      "epoch": 0.7438817543009449,
+      "grad_norm": 0.16622841358184814,
+      "learning_rate": 8.301595291208422e-05,
+      "loss": 0.0491,
+      "step": 3070
+    },
+    {
+      "epoch": 0.7463048219045312,
+      "grad_norm": 0.19702613353729248,
+      "learning_rate": 8.289159947687427e-05,
+      "loss": 0.0461,
+      "step": 3080
+    },
+    {
+      "epoch": 0.7487278895081173,
+      "grad_norm": 0.18081502616405487,
+      "learning_rate": 8.276688634440216e-05,
+      "loss": 0.0433,
+      "step": 3090
+    },
+    {
+      "epoch": 0.7511509571117034,
+      "grad_norm": 0.3770922124385834,
+      "learning_rate": 8.26418148785107e-05,
+      "loss": 0.0485,
+      "step": 3100
+    },
+    {
+      "epoch": 0.7535740247152896,
+      "grad_norm": 0.14518462121486664,
+      "learning_rate": 8.251638644696141e-05,
+      "loss": 0.0493,
+      "step": 3110
+    },
+    {
+      "epoch": 0.7559970923188757,
+      "grad_norm": 0.25315719842910767,
+      "learning_rate": 8.23906024214195e-05,
+      "loss": 0.0409,
+      "step": 3120
+    },
+    {
+      "epoch": 0.7584201599224618,
+      "grad_norm": 0.22215618193149567,
+      "learning_rate": 8.226446417743897e-05,
+      "loss": 0.0437,
+      "step": 3130
+    },
+    {
+      "epoch": 0.7608432275260479,
+      "grad_norm": 0.15837465226650238,
+      "learning_rate": 8.213797309444742e-05,
+      "loss": 0.0454,
+      "step": 3140
+    },
+    {
+      "epoch": 0.7632662951296341,
+      "grad_norm": 0.17210164666175842,
+      "learning_rate": 8.201113055573105e-05,
+      "loss": 0.0456,
+      "step": 3150
+    },
+    {
+      "epoch": 0.7656893627332203,
+      "grad_norm": 0.17189684510231018,
+      "learning_rate": 8.188393794841958e-05,
+      "loss": 0.0397,
+      "step": 3160
+    },
+    {
+      "epoch": 0.7681124303368064,
+      "grad_norm": 0.2261638641357422,
+      "learning_rate": 8.175639666347094e-05,
+      "loss": 0.0441,
+      "step": 3170
+    },
+    {
+      "epoch": 0.7705354979403926,
+      "grad_norm": 0.2089548110961914,
+      "learning_rate": 8.162850809565623e-05,
+      "loss": 0.0446,
+      "step": 3180
+    },
+    {
+      "epoch": 0.7729585655439787,
+      "grad_norm": 0.21397393941879272,
+      "learning_rate": 8.150027364354431e-05,
+      "loss": 0.0459,
+      "step": 3190
+    },
+    {
+      "epoch": 0.7753816331475648,
+      "grad_norm": 0.265156626701355,
+      "learning_rate": 8.137169470948662e-05,
+      "loss": 0.0514,
+      "step": 3200
+    },
+    {
+      "epoch": 0.7778047007511509,
+      "grad_norm": 0.19327251613140106,
+      "learning_rate": 8.124277269960179e-05,
+      "loss": 0.0436,
+      "step": 3210
+    },
+    {
+      "epoch": 0.7802277683547371,
+      "grad_norm": 0.1638793647289276,
+      "learning_rate": 8.111350902376023e-05,
+      "loss": 0.0411,
+      "step": 3220
+    },
+    {
+      "epoch": 0.7826508359583232,
+      "grad_norm": 0.2995148003101349,
+      "learning_rate": 8.098390509556883e-05,
+      "loss": 0.0429,
+      "step": 3230
+    },
+    {
+      "epoch": 0.7850739035619094,
+      "grad_norm": 0.19604302942752838,
+      "learning_rate": 8.085396233235536e-05,
+      "loss": 0.0374,
+      "step": 3240
+    },
+    {
+      "epoch": 0.7874969711654956,
+      "grad_norm": 0.13748332858085632,
+      "learning_rate": 8.072368215515306e-05,
+      "loss": 0.0316,
+      "step": 3250
+    },
+    {
+      "epoch": 0.7899200387690817,
+      "grad_norm": 0.1680527925491333,
+      "learning_rate": 8.059306598868506e-05,
+      "loss": 0.0511,
+      "step": 3260
+    },
+    {
+      "epoch": 0.7923431063726678,
+      "grad_norm": 0.21089929342269897,
+      "learning_rate": 8.046211526134888e-05,
+      "loss": 0.0427,
+      "step": 3270
+    },
+    {
+      "epoch": 0.7947661739762539,
+      "grad_norm": 0.18065360188484192,
+      "learning_rate": 8.033083140520065e-05,
+      "loss": 0.0349,
+      "step": 3280
+    },
+    {
+      "epoch": 0.7971892415798401,
+      "grad_norm": 0.12920215725898743,
+      "learning_rate": 8.019921585593962e-05,
+      "loss": 0.0464,
+      "step": 3290
+    },
+    {
+      "epoch": 0.7996123091834262,
+      "grad_norm": 0.25489136576652527,
+      "learning_rate": 8.006727005289232e-05,
+      "loss": 0.0366,
+      "step": 3300
+    },
+    {
+      "epoch": 0.8020353767870123,
+      "grad_norm": 0.20825359225273132,
+      "learning_rate": 7.993499543899692e-05,
+      "loss": 0.0411,
+      "step": 3310
+    },
+    {
+      "epoch": 0.8044584443905985,
+      "grad_norm": 0.16933447122573853,
+      "learning_rate": 7.980239346078742e-05,
+      "loss": 0.0366,
+      "step": 3320
+    },
+    {
+      "epoch": 0.8068815119941847,
+      "grad_norm": 0.1853591501712799,
+      "learning_rate": 7.966946556837778e-05,
+      "loss": 0.0424,
+      "step": 3330
+    },
+    {
+      "epoch": 0.8093045795977708,
+      "grad_norm": 0.15621642768383026,
+      "learning_rate": 7.953621321544616e-05,
+      "loss": 0.0397,
+      "step": 3340
+    },
+    {
+      "epoch": 0.8117276472013569,
+      "grad_norm": 0.13941192626953125,
+      "learning_rate": 7.940263785921896e-05,
+      "loss": 0.0442,
+      "step": 3350
+    },
+    {
+      "epoch": 0.814150714804943,
+      "grad_norm": 0.3172297775745392,
+      "learning_rate": 7.926874096045482e-05,
+      "loss": 0.0373,
+      "step": 3360
+    },
+    {
+      "epoch": 0.8165737824085292,
+      "grad_norm": 0.21606160700321198,
+      "learning_rate": 7.913452398342881e-05,
+      "loss": 0.0511,
+      "step": 3370
+    },
+    {
+      "epoch": 0.8189968500121153,
+      "grad_norm": 0.15523676574230194,
+      "learning_rate": 7.89999883959163e-05,
+      "loss": 0.048,
+      "step": 3380
+    },
+    {
+      "epoch": 0.8214199176157014,
+      "grad_norm": 0.19805888831615448,
+      "learning_rate": 7.886513566917687e-05,
+      "loss": 0.0452,
+      "step": 3390
+    },
+    {
+      "epoch": 0.8238429852192877,
+      "grad_norm": 0.21451659500598907,
+      "learning_rate": 7.872996727793838e-05,
+      "loss": 0.0444,
+      "step": 3400
+    },
+    {
+      "epoch": 0.8262660528228738,
+      "grad_norm": 0.17402143776416779,
+      "learning_rate": 7.859448470038069e-05,
+      "loss": 0.0431,
+      "step": 3410
+    },
+    {
+      "epoch": 0.8286891204264599,
+      "grad_norm": 0.14573906362056732,
+      "learning_rate": 7.845868941811956e-05,
+      "loss": 0.0428,
+      "step": 3420
+    },
+    {
+      "epoch": 0.831112188030046,
+      "grad_norm": 0.16346170008182526,
+      "learning_rate": 7.832258291619043e-05,
+      "loss": 0.0404,
