diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8203 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 11670,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001713796058269066,
+      "grad_norm": 3.5197439193725586,
+      "learning_rate": 4.9974293059125964e-05,
+      "loss": 3.229,
+      "step": 10
+    },
+    {
+      "epoch": 0.003427592116538132,
+      "grad_norm": 1.788411259651184,
+      "learning_rate": 4.994572979148815e-05,
+      "loss": 1.4466,
+      "step": 20
+    },
+    {
+      "epoch": 0.005141388174807198,
+      "grad_norm": 1.3185161352157593,
+      "learning_rate": 4.991716652385033e-05,
+      "loss": 1.0682,
+      "step": 30
+    },
+    {
+      "epoch": 0.006855184233076264,
+      "grad_norm": 0.9748584628105164,
+      "learning_rate": 4.9888603256212515e-05,
+      "loss": 0.9275,
+      "step": 40
+    },
+    {
+      "epoch": 0.00856898029134533,
+      "grad_norm": 0.9990955591201782,
+      "learning_rate": 4.9860039988574694e-05,
+      "loss": 0.8359,
+      "step": 50
+    },
+    {
+      "epoch": 0.010282776349614395,
+      "grad_norm": 1.0820547342300415,
+      "learning_rate": 4.9831476720936873e-05,
+      "loss": 0.7742,
+      "step": 60
+    },
+    {
+      "epoch": 0.011996572407883462,
+      "grad_norm": 1.0009771585464478,
+      "learning_rate": 4.980291345329906e-05,
+      "loss": 0.7429,
+      "step": 70
+    },
+    {
+      "epoch": 0.013710368466152529,
+      "grad_norm": 0.8651469349861145,
+      "learning_rate": 4.977435018566124e-05,
+      "loss": 0.7075,
+      "step": 80
+    },
+    {
+      "epoch": 0.015424164524421594,
+      "grad_norm": 1.003004789352417,
+      "learning_rate": 4.9745786918023425e-05,
+      "loss": 0.7025,
+      "step": 90
+    },
+    {
+      "epoch": 0.01713796058269066,
+      "grad_norm": 1.044442057609558,
+      "learning_rate": 4.971722365038561e-05,
+      "loss": 0.6876,
+      "step": 100
+    },
+    {
+      "epoch": 0.018851756640959727,
+      "grad_norm": 0.912819504737854,
+      "learning_rate": 4.968866038274779e-05,
+      "loss": 0.6582,
+      "step": 110
+    },
+    {
+      "epoch": 0.02056555269922879,
+      "grad_norm": 0.7841666340827942,
+      "learning_rate": 4.966009711510997e-05,
+      "loss": 0.6521,
+      "step": 120
+    },
+    {
+      "epoch": 0.022279348757497857,
+      "grad_norm": 0.8431994915008545,
+      "learning_rate": 4.9631533847472155e-05,
+      "loss": 0.6473,
+      "step": 130
+    },
+    {
+      "epoch": 0.023993144815766924,
+      "grad_norm": 0.854539692401886,
+      "learning_rate": 4.9602970579834334e-05,
+      "loss": 0.6686,
+      "step": 140
+    },
+    {
+      "epoch": 0.02570694087403599,
+      "grad_norm": 0.8027311563491821,
+      "learning_rate": 4.957440731219652e-05,
+      "loss": 0.6357,
+      "step": 150
+    },
+    {
+      "epoch": 0.027420736932305057,
+      "grad_norm": 0.7218838930130005,
+      "learning_rate": 4.95458440445587e-05,
+      "loss": 0.636,
+      "step": 160
+    },
+    {
+      "epoch": 0.02913453299057412,
+      "grad_norm": 0.8693987727165222,
+      "learning_rate": 4.9517280776920886e-05,
+      "loss": 0.6319,
+      "step": 170
+    },
+    {
+      "epoch": 0.030848329048843187,
+      "grad_norm": 0.7060816884040833,
+      "learning_rate": 4.9488717509283065e-05,
+      "loss": 0.6215,
+      "step": 180
+    },
+    {
+      "epoch": 0.032562125107112254,
+      "grad_norm": 0.8361769914627075,
+      "learning_rate": 4.9460154241645244e-05,
+      "loss": 0.6304,
+      "step": 190
+    },
+    {
+      "epoch": 0.03427592116538132,
+      "grad_norm": 1.2885167598724365,
+      "learning_rate": 4.943159097400743e-05,
+      "loss": 0.5948,
+      "step": 200
+    },
+    {
+      "epoch": 0.03598971722365039,
+      "grad_norm": 0.8330972790718079,
+      "learning_rate": 4.940302770636961e-05,
+      "loss": 0.5624,
+      "step": 210
+    },
+    {
+      "epoch": 0.037703513281919454,
+      "grad_norm": 0.8592369556427002,
+      "learning_rate": 4.9374464438731795e-05,
+      "loss": 0.6062,
+      "step": 220
+    },
+    {
+      "epoch": 0.03941730934018852,
+      "grad_norm": 0.6641691327095032,
+      "learning_rate": 4.9345901171093975e-05,
+      "loss": 0.5822,
+      "step": 230
+    },
+    {
+      "epoch": 0.04113110539845758,
+      "grad_norm": 0.7218849062919617,
+      "learning_rate": 4.9317337903456154e-05,
+      "loss": 0.5898,
+      "step": 240
+    },
+    {
+      "epoch": 0.04284490145672665,
+      "grad_norm": 0.7507635951042175,
+      "learning_rate": 4.928877463581834e-05,
+      "loss": 0.5637,
+      "step": 250
+    },
+    {
+      "epoch": 0.044558697514995714,
+      "grad_norm": 0.6507437229156494,
+      "learning_rate": 4.926021136818052e-05,
+      "loss": 0.5765,
+      "step": 260
+    },
+    {
+      "epoch": 0.04627249357326478,
+      "grad_norm": 0.6712321043014526,
+      "learning_rate": 4.9231648100542705e-05,
+      "loss": 0.5637,
+      "step": 270
+    },
+    {
+      "epoch": 0.04798628963153385,
+      "grad_norm": 0.8501739501953125,
+      "learning_rate": 4.920308483290489e-05,
+      "loss": 0.5514,
+      "step": 280
+    },
+    {
+      "epoch": 0.049700085689802914,
+      "grad_norm": 0.7565245032310486,
+      "learning_rate": 4.917452156526707e-05,
+      "loss": 0.5569,
+      "step": 290
+    },
+    {
+      "epoch": 0.05141388174807198,
+      "grad_norm": 0.7239087224006653,
+      "learning_rate": 4.914595829762925e-05,
+      "loss": 0.5468,
+      "step": 300
+    },
+    {
+      "epoch": 0.05312767780634105,
+      "grad_norm": 0.7810051441192627,
+      "learning_rate": 4.911739502999143e-05,
+      "loss": 0.5516,
+      "step": 310
+    },
+    {
+      "epoch": 0.054841473864610114,
+      "grad_norm": 0.9277384877204895,
+      "learning_rate": 4.9088831762353615e-05,
+      "loss": 0.5593,
+      "step": 320
+    },
+    {
+      "epoch": 0.056555269922879174,
+      "grad_norm": 0.7718018293380737,
+      "learning_rate": 4.90602684947158e-05,
+      "loss": 0.5402,
+      "step": 330
+    },
+    {
+      "epoch": 0.05826906598114824,
+      "grad_norm": 0.6351784467697144,
+      "learning_rate": 4.903170522707798e-05,
+      "loss": 0.5221,
+      "step": 340
+    },
+    {
+      "epoch": 0.05998286203941731,
+      "grad_norm": 0.7788419723510742,
+      "learning_rate": 4.9003141959440166e-05,
+      "loss": 0.5186,
+      "step": 350
+    },
+    {
+      "epoch": 0.061696658097686374,
+      "grad_norm": 0.7004103064537048,
+      "learning_rate": 4.897457869180234e-05,
+      "loss": 0.5543,
+      "step": 360
+    },
+    {
+      "epoch": 0.06341045415595545,
+      "grad_norm": 0.6299038529396057,
+      "learning_rate": 4.8946015424164525e-05,
+      "loss": 0.5385,
+      "step": 370
+    },
+    {
+      "epoch": 0.06512425021422451,
+      "grad_norm": 0.6537446975708008,
+      "learning_rate": 4.891745215652671e-05,
+      "loss": 0.5068,
+      "step": 380
+    },
+    {
+      "epoch": 0.06683804627249357,
+      "grad_norm": 0.7129578590393066,
+      "learning_rate": 4.888888888888889e-05,
+      "loss": 0.5386,
+      "step": 390
+    },
+    {
+      "epoch": 0.06855184233076264,
+      "grad_norm": 0.6344906687736511,
+      "learning_rate": 4.8860325621251076e-05,
+      "loss": 0.5402,
+      "step": 400
+    },
+    {
+      "epoch": 0.0702656383890317,
+      "grad_norm": 0.7094561457633972,
+      "learning_rate": 4.8831762353613255e-05,
+      "loss": 0.5,
+      "step": 410
+    },
+    {
+      "epoch": 0.07197943444730077,
+      "grad_norm": 0.7106287479400635,
+      "learning_rate": 4.8803199085975434e-05,
+      "loss": 0.5181,
+      "step": 420
+    },
+    {
+      "epoch": 0.07369323050556983,
+      "grad_norm": 0.6683750748634338,
+      "learning_rate": 4.877463581833762e-05,
+      "loss": 0.525,
+      "step": 430
+    },
+    {
+      "epoch": 0.07540702656383891,
+      "grad_norm": 0.6109952330589294,
+      "learning_rate": 4.87460725506998e-05,
+      "loss": 0.5109,
+      "step": 440
+    },
+    {
+      "epoch": 0.07712082262210797,
+      "grad_norm": 0.6429896950721741,
+      "learning_rate": 4.8717509283061986e-05,
+      "loss": 0.5295,
+      "step": 450
+    },
+    {
+      "epoch": 0.07883461868037704,
+      "grad_norm": 0.6576216816902161,
+      "learning_rate": 4.868894601542417e-05,
+      "loss": 0.5145,
+      "step": 460
+    },
+    {
+      "epoch": 0.0805484147386461,
+      "grad_norm": 0.7273991107940674,
+      "learning_rate": 4.866038274778635e-05,
+      "loss": 0.512,
+      "step": 470
+    },
+    {
+      "epoch": 0.08226221079691516,
+      "grad_norm": 0.652603268623352,
+      "learning_rate": 4.863181948014853e-05,
+      "loss": 0.521,
+      "step": 480
+    },
+    {
+      "epoch": 0.08397600685518423,
+      "grad_norm": 0.7902228832244873,
+      "learning_rate": 4.860325621251071e-05,
+      "loss": 0.5337,
+      "step": 490
+    },
+    {
+      "epoch": 0.0856898029134533,
+      "grad_norm": 0.6212602257728577,
+      "learning_rate": 4.8574692944872895e-05,
+      "loss": 0.5135,
+      "step": 500
+    },
+    {
+      "epoch": 0.08740359897172237,
+      "grad_norm": 0.6101359724998474,
+      "learning_rate": 4.854612967723508e-05,
+      "loss": 0.4812,
+      "step": 510
+    },
+    {
+      "epoch": 0.08911739502999143,
+      "grad_norm": 0.623332142829895,
+      "learning_rate": 4.851756640959726e-05,
+      "loss": 0.4979,
+      "step": 520
+    },
+    {
+      "epoch": 0.0908311910882605,
+      "grad_norm": 0.6756040453910828,
+      "learning_rate": 4.848900314195945e-05,
+      "loss": 0.488,
+      "step": 530
+    },
+    {
+      "epoch": 0.09254498714652956,
+      "grad_norm": 0.6245468854904175,
+      "learning_rate": 4.8460439874321626e-05,
+      "loss": 0.4887,
+      "step": 540
+    },
+    {
+      "epoch": 0.09425878320479864,
+      "grad_norm": 0.6777324676513672,
+      "learning_rate": 4.8431876606683805e-05,
+      "loss": 0.518,
+      "step": 550
+    },
+    {
+      "epoch": 0.0959725792630677,
+      "grad_norm": 0.6594008207321167,
+      "learning_rate": 4.840331333904599e-05,
+      "loss": 0.5027,
+      "step": 560
+    },
+    {
+      "epoch": 0.09768637532133675,
+      "grad_norm": 0.6377647519111633,
+      "learning_rate": 4.837475007140817e-05,
+      "loss": 0.4977,
+      "step": 570
+    },
+    {
+      "epoch": 0.09940017137960583,
+      "grad_norm": 0.637972354888916,
+      "learning_rate": 4.8346186803770356e-05,
+      "loss": 0.5055,
+      "step": 580
+    },
+    {
+      "epoch": 0.10111396743787489,
+      "grad_norm": 0.5835251808166504,
+      "learning_rate": 4.8317623536132536e-05,
+      "loss": 0.4866,
+      "step": 590
+    },
+    {
+      "epoch": 0.10282776349614396,
+      "grad_norm": 0.5735305547714233,
+      "learning_rate": 4.8289060268494715e-05,
+      "loss": 0.4788,
+      "step": 600
+    },
+    {
+      "epoch": 0.10454155955441302,
+      "grad_norm": 0.643041729927063,
+      "learning_rate": 4.82604970008569e-05,
+      "loss": 0.4675,
+      "step": 610
+    },
+    {
+      "epoch": 0.1062553556126821,
+      "grad_norm": 0.6759035587310791,
+      "learning_rate": 4.823193373321908e-05,
+      "loss": 0.4864,
+      "step": 620
+    },
+    {
+      "epoch": 0.10796915167095116,
+      "grad_norm": 0.6026252508163452,
+      "learning_rate": 4.8203370465581266e-05,
+      "loss": 0.5074,
+      "step": 630
+    },
+    {
+      "epoch": 0.10968294772922023,
+      "grad_norm": 0.6817514300346375,
+      "learning_rate": 4.8174807197943445e-05,
+      "loss": 0.4844,
+      "step": 640
+    },
+    {
+      "epoch": 0.11139674378748929,
+      "grad_norm": 0.5522242188453674,
+      "learning_rate": 4.814624393030563e-05,
+      "loss": 0.5001,
+      "step": 650
+    },
+    {
+      "epoch": 0.11311053984575835,
+      "grad_norm": 0.5717945098876953,
+      "learning_rate": 4.811768066266781e-05,
+      "loss": 0.4829,
+      "step": 660
+    },
+    {
+      "epoch": 0.11482433590402742,
+      "grad_norm": 0.5866938233375549,
+      "learning_rate": 4.808911739502999e-05,
+      "loss": 0.5096,
+      "step": 670
+    },
+    {
+      "epoch": 0.11653813196229648,
+      "grad_norm": 0.5821776390075684,
+      "learning_rate": 4.8060554127392176e-05,
+      "loss": 0.49,
+      "step": 680
+    },
+    {
+      "epoch": 0.11825192802056556,
+      "grad_norm": 0.6838080883026123,
+      "learning_rate": 4.8031990859754355e-05,
+      "loss": 0.4756,
+      "step": 690
+    },
+    {
+      "epoch": 0.11996572407883462,
+      "grad_norm": 0.5338395237922668,
+      "learning_rate": 4.800342759211654e-05,
+      "loss": 0.4788,
+      "step": 700
+    },
+    {
+      "epoch": 0.12167952013710369,
+      "grad_norm": 0.6005649566650391,
+      "learning_rate": 4.797486432447873e-05,
+      "loss": 0.476,
+      "step": 710
+    },
+    {
+      "epoch": 0.12339331619537275,
+      "grad_norm": 0.5598942637443542,
+      "learning_rate": 4.7946301056840906e-05,
+      "loss": 0.4721,
+      "step": 720
+    },
+    {
+      "epoch": 0.12510711225364182,
+      "grad_norm": 0.6488958597183228,
+      "learning_rate": 4.7917737789203086e-05,
+      "loss": 0.4608,
+      "step": 730
+    },
+    {
+      "epoch": 0.1268209083119109,
+      "grad_norm": 0.6719090342521667,
+      "learning_rate": 4.788917452156527e-05,
+      "loss": 0.4972,
+      "step": 740
+    },
+    {
+      "epoch": 0.12853470437017994,
+      "grad_norm": 0.722967803478241,
+      "learning_rate": 4.786061125392745e-05,
+      "loss": 0.4663,
+      "step": 750
+    },
+    {
+      "epoch": 0.13024850042844902,
+      "grad_norm": 0.6082452535629272,
+      "learning_rate": 4.783204798628964e-05,
+      "loss": 0.4922,
+      "step": 760
+    },
+    {
+      "epoch": 0.1319622964867181,
+      "grad_norm": 0.5927706956863403,
+      "learning_rate": 4.7803484718651816e-05,
+      "loss": 0.4894,
+      "step": 770
+    },
+    {
+      "epoch": 0.13367609254498714,
+      "grad_norm": 0.5587890148162842,
+      "learning_rate": 4.7774921451013995e-05,
+      "loss": 0.4895,
+      "step": 780
+    },
+    {
+      "epoch": 0.1353898886032562,
+      "grad_norm": 0.6011940240859985,
+      "learning_rate": 4.774635818337618e-05,
+      "loss": 0.4714,
+      "step": 790
+    },
+    {
+      "epoch": 0.13710368466152528,
+      "grad_norm": 0.6309077143669128,
+      "learning_rate": 4.771779491573836e-05,
+      "loss": 0.4896,
+      "step": 800
+    },
+    {
+      "epoch": 0.13881748071979436,
+      "grad_norm": 0.5483425855636597,
+      "learning_rate": 4.7689231648100547e-05,
+      "loss": 0.4906,
+      "step": 810
+    },
+    {
+      "epoch": 0.1405312767780634,
+      "grad_norm": 0.6498242020606995,
+      "learning_rate": 4.7660668380462726e-05,
+      "loss": 0.4845,
+      "step": 820
+    },
+    {
+      "epoch": 0.14224507283633248,
+      "grad_norm": 0.6233829259872437,
+      "learning_rate": 4.763210511282491e-05,
+      "loss": 0.4587,
+      "step": 830
+    },
+    {
+      "epoch": 0.14395886889460155,
+      "grad_norm": 0.586863100528717,
+      "learning_rate": 4.760354184518709e-05,
+      "loss": 0.4577,
+      "step": 840
+    },
+    {
+      "epoch": 0.1456726649528706,
+      "grad_norm": 0.5762030482292175,
+      "learning_rate": 4.757497857754927e-05,
+      "loss": 0.4692,
+      "step": 850
+    },
+    {
+      "epoch": 0.14738646101113967,
+      "grad_norm": 0.5599485635757446,
+      "learning_rate": 4.7546415309911456e-05,
+      "loss": 0.4843,
+      "step": 860
+    },
+    {
+      "epoch": 0.14910025706940874,
+      "grad_norm": 0.5596937537193298,
+      "learning_rate": 4.7517852042273636e-05,
+      "loss": 0.4819,
+      "step": 870
+    },
+    {
+      "epoch": 0.15081405312767782,
+      "grad_norm": 0.6436209678649902,
+      "learning_rate": 4.748928877463582e-05,
+      "loss": 0.4549,
+      "step": 880
+    },
+    {
+      "epoch": 0.15252784918594686,
+      "grad_norm": 0.6014467477798462,
+      "learning_rate": 4.746072550699801e-05,
+      "loss": 0.4606,
+      "step": 890
+    },
+    {
+      "epoch": 0.15424164524421594,
+      "grad_norm": 0.6054000854492188,
+      "learning_rate": 4.743216223936019e-05,
+      "loss": 0.4556,
+      "step": 900
+    },
+    {
+      "epoch": 0.155955441302485,
+      "grad_norm": 0.5195685029029846,
+      "learning_rate": 4.7403598971722366e-05,
+      "loss": 0.4582,
+      "step": 910
+    },
+    {
+      "epoch": 0.15766923736075408,
+      "grad_norm": 0.6742907762527466,
+      "learning_rate": 4.7375035704084545e-05,
+      "loss": 0.4666,
+      "step": 920
+    },
+    {
+      "epoch": 0.15938303341902313,
+      "grad_norm": 0.8041543960571289,
+      "learning_rate": 4.734647243644673e-05,
+      "loss": 0.4533,
+      "step": 930
+    },
+    {
+      "epoch": 0.1610968294772922,
+      "grad_norm": 0.5654326677322388,
+      "learning_rate": 4.731790916880892e-05,
+      "loss": 0.4534,
+      "step": 940
+    },
+    {
+      "epoch": 0.16281062553556128,
+      "grad_norm": 0.5578976273536682,
+      "learning_rate": 4.7289345901171097e-05,
+      "loss": 0.4593,
+      "step": 950
+    },
+    {
+      "epoch": 0.16452442159383032,
+      "grad_norm": 0.5508532524108887,
+      "learning_rate": 4.726078263353328e-05,
+      "loss": 0.4528,
+      "step": 960
+    },
+    {
+      "epoch": 0.1662382176520994,
+      "grad_norm": 0.5675921440124512,
+      "learning_rate": 4.7232219365895455e-05,
+      "loss": 0.4556,
+      "step": 970
+    },
+    {
+      "epoch": 0.16795201371036847,
+      "grad_norm": 0.5847021341323853,
+      "learning_rate": 4.720365609825764e-05,
+      "loss": 0.4573,
+      "step": 980
+    },
+    {
+      "epoch": 0.16966580976863754,
+      "grad_norm": 0.6159939765930176,
+      "learning_rate": 4.717509283061983e-05,
+      "loss": 0.4497,
+      "step": 990
+    },
+    {
+      "epoch": 0.1713796058269066,
+      "grad_norm": 0.5292489528656006,
+      "learning_rate": 4.7146529562982006e-05,
+      "loss": 0.466,
+      "step": 1000
+    },
+    {
+      "epoch": 0.17309340188517566,
+      "grad_norm": 0.5373420715332031,
+      "learning_rate": 4.711796629534419e-05,
+      "loss": 0.4543,
+      "step": 1010
+    },
+    {
+      "epoch": 0.17480719794344474,
+      "grad_norm": 0.5367053747177124,
+      "learning_rate": 4.708940302770637e-05,
+      "loss": 0.4555,
+      "step": 1020
+    },
+    {
+      "epoch": 0.17652099400171378,
+      "grad_norm": 0.5302269458770752,
+      "learning_rate": 4.706083976006855e-05,
+      "loss": 0.4526,
+      "step": 1030
+    },
+    {
+      "epoch": 0.17823479005998286,
+      "grad_norm": 0.5884714126586914,
+      "learning_rate": 4.703227649243074e-05,
+      "loss": 0.4533,
+      "step": 1040
+    },
+    {
+      "epoch": 0.17994858611825193,
+      "grad_norm": 0.5478530526161194,
+      "learning_rate": 4.7003713224792916e-05,
+      "loss": 0.454,
+      "step": 1050
+    },
+    {
+      "epoch": 0.181662382176521,
+      "grad_norm": 0.5300725698471069,
+      "learning_rate": 4.69751499571551e-05,
+      "loss": 0.4413,
+      "step": 1060
+    },
+    {
+      "epoch": 0.18337617823479005,
+      "grad_norm": 0.5762424468994141,
+      "learning_rate": 4.694658668951729e-05,
+      "loss": 0.472,
+      "step": 1070
+    },
+    {
+      "epoch": 0.18508997429305912,
+      "grad_norm": 0.5916193127632141,
+      "learning_rate": 4.691802342187947e-05,
+      "loss": 0.4465,
+      "step": 1080
+    },
+    {
+      "epoch": 0.1868037703513282,
+      "grad_norm": 0.5725514888763428,
+      "learning_rate": 4.6889460154241646e-05,
+      "loss": 0.4694,
+      "step": 1090
+    },
+    {
+      "epoch": 0.18851756640959727,
+      "grad_norm": 0.6294386386871338,
+      "learning_rate": 4.6860896886603826e-05,
+      "loss": 0.4513,
+      "step": 1100
+    },
+    {
+      "epoch": 0.19023136246786632,
+      "grad_norm": 0.5739086270332336,
+      "learning_rate": 4.683233361896601e-05,
+      "loss": 0.4769,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1919451585261354,
+      "grad_norm": 0.5288265347480774,
+      "learning_rate": 4.68037703513282e-05,
+      "loss": 0.4506,
+      "step": 1120
+    },
+    {
+      "epoch": 0.19365895458440446,
+      "grad_norm": 0.5287862420082092,
+      "learning_rate": 4.677520708369038e-05,
+      "loss": 0.436,
+      "step": 1130
+    },
+    {
+      "epoch": 0.1953727506426735,
+      "grad_norm": 0.6703059077262878,
+      "learning_rate": 4.674664381605256e-05,
+      "loss": 0.447,
+      "step": 1140
+    },
+    {
+      "epoch": 0.19708654670094258,
+      "grad_norm": 0.5269696116447449,
+      "learning_rate": 4.6718080548414735e-05,
+      "loss": 0.4742,
+      "step": 1150
+    },
+    {
+      "epoch": 0.19880034275921166,
+      "grad_norm": 0.5767235159873962,
+      "learning_rate": 4.668951728077692e-05,
+      "loss": 0.4422,
+      "step": 1160
+    },
+    {
+      "epoch": 0.20051413881748073,
+      "grad_norm": 0.5331224799156189,
+      "learning_rate": 4.666095401313911e-05,
+      "loss": 0.4441,
+      "step": 1170
+    },
+    {
+      "epoch": 0.20222793487574978,
+      "grad_norm": 0.6412308216094971,
+      "learning_rate": 4.663239074550129e-05,
+      "loss": 0.4354,
+      "step": 1180
+    },
+    {
+      "epoch": 0.20394173093401885,
+      "grad_norm": 0.519543468952179,
+      "learning_rate": 4.660382747786347e-05,
+      "loss": 0.4587,
+      "step": 1190
+    },
+    {
+      "epoch": 0.20565552699228792,
+      "grad_norm": 0.5464423298835754,
+      "learning_rate": 4.657526421022565e-05,
+      "loss": 0.445,
+      "step": 1200
+    },
+    {
+      "epoch": 0.207369323050557,
+      "grad_norm": 0.5126811861991882,
+      "learning_rate": 4.654670094258783e-05,
+      "loss": 0.4336,
+      "step": 1210
+    },
+    {
+      "epoch": 0.20908311910882604,
+      "grad_norm": 0.49116766452789307,
+      "learning_rate": 4.651813767495002e-05,
+      "loss": 0.4542,
+      "step": 1220
+    },
+    {
+      "epoch": 0.21079691516709512,
+      "grad_norm": 0.562121570110321,
+      "learning_rate": 4.6489574407312196e-05,
+      "loss": 0.447,
+      "step": 1230
+    },
+    {
+      "epoch": 0.2125107112253642,
+      "grad_norm": 0.5438272356987,
+      "learning_rate": 4.646101113967438e-05,
+      "loss": 0.4473,
+      "step": 1240
+    },
+    {
+      "epoch": 0.21422450728363324,
+      "grad_norm": 0.5589945316314697,
+      "learning_rate": 4.643244787203656e-05,
+      "loss": 0.4365,
+      "step": 1250
+    },
+    {
+      "epoch": 0.2159383033419023,
+      "grad_norm": 0.49309206008911133,
+      "learning_rate": 4.640388460439875e-05,
+      "loss": 0.4549,
+      "step": 1260
+    },
+    {
+      "epoch": 0.21765209940017138,
+      "grad_norm": 0.5735007524490356,
+      "learning_rate": 4.637532133676093e-05,
+      "loss": 0.434,
+      "step": 1270
+    },
+    {
+      "epoch": 0.21936589545844046,
+      "grad_norm": 0.5194084048271179,
+      "learning_rate": 4.6346758069123106e-05,
+      "loss": 0.4324,
+      "step": 1280
+    },
+    {
+      "epoch": 0.2210796915167095,
+      "grad_norm": 0.5877565741539001,
+      "learning_rate": 4.631819480148529e-05,
+      "loss": 0.4328,
+      "step": 1290
+    },
+    {
+      "epoch": 0.22279348757497858,
+      "grad_norm": 0.6073413491249084,
+      "learning_rate": 4.628963153384747e-05,
+      "loss": 0.4413,
+      "step": 1300
+    },
+    {
+      "epoch": 0.22450728363324765,
+      "grad_norm": 0.5142035484313965,
+      "learning_rate": 4.626106826620966e-05,
+      "loss": 0.4277,
+      "step": 1310
+    },
+    {
+      "epoch": 0.2262210796915167,
+      "grad_norm": 0.6134445071220398,
+      "learning_rate": 4.6232504998571843e-05,
+      "loss": 0.4399,
+      "step": 1320
+    },
+    {
+      "epoch": 0.22793487574978577,
+      "grad_norm": 0.5537267327308655,
+      "learning_rate": 4.620394173093402e-05,
+      "loss": 0.4227,
+      "step": 1330
+    },
+    {
+      "epoch": 0.22964867180805484,
+      "grad_norm": 0.5451083779335022,
+      "learning_rate": 4.61753784632962e-05,
+      "loss": 0.4462,
+      "step": 1340
+    },
+    {
+      "epoch": 0.23136246786632392,
+      "grad_norm": 0.5471403002738953,
+      "learning_rate": 4.614681519565839e-05,
+      "loss": 0.425,