+      "step": 3430
+    },
+    {
+      "epoch": 0.8335352556336322,
+      "grad_norm": 0.14306640625,
+      "learning_rate": 7.81861666830322e-05,
+      "loss": 0.0423,
+      "step": 3440
+    },
+    {
+      "epoch": 0.8359583232372183,
+      "grad_norm": 0.15425516664981842,
+      "learning_rate": 7.804944221047097e-05,
+      "loss": 0.0372,
+      "step": 3450
+    },
+    {
+      "epoch": 0.8383813908408044,
+      "grad_norm": 0.3661347031593323,
+      "learning_rate": 7.791241099370364e-05,
+      "loss": 0.0444,
+      "step": 3460
+    },
+    {
+      "epoch": 0.8408044584443906,
+      "grad_norm": 0.22419315576553345,
+      "learning_rate": 7.777507453128163e-05,
+      "loss": 0.0394,
+      "step": 3470
+    },
+    {
+      "epoch": 0.8432275260479767,
+      "grad_norm": 0.31667467951774597,
+      "learning_rate": 7.763743432509451e-05,
+      "loss": 0.043,
+      "step": 3480
+    },
+    {
+      "epoch": 0.8456505936515629,
+      "grad_norm": 0.17802435159683228,
+      "learning_rate": 7.749949188035353e-05,
+      "loss": 0.042,
+      "step": 3490
+    },
+    {
+      "epoch": 0.848073661255149,
+      "grad_norm": 0.1798791140317917,
+      "learning_rate": 7.736124870557516e-05,
+      "loss": 0.0516,
+      "step": 3500
+    },
+    {
+      "epoch": 0.8504967288587352,
+      "grad_norm": 0.17469356954097748,
+      "learning_rate": 7.722270631256459e-05,
+      "loss": 0.0472,
+      "step": 3510
+    },
+    {
+      "epoch": 0.8529197964623213,
+      "grad_norm": 0.1313830316066742,
+      "learning_rate": 7.708386621639925e-05,
+      "loss": 0.0363,
+      "step": 3520
+    },
+    {
+      "epoch": 0.8553428640659074,
+      "grad_norm": 0.3296225666999817,
+      "learning_rate": 7.694472993541219e-05,
+      "loss": 0.0386,
+      "step": 3530
+    },
+    {
+      "epoch": 0.8577659316694936,
+      "grad_norm": 0.17370855808258057,
+      "learning_rate": 7.680529899117547e-05,
+      "loss": 0.0456,
+      "step": 3540
+    },
+    {
+      "epoch": 0.8601889992730797,
+      "grad_norm": 0.18722471594810486,
+      "learning_rate": 7.666557490848358e-05,
+      "loss": 0.0441,
+      "step": 3550
+    },
+    {
+      "epoch": 0.8626120668766658,
+      "grad_norm": 0.1968887597322464,
+      "learning_rate": 7.65255592153367e-05,
+      "loss": 0.0417,
+      "step": 3560
+    },
+    {
+      "epoch": 0.865035134480252,
+      "grad_norm": 0.20979289710521698,
+      "learning_rate": 7.638525344292402e-05,
+      "loss": 0.0456,
+      "step": 3570
+    },
+    {
+      "epoch": 0.8674582020838382,
+      "grad_norm": 0.2972024083137512,
+      "learning_rate": 7.624465912560697e-05,
+      "loss": 0.0408,
+      "step": 3580
+    },
+    {
+      "epoch": 0.8698812696874243,
+      "grad_norm": 0.2382773458957672,
+      "learning_rate": 7.610377780090249e-05,
+      "loss": 0.0385,
+      "step": 3590
+    },
+    {
+      "epoch": 0.8723043372910104,
+      "grad_norm": 0.20507784187793732,
+      "learning_rate": 7.596261100946618e-05,
+      "loss": 0.0438,
+      "step": 3600
+    },
+    {
+      "epoch": 0.8747274048945965,
+      "grad_norm": 0.16031067073345184,
+      "learning_rate": 7.582116029507542e-05,
+      "loss": 0.0373,
+      "step": 3610
+    },
+    {
+      "epoch": 0.8771504724981827,
+      "grad_norm": 0.15577581524848938,
+      "learning_rate": 7.56794272046126e-05,
+      "loss": 0.0394,
+      "step": 3620
+    },
+    {
+      "epoch": 0.8795735401017688,
+      "grad_norm": 0.17504993081092834,
+      "learning_rate": 7.55374132880481e-05,
+      "loss": 0.0448,
+      "step": 3630
+    },
+    {
+      "epoch": 0.8819966077053549,
+      "grad_norm": 0.17116758227348328,
+      "learning_rate": 7.539512009842333e-05,
+      "loss": 0.0424,
+      "step": 3640
+    },
+    {
+      "epoch": 0.8844196753089412,
+      "grad_norm": 0.1967484951019287,
+      "learning_rate": 7.525254919183382e-05,
+      "loss": 0.0454,
+      "step": 3650
+    },
+    {
+      "epoch": 0.8868427429125273,
+      "grad_norm": 0.19744765758514404,
+      "learning_rate": 7.510970212741215e-05,
+      "loss": 0.0328,
+      "step": 3660
+    },
+    {
+      "epoch": 0.8892658105161134,
+      "grad_norm": 0.1328544318675995,
+      "learning_rate": 7.496658046731096e-05,
+      "loss": 0.0411,
+      "step": 3670
+    },
+    {
+      "epoch": 0.8916888781196995,
+      "grad_norm": 0.12274576723575592,
+      "learning_rate": 7.482318577668578e-05,
+      "loss": 0.0434,
+      "step": 3680
+    },
+    {
+      "epoch": 0.8941119457232857,
+      "grad_norm": 0.21458220481872559,
+      "learning_rate": 7.467951962367796e-05,
+      "loss": 0.0461,
+      "step": 3690
+    },
+    {
+      "epoch": 0.8965350133268718,
+      "grad_norm": 0.16265733540058136,
+      "learning_rate": 7.453558357939755e-05,
+      "loss": 0.0391,
+      "step": 3700
+    },
+    {
+      "epoch": 0.8989580809304579,
+      "grad_norm": 0.11289467662572861,
+      "learning_rate": 7.439137921790606e-05,
+      "loss": 0.0441,
+      "step": 3710
+    },
+    {
+      "epoch": 0.901381148534044,
+      "grad_norm": 0.2518782615661621,
+      "learning_rate": 7.42469081161993e-05,
+      "loss": 0.0435,
+      "step": 3720
+    },
+    {
+      "epoch": 0.9038042161376303,
+      "grad_norm": 0.15403211116790771,
+      "learning_rate": 7.410217185419006e-05,
+      "loss": 0.0352,
+      "step": 3730
+    },
+    {
+      "epoch": 0.9062272837412164,
+      "grad_norm": 0.14967775344848633,
+      "learning_rate": 7.395717201469095e-05,
+      "loss": 0.0419,
+      "step": 3740
+    },
+    {
+      "epoch": 0.9086503513448025,
+      "grad_norm": 0.18926765024662018,
+      "learning_rate": 7.381191018339696e-05,
+      "loss": 0.0414,
+      "step": 3750
+    },
+    {
+      "epoch": 0.9110734189483887,
+      "grad_norm": 0.13593383133411407,
+      "learning_rate": 7.36663879488682e-05,
+      "loss": 0.0388,
+      "step": 3760
+    },
+    {
+      "epoch": 0.9134964865519748,
+      "grad_norm": 0.27778002619743347,
+      "learning_rate": 7.352060690251254e-05,
+      "loss": 0.0509,
+      "step": 3770
+    },
+    {
+      "epoch": 0.9159195541555609,
+      "grad_norm": 0.13119299709796906,
+      "learning_rate": 7.337456863856811e-05,
+      "loss": 0.0392,
+      "step": 3780
+    },
+    {
+      "epoch": 0.918342621759147,
+      "grad_norm": 0.19899316132068634,
+      "learning_rate": 7.3228274754086e-05,
+      "loss": 0.0351,
+      "step": 3790
+    },
+    {
+      "epoch": 0.9207656893627332,
+      "grad_norm": 0.23209546506404877,
+      "learning_rate": 7.308172684891267e-05,
+      "loss": 0.0417,
+      "step": 3800
+    },
+    {
+      "epoch": 0.9231887569663194,
+      "grad_norm": 0.1809903234243393,
+      "learning_rate": 7.293492652567255e-05,
+      "loss": 0.0357,
+      "step": 3810
+    },
+    {
+      "epoch": 0.9256118245699055,
+      "grad_norm": 0.24020709097385406,
+      "learning_rate": 7.278787538975043e-05,
+      "loss": 0.0421,
+      "step": 3820
+    },
+    {
+      "epoch": 0.9280348921734917,
+      "grad_norm": 0.13872717320919037,
+      "learning_rate": 7.2640575049274e-05,
+      "loss": 0.0434,
+      "step": 3830
+    },
+    {
+      "epoch": 0.9304579597770778,
+      "grad_norm": 0.12468786537647247,
+      "learning_rate": 7.249302711509616e-05,
+      "loss": 0.0375,
+      "step": 3840
+    },
+    {
+      "epoch": 0.9328810273806639,
+      "grad_norm": 0.2610887885093689,
+      "learning_rate": 7.23452332007775e-05,
+      "loss": 0.0368,
+      "step": 3850
+    },
+    {
+      "epoch": 0.93530409498425,
+      "grad_norm": 0.14304566383361816,
+      "learning_rate": 7.219719492256858e-05,
+      "loss": 0.044,
+      "step": 3860
+    },
+    {
+      "epoch": 0.9377271625878362,
+      "grad_norm": 0.16474686563014984,
+      "learning_rate": 7.20489138993923e-05,
+      "loss": 0.0406,
+      "step": 3870
+    },
+    {
+      "epoch": 0.9401502301914223,
+      "grad_norm": 0.15889941155910492,
+      "learning_rate": 7.190039175282614e-05,
+      "loss": 0.0366,
+      "step": 3880
+    },
+    {
+      "epoch": 0.9425732977950085,
+      "grad_norm": 0.20691543817520142,
+      "learning_rate": 7.175163010708455e-05,
+      "loss": 0.0369,
+      "step": 3890
+    },
+    {
+      "epoch": 0.9449963653985947,
+      "grad_norm": 0.2763959467411041,
+      "learning_rate": 7.1602630589001e-05,
+      "loss": 0.0427,
+      "step": 3900
+    },
+    {
+      "epoch": 0.9474194330021808,
+      "grad_norm": 0.28597092628479004,
+      "learning_rate": 7.14533948280104e-05,
+      "loss": 0.0405,
+      "step": 3910
+    },
+    {
+      "epoch": 0.9498425006057669,
+      "grad_norm": 0.10694515705108643,
+      "learning_rate": 7.130392445613109e-05,
+      "loss": 0.0377,
+      "step": 3920
+    },
+    {
+      "epoch": 0.952265568209353,
+      "grad_norm": 0.23446550965309143,
+      "learning_rate": 7.115422110794711e-05,
+      "loss": 0.0384,
+      "step": 3930
+    },
+    {
+      "epoch": 0.9546886358129392,
+      "grad_norm": 0.20818932354450226,
+      "learning_rate": 7.100428642059033e-05,
+      "loss": 0.0447,
+      "step": 3940
+    },
+    {
+      "epoch": 0.9571117034165253,
+      "grad_norm": 0.22103236615657806,
+      "learning_rate": 7.08541220337224e-05,
+      "loss": 0.0438,
+      "step": 3950
+    },
+    {
+      "epoch": 0.9595347710201114,
+      "grad_norm": 0.29493871331214905,
+      "learning_rate": 7.070372958951706e-05,
+      "loss": 0.037,
+      "step": 3960
+    },
+    {
+      "epoch": 0.9619578386236975,
+      "grad_norm": 0.22487111389636993,
+      "learning_rate": 7.055311073264194e-05,
+      "loss": 0.044,
+      "step": 3970
+    },
+    {
+      "epoch": 0.9643809062272838,
+      "grad_norm": 0.21388165652751923,
+      "learning_rate": 7.040226711024077e-05,
+      "loss": 0.0352,
+      "step": 3980
+    },
+    {
+      "epoch": 0.9668039738308699,
+      "grad_norm": 0.13826294243335724,
+      "learning_rate": 7.02512003719152e-05,
+      "loss": 0.0342,
+      "step": 3990
+    },
+    {
+      "epoch": 0.969227041434456,
+      "grad_norm": 0.20242105424404144,
+      "learning_rate": 7.00999121697069e-05,
+      "loss": 0.0395,
+      "step": 4000
+    },
+    {
+      "epoch": 0.9716501090380422,
+      "grad_norm": 0.16864188015460968,
+      "learning_rate": 6.99484041580794e-05,
+      "loss": 0.0365,
+      "step": 4010
+    },
+    {
+      "epoch": 0.9740731766416283,
+      "grad_norm": 0.17126575112342834,
+      "learning_rate": 6.979667799390004e-05,
+      "loss": 0.0396,
+      "step": 4020
+    },
+    {
+      "epoch": 0.9764962442452144,
+      "grad_norm": 0.21484066545963287,
+      "learning_rate": 6.964473533642185e-05,
+      "loss": 0.0418,
+      "step": 4030
+    },
+    {
+      "epoch": 0.9789193118488005,
+      "grad_norm": 0.29061970114707947,
+      "learning_rate": 6.949257784726539e-05,
+      "loss": 0.0438,
+      "step": 4040
+    },
+    {
+      "epoch": 0.9813423794523867,
+      "grad_norm": 0.16736122965812683,
+      "learning_rate": 6.934020719040056e-05,
+      "loss": 0.0329,
+      "step": 4050
+    },
+    {
+      "epoch": 0.9837654470559729,
+      "grad_norm": 0.13058723509311676,
+      "learning_rate": 6.918762503212848e-05,
+      "loss": 0.0439,
+      "step": 4060
+    },
+    {
+      "epoch": 0.986188514659559,
+      "grad_norm": 0.2493453025817871,
+      "learning_rate": 6.903483304106319e-05,
+      "loss": 0.0386,
+      "step": 4070
+    },
+    {
+      "epoch": 0.9886115822631452,
+      "grad_norm": 0.21018049120903015,
+      "learning_rate": 6.888183288811341e-05,
+      "loss": 0.0359,
+      "step": 4080
+    },
+    {
+      "epoch": 0.9910346498667313,
+      "grad_norm": 0.14713706076145172,
+      "learning_rate": 6.87286262464643e-05,
+      "loss": 0.0358,
+      "step": 4090
+    },
+    {
+      "epoch": 0.9934577174703174,
+      "grad_norm": 0.13290318846702576,
+      "learning_rate": 6.857521479155915e-05,
+      "loss": 0.0454,
+      "step": 4100
+    },
+    {
+      "epoch": 0.9958807850739035,
+      "grad_norm": 0.1758163571357727,
+      "learning_rate": 6.842160020108104e-05,
+      "loss": 0.0388,
+      "step": 4110
+    },
+    {
+      "epoch": 0.9983038526774897,
+      "grad_norm": 0.2236786186695099,
+      "learning_rate": 6.826778415493455e-05,
+      "loss": 0.0375,
+      "step": 4120
+    },
+    {
+      "epoch": 1.0007269202810758,
+      "grad_norm": 0.18666422367095947,
+      "learning_rate": 6.811376833522729e-05,
+      "loss": 0.0449,
+      "step": 4130
+    },
+    {
+      "epoch": 1.003149987884662,
+      "grad_norm": 0.24746845662593842,
+      "learning_rate": 6.795955442625159e-05,
+      "loss": 0.0417,
+      "step": 4140
+    },
+    {
+      "epoch": 1.005573055488248,
+      "grad_norm": 0.18281657993793488,
+      "learning_rate": 6.780514411446608e-05,
+      "loss": 0.0349,
+      "step": 4150
+    },
+    {
+      "epoch": 1.0079961230918342,
+      "grad_norm": 0.14395718276500702,
+      "learning_rate": 6.765053908847716e-05,
+      "loss": 0.0279,
+      "step": 4160
+    },
+    {
+      "epoch": 1.0104191906954203,
+      "grad_norm": 0.19150447845458984,