+      "step": 1350
+    },
+    {
+      "epoch": 0.23307626392459296,
+      "grad_norm": 0.5239073634147644,
+      "learning_rate": 4.611825192802057e-05,
+      "loss": 0.4396,
+      "step": 1360
+    },
+    {
+      "epoch": 0.23479005998286204,
+      "grad_norm": 0.5054474472999573,
+      "learning_rate": 4.608968866038275e-05,
+      "loss": 0.4185,
+      "step": 1370
+    },
+    {
+      "epoch": 0.2365038560411311,
+      "grad_norm": 0.5151476860046387,
+      "learning_rate": 4.606112539274493e-05,
+      "loss": 0.4284,
+      "step": 1380
+    },
+    {
+      "epoch": 0.23821765209940018,
+      "grad_norm": 0.5287209749221802,
+      "learning_rate": 4.603256212510711e-05,
+      "loss": 0.4349,
+      "step": 1390
+    },
+    {
+      "epoch": 0.23993144815766923,
+      "grad_norm": 0.5235722661018372,
+      "learning_rate": 4.60039988574693e-05,
+      "loss": 0.4283,
+      "step": 1400
+    },
+    {
+      "epoch": 0.2416452442159383,
+      "grad_norm": 0.5603968501091003,
+      "learning_rate": 4.597543558983148e-05,
+      "loss": 0.4428,
+      "step": 1410
+    },
+    {
+      "epoch": 0.24335904027420738,
+      "grad_norm": 0.54140704870224,
+      "learning_rate": 4.594687232219366e-05,
+      "loss": 0.4289,
+      "step": 1420
+    },
+    {
+      "epoch": 0.24507283633247642,
+      "grad_norm": 0.5573848485946655,
+      "learning_rate": 4.591830905455584e-05,
+      "loss": 0.4377,
+      "step": 1430
+    },
+    {
+      "epoch": 0.2467866323907455,
+      "grad_norm": 0.5520525574684143,
+      "learning_rate": 4.588974578691803e-05,
+      "loss": 0.4365,
+      "step": 1440
+    },
+    {
+      "epoch": 0.24850042844901457,
+      "grad_norm": 0.6135875582695007,
+      "learning_rate": 4.586118251928021e-05,
+      "loss": 0.4573,
+      "step": 1450
+    },
+    {
+      "epoch": 0.25021422450728364,
+      "grad_norm": 0.5709384679794312,
+      "learning_rate": 4.583261925164239e-05,
+      "loss": 0.4293,
+      "step": 1460
+    },
+    {
+      "epoch": 0.2519280205655527,
+      "grad_norm": 0.49724119901657104,
+      "learning_rate": 4.580405598400457e-05,
+      "loss": 0.4393,
+      "step": 1470
+    },
+    {
+      "epoch": 0.2536418166238218,
+      "grad_norm": 0.5525110363960266,
+      "learning_rate": 4.577549271636675e-05,
+      "loss": 0.4281,
+      "step": 1480
+    },
+    {
+      "epoch": 0.25535561268209084,
+      "grad_norm": 0.5513701438903809,
+      "learning_rate": 4.574692944872894e-05,
+      "loss": 0.4395,
+      "step": 1490
+    },
+    {
+      "epoch": 0.2570694087403599,
+      "grad_norm": 0.4892454743385315,
+      "learning_rate": 4.5718366181091124e-05,
+      "loss": 0.4315,
+      "step": 1500
+    },
+    {
+      "epoch": 0.258783204798629,
+      "grad_norm": 0.5723627209663391,
+      "learning_rate": 4.56898029134533e-05,
+      "loss": 0.4402,
+      "step": 1510
+    },
+    {
+      "epoch": 0.26049700085689803,
+      "grad_norm": 0.5150974988937378,
+      "learning_rate": 4.566123964581548e-05,
+      "loss": 0.4367,
+      "step": 1520
+    },
+    {
+      "epoch": 0.2622107969151671,
+      "grad_norm": 0.5223197340965271,
+      "learning_rate": 4.563267637817766e-05,
+      "loss": 0.4217,
+      "step": 1530
+    },
+    {
+      "epoch": 0.2639245929734362,
+      "grad_norm": 0.4551326036453247,
+      "learning_rate": 4.560411311053985e-05,
+      "loss": 0.4222,
+      "step": 1540
+    },
+    {
+      "epoch": 0.2656383890317052,
+      "grad_norm": 0.5828386545181274,
+      "learning_rate": 4.5575549842902034e-05,
+      "loss": 0.4416,
+      "step": 1550
+    },
+    {
+      "epoch": 0.26735218508997427,
+      "grad_norm": 0.5964780449867249,
+      "learning_rate": 4.554698657526421e-05,
+      "loss": 0.4306,
+      "step": 1560
+    },
+    {
+      "epoch": 0.26906598114824337,
+      "grad_norm": 0.5186784267425537,
+      "learning_rate": 4.55184233076264e-05,
+      "loss": 0.4263,
+      "step": 1570
+    },
+    {
+      "epoch": 0.2707797772065124,
+      "grad_norm": 0.49257805943489075,
+      "learning_rate": 4.548986003998857e-05,
+      "loss": 0.4066,
+      "step": 1580
+    },
+    {
+      "epoch": 0.27249357326478146,
+      "grad_norm": 0.5377402901649475,
+      "learning_rate": 4.546129677235076e-05,
+      "loss": 0.4257,
+      "step": 1590
+    },
+    {
+      "epoch": 0.27420736932305056,
+      "grad_norm": 0.4789324104785919,
+      "learning_rate": 4.543273350471294e-05,
+      "loss": 0.444,
+      "step": 1600
+    },
+    {
+      "epoch": 0.2759211653813196,
+      "grad_norm": 0.5270596742630005,
+      "learning_rate": 4.540417023707512e-05,
+      "loss": 0.4391,
+      "step": 1610
+    },
+    {
+      "epoch": 0.2776349614395887,
+      "grad_norm": 0.5666368007659912,
+      "learning_rate": 4.537560696943731e-05,
+      "loss": 0.4238,
+      "step": 1620
+    },
+    {
+      "epoch": 0.27934875749785776,
+      "grad_norm": 0.5089669823646545,
+      "learning_rate": 4.534704370179949e-05,
+      "loss": 0.4322,
+      "step": 1630
+    },
+    {
+      "epoch": 0.2810625535561268,
+      "grad_norm": 0.5666882991790771,
+      "learning_rate": 4.531848043416167e-05,
+      "loss": 0.4238,
+      "step": 1640
+    },
+    {
+      "epoch": 0.2827763496143959,
+      "grad_norm": 0.48932990431785583,
+      "learning_rate": 4.528991716652385e-05,
+      "loss": 0.4275,
+      "step": 1650
+    },
+    {
+      "epoch": 0.28449014567266495,
+      "grad_norm": 0.5295417904853821,
+      "learning_rate": 4.526135389888603e-05,
+      "loss": 0.4321,
+      "step": 1660
+    },
+    {
+      "epoch": 0.286203941730934,
+      "grad_norm": 0.5406290888786316,
+      "learning_rate": 4.523279063124822e-05,
+      "loss": 0.4351,
+      "step": 1670
+    },
+    {
+      "epoch": 0.2879177377892031,
+      "grad_norm": 0.515723466873169,
+      "learning_rate": 4.5204227363610404e-05,
+      "loss": 0.4359,
+      "step": 1680
+    },
+    {
+      "epoch": 0.28963153384747214,
+      "grad_norm": 0.6831620335578918,
+      "learning_rate": 4.5175664095972584e-05,
+      "loss": 0.4173,
+      "step": 1690
+    },
+    {
+      "epoch": 0.2913453299057412,
+      "grad_norm": 0.4642072319984436,
+      "learning_rate": 4.514710082833476e-05,
+      "loss": 0.4148,
+      "step": 1700
+    },
+    {
+      "epoch": 0.2930591259640103,
+      "grad_norm": 0.5784996747970581,
+      "learning_rate": 4.511853756069694e-05,
+      "loss": 0.4344,
+      "step": 1710
+    },
+    {
+      "epoch": 0.29477292202227934,
+      "grad_norm": 0.46822839975357056,
+      "learning_rate": 4.508997429305913e-05,
+      "loss": 0.4192,
+      "step": 1720
+    },
+    {
+      "epoch": 0.29648671808054844,
+      "grad_norm": 0.5901206731796265,
+      "learning_rate": 4.5061411025421314e-05,
+      "loss": 0.4224,
+      "step": 1730
+    },
+    {
+      "epoch": 0.2982005141388175,
+      "grad_norm": 0.5184633135795593,
+      "learning_rate": 4.503284775778349e-05,
+      "loss": 0.4459,
+      "step": 1740
+    },
+    {
+      "epoch": 0.29991431019708653,
+      "grad_norm": 0.47953465580940247,
+      "learning_rate": 4.500428449014568e-05,
+      "loss": 0.4075,
+      "step": 1750
+    },
+    {
+      "epoch": 0.30162810625535563,
+      "grad_norm": 0.508063554763794,
+      "learning_rate": 4.497572122250785e-05,
+      "loss": 0.4201,
+      "step": 1760
+    },
+    {
+      "epoch": 0.3033419023136247,
+      "grad_norm": 0.5239994525909424,
+      "learning_rate": 4.494715795487004e-05,
+      "loss": 0.4097,
+      "step": 1770
+    },
+    {
+      "epoch": 0.3050556983718937,
+      "grad_norm": 0.47601497173309326,
+      "learning_rate": 4.4918594687232224e-05,
+      "loss": 0.4176,
+      "step": 1780
+    },
+    {
+      "epoch": 0.3067694944301628,
+      "grad_norm": 0.5476956367492676,
+      "learning_rate": 4.48900314195944e-05,
+      "loss": 0.4325,
+      "step": 1790
+    },
+    {
+      "epoch": 0.30848329048843187,
+      "grad_norm": 0.5456777215003967,
+      "learning_rate": 4.486146815195659e-05,
+      "loss": 0.4202,
+      "step": 1800
+    },
+    {
+      "epoch": 0.3101970865467009,
+      "grad_norm": 0.5144990086555481,
+      "learning_rate": 4.483290488431877e-05,
+      "loss": 0.4349,
+      "step": 1810
+    },
+    {
+      "epoch": 0.31191088260497,
+      "grad_norm": 0.483378142118454,
+      "learning_rate": 4.480434161668095e-05,
+      "loss": 0.4414,
+      "step": 1820
+    },
+    {
+      "epoch": 0.31362467866323906,
+      "grad_norm": 0.542691707611084,
+      "learning_rate": 4.4775778349043134e-05,
+      "loss": 0.4298,
+      "step": 1830
+    },
+    {
+      "epoch": 0.31533847472150817,
+      "grad_norm": 0.49622848629951477,
+      "learning_rate": 4.474721508140531e-05,
+      "loss": 0.4151,
+      "step": 1840
+    },
+    {
+      "epoch": 0.3170522707797772,
+      "grad_norm": 0.48325487971305847,
+      "learning_rate": 4.47186518137675e-05,
+      "loss": 0.4111,
+      "step": 1850
+    },
+    {
+      "epoch": 0.31876606683804626,
+      "grad_norm": 0.47797560691833496,
+      "learning_rate": 4.469008854612968e-05,
+      "loss": 0.4374,
+      "step": 1860
+    },
+    {
+      "epoch": 0.32047986289631536,
+      "grad_norm": 0.48413246870040894,
+      "learning_rate": 4.4661525278491864e-05,
+      "loss": 0.4144,
+      "step": 1870
+    },
+    {
+      "epoch": 0.3221936589545844,
+      "grad_norm": 0.48628735542297363,
+      "learning_rate": 4.463296201085404e-05,
+      "loss": 0.4199,
+      "step": 1880
+    },
+    {
+      "epoch": 0.32390745501285345,
+      "grad_norm": 0.5147850513458252,
+      "learning_rate": 4.460439874321622e-05,
+      "loss": 0.4185,
+      "step": 1890
+    },
+    {
+      "epoch": 0.32562125107112255,
+      "grad_norm": 0.5466386079788208,
+      "learning_rate": 4.457583547557841e-05,
+      "loss": 0.4086,
+      "step": 1900
+    },
+    {
+      "epoch": 0.3273350471293916,
+      "grad_norm": 0.5725340247154236,
+      "learning_rate": 4.454727220794059e-05,
+      "loss": 0.4036,
+      "step": 1910
+    },
+    {
+      "epoch": 0.32904884318766064,
+      "grad_norm": 0.5826147794723511,
+      "learning_rate": 4.4518708940302774e-05,
+      "loss": 0.4045,
+      "step": 1920
+    },
+    {
+      "epoch": 0.33076263924592975,
+      "grad_norm": 0.5249732136726379,
+      "learning_rate": 4.449014567266496e-05,
+      "loss": 0.4046,
+      "step": 1930
+    },
+    {
+      "epoch": 0.3324764353041988,
+      "grad_norm": 0.49851375818252563,
+      "learning_rate": 4.446158240502714e-05,
+      "loss": 0.4303,
+      "step": 1940
+    },
+    {
+      "epoch": 0.3341902313624679,
+      "grad_norm": 0.49083590507507324,
+      "learning_rate": 4.443301913738932e-05,
+      "loss": 0.4139,
+      "step": 1950
+    },
+    {
+      "epoch": 0.33590402742073694,
+      "grad_norm": 0.4749123454093933,
+      "learning_rate": 4.44044558697515e-05,
+      "loss": 0.4178,
+      "step": 1960
+    },
+    {
+      "epoch": 0.337617823479006,
+      "grad_norm": 0.474155455827713,
+      "learning_rate": 4.4375892602113684e-05,
+      "loss": 0.4198,
+      "step": 1970
+    },
+    {
+      "epoch": 0.3393316195372751,
+      "grad_norm": 0.48373639583587646,
+      "learning_rate": 4.434732933447587e-05,
+      "loss": 0.4289,
+      "step": 1980
+    },
+    {
+      "epoch": 0.34104541559554413,
+      "grad_norm": 0.4969842731952667,
+      "learning_rate": 4.431876606683805e-05,
+      "loss": 0.4071,
+      "step": 1990
+    },
+    {
+      "epoch": 0.3427592116538132,
+      "grad_norm": 0.4806075990200043,
+      "learning_rate": 4.429020279920023e-05,
+      "loss": 0.4005,
+      "step": 2000
+    },
+    {
+      "epoch": 0.3444730077120823,
+      "grad_norm": 0.5082671642303467,
+      "learning_rate": 4.4261639531562414e-05,
+      "loss": 0.4015,
+      "step": 2010
+    },
+    {
+      "epoch": 0.3461868037703513,
+      "grad_norm": 0.530406653881073,
+      "learning_rate": 4.423307626392459e-05,
+      "loss": 0.4169,
+      "step": 2020
+    },
+    {
+      "epoch": 0.34790059982862037,
+      "grad_norm": 0.4801930785179138,
+      "learning_rate": 4.420451299628678e-05,
+      "loss": 0.4212,
+      "step": 2030
+    },
+    {
+      "epoch": 0.3496143958868895,
+      "grad_norm": 0.4946390688419342,
+      "learning_rate": 4.417594972864896e-05,
+      "loss": 0.4171,
+      "step": 2040
+    },
+    {
+      "epoch": 0.3513281919451585,
+      "grad_norm": 0.4361880123615265,
+      "learning_rate": 4.4147386461011145e-05,
+      "loss": 0.4207,
+      "step": 2050
+    },
+    {
+      "epoch": 0.35304198800342756,
+      "grad_norm": 0.48112979531288147,
+      "learning_rate": 4.4118823193373324e-05,
+      "loss": 0.4189,
+      "step": 2060
+    },
+    {
+      "epoch": 0.35475578406169667,
+      "grad_norm": 0.49672552943229675,
+      "learning_rate": 4.40902599257355e-05,
+      "loss": 0.4298,
+      "step": 2070
+    },
+    {
+      "epoch": 0.3564695801199657,
+      "grad_norm": 0.4827049970626831,
+      "learning_rate": 4.406169665809769e-05,
+      "loss": 0.4138,
+      "step": 2080
+    },
+    {
+      "epoch": 0.3581833761782348,
+      "grad_norm": 0.49597135186195374,
+      "learning_rate": 4.403313339045987e-05,
+      "loss": 0.4109,
+      "step": 2090
+    },
+    {
+      "epoch": 0.35989717223650386,
+      "grad_norm": 0.5000253915786743,
+      "learning_rate": 4.4004570122822054e-05,
+      "loss": 0.4219,
+      "step": 2100
+    },
+    {
+      "epoch": 0.3616109682947729,
+      "grad_norm": 0.5186734795570374,
+      "learning_rate": 4.397600685518424e-05,
+      "loss": 0.4304,
+      "step": 2110
+    },
+    {
+      "epoch": 0.363324764353042,
+      "grad_norm": 0.5129287242889404,
+      "learning_rate": 4.394744358754642e-05,
+      "loss": 0.416,
+      "step": 2120
+    },
+    {
+      "epoch": 0.36503856041131105,
+      "grad_norm": 0.5247070789337158,
+      "learning_rate": 4.39188803199086e-05,
+      "loss": 0.414,
+      "step": 2130
+    },
+    {
+      "epoch": 0.3667523564695801,
+      "grad_norm": 0.4930817186832428,
+      "learning_rate": 4.389031705227078e-05,
+      "loss": 0.3926,
+      "step": 2140
+    },
+    {
+      "epoch": 0.3684661525278492,
+      "grad_norm": 0.49000611901283264,
+      "learning_rate": 4.3861753784632964e-05,
+      "loss": 0.4318,
+      "step": 2150
+    },
+    {
+      "epoch": 0.37017994858611825,
+      "grad_norm": 0.4887852966785431,
+      "learning_rate": 4.383319051699515e-05,
+      "loss": 0.4099,
+      "step": 2160
+    },
+    {
+      "epoch": 0.3718937446443873,
+      "grad_norm": 0.49520814418792725,
+      "learning_rate": 4.380462724935733e-05,
+      "loss": 0.4026,
+      "step": 2170
+    },
+    {
+      "epoch": 0.3736075407026564,
+      "grad_norm": 0.5411486029624939,
+      "learning_rate": 4.377606398171951e-05,
+      "loss": 0.4369,
+      "step": 2180
+    },
+    {
+      "epoch": 0.37532133676092544,
+      "grad_norm": 0.5404260158538818,
+      "learning_rate": 4.374750071408169e-05,
+      "loss": 0.4014,
+      "step": 2190
+    },
+    {
+      "epoch": 0.37703513281919454,
+      "grad_norm": 0.4977183938026428,
+      "learning_rate": 4.3718937446443874e-05,
+      "loss": 0.4055,
+      "step": 2200
+    },
+    {
+      "epoch": 0.3787489288774636,
+      "grad_norm": 0.5122350454330444,
+      "learning_rate": 4.369037417880606e-05,
+      "loss": 0.414,
+      "step": 2210
+    },
+    {
+      "epoch": 0.38046272493573263,
+      "grad_norm": 0.46838364005088806,
+      "learning_rate": 4.366181091116824e-05,
+      "loss": 0.3943,
+      "step": 2220
+    },
+    {
+      "epoch": 0.38217652099400173,
+      "grad_norm": 0.5282347798347473,
+      "learning_rate": 4.3633247643530425e-05,
+      "loss": 0.4083,
+      "step": 2230
+    },
+    {
+      "epoch": 0.3838903170522708,
+      "grad_norm": 0.5209845304489136,
+      "learning_rate": 4.3604684375892604e-05,
+      "loss": 0.4067,
+      "step": 2240
+    },
+    {
+      "epoch": 0.3856041131105398,
+      "grad_norm": 0.47322601079940796,
+      "learning_rate": 4.3576121108254783e-05,
+      "loss": 0.3973,
+      "step": 2250
+    },
+    {
+      "epoch": 0.3873179091688089,
+      "grad_norm": 0.48601487278938293,
+      "learning_rate": 4.354755784061697e-05,
+      "loss": 0.3997,
+      "step": 2260
+    },
+    {
+      "epoch": 0.389031705227078,
+      "grad_norm": 0.5330342054367065,
+      "learning_rate": 4.351899457297915e-05,
+      "loss": 0.4055,
+      "step": 2270
+    },
+    {
+      "epoch": 0.390745501285347,
+      "grad_norm": 0.5453196167945862,
+      "learning_rate": 4.3490431305341335e-05,
+      "loss": 0.4296,
+      "step": 2280
+    },
+    {
+      "epoch": 0.3924592973436161,
+      "grad_norm": 0.4657805860042572,
+      "learning_rate": 4.346186803770352e-05,
+      "loss": 0.4034,
+      "step": 2290
+    },
+    {
+      "epoch": 0.39417309340188517,
+      "grad_norm": 0.5053611993789673,
+      "learning_rate": 4.34333047700657e-05,
+      "loss": 0.3915,
+      "step": 2300
+    },
+    {
+      "epoch": 0.39588688946015427,
+      "grad_norm": 0.46003392338752747,
+      "learning_rate": 4.340474150242788e-05,
+      "loss": 0.4058,
+      "step": 2310
+    },
+    {
+      "epoch": 0.3976006855184233,
+      "grad_norm": 0.5221563577651978,
+      "learning_rate": 4.337617823479006e-05,
+      "loss": 0.3929,
+      "step": 2320
+    },
+    {
+      "epoch": 0.39931448157669236,
+      "grad_norm": 0.5037152767181396,
+      "learning_rate": 4.3347614967152244e-05,
+      "loss": 0.407,
+      "step": 2330
+    },
+    {
+      "epoch": 0.40102827763496146,
+      "grad_norm": 0.4151383340358734,
+      "learning_rate": 4.331905169951443e-05,
+      "loss": 0.4127,
+      "step": 2340
+    },
+    {
+      "epoch": 0.4027420736932305,
+      "grad_norm": 0.45278656482696533,
+      "learning_rate": 4.329048843187661e-05,
+      "loss": 0.3931,
+      "step": 2350
+    },
+    {
+      "epoch": 0.40445586975149955,
+      "grad_norm": 0.4748023748397827,
+      "learning_rate": 4.3261925164238796e-05,
+      "loss": 0.4079,
+      "step": 2360
+    },
+    {
+      "epoch": 0.40616966580976865,
+      "grad_norm": 0.4828534424304962,
+      "learning_rate": 4.323336189660097e-05,
+      "loss": 0.4322,
+      "step": 2370
+    },
+    {
+      "epoch": 0.4078834618680377,
+      "grad_norm": 0.4558246433734894,
+      "learning_rate": 4.3204798628963154e-05,
+      "loss": 0.4125,
+      "step": 2380
+    },
+    {
+      "epoch": 0.40959725792630675,
+      "grad_norm": 0.4439447522163391,
+      "learning_rate": 4.317623536132534e-05,
+      "loss": 0.4051,
+      "step": 2390
+    },
+    {
+      "epoch": 0.41131105398457585,
+      "grad_norm": 0.5642324090003967,
+      "learning_rate": 4.314767209368752e-05,
+      "loss": 0.4008,
+      "step": 2400
+    },
+    {
+      "epoch": 0.4130248500428449,
+      "grad_norm": 0.4604392945766449,
+      "learning_rate": 4.3119108826049705e-05,
+      "loss": 0.4047,
+      "step": 2410
+    },
+    {
+      "epoch": 0.414738646101114,
+      "grad_norm": 0.45364564657211304,
+      "learning_rate": 4.3090545558411885e-05,
+      "loss": 0.3789,
+      "step": 2420
+    },
+    {
+      "epoch": 0.41645244215938304,
+      "grad_norm": 0.47339892387390137,
+      "learning_rate": 4.3061982290774064e-05,
+      "loss": 0.3985,
+      "step": 2430
+    },
+    {
+      "epoch": 0.4181662382176521,
+      "grad_norm": 0.4818185269832611,
+      "learning_rate": 4.303341902313625e-05,
+      "loss": 0.4123,
+      "step": 2440
+    },
+    {
+      "epoch": 0.4198800342759212,
+      "grad_norm": 0.4778873324394226,
+      "learning_rate": 4.300485575549843e-05,
+      "loss": 0.3915,
+      "step": 2450
+    },
+    {
+      "epoch": 0.42159383033419023,
+      "grad_norm": 0.4846336841583252,
+      "learning_rate": 4.2976292487860615e-05,
+      "loss": 0.42,
+      "step": 2460
+    },
+    {
+      "epoch": 0.4233076263924593,
+      "grad_norm": 0.5048242211341858,
+      "learning_rate": 4.2947729220222794e-05,
+      "loss": 0.3926,
+      "step": 2470
+    },
+    {
+      "epoch": 0.4250214224507284,
+      "grad_norm": 0.5179741382598877,
+      "learning_rate": 4.291916595258498e-05,
+      "loss": 0.4047,
+      "step": 2480
+    },
+    {
+      "epoch": 0.4267352185089974,
+      "grad_norm": 0.4281986653804779,
+      "learning_rate": 4.289060268494716e-05,
+      "loss": 0.3953,
+      "step": 2490
+    },
+    {
+      "epoch": 0.4284490145672665,
+      "grad_norm": 0.47407853603363037,
+      "learning_rate": 4.286203941730934e-05,
+      "loss": 0.4099,
+      "step": 2500
+    },
+    {
+      "epoch": 0.4301628106255356,
+      "grad_norm": 0.4928293228149414,
+      "learning_rate": 4.2833476149671525e-05,
+      "loss": 0.4039,
+      "step": 2510
+    },
+    {
+      "epoch": 0.4318766066838046,
+      "grad_norm": 0.45097365975379944,
+      "learning_rate": 4.2804912882033704e-05,
+      "loss": 0.401,
+      "step": 2520
+    },
+    {
+      "epoch": 0.43359040274207367,
+      "grad_norm": 0.5395087003707886,
+      "learning_rate": 4.277634961439589e-05,
+      "loss": 0.4043,
+      "step": 2530
+    },
+    {
+      "epoch": 0.43530419880034277,
+      "grad_norm": 0.48334208130836487,
+      "learning_rate": 4.2747786346758076e-05,
+      "loss": 0.4065,
+      "step": 2540
+    },
+    {
+      "epoch": 0.4370179948586118,
+      "grad_norm": 0.4694972336292267,
+      "learning_rate": 4.271922307912025e-05,
+      "loss": 0.4004,
+      "step": 2550
+    },
+    {
+      "epoch": 0.4387317909168809,
+      "grad_norm": 0.45093536376953125,
+      "learning_rate": 4.2690659811482435e-05,
+      "loss": 0.3972,
+      "step": 2560
+    },
+    {
+      "epoch": 0.44044558697514996,
+      "grad_norm": 0.606305718421936,
+      "learning_rate": 4.2662096543844614e-05,
+      "loss": 0.4081,
+      "step": 2570
+    },
+    {
+      "epoch": 0.442159383033419,
+      "grad_norm": 0.5000024437904358,
+      "learning_rate": 4.26335332762068e-05,
+      "loss": 0.4007,
+      "step": 2580
+    },
+    {
+      "epoch": 0.4438731790916881,
+      "grad_norm": 0.5056565403938293,
+      "learning_rate": 4.2604970008568986e-05,
+      "loss": 0.4004,
+      "step": 2590
+    },
+    {
+      "epoch": 0.44558697514995715,
+      "grad_norm": 0.5103461146354675,
+      "learning_rate": 4.2576406740931165e-05,
+      "loss": 0.4072,
+      "step": 2600
+    },
+    {
+      "epoch": 0.4473007712082262,
+      "grad_norm": 0.5315241813659668,
+      "learning_rate": 4.2547843473293344e-05,
+      "loss": 0.4081,
+      "step": 2610
+    },
+    {
+      "epoch": 0.4490145672664953,
+      "grad_norm": 0.51112961769104,
+      "learning_rate": 4.251928020565553e-05,
+      "loss": 0.3999,
+      "step": 2620
+    },
+    {
+      "epoch": 0.45072836332476435,
+      "grad_norm": 0.5215428471565247,
+      "learning_rate": 4.249071693801771e-05,
+      "loss": 0.4108,
+      "step": 2630
+    },
+    {
+      "epoch": 0.4524421593830334,
+      "grad_norm": 0.43863195180892944,
+      "learning_rate": 4.2462153670379896e-05,
+      "loss": 0.3979,
+      "step": 2640
+    },
+    {
+      "epoch": 0.4541559554413025,
+      "grad_norm": 0.4583381116390228,
+      "learning_rate": 4.2433590402742075e-05,
+      "loss": 0.4118,
+      "step": 2650
+    },
+    {
+      "epoch": 0.45586975149957154,
+      "grad_norm": 0.4846024811267853,
+      "learning_rate": 4.240502713510426e-05,
+      "loss": 0.4165,
+      "step": 2660
+    },
+    {
+      "epoch": 0.45758354755784064,
+      "grad_norm": 0.5279461741447449,
+      "learning_rate": 4.237646386746644e-05,
+      "loss": 0.4105,
+      "step": 2670
+    },
+    {
+      "epoch": 0.4592973436161097,
+      "grad_norm": 0.4175077974796295,
+      "learning_rate": 4.234790059982862e-05,
+      "loss": 0.3939,
+      "step": 2680
+    },
+    {
+      "epoch": 0.46101113967437873,
+      "grad_norm": 0.5505842566490173,
+      "learning_rate": 4.2319337332190805e-05,
+      "loss": 0.4048,
+      "step": 2690
+    },
+    {
+      "epoch": 0.46272493573264784,
+      "grad_norm": 0.4925336539745331,
+      "learning_rate": 4.2290774064552985e-05,
+      "loss": 0.4104,
+      "step": 2700
+    },
+    {