+      "learning_rate": 6.749574103902064e-05,
+      "loss": 0.0426,
+      "step": 4170
+    },
+    {
+      "epoch": 1.0128422582990066,
+      "grad_norm": 0.17420750856399536,
+      "learning_rate": 6.734075165894317e-05,
+      "loss": 0.0412,
+      "step": 4180
+    },
+    {
+      "epoch": 1.0152653259025928,
+      "grad_norm": 0.2508639395236969,
+      "learning_rate": 6.71855726431838e-05,
+      "loss": 0.0406,
+      "step": 4190
+    },
+    {
+      "epoch": 1.017688393506179,
+      "grad_norm": 0.18227313458919525,
+      "learning_rate": 6.703020568875538e-05,
+      "loss": 0.0358,
+      "step": 4200
+    },
+    {
+      "epoch": 1.020111461109765,
+      "grad_norm": 0.13005763292312622,
+      "learning_rate": 6.687465249472603e-05,
+      "loss": 0.0344,
+      "step": 4210
+    },
+    {
+      "epoch": 1.0225345287133512,
+      "grad_norm": 0.2406199723482132,
+      "learning_rate": 6.671891476220055e-05,
+      "loss": 0.035,
+      "step": 4220
+    },
+    {
+      "epoch": 1.0249575963169373,
+      "grad_norm": 0.13731837272644043,
+      "learning_rate": 6.656299419430183e-05,
+      "loss": 0.0341,
+      "step": 4230
+    },
+    {
+      "epoch": 1.0273806639205234,
+      "grad_norm": 0.21183118224143982,
+      "learning_rate": 6.640689249615223e-05,
+      "loss": 0.0392,
+      "step": 4240
+    },
+    {
+      "epoch": 1.0298037315241095,
+      "grad_norm": 0.3791130483150482,
+      "learning_rate": 6.625061137485491e-05,
+      "loss": 0.0383,
+      "step": 4250
+    },
+    {
+      "epoch": 1.0322267991276957,
+      "grad_norm": 0.20733681321144104,
+      "learning_rate": 6.609415253947517e-05,
+      "loss": 0.0444,
+      "step": 4260
+    },
+    {
+      "epoch": 1.0346498667312818,
+      "grad_norm": 0.18938347697257996,
+      "learning_rate": 6.593751770102178e-05,
+      "loss": 0.0325,
+      "step": 4270
+    },
+    {
+      "epoch": 1.037072934334868,
+      "grad_norm": 0.15026292204856873,
+      "learning_rate": 6.578070857242823e-05,
+      "loss": 0.0343,
+      "step": 4280
+    },
+    {
+      "epoch": 1.039496001938454,
+      "grad_norm": 0.12519890069961548,
+      "learning_rate": 6.562372686853402e-05,
+      "loss": 0.0345,
+      "step": 4290
+    },
+    {
+      "epoch": 1.0419190695420402,
+      "grad_norm": 0.1458505094051361,
+      "learning_rate": 6.546657430606593e-05,
+      "loss": 0.0411,
+      "step": 4300
+    },
+    {
+      "epoch": 1.0443421371456263,
+      "grad_norm": 0.18468616902828217,
+      "learning_rate": 6.530925260361918e-05,
+      "loss": 0.033,
+      "step": 4310
+    },
+    {
+      "epoch": 1.0467652047492124,
+      "grad_norm": 0.154633030295372,
+      "learning_rate": 6.515176348163871e-05,
+      "loss": 0.0372,
+      "step": 4320
+    },
+    {
+      "epoch": 1.0491882723527985,
+      "grad_norm": 0.1589757353067398,
+      "learning_rate": 6.499410866240032e-05,
+      "loss": 0.0305,
+      "step": 4330
+    },
+    {
+      "epoch": 1.0516113399563847,
+      "grad_norm": 0.21719059348106384,
+      "learning_rate": 6.48362898699919e-05,
+      "loss": 0.0328,
+      "step": 4340
+    },
+    {
+      "epoch": 1.054034407559971,
+      "grad_norm": 0.20215091109275818,
+      "learning_rate": 6.467830883029443e-05,
+      "loss": 0.0366,
+      "step": 4350
+    },
+    {
+      "epoch": 1.0564574751635571,
+      "grad_norm": 0.15874029695987701,
+      "learning_rate": 6.452016727096326e-05,
+      "loss": 0.0301,
+      "step": 4360
+    },
+    {
+      "epoch": 1.0588805427671433,
+      "grad_norm": 0.1615975797176361,
+      "learning_rate": 6.436186692140916e-05,
+      "loss": 0.0342,
+      "step": 4370
+    },
+    {
+      "epoch": 1.0613036103707294,
+      "grad_norm": 0.16992174088954926,
+      "learning_rate": 6.420340951277938e-05,
+      "loss": 0.0365,
+      "step": 4380
+    },
+    {
+      "epoch": 1.0637266779743155,
+      "grad_norm": 0.1354915052652359,
+      "learning_rate": 6.404479677793874e-05,
+      "loss": 0.0361,
+      "step": 4390
+    },
+    {
+      "epoch": 1.0661497455779017,
+      "grad_norm": 0.24680233001708984,
+      "learning_rate": 6.388603045145075e-05,
+      "loss": 0.0466,
+      "step": 4400
+    },
+    {
+      "epoch": 1.0685728131814878,
+      "grad_norm": 0.20606473088264465,
+      "learning_rate": 6.372711226955843e-05,
+      "loss": 0.0396,
+      "step": 4410
+    },
+    {
+      "epoch": 1.070995880785074,
+      "grad_norm": 0.16568884253501892,
+      "learning_rate": 6.356804397016564e-05,
+      "loss": 0.0404,
+      "step": 4420
+    },
+    {
+      "epoch": 1.07341894838866,
+      "grad_norm": 0.1599683314561844,
+      "learning_rate": 6.340882729281779e-05,
+      "loss": 0.0384,
+      "step": 4430
+    },
+    {
+      "epoch": 1.0758420159922462,
+      "grad_norm": 0.2357213795185089,
+      "learning_rate": 6.324946397868294e-05,
+      "loss": 0.0435,
+      "step": 4440
+    },
+    {
+      "epoch": 1.0782650835958323,
+      "grad_norm": 0.2821190357208252,
+      "learning_rate": 6.308995577053276e-05,
+      "loss": 0.0359,
+      "step": 4450
+    },
+    {
+      "epoch": 1.0806881511994184,
+      "grad_norm": 0.35420918464660645,
+      "learning_rate": 6.293030441272347e-05,
+      "loss": 0.0394,
+      "step": 4460
+    },
+    {
+      "epoch": 1.0831112188030045,
+      "grad_norm": 0.17902354896068573,
+      "learning_rate": 6.277051165117677e-05,
+      "loss": 0.0383,
+      "step": 4470
+    },
+    {
+      "epoch": 1.0855342864065907,
+      "grad_norm": 0.21981704235076904,
+      "learning_rate": 6.261057923336064e-05,
+      "loss": 0.0385,
+      "step": 4480
+    },
+    {
+      "epoch": 1.0879573540101768,
+      "grad_norm": 0.14618508517742157,
+      "learning_rate": 6.245050890827042e-05,
+      "loss": 0.0385,
+      "step": 4490
+    },
+    {
+      "epoch": 1.0903804216137631,
+      "grad_norm": 0.1590287983417511,
+      "learning_rate": 6.229030242640952e-05,
+      "loss": 0.0376,
+      "step": 4500
+    },
+    {
+      "epoch": 1.0928034892173493,
+      "grad_norm": 0.12258179485797882,
+      "learning_rate": 6.212996153977037e-05,
+      "loss": 0.0411,
+      "step": 4510
+    },
+    {
+      "epoch": 1.0952265568209354,
+      "grad_norm": 0.24019291996955872,
+      "learning_rate": 6.196948800181523e-05,
+      "loss": 0.0366,
+      "step": 4520
+    },
+    {
+      "epoch": 1.0976496244245215,
+      "grad_norm": 0.19612370431423187,
+      "learning_rate": 6.180888356745695e-05,
+      "loss": 0.0382,
+      "step": 4530
+    },
+    {
+      "epoch": 1.1000726920281076,