+      "epoch": 0.4644387317909169,
+      "grad_norm": 0.5095216035842896,
+      "learning_rate": 4.226221079691517e-05,
+      "loss": 0.4181,
+      "step": 2710
+    },
+    {
+      "epoch": 0.4661525278491859,
+      "grad_norm": 0.5652039647102356,
+      "learning_rate": 4.223364752927736e-05,
+      "loss": 0.3889,
+      "step": 2720
+    },
+    {
+      "epoch": 0.46786632390745503,
+      "grad_norm": 0.49469706416130066,
+      "learning_rate": 4.2205084261639536e-05,
+      "loss": 0.4209,
+      "step": 2730
+    },
+    {
+      "epoch": 0.4695801199657241,
+      "grad_norm": 0.42379167675971985,
+      "learning_rate": 4.2176520994001715e-05,
+      "loss": 0.3982,
+      "step": 2740
+    },
+    {
+      "epoch": 0.4712939160239931,
+      "grad_norm": 0.5219408273696899,
+      "learning_rate": 4.2147957726363894e-05,
+      "loss": 0.3772,
+      "step": 2750
+    },
+    {
+      "epoch": 0.4730077120822622,
+      "grad_norm": 0.4595051407814026,
+      "learning_rate": 4.211939445872608e-05,
+      "loss": 0.3932,
+      "step": 2760
+    },
+    {
+      "epoch": 0.47472150814053127,
+      "grad_norm": 0.4492412805557251,
+      "learning_rate": 4.2090831191088266e-05,
+      "loss": 0.408,
+      "step": 2770
+    },
+    {
+      "epoch": 0.47643530419880037,
+      "grad_norm": 0.5153499841690063,
+      "learning_rate": 4.2062267923450446e-05,
+      "loss": 0.3902,
+      "step": 2780
+    },
+    {
+      "epoch": 0.4781491002570694,
+      "grad_norm": 0.49975866079330444,
+      "learning_rate": 4.2033704655812625e-05,
+      "loss": 0.3845,
+      "step": 2790
+    },
+    {
+      "epoch": 0.47986289631533846,
+      "grad_norm": 0.4870286285877228,
+      "learning_rate": 4.2005141388174804e-05,
+      "loss": 0.4034,
+      "step": 2800
+    },
+    {
+      "epoch": 0.48157669237360756,
+      "grad_norm": 0.4432912766933441,
+      "learning_rate": 4.197657812053699e-05,
+      "loss": 0.3981,
+      "step": 2810
+    },
+    {
+      "epoch": 0.4832904884318766,
+      "grad_norm": 0.3999868929386139,
+      "learning_rate": 4.1948014852899176e-05,
+      "loss": 0.374,
+      "step": 2820
+    },
+    {
+      "epoch": 0.48500428449014565,
+      "grad_norm": 0.5054958462715149,
+      "learning_rate": 4.1919451585261355e-05,
+      "loss": 0.4125,
+      "step": 2830
+    },
+    {
+      "epoch": 0.48671808054841476,
+      "grad_norm": 0.4698047935962677,
+      "learning_rate": 4.189088831762354e-05,
+      "loss": 0.3797,
+      "step": 2840
+    },
+    {
+      "epoch": 0.4884318766066838,
+      "grad_norm": 0.45830556750297546,
+      "learning_rate": 4.186232504998572e-05,
+      "loss": 0.4044,
+      "step": 2850
+    },
+    {
+      "epoch": 0.49014567266495285,
+      "grad_norm": 0.509774386882782,
+      "learning_rate": 4.18337617823479e-05,
+      "loss": 0.3836,
+      "step": 2860
+    },
+    {
+      "epoch": 0.49185946872322195,
+      "grad_norm": 0.49995678663253784,
+      "learning_rate": 4.1805198514710086e-05,
+      "loss": 0.3799,
+      "step": 2870
+    },
+    {
+      "epoch": 0.493573264781491,
+      "grad_norm": 0.49276742339134216,
+      "learning_rate": 4.1776635247072265e-05,
+      "loss": 0.4116,
+      "step": 2880
+    },
+    {
+      "epoch": 0.4952870608397601,
+      "grad_norm": 0.4521953761577606,
+      "learning_rate": 4.174807197943445e-05,
+      "loss": 0.3889,
+      "step": 2890
+    },
+    {
+      "epoch": 0.49700085689802914,
+      "grad_norm": 0.5152191519737244,
+      "learning_rate": 4.171950871179663e-05,
+      "loss": 0.3982,
+      "step": 2900
+    },
+    {
+      "epoch": 0.4987146529562982,
+      "grad_norm": 0.4679267406463623,
+      "learning_rate": 4.1690945444158816e-05,
+      "loss": 0.3947,
+      "step": 2910
+    },
+    {
+      "epoch": 0.5004284490145673,
+      "grad_norm": 0.45497140288352966,
+      "learning_rate": 4.1662382176520996e-05,
+      "loss": 0.3759,
+      "step": 2920
+    },
+    {
+      "epoch": 0.5021422450728363,
+      "grad_norm": 0.5409018993377686,
+      "learning_rate": 4.1633818908883175e-05,
+      "loss": 0.4003,
+      "step": 2930
+    },
+    {
+      "epoch": 0.5038560411311054,
+      "grad_norm": 0.4464910328388214,
+      "learning_rate": 4.160525564124536e-05,
+      "loss": 0.3915,
+      "step": 2940
+    },
+    {
+      "epoch": 0.5055698371893744,
+      "grad_norm": 0.5109448432922363,
+      "learning_rate": 4.157669237360755e-05,
+      "loss": 0.3709,
+      "step": 2950
+    },
+    {
+      "epoch": 0.5072836332476436,
+      "grad_norm": 0.4784081280231476,
+      "learning_rate": 4.1548129105969726e-05,
+      "loss": 0.3896,
+      "step": 2960
+    },
+    {
+      "epoch": 0.5089974293059126,
+      "grad_norm": 0.4331073462963104,
+      "learning_rate": 4.1519565838331905e-05,
+      "loss": 0.3739,
+      "step": 2970
+    },
+    {
+      "epoch": 0.5107112253641817,
+      "grad_norm": 0.43807411193847656,
+      "learning_rate": 4.1491002570694085e-05,
+      "loss": 0.3891,
+      "step": 2980
+    },
+    {
+      "epoch": 0.5124250214224507,
+      "grad_norm": 0.4344954788684845,
+      "learning_rate": 4.146243930305627e-05,
+      "loss": 0.4073,
+      "step": 2990
+    },
+    {
+      "epoch": 0.5141388174807198,
+      "grad_norm": 0.45541128516197205,
+      "learning_rate": 4.1433876035418457e-05,
+      "loss": 0.4017,
+      "step": 3000
+    },
+    {
+      "epoch": 0.5158526135389888,
+      "grad_norm": 0.4948163330554962,
+      "learning_rate": 4.1405312767780636e-05,
+      "loss": 0.3836,
+      "step": 3010
+    },
+    {
+      "epoch": 0.517566409597258,
+      "grad_norm": 0.5308042764663696,
+      "learning_rate": 4.137674950014282e-05,
+      "loss": 0.3793,
+      "step": 3020
+    },
+    {
+      "epoch": 0.519280205655527,
+      "grad_norm": 0.46851393580436707,
+      "learning_rate": 4.1348186232505e-05,
+      "loss": 0.3849,
+      "step": 3030
+    },
+    {
+      "epoch": 0.5209940017137961,
+      "grad_norm": 0.482305645942688,
+      "learning_rate": 4.131962296486718e-05,
+      "loss": 0.3998,
+      "step": 3040
+    },
+    {
+      "epoch": 0.5227077977720651,
+      "grad_norm": 0.41640397906303406,
+      "learning_rate": 4.1291059697229366e-05,
+      "loss": 0.3817,
+      "step": 3050
+    },
+    {
+      "epoch": 0.5244215938303342,
+      "grad_norm": 0.46141761541366577,
+      "learning_rate": 4.1262496429591546e-05,
+      "loss": 0.3731,
+      "step": 3060
+    },
+    {
+      "epoch": 0.5261353898886033,
+      "grad_norm": 0.46264177560806274,
+      "learning_rate": 4.123393316195373e-05,
+      "loss": 0.381,
+      "step": 3070
+    },
+    {
+      "epoch": 0.5278491859468724,
+      "grad_norm": 0.4680739641189575,
+      "learning_rate": 4.120536989431591e-05,
+      "loss": 0.3817,
+      "step": 3080
+    },
+    {
+      "epoch": 0.5295629820051414,
+      "grad_norm": 0.4986321032047272,
+      "learning_rate": 4.11768066266781e-05,
+      "loss": 0.3897,
+      "step": 3090
+    },
+    {
+      "epoch": 0.5312767780634104,
+      "grad_norm": 0.4697842597961426,
+      "learning_rate": 4.1148243359040276e-05,
+      "loss": 0.3843,
+      "step": 3100
+    },
+    {
+      "epoch": 0.5329905741216795,
+      "grad_norm": 0.4521206319332123,
+      "learning_rate": 4.1119680091402455e-05,
+      "loss": 0.3856,
+      "step": 3110
+    },
+    {
+      "epoch": 0.5347043701799485,
+      "grad_norm": 0.46120381355285645,
+      "learning_rate": 4.109111682376464e-05,
+      "loss": 0.3976,
+      "step": 3120
+    },
+    {
+      "epoch": 0.5364181662382177,
+      "grad_norm": 0.48239853978157043,
+      "learning_rate": 4.106255355612682e-05,
+      "loss": 0.3949,
+      "step": 3130
+    },
+    {
+      "epoch": 0.5381319622964867,
+      "grad_norm": 0.4508815109729767,
+      "learning_rate": 4.1033990288489007e-05,
+      "loss": 0.4015,
+      "step": 3140
+    },
+    {
+      "epoch": 0.5398457583547558,
+      "grad_norm": 0.4042247533798218,
+      "learning_rate": 4.100542702085119e-05,
+      "loss": 0.4032,
+      "step": 3150
+    },
+    {
+      "epoch": 0.5415595544130248,
+      "grad_norm": 0.49115559458732605,
+      "learning_rate": 4.0976863753213365e-05,
+      "loss": 0.3847,
+      "step": 3160
+    },
+    {
+      "epoch": 0.5432733504712939,
+      "grad_norm": 0.4663139581680298,
+      "learning_rate": 4.094830048557555e-05,
+      "loss": 0.3991,
+      "step": 3170
+    },
+    {
+      "epoch": 0.5449871465295629,
+      "grad_norm": 0.5062701106071472,
+      "learning_rate": 4.091973721793773e-05,
+      "loss": 0.3901,
+      "step": 3180
+    },
+    {
+      "epoch": 0.5467009425878321,
+      "grad_norm": 0.49512970447540283,
+      "learning_rate": 4.0891173950299916e-05,
+      "loss": 0.3921,
+      "step": 3190
+    },
+    {
+      "epoch": 0.5484147386461011,
+      "grad_norm": 0.6591855883598328,
+      "learning_rate": 4.08626106826621e-05,
+      "loss": 0.3768,
+      "step": 3200
+    },
+    {
+      "epoch": 0.5501285347043702,
+      "grad_norm": 0.45946204662323,
+      "learning_rate": 4.083404741502428e-05,
+      "loss": 0.3922,
+      "step": 3210
+    },
+    {
+      "epoch": 0.5518423307626392,
+      "grad_norm": 0.4640927314758301,
+      "learning_rate": 4.080548414738646e-05,
+      "loss": 0.379,
+      "step": 3220
+    },
+    {
+      "epoch": 0.5535561268209083,
+      "grad_norm": 0.4628278911113739,
+      "learning_rate": 4.077692087974865e-05,
+      "loss": 0.4021,
+      "step": 3230
+    },
+    {
+      "epoch": 0.5552699228791774,
+      "grad_norm": 0.49231502413749695,
+      "learning_rate": 4.0748357612110826e-05,
+      "loss": 0.3959,
+      "step": 3240
+    },
+    {
+      "epoch": 0.5569837189374465,
+      "grad_norm": 0.48861929774284363,
+      "learning_rate": 4.071979434447301e-05,
+      "loss": 0.3883,
+      "step": 3250
+    },
+    {
+      "epoch": 0.5586975149957155,
+      "grad_norm": 0.5438745617866516,
+      "learning_rate": 4.069123107683519e-05,
+      "loss": 0.3985,
+      "step": 3260
+    },
+    {
+      "epoch": 0.5604113110539846,
+      "grad_norm": 0.5093244314193726,
+      "learning_rate": 4.066266780919738e-05,
+      "loss": 0.3536,
+      "step": 3270
+    },
+    {
+      "epoch": 0.5621251071122536,
+      "grad_norm": 0.43954500555992126,
+      "learning_rate": 4.0634104541559556e-05,
+      "loss": 0.3826,
+      "step": 3280
+    },
+    {
+      "epoch": 0.5638389031705227,
+      "grad_norm": 0.5128121376037598,
+      "learning_rate": 4.0605541273921736e-05,
+      "loss": 0.4037,
+      "step": 3290
+    },
+    {
+      "epoch": 0.5655526992287918,
+      "grad_norm": 0.4617377519607544,
+      "learning_rate": 4.057697800628392e-05,
+      "loss": 0.4009,
+      "step": 3300
+    },
+    {
+      "epoch": 0.5672664952870609,
+      "grad_norm": 0.46787700057029724,
+      "learning_rate": 4.05484147386461e-05,
+      "loss": 0.3989,
+      "step": 3310
+    },
+    {
+      "epoch": 0.5689802913453299,
+      "grad_norm": 0.5328530669212341,
+      "learning_rate": 4.051985147100829e-05,
+      "loss": 0.3814,
+      "step": 3320
+    },
+    {
+      "epoch": 0.570694087403599,
+      "grad_norm": 0.4385411739349365,
+      "learning_rate": 4.049128820337047e-05,
+      "loss": 0.3681,
+      "step": 3330
+    },
+    {
+      "epoch": 0.572407883461868,
+      "grad_norm": 0.45673176646232605,
+      "learning_rate": 4.0462724935732645e-05,
+      "loss": 0.3829,
+      "step": 3340
+    },
+    {
+      "epoch": 0.5741216795201372,
+      "grad_norm": 0.45383307337760925,
+      "learning_rate": 4.043416166809483e-05,
+      "loss": 0.3747,
+      "step": 3350
+    },
+    {
+      "epoch": 0.5758354755784062,
+      "grad_norm": 0.4405433237552643,
+      "learning_rate": 4.040559840045701e-05,
+      "loss": 0.3827,
+      "step": 3360
+    },
+    {
+      "epoch": 0.5775492716366752,
+      "grad_norm": 0.45892763137817383,
+      "learning_rate": 4.03770351328192e-05,
+      "loss": 0.3781,
+      "step": 3370
+    },
+    {
+      "epoch": 0.5792630676949443,
+      "grad_norm": 0.48304322361946106,
+      "learning_rate": 4.034847186518138e-05,
+      "loss": 0.398,
+      "step": 3380
+    },
+    {
+      "epoch": 0.5809768637532133,
+      "grad_norm": 0.4620954692363739,
+      "learning_rate": 4.031990859754356e-05,
+      "loss": 0.3927,
+      "step": 3390
+    },
+    {
+      "epoch": 0.5826906598114824,
+      "grad_norm": 0.4997206926345825,
+      "learning_rate": 4.029134532990574e-05,
+      "loss": 0.383,
+      "step": 3400
+    },
+    {
+      "epoch": 0.5844044558697515,
+      "grad_norm": 0.5515158176422119,
+      "learning_rate": 4.026278206226792e-05,
+      "loss": 0.3983,
+      "step": 3410
+    },
+    {
+      "epoch": 0.5861182519280206,
+      "grad_norm": 0.5057517290115356,
+      "learning_rate": 4.0234218794630106e-05,
+      "loss": 0.4052,
+      "step": 3420
+    },
+    {
+      "epoch": 0.5878320479862896,
+      "grad_norm": 0.4466065764427185,
+      "learning_rate": 4.020565552699229e-05,
+      "loss": 0.4019,
+      "step": 3430
+    },
+    {
+      "epoch": 0.5895458440445587,
+      "grad_norm": 0.44154247641563416,
+      "learning_rate": 4.017709225935447e-05,
+      "loss": 0.382,
+      "step": 3440
+    },
+    {
+      "epoch": 0.5912596401028277,
+      "grad_norm": 0.46753990650177,
+      "learning_rate": 4.014852899171666e-05,
+      "loss": 0.383,
+      "step": 3450
+    },
+    {
+      "epoch": 0.5929734361610969,
+      "grad_norm": 0.4521337151527405,
+      "learning_rate": 4.011996572407884e-05,
+      "loss": 0.3822,
+      "step": 3460
+    },
+    {
+      "epoch": 0.5946872322193659,
+      "grad_norm": 0.46324121952056885,
+      "learning_rate": 4.0091402456441016e-05,
+      "loss": 0.3954,
+      "step": 3470
+    },
+    {
+      "epoch": 0.596401028277635,
+      "grad_norm": 0.499546617269516,
+      "learning_rate": 4.00628391888032e-05,
+      "loss": 0.4129,
+      "step": 3480
+    },
+    {
+      "epoch": 0.598114824335904,
+      "grad_norm": 0.44482311606407166,
+      "learning_rate": 4.003427592116538e-05,
+      "loss": 0.392,
+      "step": 3490
+    },
+    {
+      "epoch": 0.5998286203941731,
+      "grad_norm": 0.5121467113494873,
+      "learning_rate": 4.000571265352757e-05,
+      "loss": 0.3792,
+      "step": 3500
+    },
+    {
+      "epoch": 0.6015424164524421,
+      "grad_norm": 0.4905322790145874,
+      "learning_rate": 3.997714938588975e-05,
+      "loss": 0.3858,
+      "step": 3510
+    },
+    {
+      "epoch": 0.6032562125107113,
+      "grad_norm": 0.4642052948474884,
+      "learning_rate": 3.994858611825193e-05,
+      "loss": 0.3872,
+      "step": 3520
+    },
+    {
+      "epoch": 0.6049700085689803,
+      "grad_norm": 0.47320064902305603,
+      "learning_rate": 3.992002285061411e-05,
+      "loss": 0.3881,
+      "step": 3530
+    },
+    {
+      "epoch": 0.6066838046272494,
+      "grad_norm": 0.45698079466819763,
+      "learning_rate": 3.989145958297629e-05,
+      "loss": 0.3894,
+      "step": 3540
+    },
+    {
+      "epoch": 0.6083976006855184,
+      "grad_norm": 0.4781307876110077,
+      "learning_rate": 3.986289631533848e-05,
+      "loss": 0.3826,
+      "step": 3550
+    },
+    {
+      "epoch": 0.6101113967437874,
+      "grad_norm": 0.48909029364585876,
+      "learning_rate": 3.983433304770066e-05,
+      "loss": 0.3853,
+      "step": 3560
+    },
+    {
+      "epoch": 0.6118251928020566,
+      "grad_norm": 0.45835670828819275,
+      "learning_rate": 3.980576978006284e-05,
+      "loss": 0.3905,
+      "step": 3570
+    },
+    {
+      "epoch": 0.6135389888603257,
+      "grad_norm": 0.4872204065322876,
+      "learning_rate": 3.977720651242502e-05,
+      "loss": 0.3856,
+      "step": 3580
+    },
+    {
+      "epoch": 0.6152527849185947,
+      "grad_norm": 0.4030555486679077,
+      "learning_rate": 3.97486432447872e-05,
+      "loss": 0.3713,
+      "step": 3590
+    },
+    {
+      "epoch": 0.6169665809768637,
+      "grad_norm": 0.5011148452758789,
+      "learning_rate": 3.972007997714939e-05,
+      "loss": 0.3918,
+      "step": 3600
+    },
+    {
+      "epoch": 0.6186803770351328,
+      "grad_norm": 0.4357930123806,
+      "learning_rate": 3.969151670951157e-05,
+      "loss": 0.3775,
+      "step": 3610
+    },
+    {
+      "epoch": 0.6203941730934018,
+      "grad_norm": 0.47152426838874817,
+      "learning_rate": 3.966295344187375e-05,
+      "loss": 0.3727,
+      "step": 3620
+    },
+    {
+      "epoch": 0.622107969151671,
+      "grad_norm": 0.4790559709072113,
+      "learning_rate": 3.963439017423594e-05,
+      "loss": 0.3844,
+      "step": 3630
+    },
+    {
+      "epoch": 0.62382176520994,
+      "grad_norm": 0.43179333209991455,
+      "learning_rate": 3.960582690659812e-05,
+      "loss": 0.3849,
+      "step": 3640
+    },
+    {
+      "epoch": 0.6255355612682091,
+      "grad_norm": 0.4304300546646118,
+      "learning_rate": 3.95772636389603e-05,
+      "loss": 0.3839,
+      "step": 3650
+    },
+    {
+      "epoch": 0.6272493573264781,
+      "grad_norm": 0.4341460168361664,
+      "learning_rate": 3.954870037132248e-05,
+      "loss": 0.3855,
+      "step": 3660
+    },
+    {
+      "epoch": 0.6289631533847472,
+      "grad_norm": 0.4285462200641632,
+      "learning_rate": 3.952013710368466e-05,
+      "loss": 0.3673,
+      "step": 3670
+    },
+    {
+      "epoch": 0.6306769494430163,
+      "grad_norm": 0.49335575103759766,
+      "learning_rate": 3.949157383604685e-05,
+      "loss": 0.4034,
+      "step": 3680
+    },
+    {
+      "epoch": 0.6323907455012854,
+      "grad_norm": 0.44308072328567505,
+      "learning_rate": 3.946301056840903e-05,
+      "loss": 0.3802,
+      "step": 3690
+    },
+    {
+      "epoch": 0.6341045415595544,
+      "grad_norm": 0.4259876012802124,
+      "learning_rate": 3.943444730077121e-05,
+      "loss": 0.3807,
+      "step": 3700
+    },
+    {
+      "epoch": 0.6358183376178235,
+      "grad_norm": 0.4075721204280853,
+      "learning_rate": 3.940588403313339e-05,
+      "loss": 0.3853,
+      "step": 3710
+    },
+    {
+      "epoch": 0.6375321336760925,
+      "grad_norm": 0.4169134199619293,
+      "learning_rate": 3.937732076549557e-05,
+      "loss": 0.3715,
+      "step": 3720
+    },
+    {
+      "epoch": 0.6392459297343616,
+      "grad_norm": 0.45992928743362427,
+      "learning_rate": 3.934875749785776e-05,
+      "loss": 0.3808,
+      "step": 3730
+    },
+    {
+      "epoch": 0.6409597257926307,
+      "grad_norm": 0.42471954226493835,
+      "learning_rate": 3.932019423021994e-05,
+      "loss": 0.3902,
+      "step": 3740
+    },
+    {
+      "epoch": 0.6426735218508998,
+      "grad_norm": 0.4044492840766907,
+      "learning_rate": 3.929163096258212e-05,
+      "loss": 0.3775,
+      "step": 3750
+    },
+    {
+      "epoch": 0.6443873179091688,
+      "grad_norm": 0.41978341341018677,
+      "learning_rate": 3.92630676949443e-05,
+      "loss": 0.3626,
+      "step": 3760
+    },
+    {
+      "epoch": 0.6461011139674379,
+      "grad_norm": 0.4912661910057068,
+      "learning_rate": 3.923450442730648e-05,
+      "loss": 0.3698,
+      "step": 3770
+    },
+    {
+      "epoch": 0.6478149100257069,
+      "grad_norm": 0.44264456629753113,
+      "learning_rate": 3.920594115966867e-05,
+      "loss": 0.3761,
+      "step": 3780
+    },
+    {
+      "epoch": 0.6495287060839761,
+      "grad_norm": 0.45757827162742615,
+      "learning_rate": 3.9177377892030847e-05,
+      "loss": 0.3851,
+      "step": 3790
+    },
+    {
+      "epoch": 0.6512425021422451,
+      "grad_norm": 0.5428141355514526,
+      "learning_rate": 3.914881462439303e-05,
+      "loss": 0.397,
+      "step": 3800
+    },
+    {
+      "epoch": 0.6529562982005142,
+      "grad_norm": 0.45041322708129883,
+      "learning_rate": 3.912025135675522e-05,
+      "loss": 0.3966,
+      "step": 3810
+    },
+    {
+      "epoch": 0.6546700942587832,
+      "grad_norm": 0.4285386800765991,
+      "learning_rate": 3.90916880891174e-05,
+      "loss": 0.3889,
+      "step": 3820
+    },
+    {
+      "epoch": 0.6563838903170522,
+      "grad_norm": 0.45799657702445984,
+      "learning_rate": 3.906312482147958e-05,
+      "loss": 0.4048,
+      "step": 3830
+    },
+    {
+      "epoch": 0.6580976863753213,
+      "grad_norm": 0.445076048374176,
+      "learning_rate": 3.9034561553841756e-05,
+      "loss": 0.3675,
+      "step": 3840
+    },
+    {
+      "epoch": 0.6598114824335904,
+      "grad_norm": 0.4635215103626251,
+      "learning_rate": 3.900599828620394e-05,
+      "loss": 0.3991,
+      "step": 3850
+    },
+    {
+      "epoch": 0.6615252784918595,
+      "grad_norm": 0.42482051253318787,
+      "learning_rate": 3.897743501856613e-05,
+      "loss": 0.3899,
+      "step": 3860
+    },
+    {
+      "epoch": 0.6632390745501285,
+      "grad_norm": 0.4092475175857544,
+      "learning_rate": 3.894887175092831e-05,
+      "loss": 0.382,
+      "step": 3870
+    },
+    {
+      "epoch": 0.6649528706083976,
+      "grad_norm": 0.4773096740245819,
+      "learning_rate": 3.8920308483290494e-05,
+      "loss": 0.3721,
+      "step": 3880
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 0.46646052598953247,
+      "learning_rate": 3.889174521565267e-05,
+      "loss": 0.3584,
+      "step": 3890
+    },
+    {
+      "epoch": 0.6683804627249358,
+      "grad_norm": 0.4377121925354004,
+      "learning_rate": 3.886318194801485e-05,
+      "loss": 0.3715,
+      "step": 3900
+    },
+    {
+      "epoch": 0.6700942587832048,
+      "grad_norm": 0.434952974319458,
+      "learning_rate": 3.883461868037704e-05,
+      "loss": 0.3874,
+      "step": 3910
+    },
+    {
+      "epoch": 0.6718080548414739,
+      "grad_norm": 0.44789665937423706,
+      "learning_rate": 3.880605541273922e-05,
+      "loss": 0.3832,
+      "step": 3920
+    },
+    {
+      "epoch": 0.6735218508997429,
+      "grad_norm": 0.4606776833534241,
+      "learning_rate": 3.87774921451014e-05,
+      "loss": 0.3864,
+      "step": 3930
+    },
+    {
+      "epoch": 0.675235646958012,
+      "grad_norm": 0.4751772880554199,
+      "learning_rate": 3.874892887746359e-05,
+      "loss": 0.3901,
+      "step": 3940
+    },
+    {
+      "epoch": 0.676949443016281,
+      "grad_norm": 0.4270033538341522,
+      "learning_rate": 3.872036560982576e-05,
+      "loss": 0.3792,
+      "step": 3950
+    },
+    {
+      "epoch": 0.6786632390745502,
+      "grad_norm": 0.36948758363723755,
+      "learning_rate": 3.869180234218795e-05,
+      "loss": 0.3615,
+      "step": 3960
+    },
+    {
+      "epoch": 0.6803770351328192,
+      "grad_norm": 0.40578505396842957,
+      "learning_rate": 3.866323907455013e-05,
+      "loss": 0.3905,
+      "step": 3970
+    },
+    {
+      "epoch": 0.6820908311910883,
+      "grad_norm": 0.43324077129364014,
+      "learning_rate": 3.863467580691231e-05,
+      "loss": 0.3976,
+      "step": 3980
+    },
+    {
+      "epoch": 0.6838046272493573,
+      "grad_norm": 0.4060199558734894,
+      "learning_rate": 3.86061125392745e-05,
+      "loss": 0.384,
+      "step": 3990
+    },
+    {
+      "epoch": 0.6855184233076264,
+      "grad_norm": 0.41697928309440613,
+      "learning_rate": 3.857754927163668e-05,
+      "loss": 0.3907,
+      "step": 4000
+    },
+    {
+      "epoch": 0.6872322193658955,
+      "grad_norm": 0.43197640776634216,
+      "learning_rate": 3.854898600399886e-05,
+      "loss": 0.3644,
+      "step": 4010
+    },
+    {
+      "epoch": 0.6889460154241646,
+      "grad_norm": 0.4213910400867462,
+      "learning_rate": 3.852042273636104e-05,
+      "loss": 0.3828,
+      "step": 4020
+    },
+    {
+      "epoch": 0.6906598114824336,
+      "grad_norm": 0.45907771587371826,
+      "learning_rate": 3.849185946872322e-05,
+      "loss": 0.3796,
+      "step": 4030
+    },
+    {
+      "epoch": 0.6923736075407027,
+      "grad_norm": 0.4346909523010254,
+      "learning_rate": 3.846329620108541e-05,
+      "loss": 0.3819,
+      "step": 4040
+    },
+    {
+      "epoch": 0.6940874035989717,
+      "grad_norm": 0.4609411060810089,
+      "learning_rate": 3.843473293344759e-05,
+      "loss": 0.3678,
+      "step": 4050
+    },
+    {
+      "epoch": 0.6958011996572407,
+      "grad_norm": 0.4279443323612213,
+      "learning_rate": 3.8406169665809774e-05,
+      "loss": 0.3646,
+      "step": 4060
+    },
+    {
+      "epoch": 0.6975149957155099,
+      "grad_norm": 0.42237308621406555,
+      "learning_rate": 3.837760639817195e-05,
+      "loss": 0.3764,
+      "step": 4070
+    },
+    {
+      "epoch": 0.699228791773779,
+      "grad_norm": 0.40210121870040894,
+      "learning_rate": 3.834904313053413e-05,
+      "loss": 0.3715,
+      "step": 4080
+    },
+    {
+      "epoch": 0.700942587832048,
+      "grad_norm": 0.5159941911697388,
+      "learning_rate": 3.832047986289632e-05,
+      "loss": 0.3653,
+      "step": 4090
+    },
+    {
+      "epoch": 0.702656383890317,
+      "grad_norm": 0.39419978857040405,
+      "learning_rate": 3.82919165952585e-05,
+      "loss": 0.3966,
+      "step": 4100
+    },
+    {
+      "epoch": 0.7043701799485861,
+      "grad_norm": 0.459267795085907,
+      "learning_rate": 3.8263353327620684e-05,
+      "loss": 0.3784,
+      "step": 4110
+    },
+    {
+      "epoch": 0.7060839760068551,
+      "grad_norm": 0.46595674753189087,
+      "learning_rate": 3.823479005998286e-05,
+      "loss": 0.3841,
+      "step": 4120
+    },
+    {
+      "epoch": 0.7077977720651243,
+      "grad_norm": 0.42522045969963074,
+      "learning_rate": 3.820622679234505e-05,
+      "loss": 0.3847,
+      "step": 4130
+    },
+    {
+      "epoch": 0.7095115681233933,
+      "grad_norm": 0.4188786745071411,
+      "learning_rate": 3.817766352470723e-05,
+      "loss": 0.3744,
+      "step": 4140
+    },
+    {
+      "epoch": 0.7112253641816624,
+      "grad_norm": 0.46960318088531494,
+      "learning_rate": 3.814910025706941e-05,
+      "loss": 0.3566,
+      "step": 4150
+    },
+    {
+      "epoch": 0.7129391602399314,
+      "grad_norm": 0.4544188380241394,
+      "learning_rate": 3.8120536989431594e-05,
+      "loss": 0.3798,
+      "step": 4160
+    },
+    {
+      "epoch": 0.7146529562982005,
+      "grad_norm": 0.44834500551223755,
+      "learning_rate": 3.809197372179378e-05,
+      "loss": 0.377,
+      "step": 4170
+    },
+    {
+      "epoch": 0.7163667523564696,
+      "grad_norm": 0.4910108149051666,
+      "learning_rate": 3.806341045415596e-05,
+      "loss": 0.3867,
+      "step": 4180
+    },
+    {
+      "epoch": 0.7180805484147387,
+      "grad_norm": 0.4388539493083954,
+      "learning_rate": 3.803484718651814e-05,
+      "loss": 0.3739,
+      "step": 4190
+    },
+    {
+      "epoch": 0.7197943444730077,
+      "grad_norm": 0.4148114323616028,
+      "learning_rate": 3.800628391888032e-05,
+      "loss": 0.3592,
+      "step": 4200
+    },
+    {
+      "epoch": 0.7215081405312768,
+      "grad_norm": 0.399849534034729,
+      "learning_rate": 3.79777206512425e-05,
+      "loss": 0.3705,
+      "step": 4210
+    },
+    {
+      "epoch": 0.7232219365895458,
+      "grad_norm": 0.4451843202114105,
+      "learning_rate": 3.794915738360469e-05,
+      "loss": 0.3772,
+      "step": 4220
+    },
+    {
+      "epoch": 0.7249357326478149,
+      "grad_norm": 0.4321479797363281,
+      "learning_rate": 3.792059411596687e-05,
+      "loss": 0.3654,
+      "step": 4230
+    },
+    {
+      "epoch": 0.726649528706084,
+      "grad_norm": 0.444069504737854,
+      "learning_rate": 3.7892030848329055e-05,
+      "loss": 0.3614,
+      "step": 4240
+    },
+    {
+      "epoch": 0.7283633247643531,
+      "grad_norm": 0.4706447720527649,
+      "learning_rate": 3.7863467580691234e-05,
+      "loss": 0.3809,
+      "step": 4250
+    },
+    {
+      "epoch": 0.7300771208226221,
+      "grad_norm": 0.3922620713710785,
+      "learning_rate": 3.783490431305341e-05,
+      "loss": 0.3925,
+      "step": 4260
+    },
+    {
+      "epoch": 0.7317909168808912,
+      "grad_norm": 0.4120844602584839,
+      "learning_rate": 3.78063410454156e-05,
+      "loss": 0.3617,
+      "step": 4270
+    },
+    {
+      "epoch": 0.7335047129391602,
+      "grad_norm": 0.4326859414577484,
+      "learning_rate": 3.777777777777778e-05,
+      "loss": 0.3647,
+      "step": 4280
+    },
+    {
+      "epoch": 0.7352185089974294,
+      "grad_norm": 0.40129154920578003,
+      "learning_rate": 3.7749214510139964e-05,
+      "loss": 0.3914,
+      "step": 4290
+    },
+    {
+      "epoch": 0.7369323050556984,
+      "grad_norm": 0.480270117521286,
+      "learning_rate": 3.7720651242502143e-05,
+      "loss": 0.3774,
+      "step": 4300
+    },
+    {
+      "epoch": 0.7386461011139674,
+      "grad_norm": 0.42428138852119446,
+      "learning_rate": 3.769208797486433e-05,
+      "loss": 0.3895,
+      "step": 4310
+    },
+    {
+      "epoch": 0.7403598971722365,
+      "grad_norm": 0.4401046931743622,
+      "learning_rate": 3.766352470722651e-05,
+      "loss": 0.3877,
+      "step": 4320
+    },
+    {
+      "epoch": 0.7420736932305055,
+      "grad_norm": 0.44101765751838684,
+      "learning_rate": 3.763496143958869e-05,
+      "loss": 0.3568,
+      "step": 4330
+    },
+    {
+      "epoch": 0.7437874892887746,
+      "grad_norm": 0.46002933382987976,
+      "learning_rate": 3.7606398171950874e-05,
+      "loss": 0.3699,
+      "step": 4340
+    },
+    {
+      "epoch": 0.7455012853470437,
+      "grad_norm": 0.40729400515556335,
+      "learning_rate": 3.757783490431305e-05,
+      "loss": 0.3693,
+      "step": 4350
+    },
+    {
+      "epoch": 0.7472150814053128,
+      "grad_norm": 0.39425283670425415,
+      "learning_rate": 3.754927163667524e-05,
+      "loss": 0.392,
+      "step": 4360
+    },
+    {
+      "epoch": 0.7489288774635818,
+      "grad_norm": 0.44731876254081726,
+      "learning_rate": 3.752070836903742e-05,
+      "loss": 0.38,
+      "step": 4370
+    },
+    {
+      "epoch": 0.7506426735218509,
+      "grad_norm": 0.4442377984523773,
+      "learning_rate": 3.74921451013996e-05,
+      "loss": 0.3871,
+      "step": 4380
+    },
+    {
+      "epoch": 0.7523564695801199,
+      "grad_norm": 0.42616844177246094,
+      "learning_rate": 3.7463581833761784e-05,
+      "loss": 0.3663,
+      "step": 4390
+    },
+    {
+      "epoch": 0.7540702656383891,
+      "grad_norm": 0.44392848014831543,
+      "learning_rate": 3.743501856612396e-05,
+      "loss": 0.3666,
+      "step": 4400
+    },
+    {
+      "epoch": 0.7557840616966581,
+      "grad_norm": 0.41869884729385376,
+      "learning_rate": 3.740645529848615e-05,
+      "loss": 0.3715,
+      "step": 4410
+    },
+    {
+      "epoch": 0.7574978577549272,
+      "grad_norm": 0.5005581378936768,
+      "learning_rate": 3.7377892030848335e-05,
+      "loss": 0.3933,
+      "step": 4420
+    },
+    {
+      "epoch": 0.7592116538131962,
+      "grad_norm": 0.42626211047172546,
+      "learning_rate": 3.7349328763210514e-05,
+      "loss": 0.396,
+      "step": 4430
+    },
+    {
+      "epoch": 0.7609254498714653,
+      "grad_norm": 0.4285553991794586,
+      "learning_rate": 3.7320765495572693e-05,
+      "loss": 0.3865,
+      "step": 4440
+    },
+    {
+      "epoch": 0.7626392459297343,
+      "grad_norm": 0.44240063428878784,
+      "learning_rate": 3.729220222793487e-05,
+      "loss": 0.4041,
+      "step": 4450
+    },
+    {
+      "epoch": 0.7643530419880035,
+      "grad_norm": 0.44902411103248596,
+      "learning_rate": 3.726363896029706e-05,
+      "loss": 0.3851,
+      "step": 4460
+    },
+    {
+      "epoch": 0.7660668380462725,
+      "grad_norm": 0.4316774904727936,
+      "learning_rate": 3.7235075692659245e-05,
+      "loss": 0.3647,
+      "step": 4470
+    },
+    {
+      "epoch": 0.7677806341045416,
+      "grad_norm": 0.4804839491844177,
+      "learning_rate": 3.7206512425021424e-05,
+      "loss": 0.3811,
+      "step": 4480
+    },
+    {
+      "epoch": 0.7694944301628106,
+      "grad_norm": 0.44255098700523376,
+      "learning_rate": 3.717794915738361e-05,
+      "loss": 0.3731,
+      "step": 4490
+    },
+    {
+      "epoch": 0.7712082262210797,
+      "grad_norm": 0.47301185131073,
+      "learning_rate": 3.714938588974579e-05,
+      "loss": 0.3777,
+      "step": 4500
+    },
+    {
+      "epoch": 0.7729220222793488,
+      "grad_norm": 0.4279269874095917,
+      "learning_rate": 3.712082262210797e-05,
+      "loss": 0.3839,
+      "step": 4510
+    },
+    {
+      "epoch": 0.7746358183376179,
+      "grad_norm": 0.47981521487236023,
+      "learning_rate": 3.7092259354470154e-05,
+      "loss": 0.366,
+      "step": 4520
+    },
+    {
+      "epoch": 0.7763496143958869,
+      "grad_norm": 0.4463767111301422,
+      "learning_rate": 3.7063696086832334e-05,
+      "loss": 0.387,
+      "step": 4530
+    },
+    {
+      "epoch": 0.778063410454156,
+      "grad_norm": 0.4834374487400055,
+      "learning_rate": 3.703513281919452e-05,
+      "loss": 0.3689,
+      "step": 4540
+    },
+    {
+      "epoch": 0.779777206512425,
+      "grad_norm": 0.4108034074306488,
+      "learning_rate": 3.7006569551556706e-05,
+      "loss": 0.3528,
+      "step": 4550
+    },
+    {
+      "epoch": 0.781491002570694,
+      "grad_norm": 0.46595698595046997,
+      "learning_rate": 3.697800628391888e-05,
+      "loss": 0.3776,
+      "step": 4560
+    },
+    {
+      "epoch": 0.7832047986289632,
+      "grad_norm": 0.556064784526825,
+      "learning_rate": 3.6949443016281064e-05,
+      "loss": 0.3689,
+      "step": 4570
+    },
+    {
+      "epoch": 0.7849185946872322,
+      "grad_norm": 0.49466294050216675,
+      "learning_rate": 3.6920879748643243e-05,
+      "loss": 0.3773,
+      "step": 4580
+    },
+    {
+      "epoch": 0.7866323907455013,
+      "grad_norm": 0.43849724531173706,
+      "learning_rate": 3.689231648100543e-05,
+      "loss": 0.3786,
+      "step": 4590
+    },
+    {
+      "epoch": 0.7883461868037703,
+      "grad_norm": 0.4461914598941803,
+      "learning_rate": 3.6863753213367615e-05,
+      "loss": 0.3731,
+      "step": 4600
+    },
+    {
+      "epoch": 0.7900599828620394,
+      "grad_norm": 0.39898931980133057,
+      "learning_rate": 3.6835189945729795e-05,
+      "loss": 0.3694,
+      "step": 4610
+    },
+    {
+      "epoch": 0.7917737789203085,
+      "grad_norm": 0.4689505398273468,
+      "learning_rate": 3.6806626678091974e-05,
+      "loss": 0.3742,
+      "step": 4620
+    },
+    {
+      "epoch": 0.7934875749785776,
+      "grad_norm": 0.4229956269264221,
+      "learning_rate": 3.677806341045415e-05,
+      "loss": 0.3465,
+      "step": 4630
+    },
+    {
+      "epoch": 0.7952013710368466,
+      "grad_norm": 0.4562397301197052,
+      "learning_rate": 3.674950014281634e-05,
+      "loss": 0.3684,
+      "step": 4640
+    },
+    {
+      "epoch": 0.7969151670951157,
+      "grad_norm": 0.3931344151496887,
+      "learning_rate": 3.6720936875178525e-05,
+      "loss": 0.3704,
+      "step": 4650
+    },
+    {
+      "epoch": 0.7986289631533847,
+      "grad_norm": 0.42493489384651184,
+      "learning_rate": 3.6692373607540704e-05,
+      "loss": 0.3837,
+      "step": 4660
+    },
+    {
+      "epoch": 0.8003427592116538,
+      "grad_norm": 0.42890891432762146,
+      "learning_rate": 3.666381033990289e-05,
+      "loss": 0.37,
+      "step": 4670
+    },
+    {
+      "epoch": 0.8020565552699229,
+      "grad_norm": 0.3955785930156708,
+      "learning_rate": 3.663524707226507e-05,
+      "loss": 0.3758,
+      "step": 4680
+    },
+    {
+      "epoch": 0.803770351328192,
+      "grad_norm": 0.40773460268974304,
+      "learning_rate": 3.660668380462725e-05,
+      "loss": 0.3697,
+      "step": 4690
+    },
+    {
+      "epoch": 0.805484147386461,
+      "grad_norm": 0.45977774262428284,
+      "learning_rate": 3.6578120536989435e-05,
+      "loss": 0.371,
+      "step": 4700
+    },
+    {
+      "epoch": 0.8071979434447301,
+      "grad_norm": 0.44712209701538086,
+      "learning_rate": 3.6549557269351614e-05,
+      "loss": 0.3744,
+      "step": 4710
+    },
+    {
+      "epoch": 0.8089117395029991,
+      "grad_norm": 0.4498061239719391,
+      "learning_rate": 3.65209940017138e-05,
+      "loss": 0.356,
+      "step": 4720
+    },
+    {
+      "epoch": 0.8106255355612683,
+      "grad_norm": 0.42666059732437134,
+      "learning_rate": 3.649243073407598e-05,
+      "loss": 0.3752,
+      "step": 4730
+    },
+    {
+      "epoch": 0.8123393316195373,
+      "grad_norm": 0.46936237812042236,
+      "learning_rate": 3.646386746643816e-05,
+      "loss": 0.349,
+      "step": 4740
+    },
+    {
+      "epoch": 0.8140531276778064,
+      "grad_norm": 0.4472278654575348,
+      "learning_rate": 3.6435304198800345e-05,
+      "loss": 0.3826,
+      "step": 4750
+    },
+    {
+      "epoch": 0.8157669237360754,
+      "grad_norm": 0.4189947545528412,
+      "learning_rate": 3.6406740931162524e-05,
+      "loss": 0.359,
+      "step": 4760
+    },
+    {
+      "epoch": 0.8174807197943444,
+      "grad_norm": 0.43510058522224426,
+      "learning_rate": 3.637817766352471e-05,
+      "loss": 0.3757,
+      "step": 4770
+    },
+    {
+      "epoch": 0.8191945158526135,
+      "grad_norm": 0.4573146104812622,
+      "learning_rate": 3.634961439588689e-05,
+      "loss": 0.3666,
+      "step": 4780
+    },
+    {
+      "epoch": 0.8209083119108826,
+      "grad_norm": 0.4547922909259796,
+      "learning_rate": 3.6321051128249075e-05,
+      "loss": 0.3763,
+      "step": 4790
+    },
+    {
+      "epoch": 0.8226221079691517,
+      "grad_norm": 0.4520827829837799,
+      "learning_rate": 3.6292487860611254e-05,
+      "loss": 0.3812,
+      "step": 4800
+    },
+    {
+      "epoch": 0.8243359040274207,
+      "grad_norm": 0.4624135196208954,
+      "learning_rate": 3.6263924592973434e-05,
+      "loss": 0.3747,
+      "step": 4810
+    },
+    {
+      "epoch": 0.8260497000856898,
+      "grad_norm": 0.4581402838230133,
+      "learning_rate": 3.623536132533562e-05,
+      "loss": 0.3758,
+      "step": 4820
+    },
+    {
+      "epoch": 0.8277634961439588,
+      "grad_norm": 0.4327716827392578,
+      "learning_rate": 3.6206798057697806e-05,
+      "loss": 0.3677,
+      "step": 4830
+    },
+    {
+      "epoch": 0.829477292202228,
+      "grad_norm": 0.41706621646881104,
+      "learning_rate": 3.6178234790059985e-05,
+      "loss": 0.3699,
+      "step": 4840
+    },
+    {
+      "epoch": 0.831191088260497,
+      "grad_norm": 0.48654213547706604,
+      "learning_rate": 3.614967152242217e-05,
+      "loss": 0.363,
+      "step": 4850
+    },
+    {
+      "epoch": 0.8329048843187661,
+      "grad_norm": 0.9540205001831055,
+      "learning_rate": 3.612110825478435e-05,
+      "loss": 0.3613,
+      "step": 4860
+    },
+    {
+      "epoch": 0.8346186803770351,
+      "grad_norm": 0.43841347098350525,
+      "learning_rate": 3.609254498714653e-05,
+      "loss": 0.3692,
+      "step": 4870
+    },
+    {
+      "epoch": 0.8363324764353042,
+      "grad_norm": 0.49240532517433167,
+      "learning_rate": 3.6063981719508715e-05,
+      "loss": 0.3826,
+      "step": 4880
+    },
+    {
+      "epoch": 0.8380462724935732,
+      "grad_norm": 0.4343222379684448,
+      "learning_rate": 3.6035418451870895e-05,
+      "loss": 0.3619,
+      "step": 4890
+    },
+    {
+      "epoch": 0.8397600685518424,
+      "grad_norm": 0.3697074353694916,
+      "learning_rate": 3.600685518423308e-05,
+      "loss": 0.3573,
+      "step": 4900
+    },
+    {
+      "epoch": 0.8414738646101114,
+      "grad_norm": 0.44506195187568665,
+      "learning_rate": 3.597829191659526e-05,
+      "loss": 0.3831,
+      "step": 4910
+    },
+    {
+      "epoch": 0.8431876606683805,
+      "grad_norm": 0.49516722559928894,
+      "learning_rate": 3.5949728648957446e-05,
+      "loss": 0.3707,
+      "step": 4920
+    },
+    {
+      "epoch": 0.8449014567266495,
+      "grad_norm": 0.43179890513420105,
+      "learning_rate": 3.5921165381319625e-05,
+      "loss": 0.3754,
+      "step": 4930
+    },
+    {
+      "epoch": 0.8466152527849186,
+      "grad_norm": 0.4354996681213379,
+      "learning_rate": 3.5892602113681804e-05,
+      "loss": 0.376,
+      "step": 4940
+    },
+    {
+      "epoch": 0.8483290488431876,
+      "grad_norm": 0.4149874150753021,
+      "learning_rate": 3.586403884604399e-05,
+      "loss": 0.3712,
+      "step": 4950
+    },
+    {
+      "epoch": 0.8500428449014568,
+      "grad_norm": 0.40607717633247375,
+      "learning_rate": 3.583547557840617e-05,
+      "loss": 0.3775,
+      "step": 4960
+    },
+    {
+      "epoch": 0.8517566409597258,
+      "grad_norm": 0.39906200766563416,
+      "learning_rate": 3.5806912310768356e-05,
+      "loss": 0.3667,
+      "step": 4970
+    },
+    {
+      "epoch": 0.8534704370179949,
+      "grad_norm": 0.42885473370552063,
+      "learning_rate": 3.5778349043130535e-05,
+      "loss": 0.3736,
+      "step": 4980
+    },
+    {
+      "epoch": 0.8551842330762639,
+      "grad_norm": 0.45682159066200256,
+      "learning_rate": 3.5749785775492714e-05,
+      "loss": 0.3758,
+      "step": 4990
+    },
+    {
+      "epoch": 0.856898029134533,
+      "grad_norm": 0.4330593943595886,
+      "learning_rate": 3.57212225078549e-05,
+      "loss": 0.351,
+      "step": 5000
+    },
+    {
+      "epoch": 0.8586118251928021,
+      "grad_norm": 0.48278284072875977,
+      "learning_rate": 3.569265924021708e-05,
+      "loss": 0.3764,
+      "step": 5010
+    },
+    {
+      "epoch": 0.8603256212510711,
+      "grad_norm": 0.4309553802013397,
+      "learning_rate": 3.5664095972579265e-05,
+      "loss": 0.3537,
+      "step": 5020
+    },
+    {
+      "epoch": 0.8620394173093402,
+      "grad_norm": 0.43522948026657104,
+      "learning_rate": 3.563553270494145e-05,
+      "loss": 0.3737,
+      "step": 5030
+    },
+    {
+      "epoch": 0.8637532133676092,
+      "grad_norm": 0.41307273507118225,
+      "learning_rate": 3.560696943730363e-05,
+      "loss": 0.3751,
+      "step": 5040
+    },
+    {
+      "epoch": 0.8654670094258783,
+      "grad_norm": 0.4293232262134552,
+      "learning_rate": 3.557840616966581e-05,
+      "loss": 0.3806,
+      "step": 5050
+    },
+    {
+      "epoch": 0.8671808054841473,
+      "grad_norm": 0.46006080508232117,
+      "learning_rate": 3.554984290202799e-05,
+      "loss": 0.3597,
+      "step": 5060
+    },
+    {
+      "epoch": 0.8688946015424165,
+      "grad_norm": 0.439104825258255,
+      "learning_rate": 3.5521279634390175e-05,
+      "loss": 0.3648,
+      "step": 5070
+    },
+    {
+      "epoch": 0.8706083976006855,
+      "grad_norm": 0.39455896615982056,
+      "learning_rate": 3.549271636675236e-05,
+      "loss": 0.3671,
+      "step": 5080
+    },
+    {
+      "epoch": 0.8723221936589546,
+      "grad_norm": 0.41576287150382996,
+      "learning_rate": 3.546415309911454e-05,
+      "loss": 0.3744,
+      "step": 5090
+    },
+    {
+      "epoch": 0.8740359897172236,
+      "grad_norm": 0.43900826573371887,
+      "learning_rate": 3.5435589831476726e-05,
+      "loss": 0.3693,
+      "step": 5100
+    },
+    {
+      "epoch": 0.8757497857754927,
+      "grad_norm": 0.4696173369884491,
+      "learning_rate": 3.5407026563838906e-05,
+      "loss": 0.3708,
+      "step": 5110
+    },
+    {
+      "epoch": 0.8774635818337618,
+      "grad_norm": 0.436869740486145,
+      "learning_rate": 3.5378463296201085e-05,
+      "loss": 0.3684,
+      "step": 5120
+    },
+    {
+      "epoch": 0.8791773778920309,
+      "grad_norm": 0.4194677770137787,
+      "learning_rate": 3.534990002856327e-05,
+      "loss": 0.3685,
+      "step": 5130
+    },
+    {
+      "epoch": 0.8808911739502999,
+      "grad_norm": 0.41912123560905457,
+      "learning_rate": 3.532133676092545e-05,
+      "loss": 0.376,
+      "step": 5140
+    },
+    {
+      "epoch": 0.882604970008569,
+      "grad_norm": 0.410951167345047,
+      "learning_rate": 3.5292773493287636e-05,
+      "loss": 0.3623,
+      "step": 5150
+    },
+    {
+      "epoch": 0.884318766066838,
+      "grad_norm": 0.41262558102607727,
+      "learning_rate": 3.5264210225649815e-05,
+      "loss": 0.389,
+      "step": 5160
+    },
+    {
+      "epoch": 0.8860325621251071,
+      "grad_norm": 0.40964463353157043,
+      "learning_rate": 3.5235646958011995e-05,
+      "loss": 0.383,
+      "step": 5170
+    },
+    {
+      "epoch": 0.8877463581833762,
+      "grad_norm": 0.4126282334327698,
+      "learning_rate": 3.520708369037418e-05,
+      "loss": 0.3431,
+      "step": 5180
+    },
+    {
+      "epoch": 0.8894601542416453,
+      "grad_norm": 0.43680474162101746,
+      "learning_rate": 3.517852042273636e-05,
+      "loss": 0.3742,
+      "step": 5190
+    },
+    {
+      "epoch": 0.8911739502999143,
+      "grad_norm": 0.42978253960609436,
+      "learning_rate": 3.5149957155098546e-05,
+      "loss": 0.3621,
+      "step": 5200
+    },
+    {
+      "epoch": 0.8928877463581834,
+      "grad_norm": 0.4275003969669342,
+      "learning_rate": 3.512139388746073e-05,
+      "loss": 0.3784,
+      "step": 5210
+    },
+    {
+      "epoch": 0.8946015424164524,
+      "grad_norm": 0.5404401421546936,
+      "learning_rate": 3.509283061982291e-05,
+      "loss": 0.3778,
+      "step": 5220
+    },
+    {
+      "epoch": 0.8963153384747216,
+      "grad_norm": 0.4109743535518646,
+      "learning_rate": 3.506426735218509e-05,
+      "loss": 0.3597,
+      "step": 5230
+    },
+    {
+      "epoch": 0.8980291345329906,
+      "grad_norm": 0.4215233623981476,
+      "learning_rate": 3.503570408454727e-05,
+      "loss": 0.3705,
+      "step": 5240
+    },
+    {
+      "epoch": 0.8997429305912596,
+      "grad_norm": 0.5016302466392517,
+      "learning_rate": 3.5007140816909456e-05,
+      "loss": 0.3667,
+      "step": 5250
+    },
+    {
+      "epoch": 0.9014567266495287,
+      "grad_norm": 0.4061182141304016,
+      "learning_rate": 3.497857754927164e-05,
+      "loss": 0.3602,
+      "step": 5260
+    },
+    {
+      "epoch": 0.9031705227077977,
+      "grad_norm": 0.4132539629936218,
+      "learning_rate": 3.495001428163382e-05,
+      "loss": 0.3638,
+      "step": 5270
+    },
+    {
+      "epoch": 0.9048843187660668,
+      "grad_norm": 0.38507506251335144,
+      "learning_rate": 3.492145101399601e-05,
+      "loss": 0.3557,
+      "step": 5280
+    },
+    {
+      "epoch": 0.9065981148243359,
+      "grad_norm": 0.39408767223358154,
+      "learning_rate": 3.4892887746358186e-05,
+      "loss": 0.3544,
+      "step": 5290
+    },
+    {
+      "epoch": 0.908311910882605,
+      "grad_norm": 0.44993147253990173,
+      "learning_rate": 3.4864324478720365e-05,
+      "loss": 0.3619,
+      "step": 5300
+    },
+    {
+      "epoch": 0.910025706940874,
+      "grad_norm": 0.4015878736972809,
+      "learning_rate": 3.483576121108255e-05,
+      "loss": 0.3647,
+      "step": 5310
+    },
+    {
+      "epoch": 0.9117395029991431,
+      "grad_norm": 0.41548463702201843,
+      "learning_rate": 3.480719794344473e-05,
+      "loss": 0.3649,
+      "step": 5320
+    },
+    {
+      "epoch": 0.9134532990574121,
+      "grad_norm": 0.4296075105667114,
+      "learning_rate": 3.4778634675806917e-05,
+      "loss": 0.3688,
+      "step": 5330
+    },
+    {
+      "epoch": 0.9151670951156813,
+      "grad_norm": 0.40596818923950195,
+      "learning_rate": 3.4750071408169096e-05,
+      "loss": 0.3553,
+      "step": 5340
+    },
+    {
+      "epoch": 0.9168808911739503,
+      "grad_norm": 0.4294643998146057,
+      "learning_rate": 3.4721508140531275e-05,
+      "loss": 0.3624,
+      "step": 5350
+    },
+    {
+      "epoch": 0.9185946872322194,
+      "grad_norm": 0.40626177191734314,
+      "learning_rate": 3.469294487289346e-05,
+      "loss": 0.358,
+      "step": 5360
+    },
+    {
+      "epoch": 0.9203084832904884,
+      "grad_norm": 0.4296109974384308,
+      "learning_rate": 3.466438160525564e-05,
+      "loss": 0.3613,
+      "step": 5370
+    },
+    {
+      "epoch": 0.9220222793487575,
+      "grad_norm": 0.4380834400653839,
+      "learning_rate": 3.4635818337617826e-05,
+      "loss": 0.3795,
+      "step": 5380
+    },
+    {
+      "epoch": 0.9237360754070265,
+      "grad_norm": 0.41026878356933594,
+      "learning_rate": 3.4607255069980005e-05,
+      "loss": 0.3748,
+      "step": 5390
+    },
+    {
+      "epoch": 0.9254498714652957,
+      "grad_norm": 0.4788786768913269,
+      "learning_rate": 3.457869180234219e-05,
+      "loss": 0.3606,
+      "step": 5400
+    },
+    {
+      "epoch": 0.9271636675235647,
+      "grad_norm": 0.40038466453552246,
+      "learning_rate": 3.455012853470437e-05,
+      "loss": 0.3885,
+      "step": 5410
+    },
+    {