+      "grad_norm": 0.13913317024707794,
+      "learning_rate": 6.164814999303995e-05,
+      "loss": 0.0327,
+      "step": 4540
+    },
+    {
+      "epoch": 1.1024957596316938,
+      "grad_norm": 0.11045411229133606,
+      "learning_rate": 6.148728903632081e-05,
+      "loss": 0.0333,
+      "step": 4550
+    },
+    {
+      "epoch": 1.10491882723528,
+      "grad_norm": 0.22665664553642273,
+      "learning_rate": 6.132630245644921e-05,
+      "loss": 0.0304,
+      "step": 4560
+    },
+    {
+      "epoch": 1.107341894838866,
+      "grad_norm": 0.14154298603534698,
+      "learning_rate": 6.116519201394857e-05,
+      "loss": 0.0386,
+      "step": 4570
+    },
+    {
+      "epoch": 1.1097649624424522,
+      "grad_norm": 0.1647247076034546,
+      "learning_rate": 6.10039594706969e-05,
+      "loss": 0.0381,
+      "step": 4580
+    },
+    {
+      "epoch": 1.1121880300460383,
+      "grad_norm": 0.18806114792823792,
+      "learning_rate": 6.084260658990744e-05,
+      "loss": 0.0374,
+      "step": 4590
+    },
+    {
+      "epoch": 1.1146110976496244,
+      "grad_norm": 0.1477152705192566,
+      "learning_rate": 6.068113513610943e-05,
+      "loss": 0.0305,
+      "step": 4600
+    },
+    {
+      "epoch": 1.1170341652532105,
+      "grad_norm": 0.1263800710439682,
+      "learning_rate": 6.0519546875128876e-05,
+      "loss": 0.0388,
+      "step": 4610
+    },
+    {
+      "epoch": 1.1194572328567967,
+      "grad_norm": 0.15480294823646545,
+      "learning_rate": 6.035784357406906e-05,
+      "loss": 0.0406,
+      "step": 4620
+    },
+    {
+      "epoch": 1.1218803004603828,
+      "grad_norm": 0.17695686221122742,
+      "learning_rate": 6.01960270012914e-05,
+      "loss": 0.0338,
+      "step": 4630
+    },
+    {
+      "epoch": 1.124303368063969,
+      "grad_norm": 0.19480949640274048,
+      "learning_rate": 6.003409892639599e-05,
+      "loss": 0.0382,
+      "step": 4640
+    },
+    {
+      "epoch": 1.126726435667555,
+      "grad_norm": 0.20768029987812042,
+      "learning_rate": 5.9872061120202336e-05,
+      "loss": 0.0338,
+      "step": 4650
+    },
+    {
+      "epoch": 1.1291495032711412,
+      "grad_norm": 0.11325472593307495,
+      "learning_rate": 5.9709915354729914e-05,
+      "loss": 0.0481,
+      "step": 4660
+    },
+    {
+      "epoch": 1.1315725708747273,
+      "grad_norm": 0.17669686675071716,
+      "learning_rate": 5.9547663403178824e-05,
+      "loss": 0.0392,
+      "step": 4670
+    },
+    {
+      "epoch": 1.1339956384783136,
+      "grad_norm": 0.17881830036640167,
+      "learning_rate": 5.9385307039910445e-05,
+      "loss": 0.0318,
+      "step": 4680
+    },
+    {
+      "epoch": 1.1364187060818998,
+      "grad_norm": 0.21821770071983337,
+      "learning_rate": 5.922284804042792e-05,
+      "loss": 0.0395,
+      "step": 4690
+    },
+    {
+      "epoch": 1.138841773685486,
+      "grad_norm": 0.18453213572502136,
+      "learning_rate": 5.906028818135687e-05,
+      "loss": 0.0425,
+      "step": 4700
+    },
+    {
+      "epoch": 1.141264841289072,
+      "grad_norm": 0.16833944618701935,
+      "learning_rate": 5.889762924042585e-05,
+      "loss": 0.0319,
+      "step": 4710
+    },
+    {
+      "epoch": 1.1436879088926581,
+      "grad_norm": 0.16403602063655853,
+      "learning_rate": 5.873487299644699e-05,
+      "loss": 0.0382,
+      "step": 4720
+    },
+    {
+      "epoch": 1.1461109764962443,
+      "grad_norm": 0.20726455748081207,
+      "learning_rate": 5.857202122929649e-05,
+      "loss": 0.0355,
+      "step": 4730
+    },
+    {
+      "epoch": 1.1485340440998304,
+      "grad_norm": 0.16238965094089508,
+      "learning_rate": 5.840907571989518e-05,
+      "loss": 0.0334,
+      "step": 4740
+    },
+    {
+      "epoch": 1.1509571117034165,
+      "grad_norm": 0.16916604340076447,
+      "learning_rate": 5.824603825018904e-05,
+      "loss": 0.0338,
+      "step": 4750
+    },
+    {
+      "epoch": 1.1533801793070027,
+      "grad_norm": 0.21727706491947174,
+      "learning_rate": 5.808291060312975e-05,
+      "loss": 0.0383,
+      "step": 4760
+    },
+    {
+      "epoch": 1.1558032469105888,
+      "grad_norm": 0.1924837827682495,
+      "learning_rate": 5.7919694562655083e-05,
+      "loss": 0.0367,
+      "step": 4770
+    },
+    {
+      "epoch": 1.158226314514175,
+      "grad_norm": 0.13621880114078522,
+      "learning_rate": 5.775639191366954e-05,
+      "loss": 0.0375,
+      "step": 4780
+    },
+    {
+      "epoch": 1.160649382117761,
+      "grad_norm": 0.177626371383667,
+      "learning_rate": 5.75930044420247e-05,
+      "loss": 0.0342,
+      "step": 4790
+    },
+    {
+      "epoch": 1.1630724497213472,
+      "grad_norm": 0.1389589011669159,
+      "learning_rate": 5.74295339344998e-05,
+      "loss": 0.0401,
+      "step": 4800
+    },
+    {
+      "epoch": 1.1654955173249333,
+      "grad_norm": 0.21249249577522278,
+      "learning_rate": 5.726598217878211e-05,
+      "loss": 0.0346,
+      "step": 4810
+    },
+    {
+      "epoch": 1.1679185849285194,
+      "grad_norm": 0.22570757567882538,
+      "learning_rate": 5.71023509634474e-05,
+      "loss": 0.0302,
+      "step": 4820
+    },
+    {
+      "epoch": 1.1703416525321058,
+      "grad_norm": 0.25085312128067017,
+      "learning_rate": 5.693864207794049e-05,
+      "loss": 0.0382,
+      "step": 4830
+    },
+    {
+      "epoch": 1.1727647201356919,
+      "grad_norm": 0.1335025131702423,
+      "learning_rate": 5.677485731255545e-05,
+      "loss": 0.0461,
+      "step": 4840
+    },
+    {
+      "epoch": 1.175187787739278,
+      "grad_norm": 0.16283638775348663,
+      "learning_rate": 5.6610998458416296e-05,
+      "loss": 0.0332,
+      "step": 4850
+    },
+    {
+      "epoch": 1.1776108553428641,
+      "grad_norm": 0.20627540349960327,
+      "learning_rate": 5.644706730745716e-05,
+      "loss": 0.034,
+      "step": 4860
+    },
+    {
+      "epoch": 1.1800339229464503,
+      "grad_norm": 0.13472777605056763,
+      "learning_rate": 5.628306565240287e-05,
+      "loss": 0.0351,
+      "step": 4870
+    },
+    {
+      "epoch": 1.1824569905500364,
+      "grad_norm": 0.12939082086086273,
+      "learning_rate": 5.611899528674923e-05,
+      "loss": 0.0393,
+      "step": 4880
+    },
+    {
+      "epoch": 1.1848800581536225,
+      "grad_norm": 0.24731796979904175,
+      "learning_rate": 5.595485800474349e-05,
+      "loss": 0.0363,
+      "step": 4890
+    },
+    {
+      "epoch": 1.1873031257572086,
+      "grad_norm": 0.1261483132839203,
+      "learning_rate": 5.579065560136467e-05,
+      "loss": 0.0403,
+      "step": 4900