+      "epoch": 0.9288774635818338,
+      "grad_norm": 0.4130307734012604,
+      "learning_rate": 3.452156526706655e-05,
+      "loss": 0.3796,
+      "step": 5420
+    },
+    {
+      "epoch": 0.9305912596401028,
+      "grad_norm": 0.4447920024394989,
+      "learning_rate": 3.4493001999428736e-05,
+      "loss": 0.3951,
+      "step": 5430
+    },
+    {
+      "epoch": 0.9323050556983719,
+      "grad_norm": 0.41205018758773804,
+      "learning_rate": 3.446443873179092e-05,
+      "loss": 0.3549,
+      "step": 5440
+    },
+    {
+      "epoch": 0.934018851756641,
+      "grad_norm": 0.4112369418144226,
+      "learning_rate": 3.44358754641531e-05,
+      "loss": 0.3634,
+      "step": 5450
+    },
+    {
+      "epoch": 0.9357326478149101,
+      "grad_norm": 0.3979431390762329,
+      "learning_rate": 3.440731219651529e-05,
+      "loss": 0.383,
+      "step": 5460
+    },
+    {
+      "epoch": 0.9374464438731791,
+      "grad_norm": 0.3819255232810974,
+      "learning_rate": 3.4378748928877466e-05,
+      "loss": 0.3705,
+      "step": 5470
+    },
+    {
+      "epoch": 0.9391602399314481,
+      "grad_norm": 0.35595545172691345,
+      "learning_rate": 3.4350185661239646e-05,
+      "loss": 0.3624,
+      "step": 5480
+    },
+    {
+      "epoch": 0.9408740359897172,
+      "grad_norm": 0.453141450881958,
+      "learning_rate": 3.432162239360183e-05,
+      "loss": 0.357,
+      "step": 5490
+    },
+    {
+      "epoch": 0.9425878320479862,
+      "grad_norm": 0.4668983817100525,
+      "learning_rate": 3.429305912596401e-05,
+      "loss": 0.3652,
+      "step": 5500
+    },
+    {
+      "epoch": 0.9443016281062554,
+      "grad_norm": 0.4746719300746918,
+      "learning_rate": 3.42644958583262e-05,
+      "loss": 0.3726,
+      "step": 5510
+    },
+    {
+      "epoch": 0.9460154241645244,
+      "grad_norm": 0.4136699140071869,
+      "learning_rate": 3.4235932590688376e-05,
+      "loss": 0.3664,
+      "step": 5520
+    },
+    {
+      "epoch": 0.9477292202227935,
+      "grad_norm": 0.50481778383255,
+      "learning_rate": 3.4207369323050555e-05,
+      "loss": 0.3729,
+      "step": 5530
+    },
+    {
+      "epoch": 0.9494430162810625,
+      "grad_norm": 0.43926066160202026,
+      "learning_rate": 3.417880605541274e-05,
+      "loss": 0.3689,
+      "step": 5540
+    },
+    {
+      "epoch": 0.9511568123393316,
+      "grad_norm": 0.3832367956638336,
+      "learning_rate": 3.415024278777492e-05,
+      "loss": 0.3679,
+      "step": 5550
+    },
+    {
+      "epoch": 0.9528706083976007,
+      "grad_norm": 0.46387988328933716,
+      "learning_rate": 3.412167952013711e-05,
+      "loss": 0.3538,
+      "step": 5560
+    },
+    {
+      "epoch": 0.9545844044558698,
+      "grad_norm": 0.4562791585922241,
+      "learning_rate": 3.4093116252499286e-05,
+      "loss": 0.3631,
+      "step": 5570
+    },
+    {
+      "epoch": 0.9562982005141388,
+      "grad_norm": 0.436279833316803,
+      "learning_rate": 3.406455298486147e-05,
+      "loss": 0.373,
+      "step": 5580
+    },
+    {
+      "epoch": 0.9580119965724079,
+      "grad_norm": 0.4309545159339905,
+      "learning_rate": 3.403598971722365e-05,
+      "loss": 0.3688,
+      "step": 5590
+    },
+    {
+      "epoch": 0.9597257926306769,
+      "grad_norm": 0.4345686435699463,
+      "learning_rate": 3.400742644958583e-05,
+      "loss": 0.3767,
+      "step": 5600
+    },
+    {
+      "epoch": 0.961439588688946,
+      "grad_norm": 0.47498658299446106,
+      "learning_rate": 3.3978863181948016e-05,
+      "loss": 0.3624,
+      "step": 5610
+    },
+    {
+      "epoch": 0.9631533847472151,
+      "grad_norm": 0.5105198621749878,
+      "learning_rate": 3.3950299914310196e-05,
+      "loss": 0.3579,
+      "step": 5620
+    },
+    {
+      "epoch": 0.9648671808054842,
+      "grad_norm": 0.42196258902549744,
+      "learning_rate": 3.392173664667238e-05,
+      "loss": 0.3641,
+      "step": 5630
+    },
+    {
+      "epoch": 0.9665809768637532,
+      "grad_norm": 0.4690169095993042,
+      "learning_rate": 3.389317337903457e-05,
+      "loss": 0.385,
+      "step": 5640
+    },
+    {
+      "epoch": 0.9682947729220223,
+      "grad_norm": 0.3849544823169708,
+      "learning_rate": 3.386461011139675e-05,
+      "loss": 0.3712,
+      "step": 5650
+    },
+    {
+      "epoch": 0.9700085689802913,
+      "grad_norm": 0.459349125623703,
+      "learning_rate": 3.3836046843758926e-05,
+      "loss": 0.3841,
+      "step": 5660
+    },
+    {
+      "epoch": 0.9717223650385605,
+      "grad_norm": 0.4056195020675659,
+      "learning_rate": 3.3807483576121105e-05,
+      "loss": 0.3581,
+      "step": 5670
+    },
+    {
+      "epoch": 0.9734361610968295,
+      "grad_norm": 0.46903523802757263,
+      "learning_rate": 3.377892030848329e-05,
+      "loss": 0.3661,
+      "step": 5680
+    },
+    {
+      "epoch": 0.9751499571550986,
+      "grad_norm": 0.41805553436279297,
+      "learning_rate": 3.375035704084548e-05,
+      "loss": 0.3711,
+      "step": 5690
+    },
+    {
+      "epoch": 0.9768637532133676,
+      "grad_norm": 0.4520112872123718,
+      "learning_rate": 3.372179377320766e-05,
+      "loss": 0.3554,
+      "step": 5700
+    },
+    {
+      "epoch": 0.9785775492716366,
+      "grad_norm": 0.4656761884689331,
+      "learning_rate": 3.369323050556984e-05,
+      "loss": 0.3681,
+      "step": 5710
+    },
+    {
+      "epoch": 0.9802913453299057,
+      "grad_norm": 0.4471518099308014,
+      "learning_rate": 3.3664667237932015e-05,
+      "loss": 0.3576,
+      "step": 5720
+    },
+    {
+      "epoch": 0.9820051413881749,
+      "grad_norm": 0.4089897572994232,
+      "learning_rate": 3.36361039702942e-05,
+      "loss": 0.3647,
+      "step": 5730
+    },
+    {
+      "epoch": 0.9837189374464439,
+      "grad_norm": 0.40042605996131897,
+      "learning_rate": 3.360754070265639e-05,
+      "loss": 0.3538,
+      "step": 5740
+    },
+    {
+      "epoch": 0.9854327335047129,
+      "grad_norm": 0.4252929985523224,
+      "learning_rate": 3.3578977435018566e-05,
+      "loss": 0.3705,
+      "step": 5750
+    },
+    {
+      "epoch": 0.987146529562982,
+      "grad_norm": 0.4239642024040222,
+      "learning_rate": 3.355041416738075e-05,
+      "loss": 0.3459,
+      "step": 5760
+    },
+    {
+      "epoch": 0.988860325621251,
+      "grad_norm": 0.42976200580596924,
+      "learning_rate": 3.352185089974293e-05,
+      "loss": 0.3481,
+      "step": 5770
+    },
+    {
+      "epoch": 0.9905741216795202,
+      "grad_norm": 0.4923042058944702,
+      "learning_rate": 3.349328763210511e-05,
+      "loss": 0.3704,
+      "step": 5780
+    },
+    {
+      "epoch": 0.9922879177377892,
+      "grad_norm": 0.44239214062690735,
+      "learning_rate": 3.34647243644673e-05,
+      "loss": 0.3615,
+      "step": 5790
+    },
+    {
+      "epoch": 0.9940017137960583,
+      "grad_norm": 0.4351910948753357,
+      "learning_rate": 3.3436161096829476e-05,
+      "loss": 0.3767,
+      "step": 5800
+    },
+    {
+      "epoch": 0.9957155098543273,
+      "grad_norm": 0.4020942151546478,
+      "learning_rate": 3.340759782919166e-05,
+      "loss": 0.3824,
+      "step": 5810
+    },
+    {
+      "epoch": 0.9974293059125964,
+      "grad_norm": 0.4038684070110321,
+      "learning_rate": 3.337903456155385e-05,
+      "loss": 0.3643,
+      "step": 5820
+    },
+    {
+      "epoch": 0.9991431019708654,
+      "grad_norm": 0.3742600679397583,
+      "learning_rate": 3.335047129391603e-05,
+      "loss": 0.3374,
+      "step": 5830
+    },
+    {
+      "epoch": 1.0008568980291346,
+      "grad_norm": 0.45896250009536743,
+      "learning_rate": 3.332190802627821e-05,
+      "loss": 0.3483,
+      "step": 5840
+    },
+    {
+      "epoch": 1.0025706940874035,
+      "grad_norm": 0.43760430812835693,
+      "learning_rate": 3.3293344758640386e-05,
+      "loss": 0.3359,
+      "step": 5850
+    },
+    {
+      "epoch": 1.0042844901456727,
+      "grad_norm": 0.43930524587631226,
+      "learning_rate": 3.326478149100257e-05,
+      "loss": 0.3547,
+      "step": 5860
+    },
+    {
+      "epoch": 1.0059982862039418,
+      "grad_norm": 0.47293394804000854,
+      "learning_rate": 3.323621822336476e-05,
+      "loss": 0.3548,
+      "step": 5870
+    },
+    {
+      "epoch": 1.0077120822622108,
+      "grad_norm": 0.422246515750885,
+      "learning_rate": 3.320765495572694e-05,
+      "loss": 0.3536,
+      "step": 5880
+    },
+    {
+      "epoch": 1.00942587832048,
+      "grad_norm": 0.3768950402736664,
+      "learning_rate": 3.317909168808912e-05,
+      "loss": 0.3375,
+      "step": 5890
+    },
+    {
+      "epoch": 1.0111396743787489,
+      "grad_norm": 0.40679293870925903,
+      "learning_rate": 3.3150528420451296e-05,
+      "loss": 0.3557,
+      "step": 5900
+    },
+    {
+      "epoch": 1.012853470437018,
+      "grad_norm": 0.42642706632614136,
+      "learning_rate": 3.312196515281348e-05,
+      "loss": 0.372,
+      "step": 5910
+    },
+    {
+      "epoch": 1.0145672664952872,
+      "grad_norm": 0.46030041575431824,
+      "learning_rate": 3.309340188517567e-05,
+      "loss": 0.3575,
+      "step": 5920
+    },
+    {
+      "epoch": 1.016281062553556,
+      "grad_norm": 0.40845292806625366,
+      "learning_rate": 3.306483861753785e-05,
+      "loss": 0.3424,
+      "step": 5930
+    },
+    {
+      "epoch": 1.0179948586118253,
+      "grad_norm": 0.4577491879463196,
+      "learning_rate": 3.303627534990003e-05,
+      "loss": 0.3494,
+      "step": 5940
+    },
+    {
+      "epoch": 1.0197086546700942,
+      "grad_norm": 0.42184576392173767,
+      "learning_rate": 3.300771208226221e-05,
+      "loss": 0.3597,
+      "step": 5950
+    },
+    {
+      "epoch": 1.0214224507283634,
+      "grad_norm": 0.44896042346954346,
+      "learning_rate": 3.297914881462439e-05,
+      "loss": 0.3453,
+      "step": 5960
+    },
+    {
+      "epoch": 1.0231362467866323,
+      "grad_norm": 0.41005054116249084,
+      "learning_rate": 3.295058554698658e-05,
+      "loss": 0.353,
+      "step": 5970
+    },
+    {
+      "epoch": 1.0248500428449014,
+      "grad_norm": 0.38638490438461304,
+      "learning_rate": 3.2922022279348757e-05,
+      "loss": 0.339,
+      "step": 5980
+    },
+    {
+      "epoch": 1.0265638389031706,
+      "grad_norm": 0.4330446422100067,
+      "learning_rate": 3.289345901171094e-05,
+      "loss": 0.3291,
+      "step": 5990
+    },
+    {
+      "epoch": 1.0282776349614395,
+      "grad_norm": 0.38331279158592224,
+      "learning_rate": 3.286489574407312e-05,
+      "loss": 0.3667,
+      "step": 6000
+    },
+    {
+      "epoch": 1.0299914310197087,
+      "grad_norm": 0.42192813754081726,
+      "learning_rate": 3.283633247643531e-05,
+      "loss": 0.3419,
+      "step": 6010
+    },
+    {
+      "epoch": 1.0317052270779776,
+      "grad_norm": 0.467171847820282,
+      "learning_rate": 3.280776920879749e-05,
+      "loss": 0.3398,
+      "step": 6020
+    },
+    {
+      "epoch": 1.0334190231362468,
+      "grad_norm": 0.4578188359737396,
+      "learning_rate": 3.2779205941159666e-05,
+      "loss": 0.3555,
+      "step": 6030
+    },
+    {
+      "epoch": 1.035132819194516,
+      "grad_norm": 0.3848859369754791,
+      "learning_rate": 3.275064267352185e-05,
+      "loss": 0.3585,
+      "step": 6040
+    },
+    {
+      "epoch": 1.0368466152527849,
+      "grad_norm": 0.41844481229782104,
+      "learning_rate": 3.272207940588404e-05,
+      "loss": 0.3536,
+      "step": 6050
+    },
+    {
+      "epoch": 1.038560411311054,
+      "grad_norm": 0.4219590425491333,
+      "learning_rate": 3.269351613824622e-05,
+      "loss": 0.3346,
+      "step": 6060
+    },
+    {
+      "epoch": 1.040274207369323,
+      "grad_norm": 0.397193044424057,
+      "learning_rate": 3.2664952870608404e-05,
+      "loss": 0.3524,
+      "step": 6070
+    },
+    {
+      "epoch": 1.0419880034275921,
+      "grad_norm": 0.3962811231613159,
+      "learning_rate": 3.263638960297058e-05,
+      "loss": 0.3528,
+      "step": 6080
+    },
+    {
+      "epoch": 1.0437017994858613,
+      "grad_norm": 0.42967748641967773,
+      "learning_rate": 3.260782633533276e-05,
+      "loss": 0.3452,
+      "step": 6090
+    },
+    {
+      "epoch": 1.0454155955441302,
+      "grad_norm": 0.46409425139427185,
+      "learning_rate": 3.257926306769495e-05,
+      "loss": 0.3503,
+      "step": 6100
+    },
+    {
+      "epoch": 1.0471293916023994,
+      "grad_norm": 0.49521201848983765,
+      "learning_rate": 3.255069980005713e-05,
+      "loss": 0.3545,
+      "step": 6110
+    },
+    {
+      "epoch": 1.0488431876606683,
+      "grad_norm": 0.4241666793823242,
+      "learning_rate": 3.252213653241931e-05,
+      "loss": 0.351,
+      "step": 6120
+    },
+    {
+      "epoch": 1.0505569837189375,
+      "grad_norm": 0.47557422518730164,
+      "learning_rate": 3.249357326478149e-05,
+      "loss": 0.3474,
+      "step": 6130
+    },
+    {
+      "epoch": 1.0522707797772064,
+      "grad_norm": 0.4359321594238281,
+      "learning_rate": 3.246500999714367e-05,
+      "loss": 0.3382,
+      "step": 6140
+    },
+    {
+      "epoch": 1.0539845758354756,
+      "grad_norm": 0.4463767409324646,
+      "learning_rate": 3.243644672950586e-05,
+      "loss": 0.3409,
+      "step": 6150
+    },
+    {
+      "epoch": 1.0556983718937447,
+      "grad_norm": 0.3884042501449585,
+      "learning_rate": 3.240788346186804e-05,
+      "loss": 0.3382,
+      "step": 6160
+    },
+    {
+      "epoch": 1.0574121679520136,
+      "grad_norm": 0.47423774003982544,
+      "learning_rate": 3.237932019423022e-05,
+      "loss": 0.3439,
+      "step": 6170
+    },
+    {
+      "epoch": 1.0591259640102828,
+      "grad_norm": 0.4362853765487671,
+      "learning_rate": 3.23507569265924e-05,
+      "loss": 0.345,
+      "step": 6180
+    },
+    {
+      "epoch": 1.0608397600685517,
+      "grad_norm": 0.43055716156959534,
+      "learning_rate": 3.232219365895459e-05,
+      "loss": 0.3622,
+      "step": 6190
+    },
+    {
+      "epoch": 1.062553556126821,
+      "grad_norm": 0.4222048819065094,
+      "learning_rate": 3.229363039131677e-05,
+      "loss": 0.353,
+      "step": 6200
+    },
+    {
+      "epoch": 1.06426735218509,
+      "grad_norm": 0.42239975929260254,
+      "learning_rate": 3.226506712367895e-05,
+      "loss": 0.3646,
+      "step": 6210
+    },
+    {
+      "epoch": 1.065981148243359,
+      "grad_norm": 0.43706634640693665,
+      "learning_rate": 3.223650385604113e-05,
+      "loss": 0.3551,
+      "step": 6220
+    },
+    {
+      "epoch": 1.0676949443016281,
+      "grad_norm": 0.4561137557029724,
+      "learning_rate": 3.220794058840331e-05,
+      "loss": 0.3494,
+      "step": 6230
+    },
+    {
+      "epoch": 1.069408740359897,
+      "grad_norm": 0.4455920159816742,
+      "learning_rate": 3.21793773207655e-05,
+      "loss": 0.3396,
+      "step": 6240
+    },
+    {
+      "epoch": 1.0711225364181662,
+      "grad_norm": 0.4367483854293823,
+      "learning_rate": 3.2150814053127684e-05,
+      "loss": 0.3551,
+      "step": 6250
+    },
+    {
+      "epoch": 1.0728363324764354,
+      "grad_norm": 0.40412530303001404,
+      "learning_rate": 3.212225078548986e-05,
+      "loss": 0.3404,
+      "step": 6260
+    },
+    {
+      "epoch": 1.0745501285347043,
+      "grad_norm": 0.4885301887989044,
+      "learning_rate": 3.209368751785204e-05,
+      "loss": 0.3493,
+      "step": 6270
+    },
+    {
+      "epoch": 1.0762639245929735,
+      "grad_norm": 0.4218777120113373,
+      "learning_rate": 3.206512425021422e-05,
+      "loss": 0.371,
+      "step": 6280
+    },
+    {
+      "epoch": 1.0779777206512424,
+      "grad_norm": 0.423092782497406,
+      "learning_rate": 3.203656098257641e-05,
+      "loss": 0.3632,
+      "step": 6290
+    },
+    {
+      "epoch": 1.0796915167095116,
+      "grad_norm": 0.4192599654197693,
+      "learning_rate": 3.2007997714938594e-05,
+      "loss": 0.3495,
+      "step": 6300
+    },
+    {
+      "epoch": 1.0814053127677807,
+      "grad_norm": 0.4132305085659027,
+      "learning_rate": 3.197943444730077e-05,
+      "loss": 0.3623,
+      "step": 6310
+    },
+    {
+      "epoch": 1.0831191088260497,
+      "grad_norm": 0.4898054897785187,
+      "learning_rate": 3.195087117966295e-05,
+      "loss": 0.3417,
+      "step": 6320
+    },
+    {
+      "epoch": 1.0848329048843188,
+      "grad_norm": 0.4401434063911438,
+      "learning_rate": 3.192230791202513e-05,
+      "loss": 0.361,
+      "step": 6330
+    },
+    {
+      "epoch": 1.0865467009425878,
+      "grad_norm": 0.41742950677871704,
+      "learning_rate": 3.189374464438732e-05,
+      "loss": 0.3634,
+      "step": 6340
+    },
+    {
+      "epoch": 1.088260497000857,
+      "grad_norm": 0.3645254075527191,
+      "learning_rate": 3.1865181376749504e-05,
+      "loss": 0.3335,
+      "step": 6350
+    },
+    {
+      "epoch": 1.089974293059126,
+      "grad_norm": 0.3742375671863556,
+      "learning_rate": 3.183661810911168e-05,
+      "loss": 0.3526,
+      "step": 6360
+    },
+    {
+      "epoch": 1.091688089117395,
+      "grad_norm": 0.42631155252456665,
+      "learning_rate": 3.180805484147387e-05,
+      "loss": 0.3387,
+      "step": 6370
+    },
+    {
+      "epoch": 1.0934018851756642,
+      "grad_norm": 0.41597044467926025,
+      "learning_rate": 3.177949157383605e-05,
+      "loss": 0.3566,
+      "step": 6380
+    },
+    {
+      "epoch": 1.095115681233933,
+      "grad_norm": 0.41658124327659607,
+      "learning_rate": 3.175092830619823e-05,
+      "loss": 0.3534,
+      "step": 6390
+    },
+    {
+      "epoch": 1.0968294772922023,
+      "grad_norm": 0.4091750681400299,
+      "learning_rate": 3.172236503856041e-05,
+      "loss": 0.3526,
+      "step": 6400
+    },
+    {
+      "epoch": 1.0985432733504712,
+      "grad_norm": 0.4087865948677063,
+      "learning_rate": 3.169380177092259e-05,
+      "loss": 0.3258,
+      "step": 6410
+    },
+    {
+      "epoch": 1.1002570694087404,
+      "grad_norm": 0.4153346121311188,
+      "learning_rate": 3.166523850328478e-05,
+      "loss": 0.3547,
+      "step": 6420
+    },
+    {
+      "epoch": 1.1019708654670095,
+      "grad_norm": 0.43321630358695984,
+      "learning_rate": 3.1636675235646965e-05,
+      "loss": 0.3489,
+      "step": 6430
+    },
+    {
+      "epoch": 1.1036846615252784,
+      "grad_norm": 0.42528247833251953,
+      "learning_rate": 3.1608111968009144e-05,
+      "loss": 0.3463,
+      "step": 6440
+    },
+    {
+      "epoch": 1.1053984575835476,
+      "grad_norm": 0.3978523015975952,
+      "learning_rate": 3.157954870037132e-05,
+      "loss": 0.3506,
+      "step": 6450
+    },
+    {
+      "epoch": 1.1071122536418165,
+      "grad_norm": 0.48159298300743103,
+      "learning_rate": 3.15509854327335e-05,
+      "loss": 0.3726,
+      "step": 6460
+    },
+    {
+      "epoch": 1.1088260497000857,
+      "grad_norm": 0.4103160798549652,
+      "learning_rate": 3.152242216509569e-05,
+      "loss": 0.3409,
+      "step": 6470
+    },
+    {
+      "epoch": 1.1105398457583548,
+      "grad_norm": 0.496094673871994,
+      "learning_rate": 3.1493858897457874e-05,
+      "loss": 0.3455,
+      "step": 6480
+    },
+    {
+      "epoch": 1.1122536418166238,
+      "grad_norm": 0.4564761519432068,
+      "learning_rate": 3.1465295629820053e-05,
+      "loss": 0.3513,
+      "step": 6490
+    },
+    {
+      "epoch": 1.113967437874893,
+      "grad_norm": 0.4227219223976135,
+      "learning_rate": 3.143673236218224e-05,
+      "loss": 0.3581,
+      "step": 6500
+    },
+    {
+      "epoch": 1.1156812339331619,
+      "grad_norm": 0.42185524106025696,
+      "learning_rate": 3.140816909454441e-05,
+      "loss": 0.3514,
+      "step": 6510
+    },
+    {
+      "epoch": 1.117395029991431,
+      "grad_norm": 0.4504411816596985,
+      "learning_rate": 3.13796058269066e-05,
+      "loss": 0.3564,
+      "step": 6520
+    },
+    {
+      "epoch": 1.1191088260497002,
+      "grad_norm": 0.4239251911640167,
+      "learning_rate": 3.1351042559268784e-05,
+      "loss": 0.3483,
+      "step": 6530
+    },
+    {
+      "epoch": 1.1208226221079691,
+      "grad_norm": 0.42364707589149475,
+      "learning_rate": 3.132247929163096e-05,
+      "loss": 0.3585,
+      "step": 6540
+    },
+    {
+      "epoch": 1.1225364181662383,
+      "grad_norm": 0.40176278352737427,
+      "learning_rate": 3.129391602399315e-05,
+      "loss": 0.3432,
+      "step": 6550
+    },
+    {
+      "epoch": 1.1242502142245072,
+      "grad_norm": 0.42972254753112793,
+      "learning_rate": 3.126535275635533e-05,
+      "loss": 0.3632,
+      "step": 6560
+    },
+    {
+      "epoch": 1.1259640102827764,
+      "grad_norm": 0.4208223223686218,
+      "learning_rate": 3.123678948871751e-05,
+      "loss": 0.3487,
+      "step": 6570
+    },
+    {
+      "epoch": 1.1276778063410453,
+      "grad_norm": 0.41982948780059814,
+      "learning_rate": 3.1208226221079694e-05,
+      "loss": 0.3448,
+      "step": 6580
+    },
+    {
+      "epoch": 1.1293916023993145,
+      "grad_norm": 0.3786119222640991,
+      "learning_rate": 3.117966295344187e-05,
+      "loss": 0.3532,
+      "step": 6590
+    },
+    {
+      "epoch": 1.1311053984575836,
+      "grad_norm": 0.39533334970474243,
+      "learning_rate": 3.115109968580406e-05,
+      "loss": 0.3499,
+      "step": 6600
+    },
+    {
+      "epoch": 1.1328191945158526,
+      "grad_norm": 0.4255177676677704,
+      "learning_rate": 3.112253641816624e-05,
+      "loss": 0.3401,
+      "step": 6610
+    },
+    {
+      "epoch": 1.1345329905741217,
+      "grad_norm": 0.4227048456668854,
+      "learning_rate": 3.1093973150528424e-05,
+      "loss": 0.3589,
+      "step": 6620
+    },
+    {
+      "epoch": 1.1362467866323906,
+      "grad_norm": 0.4483916461467743,
+      "learning_rate": 3.1065409882890603e-05,
+      "loss": 0.3547,
+      "step": 6630
+    },
+    {
+      "epoch": 1.1379605826906598,
+      "grad_norm": 0.4231320917606354,
+      "learning_rate": 3.103684661525278e-05,
+      "loss": 0.3588,
+      "step": 6640
+    },
+    {
+      "epoch": 1.139674378748929,
+      "grad_norm": 0.4227351248264313,
+      "learning_rate": 3.100828334761497e-05,
+      "loss": 0.3455,
+      "step": 6650
+    },
+    {
+      "epoch": 1.141388174807198,
+      "grad_norm": 0.4091714322566986,
+      "learning_rate": 3.0979720079977155e-05,
+      "loss": 0.3476,
+      "step": 6660
+    },
+    {
+      "epoch": 1.143101970865467,
+      "grad_norm": 0.4448917508125305,
+      "learning_rate": 3.0951156812339334e-05,
+      "loss": 0.3498,
+      "step": 6670
+    },
+    {
+      "epoch": 1.144815766923736,
+      "grad_norm": 0.39556825160980225,
+      "learning_rate": 3.092259354470152e-05,
+      "loss": 0.3586,
+      "step": 6680
+    },
+    {
+      "epoch": 1.1465295629820051,
+      "grad_norm": 0.44014015793800354,
+      "learning_rate": 3.089403027706369e-05,
+      "loss": 0.3523,
+      "step": 6690
+    },
+    {
+      "epoch": 1.1482433590402743,
+      "grad_norm": 0.4422236382961273,
+      "learning_rate": 3.086546700942588e-05,
+      "loss": 0.3433,
+      "step": 6700
+    },
+    {
+      "epoch": 1.1499571550985432,
+      "grad_norm": 0.4930676519870758,
+      "learning_rate": 3.0836903741788064e-05,
+      "loss": 0.3565,
+      "step": 6710
+    },
+    {
+      "epoch": 1.1516709511568124,
+      "grad_norm": 0.4265972077846527,
+      "learning_rate": 3.0808340474150244e-05,
+      "loss": 0.3353,
+      "step": 6720
+    },
+    {
+      "epoch": 1.1533847472150813,
+      "grad_norm": 0.4512198865413666,
+      "learning_rate": 3.077977720651243e-05,
+      "loss": 0.3382,
+      "step": 6730
+    },
+    {
+      "epoch": 1.1550985432733505,
+      "grad_norm": 0.435465931892395,
+      "learning_rate": 3.075121393887461e-05,
+      "loss": 0.3454,
+      "step": 6740
+    },
+    {
+      "epoch": 1.1568123393316196,
+      "grad_norm": 0.42844897508621216,
+      "learning_rate": 3.072265067123679e-05,
+      "loss": 0.3518,
+      "step": 6750
+    },
+    {
+      "epoch": 1.1585261353898886,
+      "grad_norm": 0.3897634446620941,
+      "learning_rate": 3.0694087403598974e-05,
+      "loss": 0.3495,
+      "step": 6760
+    },
+    {
+      "epoch": 1.1602399314481577,
+      "grad_norm": 0.46333223581314087,
+      "learning_rate": 3.0665524135961153e-05,