+    },
+    {
+      "epoch": 1.1897261933607948,
+      "grad_norm": 0.18301163613796234,
+      "learning_rate": 5.562638987230392e-05,
+      "loss": 0.0362,
+      "step": 4910
+    },
+    {
+      "epoch": 1.192149260964381,
+      "grad_norm": 0.28962644934654236,
+      "learning_rate": 5.546206261394498e-05,
+      "loss": 0.0304,
+      "step": 4920
+    },
+    {
+      "epoch": 1.194572328567967,
+      "grad_norm": 0.19975295662879944,
+      "learning_rate": 5.529767562334437e-05,
+      "loss": 0.0385,
+      "step": 4930
+    },
+    {
+      "epoch": 1.1969953961715531,
+      "grad_norm": 0.17144648730754852,
+      "learning_rate": 5.5133230698211926e-05,
+      "loss": 0.0334,
+      "step": 4940
+    },
+    {
+      "epoch": 1.1994184637751393,
+      "grad_norm": 0.1319882720708847,
+      "learning_rate": 5.496872963689096e-05,
+      "loss": 0.0375,
+      "step": 4950
+    },
+    {
+      "epoch": 1.2018415313787254,
+      "grad_norm": 0.1552356332540512,
+      "learning_rate": 5.4804174238338756e-05,
+      "loss": 0.0314,
+      "step": 4960
+    },
+    {
+      "epoch": 1.2042645989823115,
+      "grad_norm": 0.18072573840618134,
+      "learning_rate": 5.463956630210678e-05,
+      "loss": 0.0436,
+      "step": 4970
+    },
+    {
+      "epoch": 1.2066876665858977,
+      "grad_norm": 0.18687626719474792,
+      "learning_rate": 5.4474907628321046e-05,
+      "loss": 0.0391,
+      "step": 4980
+    },
+    {
+      "epoch": 1.2091107341894838,
+      "grad_norm": 0.25423336029052734,
+      "learning_rate": 5.431020001766244e-05,
+      "loss": 0.0332,
+      "step": 4990
+    },
+    {
+      "epoch": 1.21153380179307,
+      "grad_norm": 0.27445152401924133,
+      "learning_rate": 5.4145445271346986e-05,
+      "loss": 0.0332,
+      "step": 5000
+    },
+    {
+      "epoch": 1.2139568693966563,
+      "grad_norm": 0.21275608241558075,
+      "learning_rate": 5.398064519110622e-05,
+      "loss": 0.0368,
+      "step": 5010
+    },
+    {
+      "epoch": 1.2163799370002424,
+      "grad_norm": 0.15624688565731049,
+      "learning_rate": 5.3815801579167394e-05,
+      "loss": 0.0357,
+      "step": 5020
+    },
+    {
+      "epoch": 1.2188030046038285,
+      "grad_norm": 0.11576736718416214,
+      "learning_rate": 5.365091623823382e-05,
+      "loss": 0.0363,
+      "step": 5030
+    },
+    {
+      "epoch": 1.2212260722074146,
+      "grad_norm": 0.1762501299381256,
+      "learning_rate": 5.348599097146521e-05,
+      "loss": 0.0338,
+      "step": 5040
+    },
+    {
+      "epoch": 1.2236491398110008,
+      "grad_norm": 0.2148737758398056,
+      "learning_rate": 5.3321027582457836e-05,
+      "loss": 0.0348,
+      "step": 5050
+    },
+    {
+      "epoch": 1.226072207414587,
+      "grad_norm": 0.20553722977638245,
+      "learning_rate": 5.315602787522491e-05,
+      "loss": 0.035,
+      "step": 5060
+    },
+    {
+      "epoch": 1.228495275018173,
+      "grad_norm": 0.20314088463783264,
+      "learning_rate": 5.299099365417678e-05,
+      "loss": 0.0269,
+      "step": 5070
+    },
+    {
+      "epoch": 1.2309183426217591,
+      "grad_norm": 0.14651304483413696,
+      "learning_rate": 5.2825926724101236e-05,
+      "loss": 0.035,
+      "step": 5080
+    },
+    {
+      "epoch": 1.2333414102253453,
+      "grad_norm": 0.18988747894763947,
+      "learning_rate": 5.26608288901438e-05,
+      "loss": 0.0399,
+      "step": 5090
+    },
+    {
+      "epoch": 1.2357644778289314,
+      "grad_norm": 0.1370977759361267,
+      "learning_rate": 5.24957019577879e-05,
+      "loss": 0.0326,
+      "step": 5100
+    },
+    {
+      "epoch": 1.2381875454325175,
+      "grad_norm": 0.1975657343864441,
+      "learning_rate": 5.2330547732835266e-05,
+      "loss": 0.0344,
+      "step": 5110
+    },
+    {
+      "epoch": 1.2406106130361036,
+      "grad_norm": 0.1782601922750473,
+      "learning_rate": 5.2165368021385996e-05,
+      "loss": 0.044,
+      "step": 5120
+    },
+    {
+      "epoch": 1.2430336806396898,
+      "grad_norm": 0.1720588505268097,
+      "learning_rate": 5.200016462981897e-05,
+      "loss": 0.0293,
+      "step": 5130
+    },
+    {
+      "epoch": 1.245456748243276,
+      "grad_norm": 0.2118971198797226,
+      "learning_rate": 5.1834939364772015e-05,
+      "loss": 0.0307,
+      "step": 5140
+    },
+    {
+      "epoch": 1.2478798158468623,
+      "grad_norm": 0.17877288162708282,
+      "learning_rate": 5.166969403312214e-05,
+      "loss": 0.0307,
+      "step": 5150
+    },
+    {
+      "epoch": 1.2503028834504484,
+      "grad_norm": 0.18469521403312683,
+      "learning_rate": 5.1504430441965844e-05,
+      "loss": 0.0368,
+      "step": 5160
+    },
+    {
+      "epoch": 1.2527259510540345,
+      "grad_norm": 0.1583702713251114,
+      "learning_rate": 5.133915039859923e-05,
+      "loss": 0.0292,
+      "step": 5170
+    },
+    {
+      "epoch": 1.2551490186576206,
+      "grad_norm": 0.18019641935825348,
+      "learning_rate": 5.1173855710498444e-05,
+      "loss": 0.0309,
+      "step": 5180
+    },
+    {
+      "epoch": 1.2575720862612068,
+      "grad_norm": 0.18975822627544403,
+      "learning_rate": 5.100854818529967e-05,
+      "loss": 0.034,
+      "step": 5190
+    },
+    {
+      "epoch": 1.2599951538647929,
+      "grad_norm": 0.1740541160106659,
+      "learning_rate": 5.084322963077951e-05,
+      "loss": 0.0424,
+      "step": 5200
+    },
+    {
+      "epoch": 1.262418221468379,
+      "grad_norm": 0.20315106213092804,
+      "learning_rate": 5.067790185483522e-05,
+      "loss": 0.0345,
+      "step": 5210
+    },
+    {
+      "epoch": 1.2648412890719651,
+      "grad_norm": 0.2409934550523758,
+      "learning_rate": 5.0512566665464844e-05,
+      "loss": 0.0362,
+      "step": 5220
+    },
+    {
+      "epoch": 1.2672643566755513,
+      "grad_norm": 0.08300124853849411,
+      "learning_rate": 5.034722587074755e-05,
+      "loss": 0.0331,
+      "step": 5230
+    },
+    {
+      "epoch": 1.2696874242791374,
+      "grad_norm": 0.1453070491552353,
+      "learning_rate": 5.018188127882375e-05,
+      "loss": 0.0299,
+      "step": 5240
+    },
+    {
+      "epoch": 1.2721104918827235,
+      "grad_norm": 0.1614101231098175,
+      "learning_rate": 5.0016534697875417e-05,
+      "loss": 0.0285,
+      "step": 5250
+    },
+    {
+      "epoch": 1.2745335594863096,
+      "grad_norm": 0.13010790944099426,
+      "learning_rate": 4.9851187936106294e-05,
+      "loss": 0.0385,
+      "step": 5260
+    },
+    {
+      "epoch": 1.2769566270898958,
+      "grad_norm": 0.19348230957984924,
+      "learning_rate": 4.968584280172206e-05,