+      "loss": 0.3489,
+      "step": 6770
+    },
+    {
+      "epoch": 1.1619537275064267,
+      "grad_norm": 0.4132658541202545,
+      "learning_rate": 3.063696086832334e-05,
+      "loss": 0.3506,
+      "step": 6780
+    },
+    {
+      "epoch": 1.1636675235646958,
+      "grad_norm": 0.46667787432670593,
+      "learning_rate": 3.060839760068552e-05,
+      "loss": 0.3551,
+      "step": 6790
+    },
+    {
+      "epoch": 1.165381319622965,
+      "grad_norm": 0.3969384729862213,
+      "learning_rate": 3.0579834333047705e-05,
+      "loss": 0.3578,
+      "step": 6800
+    },
+    {
+      "epoch": 1.167095115681234,
+      "grad_norm": 0.39458054304122925,
+      "learning_rate": 3.0551271065409884e-05,
+      "loss": 0.3459,
+      "step": 6810
+    },
+    {
+      "epoch": 1.168808911739503,
+      "grad_norm": 0.40578460693359375,
+      "learning_rate": 3.052270779777206e-05,
+      "loss": 0.3339,
+      "step": 6820
+    },
+    {
+      "epoch": 1.170522707797772,
+      "grad_norm": 0.3973940908908844,
+      "learning_rate": 3.049414453013425e-05,
+      "loss": 0.3537,
+      "step": 6830
+    },
+    {
+      "epoch": 1.1722365038560412,
+      "grad_norm": 0.4285309314727783,
+      "learning_rate": 3.046558126249643e-05,
+      "loss": 0.3503,
+      "step": 6840
+    },
+    {
+      "epoch": 1.17395029991431,
+      "grad_norm": 0.40150052309036255,
+      "learning_rate": 3.0437017994858614e-05,
+      "loss": 0.3427,
+      "step": 6850
+    },
+    {
+      "epoch": 1.1756640959725793,
+      "grad_norm": 0.4046010971069336,
+      "learning_rate": 3.0408454727220797e-05,
+      "loss": 0.3417,
+      "step": 6860
+    },
+    {
+      "epoch": 1.1773778920308484,
+      "grad_norm": 0.42314770817756653,
+      "learning_rate": 3.0379891459582976e-05,
+      "loss": 0.3351,
+      "step": 6870
+    },
+    {
+      "epoch": 1.1790916880891174,
+      "grad_norm": 0.42427992820739746,
+      "learning_rate": 3.0351328191945162e-05,
+      "loss": 0.3499,
+      "step": 6880
+    },
+    {
+      "epoch": 1.1808054841473865,
+      "grad_norm": 0.42444947361946106,
+      "learning_rate": 3.0322764924307338e-05,
+      "loss": 0.348,
+      "step": 6890
+    },
+    {
+      "epoch": 1.1825192802056554,
+      "grad_norm": 0.41960152983665466,
+      "learning_rate": 3.0294201656669524e-05,
+      "loss": 0.3464,
+      "step": 6900
+    },
+    {
+      "epoch": 1.1842330762639246,
+      "grad_norm": 0.42055803537368774,
+      "learning_rate": 3.026563838903171e-05,
+      "loss": 0.3442,
+      "step": 6910
+    },
+    {
+      "epoch": 1.1859468723221935,
+      "grad_norm": 0.3928168714046478,
+      "learning_rate": 3.0237075121393886e-05,
+      "loss": 0.3548,
+      "step": 6920
+    },
+    {
+      "epoch": 1.1876606683804627,
+      "grad_norm": 0.5035582184791565,
+      "learning_rate": 3.0208511853756072e-05,
+      "loss": 0.3495,
+      "step": 6930
+    },
+    {
+      "epoch": 1.1893744644387318,
+      "grad_norm": 0.4152749180793762,
+      "learning_rate": 3.017994858611825e-05,
+      "loss": 0.3502,
+      "step": 6940
+    },
+    {
+      "epoch": 1.1910882604970008,
+      "grad_norm": 0.45122581720352173,
+      "learning_rate": 3.0151385318480434e-05,
+      "loss": 0.3553,
+      "step": 6950
+    },
+    {
+      "epoch": 1.19280205655527,
+      "grad_norm": 0.401976078748703,
+      "learning_rate": 3.012282205084262e-05,
+      "loss": 0.3457,
+      "step": 6960
+    },
+    {
+      "epoch": 1.1945158526135389,
+      "grad_norm": 0.44023826718330383,
+      "learning_rate": 3.00942587832048e-05,
+      "loss": 0.344,
+      "step": 6970
+    },
+    {
+      "epoch": 1.196229648671808,
+      "grad_norm": 0.4113614559173584,
+      "learning_rate": 3.0065695515566982e-05,
+      "loss": 0.3554,
+      "step": 6980
+    },
+    {
+      "epoch": 1.1979434447300772,
+      "grad_norm": 0.43389397859573364,
+      "learning_rate": 3.0037132247929168e-05,
+      "loss": 0.337,
+      "step": 6990
+    },
+    {
+      "epoch": 1.1996572407883461,
+      "grad_norm": 0.45042628049850464,
+      "learning_rate": 3.0008568980291347e-05,
+      "loss": 0.3447,
+      "step": 7000
+    },
+    {
+      "epoch": 1.2013710368466153,
+      "grad_norm": 0.4491696059703827,
+      "learning_rate": 2.998000571265353e-05,
+      "loss": 0.3471,
+      "step": 7010
+    },
+    {
+      "epoch": 1.2030848329048842,
+      "grad_norm": 0.41832613945007324,
+      "learning_rate": 2.995144244501571e-05,
+      "loss": 0.3537,
+      "step": 7020
+    },
+    {
+      "epoch": 1.2047986289631534,
+      "grad_norm": 0.4287663996219635,
+      "learning_rate": 2.9922879177377895e-05,
+      "loss": 0.3501,
+      "step": 7030
+    },
+    {
+      "epoch": 1.2065124250214225,
+      "grad_norm": 0.41054508090019226,
+      "learning_rate": 2.9894315909740077e-05,
+      "loss": 0.3476,
+      "step": 7040
+    },
+    {
+      "epoch": 1.2082262210796915,
+      "grad_norm": 0.4107612371444702,
+      "learning_rate": 2.9865752642102257e-05,
+      "loss": 0.3445,
+      "step": 7050
+    },
+    {
+      "epoch": 1.2099400171379606,
+      "grad_norm": 0.4279773235321045,
+      "learning_rate": 2.9837189374464443e-05,
+      "loss": 0.3731,
+      "step": 7060
+    },
+    {
+      "epoch": 1.2116538131962296,
+      "grad_norm": 0.5119055509567261,
+      "learning_rate": 2.9808626106826622e-05,
+      "loss": 0.3352,
+      "step": 7070
+    },
+    {
+      "epoch": 1.2133676092544987,
+      "grad_norm": 0.41452860832214355,
+      "learning_rate": 2.9780062839188805e-05,
+      "loss": 0.3418,
+      "step": 7080
+    },
+    {
+      "epoch": 1.2150814053127679,
+      "grad_norm": 0.42358824610710144,
+      "learning_rate": 2.975149957155099e-05,
+      "loss": 0.349,
+      "step": 7090
+    },
+    {
+      "epoch": 1.2167952013710368,
+      "grad_norm": 0.3853763937950134,
+      "learning_rate": 2.9722936303913166e-05,
+      "loss": 0.3491,
+      "step": 7100
+    },
+    {
+      "epoch": 1.218508997429306,
+      "grad_norm": 0.4241725206375122,
+      "learning_rate": 2.9694373036275352e-05,
+      "loss": 0.3396,
+      "step": 7110
+    },
+    {
+      "epoch": 1.220222793487575,
+      "grad_norm": 0.4051187336444855,
+      "learning_rate": 2.9665809768637532e-05,
+      "loss": 0.3477,
+      "step": 7120
+    },
+    {
+      "epoch": 1.221936589545844,
+      "grad_norm": 0.41006556153297424,
+      "learning_rate": 2.9637246500999714e-05,
+      "loss": 0.3374,
+      "step": 7130
+    },
+    {
+      "epoch": 1.2236503856041132,
+      "grad_norm": 0.4366990327835083,
+      "learning_rate": 2.96086832333619e-05,
+      "loss": 0.3537,
+      "step": 7140
+    },
+    {
+      "epoch": 1.2253641816623821,
+      "grad_norm": 0.3898869454860687,
+      "learning_rate": 2.958011996572408e-05,
+      "loss": 0.3332,
+      "step": 7150
+    },
+    {
+      "epoch": 1.2270779777206513,
+      "grad_norm": 0.4259560704231262,
+      "learning_rate": 2.9551556698086262e-05,
+      "loss": 0.3458,
+      "step": 7160
+    },
+    {
+      "epoch": 1.2287917737789202,
+      "grad_norm": 0.40676066279411316,
+      "learning_rate": 2.952299343044844e-05,
+      "loss": 0.3503,
+      "step": 7170
+    },
+    {
+      "epoch": 1.2305055698371894,
+      "grad_norm": 0.4602939486503601,
+      "learning_rate": 2.9494430162810627e-05,
+      "loss": 0.3307,
+      "step": 7180
+    },
+    {
+      "epoch": 1.2322193658954586,
+      "grad_norm": 0.4000133275985718,
+      "learning_rate": 2.946586689517281e-05,
+      "loss": 0.3699,
+      "step": 7190
+    },
+    {
+      "epoch": 1.2339331619537275,
+      "grad_norm": 0.4689088463783264,
+      "learning_rate": 2.943730362753499e-05,
+      "loss": 0.342,
+      "step": 7200
+    },
+    {
+      "epoch": 1.2356469580119966,
+      "grad_norm": 0.39080819487571716,
+      "learning_rate": 2.9408740359897175e-05,
+      "loss": 0.3325,
+      "step": 7210
+    },
+    {
+      "epoch": 1.2373607540702656,
+      "grad_norm": 0.42110803723335266,
+      "learning_rate": 2.9380177092259355e-05,
+      "loss": 0.3399,
+      "step": 7220
+    },
+    {
+      "epoch": 1.2390745501285347,
+      "grad_norm": 0.3990210294723511,
+      "learning_rate": 2.9351613824621537e-05,
+      "loss": 0.3479,
+      "step": 7230
+    },
+    {
+      "epoch": 1.240788346186804,
+      "grad_norm": 0.4055481553077698,
+      "learning_rate": 2.9323050556983723e-05,
+      "loss": 0.3596,
+      "step": 7240
+    },
+    {
+      "epoch": 1.2425021422450728,
+      "grad_norm": 0.46355926990509033,
+      "learning_rate": 2.9294487289345902e-05,
+      "loss": 0.3534,
+      "step": 7250
+    },
+    {
+      "epoch": 1.244215938303342,
+      "grad_norm": 0.3957843780517578,
+      "learning_rate": 2.9265924021708085e-05,
+      "loss": 0.3328,
+      "step": 7260
+    },
+    {
+      "epoch": 1.245929734361611,
+      "grad_norm": 0.38406893610954285,
+      "learning_rate": 2.9237360754070264e-05,
+      "loss": 0.3633,
+      "step": 7270
+    },
+    {
+      "epoch": 1.24764353041988,
+      "grad_norm": 0.4850548505783081,
+      "learning_rate": 2.920879748643245e-05,
+      "loss": 0.3695,
+      "step": 7280
+    },
+    {
+      "epoch": 1.249357326478149,
+      "grad_norm": 0.43643587827682495,
+      "learning_rate": 2.9180234218794633e-05,
+      "loss": 0.3476,
+      "step": 7290
+    },
+    {
+      "epoch": 1.2510711225364182,
+      "grad_norm": 0.429887056350708,
+      "learning_rate": 2.9151670951156812e-05,
+      "loss": 0.3645,
+      "step": 7300
+    },
+    {
+      "epoch": 1.252784918594687,
+      "grad_norm": 0.4277980625629425,
+      "learning_rate": 2.9123107683518995e-05,
+      "loss": 0.333,
+      "step": 7310
+    },
+    {
+      "epoch": 1.2544987146529563,
+      "grad_norm": 0.442796915769577,
+      "learning_rate": 2.909454441588118e-05,
+      "loss": 0.3382,
+      "step": 7320
+    },
+    {
+      "epoch": 1.2562125107112254,
+      "grad_norm": 0.4223655164241791,
+      "learning_rate": 2.906598114824336e-05,
+      "loss": 0.3503,
+      "step": 7330
+    },
+    {
+      "epoch": 1.2579263067694944,
+      "grad_norm": 0.38975775241851807,
+      "learning_rate": 2.9037417880605543e-05,
+      "loss": 0.3382,
+      "step": 7340
+    },
+    {
+      "epoch": 1.2596401028277635,
+      "grad_norm": 0.3538907468318939,
+      "learning_rate": 2.9008854612967722e-05,
+      "loss": 0.3502,
+      "step": 7350
+    },
+    {
+      "epoch": 1.2613538988860324,
+      "grad_norm": 0.4272812008857727,
+      "learning_rate": 2.8980291345329908e-05,
+      "loss": 0.3492,
+      "step": 7360
+    },
+    {
+      "epoch": 1.2630676949443016,
+      "grad_norm": 0.42171192169189453,
+      "learning_rate": 2.895172807769209e-05,
+      "loss": 0.3447,
+      "step": 7370
+    },
+    {
+      "epoch": 1.2647814910025708,
+      "grad_norm": 0.4208640456199646,
+      "learning_rate": 2.892316481005427e-05,
+      "loss": 0.3395,
+      "step": 7380
+    },
+    {
+      "epoch": 1.2664952870608397,
+      "grad_norm": 0.3956572711467743,
+      "learning_rate": 2.8894601542416456e-05,
+      "loss": 0.3273,
+      "step": 7390
+    },
+    {
+      "epoch": 1.2682090831191088,
+      "grad_norm": 0.3806025981903076,
+      "learning_rate": 2.8866038274778635e-05,
+      "loss": 0.3485,
+      "step": 7400
+    },
+    {
+      "epoch": 1.2699228791773778,
+      "grad_norm": 0.3972262442111969,
+      "learning_rate": 2.8837475007140818e-05,
+      "loss": 0.3521,
+      "step": 7410
+    },
+    {
+      "epoch": 1.271636675235647,
+      "grad_norm": 0.43498408794403076,
+      "learning_rate": 2.8808911739503004e-05,
+      "loss": 0.344,
+      "step": 7420
+    },
+    {
+      "epoch": 1.273350471293916,
+      "grad_norm": 0.46216341853141785,
+      "learning_rate": 2.8780348471865183e-05,
+      "loss": 0.3489,
+      "step": 7430
+    },
+    {
+      "epoch": 1.275064267352185,
+      "grad_norm": 0.5822446942329407,
+      "learning_rate": 2.8751785204227366e-05,
+      "loss": 0.3432,
+      "step": 7440
+    },
+    {
+      "epoch": 1.2767780634104542,
+      "grad_norm": 0.43710097670555115,
+      "learning_rate": 2.8723221936589545e-05,
+      "loss": 0.3449,
+      "step": 7450
+    },
+    {
+      "epoch": 1.2784918594687231,
+      "grad_norm": 0.5048580765724182,
+      "learning_rate": 2.869465866895173e-05,
+      "loss": 0.3395,
+      "step": 7460
+    },
+    {
+      "epoch": 1.2802056555269923,
+      "grad_norm": 0.3946685492992401,
+      "learning_rate": 2.8666095401313913e-05,
+      "loss": 0.3549,
+      "step": 7470
+    },
+    {
+      "epoch": 1.2819194515852614,
+      "grad_norm": 0.4065665006637573,
+      "learning_rate": 2.8637532133676093e-05,
+      "loss": 0.3169,
+      "step": 7480
+    },
+    {
+      "epoch": 1.2836332476435304,
+      "grad_norm": 0.40248727798461914,
+      "learning_rate": 2.860896886603828e-05,
+      "loss": 0.3255,
+      "step": 7490
+    },
+    {
+      "epoch": 1.2853470437017995,
+      "grad_norm": 0.370870977640152,
+      "learning_rate": 2.8580405598400454e-05,
+      "loss": 0.3589,
+      "step": 7500
+    },
+    {
+      "epoch": 1.2870608397600685,
+      "grad_norm": 0.4085274338722229,
+      "learning_rate": 2.855184233076264e-05,
+      "loss": 0.343,
+      "step": 7510
+    },
+    {
+      "epoch": 1.2887746358183376,
+      "grad_norm": 0.47734397649765015,
+      "learning_rate": 2.8523279063124823e-05,
+      "loss": 0.3481,
+      "step": 7520
+    },
+    {
+      "epoch": 1.2904884318766068,
+      "grad_norm": 0.4094698429107666,
+      "learning_rate": 2.8494715795487002e-05,
+      "loss": 0.3519,
+      "step": 7530
+    },
+    {
+      "epoch": 1.2922022279348757,
+      "grad_norm": 0.3843911290168762,
+      "learning_rate": 2.846615252784919e-05,
+      "loss": 0.3539,
+      "step": 7540
+    },
+    {
+      "epoch": 1.2939160239931449,
+      "grad_norm": 0.44203248620033264,
+      "learning_rate": 2.8437589260211368e-05,
+      "loss": 0.3478,
+      "step": 7550
+    },
+    {
+      "epoch": 1.2956298200514138,
+      "grad_norm": 0.37155306339263916,
+      "learning_rate": 2.840902599257355e-05,
+      "loss": 0.331,
+      "step": 7560
+    },
+    {
+      "epoch": 1.297343616109683,
+      "grad_norm": 0.3852187395095825,
+      "learning_rate": 2.8380462724935736e-05,
+      "loss": 0.3325,
+      "step": 7570
+    },
+    {
+      "epoch": 1.2990574121679521,
+      "grad_norm": 0.391021192073822,
+      "learning_rate": 2.8351899457297915e-05,
+      "loss": 0.3501,
+      "step": 7580
+    },
+    {
+      "epoch": 1.300771208226221,
+      "grad_norm": 0.43556347489356995,
+      "learning_rate": 2.8323336189660098e-05,
+      "loss": 0.3458,
+      "step": 7590
+    },
+    {
+      "epoch": 1.3024850042844902,
+      "grad_norm": 0.4070069491863251,
+      "learning_rate": 2.8294772922022284e-05,
+      "loss": 0.3507,
+      "step": 7600
+    },
+    {
+      "epoch": 1.3041988003427591,
+      "grad_norm": 0.4203561842441559,
+      "learning_rate": 2.8266209654384463e-05,
+      "loss": 0.3516,
+      "step": 7610
+    },
+    {
+      "epoch": 1.3059125964010283,
+      "grad_norm": 0.4148425757884979,
+      "learning_rate": 2.8237646386746646e-05,
+      "loss": 0.3302,
+      "step": 7620
+    },
+    {
+      "epoch": 1.3076263924592975,
+      "grad_norm": 0.4449089467525482,
+      "learning_rate": 2.8209083119108825e-05,
+      "loss": 0.3508,
+      "step": 7630
+    },
+    {
+      "epoch": 1.3093401885175664,
+      "grad_norm": 0.4570872187614441,
+      "learning_rate": 2.818051985147101e-05,
+      "loss": 0.35,
+      "step": 7640
+    },
+    {
+      "epoch": 1.3110539845758356,
+      "grad_norm": 0.38971251249313354,
+      "learning_rate": 2.8151956583833194e-05,
+      "loss": 0.3423,
+      "step": 7650
+    },
+    {
+      "epoch": 1.3127677806341045,
+      "grad_norm": 0.3867109715938568,
+      "learning_rate": 2.8123393316195373e-05,
+      "loss": 0.3389,
+      "step": 7660
+    },
+    {
+      "epoch": 1.3144815766923736,
+      "grad_norm": 0.44314730167388916,
+      "learning_rate": 2.809483004855756e-05,
+      "loss": 0.3447,
+      "step": 7670
+    },
+    {
+      "epoch": 1.3161953727506428,
+      "grad_norm": 0.3955530822277069,
+      "learning_rate": 2.8066266780919735e-05,
+      "loss": 0.3385,
+      "step": 7680
+    },
+    {
+      "epoch": 1.3179091688089117,
+      "grad_norm": 0.44573432207107544,
+      "learning_rate": 2.803770351328192e-05,
+      "loss": 0.3407,
+      "step": 7690
+    },
+    {
+      "epoch": 1.3196229648671807,
+      "grad_norm": 0.38774076104164124,
+      "learning_rate": 2.8009140245644107e-05,
+      "loss": 0.3463,
+      "step": 7700
+    },
+    {
+      "epoch": 1.3213367609254498,
+      "grad_norm": 0.44221237301826477,
+      "learning_rate": 2.7980576978006283e-05,
+      "loss": 0.3454,
+      "step": 7710
+    },
+    {
+      "epoch": 1.323050556983719,
+      "grad_norm": 0.38349688053131104,
+      "learning_rate": 2.7954870037132248e-05,
+      "loss": 0.349,
+      "step": 7720
+    },
+    {
+      "epoch": 1.3247643530419881,
+      "grad_norm": 0.6474334597587585,
+      "learning_rate": 2.792630676949443e-05,
+      "loss": 0.3639,
+      "step": 7730
+    },
+    {
+      "epoch": 1.326478149100257,
+      "grad_norm": 0.39967530965805054,
+      "learning_rate": 2.7897743501856616e-05,
+      "loss": 0.3481,
+      "step": 7740
+    },
+    {
+      "epoch": 1.328191945158526,
+      "grad_norm": 0.463186115026474,
+      "learning_rate": 2.7869180234218795e-05,
+      "loss": 0.351,
+      "step": 7750
+    },
+    {
+      "epoch": 1.3299057412167952,
+      "grad_norm": 0.4431716799736023,
+      "learning_rate": 2.7840616966580978e-05,
+      "loss": 0.3315,
+      "step": 7760
+    },
+    {
+      "epoch": 1.3316195372750643,
+      "grad_norm": 0.42048847675323486,
+      "learning_rate": 2.7812053698943157e-05,
+      "loss": 0.3426,
+      "step": 7770
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 0.41159260272979736,
+      "learning_rate": 2.7783490431305343e-05,
+      "loss": 0.3411,
+      "step": 7780
+    },
+    {
+      "epoch": 1.3350471293916024,
+      "grad_norm": 0.42812713980674744,
+      "learning_rate": 2.7754927163667526e-05,
+      "loss": 0.3561,
+      "step": 7790
+    },
+    {
+      "epoch": 1.3367609254498714,
+      "grad_norm": 0.40041008591651917,
+      "learning_rate": 2.7726363896029705e-05,
+      "loss": 0.3348,
+      "step": 7800
+    },
+    {
+      "epoch": 1.3384747215081405,
+      "grad_norm": 0.4039687514305115,
+      "learning_rate": 2.769780062839189e-05,
+      "loss": 0.3519,
+      "step": 7810
+    },
+    {
+      "epoch": 1.3401885175664097,
+      "grad_norm": 0.38299936056137085,
+      "learning_rate": 2.766923736075407e-05,
+      "loss": 0.3485,
+      "step": 7820
+    },
+    {
+      "epoch": 1.3419023136246786,
+      "grad_norm": 0.47175827622413635,
+      "learning_rate": 2.7640674093116253e-05,
+      "loss": 0.3628,
+      "step": 7830
+    },
+    {
+      "epoch": 1.3436161096829478,
+      "grad_norm": 0.3974986970424652,
+      "learning_rate": 2.761211082547844e-05,
+      "loss": 0.3462,
+      "step": 7840
+    },
+    {
+      "epoch": 1.3453299057412167,
+      "grad_norm": 0.4005205035209656,
+      "learning_rate": 2.7583547557840618e-05,
+      "loss": 0.3528,
+      "step": 7850
+    },
+    {
+      "epoch": 1.3470437017994858,
+      "grad_norm": 0.46165525913238525,
+      "learning_rate": 2.75549842902028e-05,
+      "loss": 0.3423,
+      "step": 7860
+    },
+    {
+      "epoch": 1.348757497857755,
+      "grad_norm": 0.4195481240749359,
+      "learning_rate": 2.7526421022564987e-05,
+      "loss": 0.346,
+      "step": 7870
+    },
+    {
+      "epoch": 1.350471293916024,
+      "grad_norm": 0.41186022758483887,
+      "learning_rate": 2.7497857754927163e-05,
+      "loss": 0.3587,
+      "step": 7880
+    },
+    {
+      "epoch": 1.352185089974293,
+      "grad_norm": 0.39691275358200073,
+      "learning_rate": 2.746929448728935e-05,
+      "loss": 0.3292,
+      "step": 7890
+    },
+    {
+      "epoch": 1.353898886032562,
+      "grad_norm": 0.43522560596466064,
+      "learning_rate": 2.7440731219651528e-05,
+      "loss": 0.3455,
+      "step": 7900
+    },
+    {
+      "epoch": 1.3556126820908312,
+      "grad_norm": 0.41686180233955383,
+      "learning_rate": 2.741216795201371e-05,
+      "loss": 0.3319,
+      "step": 7910
+    },
+    {
+      "epoch": 1.3573264781491003,
+      "grad_norm": 0.4145779311656952,
+      "learning_rate": 2.7383604684375897e-05,
+      "loss": 0.3413,
+      "step": 7920
+    },
+    {
+      "epoch": 1.3590402742073693,
+      "grad_norm": 0.4023870825767517,
+      "learning_rate": 2.7355041416738076e-05,
+      "loss": 0.3451,
+      "step": 7930
+    },
+    {
+      "epoch": 1.3607540702656384,
+      "grad_norm": 0.3804696798324585,
+      "learning_rate": 2.732647814910026e-05,
+      "loss": 0.3454,
+      "step": 7940
+    },
+    {
+      "epoch": 1.3624678663239074,
+      "grad_norm": 0.41007333993911743,
+      "learning_rate": 2.7297914881462438e-05,
+      "loss": 0.327,
+      "step": 7950
+    },
+    {
+      "epoch": 1.3641816623821765,
+      "grad_norm": 0.37983033061027527,
+      "learning_rate": 2.7269351613824624e-05,
+      "loss": 0.3481,
+      "step": 7960
+    },
+    {
+      "epoch": 1.3658954584404457,
+      "grad_norm": 0.4238635301589966,
+      "learning_rate": 2.7240788346186806e-05,
+      "loss": 0.3325,
+      "step": 7970
+    },
+    {
+      "epoch": 1.3676092544987146,
+      "grad_norm": 0.40278929471969604,
+      "learning_rate": 2.7212225078548986e-05,
+      "loss": 0.3391,
+      "step": 7980
+    },
+    {
+      "epoch": 1.3693230505569838,
+      "grad_norm": 0.42544665932655334,
+      "learning_rate": 2.718366181091117e-05,
+      "loss": 0.3436,
+      "step": 7990
+    },
+    {
+      "epoch": 1.3710368466152527,
+      "grad_norm": 0.3899822235107422,
+      "learning_rate": 2.715509854327335e-05,
+      "loss": 0.3608,
+      "step": 8000
+    },
+    {
+      "epoch": 1.3727506426735219,
+      "grad_norm": 0.40118059515953064,
+      "learning_rate": 2.7126535275635533e-05,
+      "loss": 0.3421,
+      "step": 8010
+    },
+    {
+      "epoch": 1.374464438731791,
+      "grad_norm": 0.3830666244029999,
+      "learning_rate": 2.709797200799772e-05,
+      "loss": 0.3455,
+      "step": 8020
+    },
+    {
+      "epoch": 1.37617823479006,
+      "grad_norm": 0.4546830356121063,
+      "learning_rate": 2.70694087403599e-05,
+      "loss": 0.3413,
+      "step": 8030
+    },
+    {
+      "epoch": 1.3778920308483291,
+      "grad_norm": 0.4319319725036621,
+      "learning_rate": 2.704084547272208e-05,
+      "loss": 0.3434,
+      "step": 8040
+    },
+    {
+      "epoch": 1.379605826906598,
+      "grad_norm": 0.4801386296749115,
+      "learning_rate": 2.701228220508426e-05,
+      "loss": 0.3475,
+      "step": 8050
+    },
+    {
+      "epoch": 1.3813196229648672,
+      "grad_norm": 0.4308033287525177,
+      "learning_rate": 2.6983718937446447e-05,
+      "loss": 0.3524,
+      "step": 8060
+    },
+    {
+      "epoch": 1.3830334190231364,
+      "grad_norm": 0.37144121527671814,
+      "learning_rate": 2.695515566980863e-05,
+      "loss": 0.3225,
+      "step": 8070
+    },
+    {
+      "epoch": 1.3847472150814053,
+      "grad_norm": 0.43441927433013916,
+      "learning_rate": 2.692659240217081e-05,
+      "loss": 0.3448,
+      "step": 8080
+    },
+    {
+      "epoch": 1.3864610111396745,
+      "grad_norm": 0.4089992046356201,
+      "learning_rate": 2.689802913453299e-05,
+      "loss": 0.3345,
+      "step": 8090
+    },
+    {
+      "epoch": 1.3881748071979434,
+      "grad_norm": 0.387023001909256,
+      "learning_rate": 2.686946586689517e-05,
+      "loss": 0.3544,
+      "step": 8100
+    },
+    {
+      "epoch": 1.3898886032562126,
+      "grad_norm": 0.4173690974712372,
+      "learning_rate": 2.6840902599257356e-05,
+      "loss": 0.3324,
+      "step": 8110
+    },
+    {
+      "epoch": 1.3916023993144817,
+      "grad_norm": 0.3853393793106079,
+      "learning_rate": 2.681233933161954e-05,
+      "loss": 0.3487,
+      "step": 8120
+    },
+    {
+      "epoch": 1.3933161953727506,