+      "loss": 0.0369,
+      "step": 5270
+    },
+    {
+      "epoch": 1.279379694693482,
+      "grad_norm": 0.18507100641727448,
+      "learning_rate": 4.95205011029106e-05,
+      "loss": 0.0354,
+      "step": 5280
+    },
+    {
+      "epoch": 1.281802762297068,
+      "grad_norm": 0.16559474170207977,
+      "learning_rate": 4.935516464782227e-05,
+      "loss": 0.0365,
+      "step": 5290
+    },
+    {
+      "epoch": 1.2842258299006541,
+      "grad_norm": 0.1510300487279892,
+      "learning_rate": 4.918983524455003e-05,
+      "loss": 0.0336,
+      "step": 5300
+    },
+    {
+      "epoch": 1.2866488975042403,
+      "grad_norm": 0.17999106645584106,
+      "learning_rate": 4.9024514701109766e-05,
+      "loss": 0.035,
+      "step": 5310
+    },
+    {
+      "epoch": 1.2890719651078264,
+      "grad_norm": 0.1714041531085968,
+      "learning_rate": 4.885920482542043e-05,
+      "loss": 0.0364,
+      "step": 5320
+    },
+    {
+      "epoch": 1.2914950327114125,
+      "grad_norm": 0.12049345672130585,
+      "learning_rate": 4.869390742528438e-05,
+      "loss": 0.032,
+      "step": 5330
+    },
+    {
+      "epoch": 1.2939181003149987,
+      "grad_norm": 0.12734355032444,
+      "learning_rate": 4.852862430836744e-05,
+      "loss": 0.0362,
+      "step": 5340
+    },
+    {
+      "epoch": 1.296341167918585,
+      "grad_norm": 0.15307103097438812,
+      "learning_rate": 4.836335728217933e-05,
+      "loss": 0.0363,
+      "step": 5350
+    },
+    {
+      "epoch": 1.2987642355221711,
+      "grad_norm": 0.1784488409757614,
+      "learning_rate": 4.819810815405379e-05,
+      "loss": 0.0426,
+      "step": 5360
+    },
+    {
+      "epoch": 1.3011873031257573,
+      "grad_norm": 0.13317061960697174,
+      "learning_rate": 4.803287873112877e-05,
+      "loss": 0.0387,
+      "step": 5370
+    },
+    {
+      "epoch": 1.3036103707293434,
+      "grad_norm": 0.16432419419288635,
+      "learning_rate": 4.786767082032681e-05,
+      "loss": 0.0349,
+      "step": 5380
+    },
+    {
+      "epoch": 1.3060334383329295,
+      "grad_norm": 0.15296678245067596,
+      "learning_rate": 4.77024862283351e-05,
+      "loss": 0.0324,
+      "step": 5390
+    },
+    {
+      "epoch": 1.3084565059365156,
+      "grad_norm": 0.2578441798686981,
+      "learning_rate": 4.753732676158593e-05,
+      "loss": 0.037,
+      "step": 5400
+    },
+    {
+      "epoch": 1.3108795735401018,
+      "grad_norm": 0.16813112795352936,
+      "learning_rate": 4.737219422623672e-05,
+      "loss": 0.0349,
+      "step": 5410
+    },
+    {
+      "epoch": 1.3133026411436879,
+      "grad_norm": 0.10465259104967117,
+      "learning_rate": 4.720709042815044e-05,
+      "loss": 0.027,
+      "step": 5420
+    },
+    {
+      "epoch": 1.315725708747274,
+      "grad_norm": 0.15062761306762695,
+      "learning_rate": 4.704201717287578e-05,
+      "loss": 0.0333,
+      "step": 5430
+    },
+    {
+      "epoch": 1.3181487763508601,
+      "grad_norm": 0.1879766434431076,
+      "learning_rate": 4.6876976265627404e-05,
+      "loss": 0.0355,
+      "step": 5440
+    },
+    {
+      "epoch": 1.3205718439544463,
+      "grad_norm": 0.2238711267709732,
+      "learning_rate": 4.671196951126626e-05,
+      "loss": 0.0358,
+      "step": 5450
+    },
+    {
+      "epoch": 1.3229949115580324,
+      "grad_norm": 0.3197612464427948,
+      "learning_rate": 4.654699871427971e-05,
+      "loss": 0.0356,
+      "step": 5460
+    },
+    {
+      "epoch": 1.3254179791616187,
+      "grad_norm": 0.14447778463363647,
+      "learning_rate": 4.6382065678762034e-05,
+      "loss": 0.0301,
+      "step": 5470
+    },
+    {
+      "epoch": 1.3278410467652049,
+      "grad_norm": 0.13374896347522736,
+      "learning_rate": 4.6217172208394424e-05,
+      "loss": 0.0288,
+      "step": 5480
+    },
+    {
+      "epoch": 1.330264114368791,
+      "grad_norm": 0.15201738476753235,
+      "learning_rate": 4.605232010642549e-05,
+      "loss": 0.0288,
+      "step": 5490
+    },
+    {
+      "epoch": 1.3326871819723771,
+      "grad_norm": 0.21614566445350647,
+      "learning_rate": 4.588751117565142e-05,
+      "loss": 0.0304,
+      "step": 5500
+    },
+    {
+      "epoch": 1.3351102495759632,
+      "grad_norm": 0.1442033052444458,
+      "learning_rate": 4.5722747218396214e-05,
+      "loss": 0.0315,
+      "step": 5510
+    },
+    {
+      "epoch": 1.3375333171795494,
+      "grad_norm": 0.18296325206756592,
+      "learning_rate": 4.5558030036492194e-05,
+      "loss": 0.029,
+      "step": 5520
+    },
+    {
+      "epoch": 1.3399563847831355,
+      "grad_norm": 0.25480255484580994,
+      "learning_rate": 4.539336143125999e-05,
+      "loss": 0.0399,
+      "step": 5530
+    },
+    {
+      "epoch": 1.3423794523867216,
+      "grad_norm": 0.18130719661712646,
+      "learning_rate": 4.522874320348916e-05,
+      "loss": 0.0331,
+      "step": 5540
+    },
+    {
+      "epoch": 1.3448025199903078,
+      "grad_norm": 0.1488102525472641,
+      "learning_rate": 4.506417715341821e-05,
+      "loss": 0.0268,
+      "step": 5550
+    },
+    {
+      "epoch": 1.3472255875938939,
+      "grad_norm": 0.11495231091976166,
+      "learning_rate": 4.489966508071511e-05,
+      "loss": 0.0327,
+      "step": 5560
+    },
+    {
+      "epoch": 1.34964865519748,
+      "grad_norm": 0.17555421590805054,
+      "learning_rate": 4.4735208784457575e-05,
+      "loss": 0.0306,
+      "step": 5570
+    },
+    {
+      "epoch": 1.3520717228010661,
+      "grad_norm": 0.17269304394721985,
+      "learning_rate": 4.457081006311325e-05,
+      "loss": 0.0322,
+      "step": 5580
+    },
+    {
+      "epoch": 1.3544947904046523,
+      "grad_norm": 0.13520239293575287,
+      "learning_rate": 4.440647071452027e-05,
+      "loss": 0.0313,
+      "step": 5590
+    },
+    {
+      "epoch": 1.3569178580082384,
+      "grad_norm": 0.20179545879364014,
+      "learning_rate": 4.424219253586737e-05,
+      "loss": 0.0355,
+      "step": 5600
+    },
+    {
+      "epoch": 1.3593409256118245,
+      "grad_norm": 0.18395577371120453,
+      "learning_rate": 4.407797732367443e-05,
+      "loss": 0.0394,
+      "step": 5610
+    },
+    {
+      "epoch": 1.3617639932154106,
+      "grad_norm": 0.18193255364894867,
+      "learning_rate": 4.391382687377268e-05,
+      "loss": 0.0326,
+      "step": 5620
+    },
+    {
+      "epoch": 1.3641870608189968,
+      "grad_norm": 0.11756373196840286,
+      "learning_rate": 4.374974298128512e-05,
+      "loss": 0.0323,
+      "step": 5630
+    },
+    {
+      "epoch": 1.366610128422583,
+      "grad_norm": 0.1866205334663391,