+      "grad_norm": 0.48058122396469116,
+      "learning_rate": 2.6783776063981718e-05,
+      "loss": 0.3375,
+      "step": 8130
+    },
+    {
+      "epoch": 1.3950299914310196,
+      "grad_norm": 0.43440964818000793,
+      "learning_rate": 2.6755212796343904e-05,
+      "loss": 0.3497,
+      "step": 8140
+    },
+    {
+      "epoch": 1.3967437874892887,
+      "grad_norm": 0.4103083312511444,
+      "learning_rate": 2.6726649528706083e-05,
+      "loss": 0.3348,
+      "step": 8150
+    },
+    {
+      "epoch": 1.398457583547558,
+      "grad_norm": 0.5796851515769958,
+      "learning_rate": 2.6698086261068266e-05,
+      "loss": 0.3518,
+      "step": 8160
+    },
+    {
+      "epoch": 1.4001713796058268,
+      "grad_norm": 0.4238094389438629,
+      "learning_rate": 2.6669522993430452e-05,
+      "loss": 0.3414,
+      "step": 8170
+    },
+    {
+      "epoch": 1.401885175664096,
+      "grad_norm": 0.4287961721420288,
+      "learning_rate": 2.664095972579263e-05,
+      "loss": 0.3369,
+      "step": 8180
+    },
+    {
+      "epoch": 1.403598971722365,
+      "grad_norm": 0.41233572363853455,
+      "learning_rate": 2.6612396458154814e-05,
+      "loss": 0.363,
+      "step": 8190
+    },
+    {
+      "epoch": 1.405312767780634,
+      "grad_norm": 0.4112406373023987,
+      "learning_rate": 2.6583833190517e-05,
+      "loss": 0.3507,
+      "step": 8200
+    },
+    {
+      "epoch": 1.4070265638389032,
+      "grad_norm": 0.3922879695892334,
+      "learning_rate": 2.655526992287918e-05,
+      "loss": 0.3444,
+      "step": 8210
+    },
+    {
+      "epoch": 1.4087403598971722,
+      "grad_norm": 0.41223299503326416,
+      "learning_rate": 2.6526706655241362e-05,
+      "loss": 0.3561,
+      "step": 8220
+    },
+    {
+      "epoch": 1.4104541559554413,
+      "grad_norm": 0.39642488956451416,
+      "learning_rate": 2.649814338760354e-05,
+      "loss": 0.3528,
+      "step": 8230
+    },
+    {
+      "epoch": 1.4121679520137103,
+      "grad_norm": 0.43318673968315125,
+      "learning_rate": 2.6469580119965727e-05,
+      "loss": 0.3368,
+      "step": 8240
+    },
+    {
+      "epoch": 1.4138817480719794,
+      "grad_norm": 0.4054166376590729,
+      "learning_rate": 2.644101685232791e-05,
+      "loss": 0.3388,
+      "step": 8250
+    },
+    {
+      "epoch": 1.4155955441302486,
+      "grad_norm": 0.42911553382873535,
+      "learning_rate": 2.641245358469009e-05,
+      "loss": 0.3404,
+      "step": 8260
+    },
+    {
+      "epoch": 1.4173093401885175,
+      "grad_norm": 0.4328351318836212,
+      "learning_rate": 2.6383890317052275e-05,
+      "loss": 0.3388,
+      "step": 8270
+    },
+    {
+      "epoch": 1.4190231362467867,
+      "grad_norm": 0.37425297498703003,
+      "learning_rate": 2.635532704941445e-05,
+      "loss": 0.3482,
+      "step": 8280
+    },
+    {
+      "epoch": 1.4207369323050556,
+      "grad_norm": 0.4178659915924072,
+      "learning_rate": 2.6326763781776637e-05,
+      "loss": 0.3375,
+      "step": 8290
+    },
+    {
+      "epoch": 1.4224507283633248,
+      "grad_norm": 0.4292641580104828,
+      "learning_rate": 2.629820051413882e-05,
+      "loss": 0.3342,
+      "step": 8300
+    },
+    {
+      "epoch": 1.424164524421594,
+      "grad_norm": 0.39173051714897156,
+      "learning_rate": 2.6269637246501e-05,
+      "loss": 0.3351,
+      "step": 8310
+    },
+    {
+      "epoch": 1.4258783204798628,
+      "grad_norm": 0.39436325430870056,
+      "learning_rate": 2.6241073978863185e-05,
+      "loss": 0.3413,
+      "step": 8320
+    },
+    {
+      "epoch": 1.427592116538132,
+      "grad_norm": 0.412300169467926,
+      "learning_rate": 2.6212510711225364e-05,
+      "loss": 0.3363,
+      "step": 8330
+    },
+    {
+      "epoch": 1.429305912596401,
+      "grad_norm": 0.45384445786476135,
+      "learning_rate": 2.6183947443587547e-05,
+      "loss": 0.3558,
+      "step": 8340
+    },
+    {
+      "epoch": 1.43101970865467,
+      "grad_norm": 0.3842832148075104,
+      "learning_rate": 2.6155384175949733e-05,
+      "loss": 0.339,
+      "step": 8350
+    },
+    {
+      "epoch": 1.4327335047129393,
+      "grad_norm": 0.4295053482055664,
+      "learning_rate": 2.6126820908311912e-05,
+      "loss": 0.3447,
+      "step": 8360
+    },
+    {
+      "epoch": 1.4344473007712082,
+      "grad_norm": 0.4409741163253784,
+      "learning_rate": 2.6098257640674094e-05,
+      "loss": 0.3509,
+      "step": 8370
+    },
+    {
+      "epoch": 1.4361610968294773,
+      "grad_norm": 0.41879725456237793,
+      "learning_rate": 2.6069694373036274e-05,
+      "loss": 0.3268,
+      "step": 8380
+    },
+    {
+      "epoch": 1.4378748928877463,
+      "grad_norm": 0.4443173110485077,
+      "learning_rate": 2.604113110539846e-05,
+      "loss": 0.333,
+      "step": 8390
+    },
+    {
+      "epoch": 1.4395886889460154,
+      "grad_norm": 0.5012508630752563,
+      "learning_rate": 2.6012567837760642e-05,
+      "loss": 0.3554,
+      "step": 8400
+    },
+    {
+      "epoch": 1.4413024850042846,
+      "grad_norm": 0.400483101606369,
+      "learning_rate": 2.598400457012282e-05,
+      "loss": 0.3405,
+      "step": 8410
+    },
+    {
+      "epoch": 1.4430162810625535,
+      "grad_norm": 0.4024072587490082,
+      "learning_rate": 2.5955441302485007e-05,
+      "loss": 0.343,
+      "step": 8420
+    },
+    {
+      "epoch": 1.4447300771208227,
+      "grad_norm": 0.3965204358100891,
+      "learning_rate": 2.5926878034847187e-05,
+      "loss": 0.3459,
+      "step": 8430
+    },
+    {
+      "epoch": 1.4464438731790916,
+      "grad_norm": 0.3904626965522766,
+      "learning_rate": 2.589831476720937e-05,
+      "loss": 0.3362,
+      "step": 8440
+    },
+    {
+      "epoch": 1.4481576692373608,
+      "grad_norm": 0.45727992057800293,
+      "learning_rate": 2.5869751499571555e-05,
+      "loss": 0.3298,
+      "step": 8450
+    },
+    {
+      "epoch": 1.44987146529563,
+      "grad_norm": 0.40625882148742676,
+      "learning_rate": 2.584118823193373e-05,
+      "loss": 0.3304,
+      "step": 8460
+    },
+    {
+      "epoch": 1.4515852613538989,
+      "grad_norm": 0.404329776763916,
+      "learning_rate": 2.5812624964295917e-05,
+      "loss": 0.3203,
+      "step": 8470
+    },
+    {
+      "epoch": 1.453299057412168,
+      "grad_norm": 0.377429723739624,
+      "learning_rate": 2.5784061696658103e-05,
+      "loss": 0.3377,
+      "step": 8480
+    },
+    {
+      "epoch": 1.455012853470437,
+      "grad_norm": 0.41216394305229187,
+      "learning_rate": 2.575549842902028e-05,
+      "loss": 0.3363,
+      "step": 8490
+    },
+    {
+      "epoch": 1.4567266495287061,
+      "grad_norm": 0.4194747805595398,
+      "learning_rate": 2.5726935161382465e-05,
+      "loss": 0.3286,
+      "step": 8500
+    },
+    {
+      "epoch": 1.4584404455869753,
+      "grad_norm": 0.4310661554336548,
+      "learning_rate": 2.5698371893744644e-05,
+      "loss": 0.3413,
+      "step": 8510
+    },
+    {
+      "epoch": 1.4601542416452442,
+      "grad_norm": 0.40952980518341064,
+      "learning_rate": 2.5669808626106827e-05,
+      "loss": 0.3529,
+      "step": 8520
+    },
+    {
+      "epoch": 1.4618680377035131,
+      "grad_norm": 0.3957708477973938,
+      "learning_rate": 2.5641245358469013e-05,
+      "loss": 0.3491,
+      "step": 8530
+    },
+    {
+      "epoch": 1.4635818337617823,
+      "grad_norm": 0.3769155740737915,
+      "learning_rate": 2.5612682090831192e-05,
+      "loss": 0.35,
+      "step": 8540
+    },
+    {
+      "epoch": 1.4652956298200515,
+      "grad_norm": 0.3695950210094452,
+      "learning_rate": 2.5584118823193375e-05,
+      "loss": 0.3188,
+      "step": 8550
+    },
+    {
+      "epoch": 1.4670094258783206,
+      "grad_norm": 0.4204133152961731,
+      "learning_rate": 2.5555555555555554e-05,
+      "loss": 0.3462,
+      "step": 8560
+    },
+    {
+      "epoch": 1.4687232219365896,
+      "grad_norm": 0.4069593548774719,
+      "learning_rate": 2.552699228791774e-05,
+      "loss": 0.3431,
+      "step": 8570
+    },
+    {
+      "epoch": 1.4704370179948585,
+      "grad_norm": 0.44790759682655334,
+      "learning_rate": 2.5498429020279923e-05,
+      "loss": 0.3304,
+      "step": 8580
+    },
+    {
+      "epoch": 1.4721508140531276,
+      "grad_norm": 0.42790570855140686,
+      "learning_rate": 2.5469865752642102e-05,
+      "loss": 0.3479,
+      "step": 8590
+    },
+    {
+      "epoch": 1.4738646101113968,
+      "grad_norm": 0.40536096692085266,
+      "learning_rate": 2.5441302485004288e-05,
+      "loss": 0.3492,
+      "step": 8600
+    },
+    {
+      "epoch": 1.4755784061696657,
+      "grad_norm": 0.41925767064094543,
+      "learning_rate": 2.5412739217366467e-05,
+      "loss": 0.3494,
+      "step": 8610
+    },
+    {
+      "epoch": 1.477292202227935,
+      "grad_norm": 0.44192054867744446,
+      "learning_rate": 2.538417594972865e-05,
+      "loss": 0.342,
+      "step": 8620
+    },
+    {
+      "epoch": 1.4790059982862038,
+      "grad_norm": 0.38461121916770935,
+      "learning_rate": 2.5355612682090836e-05,
+      "loss": 0.3346,
+      "step": 8630
+    },
+    {
+      "epoch": 1.480719794344473,
+      "grad_norm": 0.38452473282814026,
+      "learning_rate": 2.5327049414453015e-05,
+      "loss": 0.3392,
+      "step": 8640
+    },
+    {
+      "epoch": 1.4824335904027421,
+      "grad_norm": 0.42153528332710266,
+      "learning_rate": 2.5298486146815198e-05,
+      "loss": 0.3325,
+      "step": 8650
+    },
+    {
+      "epoch": 1.484147386461011,
+      "grad_norm": 0.4100632965564728,
+      "learning_rate": 2.5269922879177377e-05,
+      "loss": 0.3456,
+      "step": 8660
+    },
+    {
+      "epoch": 1.4858611825192802,
+      "grad_norm": 0.44015470147132874,
+      "learning_rate": 2.524135961153956e-05,
+      "loss": 0.3405,
+      "step": 8670
+    },
+    {
+      "epoch": 1.4875749785775492,
+      "grad_norm": 0.3868488669395447,
+      "learning_rate": 2.5212796343901746e-05,
+      "loss": 0.3526,
+      "step": 8680
+    },
+    {
+      "epoch": 1.4892887746358183,
+      "grad_norm": 0.4163185656070709,
+      "learning_rate": 2.5184233076263925e-05,
+      "loss": 0.3439,
+      "step": 8690
+    },
+    {
+      "epoch": 1.4910025706940875,
+      "grad_norm": 0.3832201361656189,
+      "learning_rate": 2.5155669808626107e-05,
+      "loss": 0.3302,
+      "step": 8700
+    },
+    {
+      "epoch": 1.4927163667523564,
+      "grad_norm": 0.38806959986686707,
+      "learning_rate": 2.5127106540988287e-05,
+      "loss": 0.3383,
+      "step": 8710
+    },
+    {
+      "epoch": 1.4944301628106256,
+      "grad_norm": 0.3635774552822113,
+      "learning_rate": 2.5098543273350473e-05,
+      "loss": 0.3396,
+      "step": 8720
+    },
+    {
+      "epoch": 1.4961439588688945,
+      "grad_norm": 0.40076860785484314,
+      "learning_rate": 2.5069980005712655e-05,
+      "loss": 0.3407,
+      "step": 8730
+    },
+    {
+      "epoch": 1.4978577549271637,
+      "grad_norm": 0.419278621673584,
+      "learning_rate": 2.5041416738074835e-05,
+      "loss": 0.3395,
+      "step": 8740
+    },
+    {
+      "epoch": 1.4995715509854328,
+      "grad_norm": 0.4140276610851288,
+      "learning_rate": 2.501285347043702e-05,
+      "loss": 0.3302,
+      "step": 8750
+    },
+    {
+      "epoch": 1.5012853470437018,
+      "grad_norm": 0.47987890243530273,
+      "learning_rate": 2.4984290202799203e-05,
+      "loss": 0.342,
+      "step": 8760
+    },
+    {
+      "epoch": 1.502999143101971,
+      "grad_norm": 0.4513384997844696,
+      "learning_rate": 2.4955726935161382e-05,
+      "loss": 0.3374,
+      "step": 8770
+    },
+    {
+      "epoch": 1.5047129391602398,
+      "grad_norm": 0.44189688563346863,
+      "learning_rate": 2.4927163667523565e-05,
+      "loss": 0.3433,
+      "step": 8780
+    },
+    {
+      "epoch": 1.506426735218509,
+      "grad_norm": 0.3951323330402374,
+      "learning_rate": 2.4898600399885748e-05,
+      "loss": 0.3466,
+      "step": 8790
+    },
+    {
+      "epoch": 1.5081405312767782,
+      "grad_norm": 0.4419739842414856,
+      "learning_rate": 2.487003713224793e-05,
+      "loss": 0.3393,
+      "step": 8800
+    },
+    {
+      "epoch": 1.509854327335047,
+      "grad_norm": 0.3944937586784363,
+      "learning_rate": 2.4841473864610113e-05,
+      "loss": 0.3378,
+      "step": 8810
+    },
+    {
+      "epoch": 1.511568123393316,
+      "grad_norm": 0.4794803857803345,
+      "learning_rate": 2.4812910596972296e-05,
+      "loss": 0.3355,
+      "step": 8820
+    },
+    {
+      "epoch": 1.5132819194515852,
+      "grad_norm": 0.44282740354537964,
+      "learning_rate": 2.4784347329334475e-05,
+      "loss": 0.3376,
+      "step": 8830
+    },
+    {
+      "epoch": 1.5149957155098543,
+      "grad_norm": 0.4469907581806183,
+      "learning_rate": 2.475578406169666e-05,
+      "loss": 0.353,
+      "step": 8840
+    },
+    {
+      "epoch": 1.5167095115681235,
+      "grad_norm": 0.43816834688186646,
+      "learning_rate": 2.4727220794058843e-05,
+      "loss": 0.3479,
+      "step": 8850
+    },
+    {
+      "epoch": 1.5184233076263924,
+      "grad_norm": 0.36124739050865173,
+      "learning_rate": 2.4698657526421023e-05,
+      "loss": 0.3472,
+      "step": 8860
+    },
+    {
+      "epoch": 1.5201371036846614,
+      "grad_norm": 0.4093656539916992,
+      "learning_rate": 2.4670094258783205e-05,
+      "loss": 0.3358,
+      "step": 8870
+    },
+    {
+      "epoch": 1.5218508997429305,
+      "grad_norm": 0.4121876358985901,
+      "learning_rate": 2.4641530991145388e-05,
+      "loss": 0.3348,
+      "step": 8880
+    },
+    {
+      "epoch": 1.5235646958011997,
+      "grad_norm": 0.38930338621139526,
+      "learning_rate": 2.461296772350757e-05,
+      "loss": 0.358,
+      "step": 8890
+    },
+    {
+      "epoch": 1.5252784918594688,
+      "grad_norm": 0.3835488259792328,
+      "learning_rate": 2.4584404455869753e-05,
+      "loss": 0.3419,
+      "step": 8900
+    },
+    {
+      "epoch": 1.5269922879177378,
+      "grad_norm": 0.387956827878952,
+      "learning_rate": 2.4555841188231936e-05,
+      "loss": 0.3288,
+      "step": 8910
+    },
+    {
+      "epoch": 1.5287060839760067,
+      "grad_norm": 0.39550501108169556,
+      "learning_rate": 2.4527277920594115e-05,
+      "loss": 0.348,
+      "step": 8920
+    },
+    {
+      "epoch": 1.5304198800342759,
+      "grad_norm": 0.44232356548309326,
+      "learning_rate": 2.44987146529563e-05,
+      "loss": 0.3407,
+      "step": 8930
+    },
+    {
+      "epoch": 1.532133676092545,
+      "grad_norm": 0.41980692744255066,
+      "learning_rate": 2.4470151385318484e-05,
+      "loss": 0.3244,
+      "step": 8940
+    },
+    {
+      "epoch": 1.5338474721508142,
+      "grad_norm": 0.40649715065956116,
+      "learning_rate": 2.4441588117680663e-05,
+      "loss": 0.3367,
+      "step": 8950
+    },
+    {
+      "epoch": 1.5355612682090831,
+      "grad_norm": 0.41725024580955505,
+      "learning_rate": 2.4413024850042845e-05,
+      "loss": 0.3521,
+      "step": 8960
+    },
+    {
+      "epoch": 1.537275064267352,
+      "grad_norm": 0.45514604449272156,
+      "learning_rate": 2.4384461582405028e-05,
+      "loss": 0.3363,
+      "step": 8970
+    },
+    {
+      "epoch": 1.5389888603256212,
+      "grad_norm": 0.40341225266456604,
+      "learning_rate": 2.435589831476721e-05,
+      "loss": 0.3551,
+      "step": 8980
+    },
+    {
+      "epoch": 1.5407026563838904,
+      "grad_norm": 0.45389923453330994,
+      "learning_rate": 2.4327335047129393e-05,
+      "loss": 0.3458,
+      "step": 8990
+    },
+    {
+      "epoch": 1.5424164524421595,
+      "grad_norm": 0.4399547576904297,
+      "learning_rate": 2.4298771779491576e-05,
+      "loss": 0.3328,
+      "step": 9000
+    },
+    {
+      "epoch": 1.5441302485004285,
+      "grad_norm": 0.39244237542152405,
+      "learning_rate": 2.4270208511853755e-05,
+      "loss": 0.3318,
+      "step": 9010
+    },
+    {
+      "epoch": 1.5458440445586974,
+      "grad_norm": 0.46719881892204285,
+      "learning_rate": 2.4241645244215938e-05,
+      "loss": 0.325,
+      "step": 9020
+    },
+    {
+      "epoch": 1.5475578406169666,
+      "grad_norm": 0.4076116681098938,
+      "learning_rate": 2.4213081976578124e-05,
+      "loss": 0.3152,
+      "step": 9030
+    },
+    {
+      "epoch": 1.5492716366752357,
+      "grad_norm": 0.4262393116950989,
+      "learning_rate": 2.4184518708940303e-05,
+      "loss": 0.3273,
+      "step": 9040
+    },
+    {
+      "epoch": 1.5509854327335049,
+      "grad_norm": 0.415228933095932,
+      "learning_rate": 2.4155955441302486e-05,
+      "loss": 0.3377,
+      "step": 9050
+    },
+    {
+      "epoch": 1.5526992287917738,
+      "grad_norm": 0.43530064821243286,
+      "learning_rate": 2.412739217366467e-05,
+      "loss": 0.3534,
+      "step": 9060
+    },
+    {
+      "epoch": 1.5544130248500427,
+      "grad_norm": 0.4395502507686615,
+      "learning_rate": 2.409882890602685e-05,
+      "loss": 0.3339,
+      "step": 9070
+    },
+    {
+      "epoch": 1.556126820908312,
+      "grad_norm": 0.41547203063964844,
+      "learning_rate": 2.4070265638389034e-05,
+      "loss": 0.3324,
+      "step": 9080
+    },
+    {
+      "epoch": 1.557840616966581,
+      "grad_norm": 0.39110681414604187,
+      "learning_rate": 2.4041702370751216e-05,
+      "loss": 0.3448,
+      "step": 9090
+    },
+    {
+      "epoch": 1.5595544130248502,
+      "grad_norm": 0.46008628606796265,
+      "learning_rate": 2.4013139103113395e-05,
+      "loss": 0.3481,
+      "step": 9100
+    },
+    {
+      "epoch": 1.5612682090831191,
+      "grad_norm": 0.41481733322143555,
+      "learning_rate": 2.3984575835475578e-05,
+      "loss": 0.3323,
+      "step": 9110
+    },
+    {
+      "epoch": 1.562982005141388,
+      "grad_norm": 0.40251603722572327,
+      "learning_rate": 2.3956012567837764e-05,
+      "loss": 0.3313,
+      "step": 9120
+    },
+    {
+      "epoch": 1.5646958011996572,
+      "grad_norm": 0.37586021423339844,
+      "learning_rate": 2.3927449300199943e-05,
+      "loss": 0.3328,
+      "step": 9130
+    },
+    {
+      "epoch": 1.5664095972579264,
+      "grad_norm": 0.39882418513298035,
+      "learning_rate": 2.3898886032562126e-05,
+      "loss": 0.3431,
+      "step": 9140
+    },
+    {
+      "epoch": 1.5681233933161953,
+      "grad_norm": 0.393945574760437,
+      "learning_rate": 2.387032276492431e-05,
+      "loss": 0.3442,
+      "step": 9150
+    },
+    {
+      "epoch": 1.5698371893744645,
+      "grad_norm": 0.3542807698249817,
+      "learning_rate": 2.3841759497286488e-05,
+      "loss": 0.3249,
+      "step": 9160
+    },
+    {
+      "epoch": 1.5715509854327334,
+      "grad_norm": 0.4241867959499359,
+      "learning_rate": 2.3813196229648674e-05,
+      "loss": 0.3391,
+      "step": 9170
+    },
+    {
+      "epoch": 1.5732647814910026,
+      "grad_norm": 0.44771257042884827,
+      "learning_rate": 2.3784632962010856e-05,
+      "loss": 0.3346,
+      "step": 9180
+    },
+    {
+      "epoch": 1.5749785775492717,
+      "grad_norm": 0.39804545044898987,
+      "learning_rate": 2.3756069694373036e-05,
+      "loss": 0.3251,
+      "step": 9190
+    },
+    {
+      "epoch": 1.5766923736075407,
+      "grad_norm": 0.47698989510536194,
+      "learning_rate": 2.3727506426735218e-05,
+      "loss": 0.3327,
+      "step": 9200
+    },
+    {
+      "epoch": 1.5784061696658098,
+      "grad_norm": 0.43843311071395874,
+      "learning_rate": 2.3698943159097404e-05,
+      "loss": 0.3272,
+      "step": 9210
+    },
+    {
+      "epoch": 1.5801199657240788,
+      "grad_norm": 0.4098581373691559,
+      "learning_rate": 2.3670379891459584e-05,
+      "loss": 0.3346,
+      "step": 9220
+    },
+    {
+      "epoch": 1.581833761782348,
+      "grad_norm": 0.43909481167793274,
+      "learning_rate": 2.3641816623821766e-05,
+      "loss": 0.3249,
+      "step": 9230
+    },
+    {
+      "epoch": 1.583547557840617,
+      "grad_norm": 0.39052918553352356,
+      "learning_rate": 2.361325335618395e-05,
+      "loss": 0.3287,
+      "step": 9240
+    },
+    {
+      "epoch": 1.585261353898886,
+      "grad_norm": 0.4083861708641052,
+      "learning_rate": 2.3584690088546128e-05,
+      "loss": 0.3329,
+      "step": 9250
+    },
+    {
+      "epoch": 1.586975149957155,
+      "grad_norm": 0.41227346658706665,
+      "learning_rate": 2.3556126820908314e-05,
+      "loss": 0.3269,
+      "step": 9260
+    },
+    {
+      "epoch": 1.588688946015424,
+      "grad_norm": 0.4148152768611908,
+      "learning_rate": 2.3527563553270497e-05,
+      "loss": 0.322,
+      "step": 9270
+    },
+    {
+      "epoch": 1.5904027420736933,
+      "grad_norm": 0.41056588292121887,
+      "learning_rate": 2.3499000285632676e-05,
+      "loss": 0.3441,
+      "step": 9280
+    },
+    {
+      "epoch": 1.5921165381319624,
+      "grad_norm": 0.3809565603733063,
+      "learning_rate": 2.347043701799486e-05,
+      "loss": 0.3447,
+      "step": 9290
+    },
+    {
+      "epoch": 1.5938303341902313,
+      "grad_norm": 0.3820127844810486,
+      "learning_rate": 2.344187375035704e-05,
+      "loss": 0.3497,
+      "step": 9300
+    },
+    {
+      "epoch": 1.5955441302485003,
+      "grad_norm": 0.40932685136795044,
+      "learning_rate": 2.3413310482719224e-05,
+      "loss": 0.3725,
+      "step": 9310
+    },
+    {
+      "epoch": 1.5972579263067694,
+      "grad_norm": 0.40701672434806824,
+      "learning_rate": 2.3384747215081406e-05,
+      "loss": 0.3167,
+      "step": 9320
+    },
+    {
+      "epoch": 1.5989717223650386,
+      "grad_norm": 0.43671759963035583,
+      "learning_rate": 2.335618394744359e-05,
+      "loss": 0.3232,
+      "step": 9330
+    },
+    {
+      "epoch": 1.6006855184233078,
+      "grad_norm": 0.4282389283180237,
+      "learning_rate": 2.332762067980577e-05,
+      "loss": 0.3399,
+      "step": 9340
+    },
+    {
+      "epoch": 1.6023993144815767,
+      "grad_norm": 0.4553786814212799,
+      "learning_rate": 2.3299057412167954e-05,
+      "loss": 0.3387,
+      "step": 9350
+    },
+    {
+      "epoch": 1.6041131105398456,
+      "grad_norm": 0.40568017959594727,
+      "learning_rate": 2.3270494144530137e-05,
+      "loss": 0.345,
+      "step": 9360
+    },
+    {
+      "epoch": 1.6058269065981148,
+      "grad_norm": 0.4190521538257599,
+      "learning_rate": 2.3241930876892316e-05,
+      "loss": 0.3347,
+      "step": 9370
+    },
+    {
+      "epoch": 1.607540702656384,
+      "grad_norm": 0.4039881229400635,
+      "learning_rate": 2.32133676092545e-05,
+      "loss": 0.3252,
+      "step": 9380
+    },
+    {
+      "epoch": 1.609254498714653,
+      "grad_norm": 0.4178449511528015,
+      "learning_rate": 2.318480434161668e-05,
+      "loss": 0.333,
+      "step": 9390
+    },
+    {
+      "epoch": 1.610968294772922,
+      "grad_norm": 0.4503561854362488,
+      "learning_rate": 2.3156241073978864e-05,
+      "loss": 0.3282,
+      "step": 9400
+    },
+    {
+      "epoch": 1.612682090831191,
+      "grad_norm": 0.43098393082618713,
+      "learning_rate": 2.3127677806341047e-05,
+      "loss": 0.337,
+      "step": 9410
+    },
+    {
+      "epoch": 1.6143958868894601,
+      "grad_norm": 0.3916527330875397,
+      "learning_rate": 2.309911453870323e-05,
+      "loss": 0.331,
+      "step": 9420
+    },
+    {
+      "epoch": 1.6161096829477293,
+      "grad_norm": 0.43365949392318726,
+      "learning_rate": 2.3070551271065412e-05,
+      "loss": 0.3325,
+      "step": 9430
+    },
+    {
+      "epoch": 1.6178234790059984,
+      "grad_norm": 0.3826988935470581,
+      "learning_rate": 2.304198800342759e-05,
+      "loss": 0.3378,
+      "step": 9440
+    },
+    {
+      "epoch": 1.6195372750642674,
+      "grad_norm": 0.37719157338142395,
+      "learning_rate": 2.3013424735789777e-05,
+      "loss": 0.3308,
+      "step": 9450
+    },
+    {
+      "epoch": 1.6212510711225363,
+      "grad_norm": 0.39002153277397156,
+      "learning_rate": 2.298486146815196e-05,
+      "loss": 0.3458,
+      "step": 9460
+    },
+    {
+      "epoch": 1.6229648671808055,
+      "grad_norm": 0.4165000021457672,
+      "learning_rate": 2.295629820051414e-05,
+      "loss": 0.3293,
+      "step": 9470
+    },
+    {
+      "epoch": 1.6246786632390746,
+      "grad_norm": 0.41711005568504333,