+      "learning_rate": 4.358572744060699e-05,
+      "loss": 0.0351,
+      "step": 5640
+    },
+    {
+      "epoch": 1.369033196026169,
+      "grad_norm": 0.21558016538619995,
+      "learning_rate": 4.342178204538588e-05,
+      "loss": 0.029,
+      "step": 5650
+    },
+    {
+      "epoch": 1.3714562636297551,
+      "grad_norm": 0.17946316301822662,
+      "learning_rate": 4.325790858850241e-05,
+      "loss": 0.0358,
+      "step": 5660
+    },
+    {
+      "epoch": 1.3738793312333413,
+      "grad_norm": 0.19241704046726227,
+      "learning_rate": 4.309410886205043e-05,
+      "loss": 0.0344,
+      "step": 5670
+    },
+    {
+      "epoch": 1.3763023988369276,
+      "grad_norm": 0.09524490684270859,
+      "learning_rate": 4.293038465731752e-05,
+      "loss": 0.0306,
+      "step": 5680
+    },
+    {
+      "epoch": 1.3787254664405137,
+      "grad_norm": 0.20127789676189423,
+      "learning_rate": 4.276673776476533e-05,
+      "loss": 0.037,
+      "step": 5690
+    },
+    {
+      "epoch": 1.3811485340440999,
+      "grad_norm": 0.22112242877483368,
+      "learning_rate": 4.260316997401007e-05,
+      "loss": 0.0315,
+      "step": 5700
+    },
+    {
+      "epoch": 1.383571601647686,
+      "grad_norm": 0.21605165302753448,
+      "learning_rate": 4.243968307380293e-05,
+      "loss": 0.035,
+      "step": 5710
+    },
+    {
+      "epoch": 1.3859946692512721,
+      "grad_norm": 0.16192230582237244,
+      "learning_rate": 4.22762788520104e-05,
+      "loss": 0.0358,
+      "step": 5720
+    },
+    {
+      "epoch": 1.3884177368548583,
+      "grad_norm": 0.13257241249084473,
+      "learning_rate": 4.211295909559491e-05,
+      "loss": 0.0412,
+      "step": 5730
+    },
+    {
+      "epoch": 1.3908408044584444,
+      "grad_norm": 0.1690298169851303,
+      "learning_rate": 4.194972559059511e-05,
+      "loss": 0.0298,
+      "step": 5740
+    },
+    {
+      "epoch": 1.3932638720620305,
+      "grad_norm": 0.2850973308086395,
+      "learning_rate": 4.178658012210651e-05,
+      "loss": 0.0301,
+      "step": 5750
+    },
+    {
+      "epoch": 1.3956869396656166,
+      "grad_norm": 0.11184634268283844,
+      "learning_rate": 4.162352447426177e-05,
+      "loss": 0.0342,
+      "step": 5760
+    },
+    {
+      "epoch": 1.3981100072692028,
+      "grad_norm": 0.1616874635219574,
+      "learning_rate": 4.146056043021135e-05,
+      "loss": 0.0322,
+      "step": 5770
+    },
+    {
+      "epoch": 1.4005330748727889,
+      "grad_norm": 0.1666002720594406,
+      "learning_rate": 4.1297689772103944e-05,
+      "loss": 0.036,
+      "step": 5780
+    },
+    {
+      "epoch": 1.402956142476375,
+      "grad_norm": 0.10762447863817215,
+      "learning_rate": 4.113491428106694e-05,
+      "loss": 0.0318,
+      "step": 5790
+    },
+    {
+      "epoch": 1.4053792100799614,
+      "grad_norm": 0.12863484025001526,
+      "learning_rate": 4.0972235737187055e-05,
+      "loss": 0.0315,
+      "step": 5800
+    },
+    {
+      "epoch": 1.4078022776835475,
+      "grad_norm": 0.1334962099790573,
+      "learning_rate": 4.080965591949076e-05,
+      "loss": 0.0294,
+      "step": 5810
+    },
+    {
+      "epoch": 1.4102253452871336,
+      "grad_norm": 0.15371330082416534,
+      "learning_rate": 4.0647176605924924e-05,
+      "loss": 0.0261,
+      "step": 5820
+    },
+    {
+      "epoch": 1.4126484128907197,
+      "grad_norm": 0.13973890244960785,
+      "learning_rate": 4.0484799573337255e-05,
+      "loss": 0.038,
+      "step": 5830
+    },
+    {
+      "epoch": 1.4150714804943059,
+      "grad_norm": 0.1423688679933548,
+      "learning_rate": 4.032252659745699e-05,
+      "loss": 0.034,
+      "step": 5840
+    },
+    {
+      "epoch": 1.417494548097892,
+      "grad_norm": 0.29787349700927734,
+      "learning_rate": 4.016035945287539e-05,
+      "loss": 0.0421,
+      "step": 5850
+    },
+    {
+      "epoch": 1.4199176157014781,
+      "grad_norm": 0.16662997007369995,
+      "learning_rate": 3.999829991302635e-05,
+      "loss": 0.0321,
+      "step": 5860
+    },
+    {
+      "epoch": 1.4223406833050642,
+      "grad_norm": 0.13844655454158783,
+      "learning_rate": 3.983634975016707e-05,
+      "loss": 0.0317,
+      "step": 5870
+    },
+    {
+      "epoch": 1.4247637509086504,
+      "grad_norm": 0.14252415299415588,
+      "learning_rate": 3.967451073535854e-05,
+      "loss": 0.0436,
+      "step": 5880
+    },
+    {
+      "epoch": 1.4271868185122365,
+      "grad_norm": 0.23531639575958252,
+      "learning_rate": 3.951278463844633e-05,
+      "loss": 0.0297,
+      "step": 5890
+    },
+    {
+      "epoch": 1.4296098861158226,
+      "grad_norm": 0.1339505910873413,
+      "learning_rate": 3.935117322804111e-05,
+      "loss": 0.034,
+      "step": 5900
+    },
+    {
+      "epoch": 1.4320329537194088,
+      "grad_norm": 0.2176676243543625,
+      "learning_rate": 3.918967827149938e-05,
+      "loss": 0.0339,
+      "step": 5910
+    },
+    {
+      "epoch": 1.4344560213229949,
+      "grad_norm": 0.228776216506958,
+      "learning_rate": 3.9028301534904094e-05,
+      "loss": 0.0349,
+      "step": 5920
+    },
+    {
+      "epoch": 1.436879088926581,
+      "grad_norm": 0.1379360854625702,
+      "learning_rate": 3.88670447830454e-05,
+      "loss": 0.0296,
+      "step": 5930
+    },
+    {
+      "epoch": 1.4393021565301671,
+      "grad_norm": 0.2305406630039215,
+      "learning_rate": 3.870590977940132e-05,
+      "loss": 0.0319,
+      "step": 5940
+    },
+    {
+      "epoch": 1.4417252241337533,
+      "grad_norm": 0.15086810290813446,
+      "learning_rate": 3.8544898286118404e-05,
+      "loss": 0.0303,
+      "step": 5950
+    },
+    {
+      "epoch": 1.4441482917373394,
+      "grad_norm": 0.22204427421092987,
+      "learning_rate": 3.838401206399257e-05,
+      "loss": 0.029,
+      "step": 5960
+    },
+    {
+      "epoch": 1.4465713593409255,
+      "grad_norm": 0.12760701775550842,
+      "learning_rate": 3.822325287244975e-05,
+      "loss": 0.0284,
+      "step": 5970
+    },
+    {
+      "epoch": 1.4489944269445116,
+      "grad_norm": 0.19484172761440277,
+      "learning_rate": 3.8062622469526725e-05,
+      "loss": 0.0301,
+      "step": 5980
+    },
+    {
+      "epoch": 1.4514174945480978,
+      "grad_norm": 0.2005198746919632,
+      "learning_rate": 3.790212261185183e-05,
+      "loss": 0.0272,
+      "step": 5990
+    },
+    {
+      "epoch": 1.4538405621516841,
+      "grad_norm": 0.11993716657161713,
+      "learning_rate": 3.7741755054625794e-05,
+      "loss": 0.0331,
+      "step": 6000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}