+      "learning_rate": 2.292773493287632e-05,
+      "loss": 0.3422,
+      "step": 9480
+    },
+    {
+      "epoch": 1.6263924592973438,
+      "grad_norm": 0.3922407925128937,
+      "learning_rate": 2.2899171665238504e-05,
+      "loss": 0.3314,
+      "step": 9490
+    },
+    {
+      "epoch": 1.6281062553556127,
+      "grad_norm": 0.40335458517074585,
+      "learning_rate": 2.2870608397600687e-05,
+      "loss": 0.3454,
+      "step": 9500
+    },
+    {
+      "epoch": 1.6298200514138816,
+      "grad_norm": 0.40134212374687195,
+      "learning_rate": 2.284204512996287e-05,
+      "loss": 0.3302,
+      "step": 9510
+    },
+    {
+      "epoch": 1.6315338474721508,
+      "grad_norm": 0.4334363639354706,
+      "learning_rate": 2.2813481862325052e-05,
+      "loss": 0.3463,
+      "step": 9520
+    },
+    {
+      "epoch": 1.63324764353042,
+      "grad_norm": 0.4288492798805237,
+      "learning_rate": 2.278491859468723e-05,
+      "loss": 0.3492,
+      "step": 9530
+    },
+    {
+      "epoch": 1.634961439588689,
+      "grad_norm": 0.41050291061401367,
+      "learning_rate": 2.2756355327049417e-05,
+      "loss": 0.3277,
+      "step": 9540
+    },
+    {
+      "epoch": 1.636675235646958,
+      "grad_norm": 0.4153129458427429,
+      "learning_rate": 2.27277920594116e-05,
+      "loss": 0.3383,
+      "step": 9550
+    },
+    {
+      "epoch": 1.638389031705227,
+      "grad_norm": 0.4749390482902527,
+      "learning_rate": 2.269922879177378e-05,
+      "loss": 0.3357,
+      "step": 9560
+    },
+    {
+      "epoch": 1.6401028277634961,
+      "grad_norm": 0.3886820077896118,
+      "learning_rate": 2.2670665524135962e-05,
+      "loss": 0.3501,
+      "step": 9570
+    },
+    {
+      "epoch": 1.6418166238217653,
+      "grad_norm": 0.42018669843673706,
+      "learning_rate": 2.2642102256498144e-05,
+      "loss": 0.3492,
+      "step": 9580
+    },
+    {
+      "epoch": 1.6435304198800342,
+      "grad_norm": 0.3968152403831482,
+      "learning_rate": 2.2613538988860327e-05,
+      "loss": 0.3411,
+      "step": 9590
+    },
+    {
+      "epoch": 1.6452442159383034,
+      "grad_norm": 0.4629608392715454,
+      "learning_rate": 2.258497572122251e-05,
+      "loss": 0.3495,
+      "step": 9600
+    },
+    {
+      "epoch": 1.6469580119965723,
+      "grad_norm": 0.43643638491630554,
+      "learning_rate": 2.2556412453584692e-05,
+      "loss": 0.3406,
+      "step": 9610
+    },
+    {
+      "epoch": 1.6486718080548415,
+      "grad_norm": 0.40299829840660095,
+      "learning_rate": 2.252784918594687e-05,
+      "loss": 0.3194,
+      "step": 9620
+    },
+    {
+      "epoch": 1.6503856041131106,
+      "grad_norm": 0.39634135365486145,
+      "learning_rate": 2.2499285918309054e-05,
+      "loss": 0.3648,
+      "step": 9630
+    },
+    {
+      "epoch": 1.6520994001713796,
+      "grad_norm": 0.37745097279548645,
+      "learning_rate": 2.247072265067124e-05,
+      "loss": 0.3309,
+      "step": 9640
+    },
+    {
+      "epoch": 1.6538131962296485,
+      "grad_norm": 0.3941361606121063,
+      "learning_rate": 2.244215938303342e-05,
+      "loss": 0.3565,
+      "step": 9650
+    },
+    {
+      "epoch": 1.6555269922879177,
+      "grad_norm": 0.3889491856098175,
+      "learning_rate": 2.2413596115395602e-05,
+      "loss": 0.3213,
+      "step": 9660
+    },
+    {
+      "epoch": 1.6572407883461868,
+      "grad_norm": 0.40471869707107544,
+      "learning_rate": 2.2385032847757785e-05,
+      "loss": 0.3422,
+      "step": 9670
+    },
+    {
+      "epoch": 1.658954584404456,
+      "grad_norm": 0.4090235233306885,
+      "learning_rate": 2.2356469580119967e-05,
+      "loss": 0.3612,
+      "step": 9680
+    },
+    {
+      "epoch": 1.660668380462725,
+      "grad_norm": 0.42937588691711426,
+      "learning_rate": 2.232790631248215e-05,
+      "loss": 0.3259,
+      "step": 9690
+    },
+    {
+      "epoch": 1.6623821765209938,
+      "grad_norm": 0.4077993631362915,
+      "learning_rate": 2.2299343044844333e-05,
+      "loss": 0.3343,
+      "step": 9700
+    },
+    {
+      "epoch": 1.664095972579263,
+      "grad_norm": 0.4152628183364868,
+      "learning_rate": 2.2270779777206512e-05,
+      "loss": 0.3338,
+      "step": 9710
+    },
+    {
+      "epoch": 1.6658097686375322,
+      "grad_norm": 0.4114777147769928,
+      "learning_rate": 2.2242216509568694e-05,
+      "loss": 0.3418,
+      "step": 9720
+    },
+    {
+      "epoch": 1.6675235646958013,
+      "grad_norm": 0.46183767914772034,
+      "learning_rate": 2.221365324193088e-05,
+      "loss": 0.3493,
+      "step": 9730
+    },
+    {
+      "epoch": 1.6692373607540703,
+      "grad_norm": 0.4328729808330536,
+      "learning_rate": 2.218508997429306e-05,
+      "loss": 0.3473,
+      "step": 9740
+    },
+    {
+      "epoch": 1.6709511568123392,
+      "grad_norm": 0.42931821942329407,
+      "learning_rate": 2.2156526706655242e-05,
+      "loss": 0.3313,
+      "step": 9750
+    },
+    {
+      "epoch": 1.6726649528706083,
+      "grad_norm": 0.4109592139720917,
+      "learning_rate": 2.2127963439017425e-05,
+      "loss": 0.3453,
+      "step": 9760
+    },
+    {
+      "epoch": 1.6743787489288775,
+      "grad_norm": 0.37919366359710693,
+      "learning_rate": 2.2099400171379604e-05,
+      "loss": 0.3242,
+      "step": 9770
+    },
+    {
+      "epoch": 1.6760925449871467,
+      "grad_norm": 0.4012930989265442,
+      "learning_rate": 2.207083690374179e-05,
+      "loss": 0.3469,
+      "step": 9780
+    },
+    {
+      "epoch": 1.6778063410454156,
+      "grad_norm": 0.3762998878955841,
+      "learning_rate": 2.2042273636103973e-05,
+      "loss": 0.3517,
+      "step": 9790
+    },
+    {
+      "epoch": 1.6795201371036845,
+      "grad_norm": 0.44242531061172485,
+      "learning_rate": 2.2013710368466152e-05,
+      "loss": 0.3469,
+      "step": 9800
+    },
+    {
+      "epoch": 1.6812339331619537,
+      "grad_norm": 0.38373252749443054,
+      "learning_rate": 2.1985147100828335e-05,
+      "loss": 0.3306,
+      "step": 9810
+    },
+    {
+      "epoch": 1.6829477292202228,
+      "grad_norm": 0.399817556142807,
+      "learning_rate": 2.195658383319052e-05,
+      "loss": 0.3196,
+      "step": 9820
+    },
+    {
+      "epoch": 1.684661525278492,
+      "grad_norm": 0.4670737385749817,
+      "learning_rate": 2.19280205655527e-05,
+      "loss": 0.339,
+      "step": 9830
+    },
+    {
+      "epoch": 1.686375321336761,
+      "grad_norm": 0.4311138391494751,
+      "learning_rate": 2.1899457297914883e-05,
+      "loss": 0.3213,
+      "step": 9840
+    },
+    {
+      "epoch": 1.6880891173950299,
+      "grad_norm": 0.42144256830215454,
+      "learning_rate": 2.1870894030277065e-05,
+      "loss": 0.3275,
+      "step": 9850
+    },
+    {
+      "epoch": 1.689802913453299,
+      "grad_norm": 0.4103786051273346,
+      "learning_rate": 2.1842330762639244e-05,
+      "loss": 0.3373,
+      "step": 9860
+    },
+    {
+      "epoch": 1.6915167095115682,
+      "grad_norm": 0.41490599513053894,
+      "learning_rate": 2.181376749500143e-05,
+      "loss": 0.3235,
+      "step": 9870
+    },
+    {
+      "epoch": 1.6932305055698373,
+      "grad_norm": 0.43073931336402893,
+      "learning_rate": 2.1785204227363613e-05,
+      "loss": 0.325,
+      "step": 9880
+    },
+    {
+      "epoch": 1.6949443016281063,
+      "grad_norm": 0.4372043013572693,
+      "learning_rate": 2.1756640959725792e-05,
+      "loss": 0.3336,
+      "step": 9890
+    },
+    {
+      "epoch": 1.6966580976863752,
+      "grad_norm": 0.3718228340148926,
+      "learning_rate": 2.1728077692087975e-05,
+      "loss": 0.3287,
+      "step": 9900
+    },
+    {
+      "epoch": 1.6983718937446444,
+      "grad_norm": 0.41269469261169434,
+      "learning_rate": 2.1699514424450158e-05,
+      "loss": 0.3324,
+      "step": 9910
+    },
+    {
+      "epoch": 1.7000856898029135,
+      "grad_norm": 0.39504536986351013,
+      "learning_rate": 2.167095115681234e-05,
+      "loss": 0.352,
+      "step": 9920
+    },
+    {
+      "epoch": 1.7017994858611827,
+      "grad_norm": 0.520569920539856,
+      "learning_rate": 2.1642387889174523e-05,
+      "loss": 0.3444,
+      "step": 9930
+    },
+    {
+      "epoch": 1.7035132819194516,
+      "grad_norm": 0.44295620918273926,
+      "learning_rate": 2.1613824621536705e-05,
+      "loss": 0.3373,
+      "step": 9940
+    },
+    {
+      "epoch": 1.7052270779777206,
+      "grad_norm": 0.38542020320892334,
+      "learning_rate": 2.1585261353898885e-05,
+      "loss": 0.3251,
+      "step": 9950
+    },
+    {
+      "epoch": 1.7069408740359897,
+      "grad_norm": 0.4100322425365448,
+      "learning_rate": 2.155669808626107e-05,
+      "loss": 0.3445,
+      "step": 9960
+    },
+    {
+      "epoch": 1.7086546700942589,
+      "grad_norm": 0.37752169370651245,
+      "learning_rate": 2.1528134818623253e-05,
+      "loss": 0.3429,
+      "step": 9970
+    },
+    {
+      "epoch": 1.7103684661525278,
+      "grad_norm": 0.4206605553627014,
+      "learning_rate": 2.1499571550985432e-05,
+      "loss": 0.3295,
+      "step": 9980
+    },
+    {
+      "epoch": 1.712082262210797,
+      "grad_norm": 0.3830428421497345,
+      "learning_rate": 2.1471008283347615e-05,
+      "loss": 0.3298,
+      "step": 9990
+    },
+    {
+      "epoch": 1.713796058269066,
+      "grad_norm": 0.40393152832984924,
+      "learning_rate": 2.1442445015709798e-05,
+      "loss": 0.3349,
+      "step": 10000
+    },
+    {
+      "epoch": 1.715509854327335,
+      "grad_norm": 0.37100762128829956,
+      "learning_rate": 2.141388174807198e-05,
+      "loss": 0.3354,
+      "step": 10010
+    },
+    {
+      "epoch": 1.7172236503856042,
+      "grad_norm": 0.3826449513435364,
+      "learning_rate": 2.1385318480434163e-05,
+      "loss": 0.3452,
+      "step": 10020
+    },
+    {
+      "epoch": 1.7189374464438731,
+      "grad_norm": 0.39858880639076233,
+      "learning_rate": 2.1356755212796346e-05,
+      "loss": 0.3416,
+      "step": 10030
+    },
+    {
+      "epoch": 1.7206512425021423,
+      "grad_norm": 0.38912972807884216,
+      "learning_rate": 2.1328191945158528e-05,
+      "loss": 0.3357,
+      "step": 10040
+    },
+    {
+      "epoch": 1.7223650385604112,
+      "grad_norm": 0.39427903294563293,
+      "learning_rate": 2.1299628677520707e-05,
+      "loss": 0.3309,
+      "step": 10050
+    },
+    {
+      "epoch": 1.7240788346186804,
+      "grad_norm": 0.4788413941860199,
+      "learning_rate": 2.1271065409882893e-05,
+      "loss": 0.3395,
+      "step": 10060
+    },
+    {
+      "epoch": 1.7257926306769495,
+      "grad_norm": 0.36351656913757324,
+      "learning_rate": 2.1242502142245073e-05,
+      "loss": 0.3499,
+      "step": 10070
+    },
+    {
+      "epoch": 1.7275064267352185,
+      "grad_norm": 0.42046666145324707,
+      "learning_rate": 2.1213938874607255e-05,
+      "loss": 0.3257,
+      "step": 10080
+    },
+    {
+      "epoch": 1.7292202227934874,
+      "grad_norm": 0.41890949010849,
+      "learning_rate": 2.1185375606969438e-05,
+      "loss": 0.3186,
+      "step": 10090
+    },
+    {
+      "epoch": 1.7309340188517566,
+      "grad_norm": 0.3858489990234375,
+      "learning_rate": 2.115681233933162e-05,
+      "loss": 0.3455,
+      "step": 10100
+    },
+    {
+      "epoch": 1.7326478149100257,
+      "grad_norm": 0.41216158866882324,
+      "learning_rate": 2.1128249071693803e-05,
+      "loss": 0.3313,
+      "step": 10110
+    },
+    {
+      "epoch": 1.7343616109682949,
+      "grad_norm": 0.4424970746040344,
+      "learning_rate": 2.1099685804055986e-05,
+      "loss": 0.335,
+      "step": 10120
+    },
+    {
+      "epoch": 1.7360754070265638,
+      "grad_norm": 0.4221234619617462,
+      "learning_rate": 2.107112253641817e-05,
+      "loss": 0.3321,
+      "step": 10130
+    },
+    {
+      "epoch": 1.7377892030848328,
+      "grad_norm": 0.4379345178604126,
+      "learning_rate": 2.1042559268780348e-05,
+      "loss": 0.3264,
+      "step": 10140
+    },
+    {
+      "epoch": 1.739502999143102,
+      "grad_norm": 0.38541027903556824,
+      "learning_rate": 2.1013996001142534e-05,
+      "loss": 0.333,
+      "step": 10150
+    },
+    {
+      "epoch": 1.741216795201371,
+      "grad_norm": 0.43533679842948914,
+      "learning_rate": 2.0985432733504713e-05,
+      "loss": 0.349,
+      "step": 10160
+    },
+    {
+      "epoch": 1.7429305912596402,
+      "grad_norm": 0.4305630326271057,
+      "learning_rate": 2.0956869465866896e-05,
+      "loss": 0.331,
+      "step": 10170
+    },
+    {
+      "epoch": 1.7446443873179092,
+      "grad_norm": 0.4137646555900574,
+      "learning_rate": 2.0928306198229078e-05,
+      "loss": 0.3435,
+      "step": 10180
+    },
+    {
+      "epoch": 1.746358183376178,
+      "grad_norm": 0.391099214553833,
+      "learning_rate": 2.089974293059126e-05,
+      "loss": 0.325,
+      "step": 10190
+    },
+    {
+      "epoch": 1.7480719794344473,
+      "grad_norm": 0.39120209217071533,
+      "learning_rate": 2.0871179662953443e-05,
+      "loss": 0.3359,
+      "step": 10200
+    },
+    {
+      "epoch": 1.7497857754927164,
+      "grad_norm": 0.37801820039749146,
+      "learning_rate": 2.0842616395315626e-05,
+      "loss": 0.3211,
+      "step": 10210
+    },
+    {
+      "epoch": 1.7514995715509856,
+      "grad_norm": 0.39814886450767517,
+      "learning_rate": 2.081405312767781e-05,
+      "loss": 0.3234,
+      "step": 10220
+    },
+    {
+      "epoch": 1.7532133676092545,
+      "grad_norm": 0.39131447672843933,
+      "learning_rate": 2.0785489860039988e-05,
+      "loss": 0.3352,
+      "step": 10230
+    },
+    {
+      "epoch": 1.7549271636675234,
+      "grad_norm": 0.4148966372013092,
+      "learning_rate": 2.075692659240217e-05,
+      "loss": 0.3339,
+      "step": 10240
+    },
+    {
+      "epoch": 1.7566409597257926,
+      "grad_norm": 0.40360793471336365,
+      "learning_rate": 2.0728363324764357e-05,
+      "loss": 0.3372,
+      "step": 10250
+    },
+    {
+      "epoch": 1.7583547557840618,
+      "grad_norm": 0.4748077690601349,
+      "learning_rate": 2.0699800057126536e-05,
+      "loss": 0.3256,
+      "step": 10260
+    },
+    {
+      "epoch": 1.760068551842331,
+      "grad_norm": 0.3388690650463104,
+      "learning_rate": 2.067123678948872e-05,
+      "loss": 0.3532,
+      "step": 10270
+    },
+    {
+      "epoch": 1.7617823479005998,
+      "grad_norm": 0.4345872402191162,
+      "learning_rate": 2.06426735218509e-05,
+      "loss": 0.3431,
+      "step": 10280
+    },
+    {
+      "epoch": 1.7634961439588688,
+      "grad_norm": 0.4032730460166931,
+      "learning_rate": 2.0614110254213084e-05,
+      "loss": 0.34,
+      "step": 10290
+    },
+    {
+      "epoch": 1.765209940017138,
+      "grad_norm": 0.3942527770996094,
+      "learning_rate": 2.0585546986575266e-05,
+      "loss": 0.3226,
+      "step": 10300
+    },
+    {
+      "epoch": 1.766923736075407,
+      "grad_norm": 0.38292983174324036,
+      "learning_rate": 2.055698371893745e-05,
+      "loss": 0.3176,
+      "step": 10310
+    },
+    {
+      "epoch": 1.7686375321336762,
+      "grad_norm": 0.407644122838974,
+      "learning_rate": 2.0528420451299628e-05,
+      "loss": 0.3395,
+      "step": 10320
+    },
+    {
+      "epoch": 1.7703513281919452,
+      "grad_norm": 0.42464375495910645,
+      "learning_rate": 2.049985718366181e-05,
+      "loss": 0.3213,
+      "step": 10330
+    },
+    {
+      "epoch": 1.7720651242502141,
+      "grad_norm": 0.3565274178981781,
+      "learning_rate": 2.0471293916023997e-05,
+      "loss": 0.3362,
+      "step": 10340
+    },
+    {
+      "epoch": 1.7737789203084833,
+      "grad_norm": 0.37619078159332275,
+      "learning_rate": 2.0442730648386176e-05,
+      "loss": 0.3526,
+      "step": 10350
+    },
+    {
+      "epoch": 1.7754927163667524,
+      "grad_norm": 0.43911996483802795,
+      "learning_rate": 2.041416738074836e-05,
+      "loss": 0.3256,
+      "step": 10360
+    },
+    {
+      "epoch": 1.7772065124250214,
+      "grad_norm": 0.3840199410915375,
+      "learning_rate": 2.038560411311054e-05,
+      "loss": 0.3259,
+      "step": 10370
+    },
+    {
+      "epoch": 1.7789203084832905,
+      "grad_norm": 0.4230378568172455,
+      "learning_rate": 2.035704084547272e-05,
+      "loss": 0.3282,
+      "step": 10380
+    },
+    {
+      "epoch": 1.7806341045415595,
+      "grad_norm": 0.4155026376247406,
+      "learning_rate": 2.0328477577834907e-05,
+      "loss": 0.335,
+      "step": 10390
+    },
+    {
+      "epoch": 1.7823479005998286,
+      "grad_norm": 0.46618354320526123,
+      "learning_rate": 2.029991431019709e-05,
+      "loss": 0.3474,
+      "step": 10400
+    },
+    {
+      "epoch": 1.7840616966580978,
+      "grad_norm": 0.3844912052154541,
+      "learning_rate": 2.027135104255927e-05,
+      "loss": 0.3218,
+      "step": 10410
+    },
+    {
+      "epoch": 1.7857754927163667,
+      "grad_norm": 0.37245050072669983,
+      "learning_rate": 2.024278777492145e-05,
+      "loss": 0.3367,
+      "step": 10420
+    },
+    {
+      "epoch": 1.7874892887746359,
+      "grad_norm": 0.3999907970428467,
+      "learning_rate": 2.0214224507283637e-05,
+      "loss": 0.3493,
+      "step": 10430
+    },
+    {
+      "epoch": 1.7892030848329048,
+      "grad_norm": 0.39571475982666016,
+      "learning_rate": 2.0185661239645816e-05,
+      "loss": 0.3301,
+      "step": 10440
+    },
+    {
+      "epoch": 1.790916880891174,
+      "grad_norm": 0.38127267360687256,
+      "learning_rate": 2.0157097972008e-05,
+      "loss": 0.3419,
+      "step": 10450
+    },
+    {
+      "epoch": 1.7926306769494431,
+      "grad_norm": 0.36275482177734375,
+      "learning_rate": 2.012853470437018e-05,
+      "loss": 0.3284,
+      "step": 10460
+    },
+    {
+      "epoch": 1.794344473007712,
+      "grad_norm": 0.3948185443878174,
+      "learning_rate": 2.009997143673236e-05,
+      "loss": 0.3473,
+      "step": 10470
+    },
+    {
+      "epoch": 1.796058269065981,
+      "grad_norm": 0.4034388065338135,
+      "learning_rate": 2.0071408169094547e-05,
+      "loss": 0.3299,
+      "step": 10480
+    },
+    {
+      "epoch": 1.7977720651242501,
+      "grad_norm": 0.3925109803676605,
+      "learning_rate": 2.004284490145673e-05,
+      "loss": 0.3232,
+      "step": 10490
+    },
+    {
+      "epoch": 1.7994858611825193,
+      "grad_norm": 0.39803266525268555,
+      "learning_rate": 2.001428163381891e-05,
+      "loss": 0.3428,
+      "step": 10500
+    },
+    {
+      "epoch": 1.8011996572407885,
+      "grad_norm": 0.39179807901382446,
+      "learning_rate": 1.998571836618109e-05,
+      "loss": 0.3123,
+      "step": 10510
+    },
+    {
+      "epoch": 1.8029134532990574,
+      "grad_norm": 0.40016594529151917,
+      "learning_rate": 1.9957155098543274e-05,
+      "loss": 0.3334,
+      "step": 10520
+    },
+    {
+      "epoch": 1.8046272493573263,
+      "grad_norm": 0.41052624583244324,
+      "learning_rate": 1.9928591830905457e-05,
+      "loss": 0.342,
+      "step": 10530
+    },
+    {
+      "epoch": 1.8063410454155955,
+      "grad_norm": 0.395678848028183,
+      "learning_rate": 1.990002856326764e-05,
+      "loss": 0.3374,
+      "step": 10540
+    },
+    {
+      "epoch": 1.8080548414738646,
+      "grad_norm": 0.40321213006973267,
+      "learning_rate": 1.9871465295629822e-05,
+      "loss": 0.3424,
+      "step": 10550
+    },
+    {
+      "epoch": 1.8097686375321338,
+      "grad_norm": 0.38921666145324707,
+      "learning_rate": 1.9842902027992e-05,
+      "loss": 0.3249,
+      "step": 10560
+    },
+    {
+      "epoch": 1.8114824335904027,
+      "grad_norm": 0.4000155031681061,
+      "learning_rate": 1.9814338760354184e-05,
+      "loss": 0.3397,
+      "step": 10570
+    },
+    {
+      "epoch": 1.8131962296486717,
+      "grad_norm": 0.4053511619567871,
+      "learning_rate": 1.978577549271637e-05,
+      "loss": 0.3315,
+      "step": 10580
+    },
+    {
+      "epoch": 1.8149100257069408,
+      "grad_norm": 0.3945310115814209,
+      "learning_rate": 1.975721222507855e-05,
+      "loss": 0.3419,
+      "step": 10590
+    },
+    {
+      "epoch": 1.81662382176521,
+      "grad_norm": 0.42978984117507935,
+      "learning_rate": 1.972864895744073e-05,
+      "loss": 0.3515,
+      "step": 10600
+    },
+    {
+      "epoch": 1.8183376178234791,
+      "grad_norm": 0.3619568943977356,
+      "learning_rate": 1.9700085689802914e-05,
+      "loss": 0.3338,
+      "step": 10610
+    },
+    {
+      "epoch": 1.820051413881748,
+      "grad_norm": 0.3962588310241699,
+      "learning_rate": 1.9671522422165097e-05,
+      "loss": 0.3329,
+      "step": 10620
+    },
+    {
+      "epoch": 1.821765209940017,
+      "grad_norm": 0.3794926106929779,
+      "learning_rate": 1.964295915452728e-05,
+      "loss": 0.3274,
+      "step": 10630
+    },
+    {
+      "epoch": 1.8234790059982862,
+      "grad_norm": 0.489555299282074,
+      "learning_rate": 1.9614395886889462e-05,
+      "loss": 0.3547,
+      "step": 10640
+    },
+    {
+      "epoch": 1.8251928020565553,
+      "grad_norm": 0.4101254642009735,
+      "learning_rate": 1.958583261925164e-05,
+      "loss": 0.3262,
+      "step": 10650
+    },
+    {
+      "epoch": 1.8269065981148245,
+      "grad_norm": 0.408819317817688,
+      "learning_rate": 1.9557269351613824e-05,
+      "loss": 0.3394,
+      "step": 10660
+    },
+    {
+      "epoch": 1.8286203941730934,
+      "grad_norm": 0.38574379682540894,
+      "learning_rate": 1.952870608397601e-05,
+      "loss": 0.3432,
+      "step": 10670
+    },
+    {
+      "epoch": 1.8303341902313623,
+      "grad_norm": 0.38276317715644836,
+      "learning_rate": 1.950014281633819e-05,
+      "loss": 0.3439,
+      "step": 10680
+    },
+    {
+      "epoch": 1.8320479862896315,
+      "grad_norm": 0.4008176922798157,
+      "learning_rate": 1.9471579548700372e-05,
+      "loss": 0.348,
+      "step": 10690
+    },
+    {
+      "epoch": 1.8337617823479007,
+      "grad_norm": 0.4130050837993622,
+      "learning_rate": 1.9443016281062554e-05,
+      "loss": 0.3327,
+      "step": 10700
+    },
+    {
+      "epoch": 1.8354755784061698,
+      "grad_norm": 0.39769622683525085,
+      "learning_rate": 1.9414453013424737e-05,
+      "loss": 0.3314,
+      "step": 10710
+    },
+    {
+      "epoch": 1.8371893744644388,
+      "grad_norm": 0.3474248945713043,
+      "learning_rate": 1.938588974578692e-05,
+      "loss": 0.3245,
+      "step": 10720
+    },
+    {
+      "epoch": 1.8389031705227077,
+      "grad_norm": 0.3834674656391144,
+      "learning_rate": 1.9357326478149102e-05,
+      "loss": 0.3327,
+      "step": 10730
+    },
+    {
+      "epoch": 1.8406169665809768,
+      "grad_norm": 0.4144718647003174,
+      "learning_rate": 1.932876321051128e-05,
+      "loss": 0.345,
+      "step": 10740
+    },
+    {
+      "epoch": 1.842330762639246,
+      "grad_norm": 0.41075778007507324,
+      "learning_rate": 1.9300199942873464e-05,
+      "loss": 0.3497,
+      "step": 10750
+    },
+    {
+      "epoch": 1.8440445586975152,
+      "grad_norm": 0.4072783291339874,
+      "learning_rate": 1.927163667523565e-05,
+      "loss": 0.329,
+      "step": 10760
+    },
+    {
+      "epoch": 1.845758354755784,
+      "grad_norm": 0.4069250226020813,
+      "learning_rate": 1.924307340759783e-05,
+      "loss": 0.3254,
+      "step": 10770
+    },
+    {
+      "epoch": 1.847472150814053,
+      "grad_norm": 0.40920811891555786,
+      "learning_rate": 1.9214510139960012e-05,
+      "loss": 0.339,
+      "step": 10780
+    },
+    {
+      "epoch": 1.8491859468723222,
+      "grad_norm": 0.4345456659793854,
+      "learning_rate": 1.9185946872322195e-05,
+      "loss": 0.3486,
+      "step": 10790
+    },
+    {
+      "epoch": 1.8508997429305913,
+      "grad_norm": 0.3994877338409424,
+      "learning_rate": 1.9157383604684377e-05,
+      "loss": 0.3352,
+      "step": 10800
+    },
+    {
+      "epoch": 1.8526135389888603,
+      "grad_norm": 0.38893967866897583,
+      "learning_rate": 1.912882033704656e-05,
+      "loss": 0.3338,
+      "step": 10810
+    },
+    {
+      "epoch": 1.8543273350471294,
+      "grad_norm": 0.37634167075157166,
+      "learning_rate": 1.9100257069408742e-05,
+      "loss": 0.3379,
+      "step": 10820
+    },
+    {
+      "epoch": 1.8560411311053984,
+      "grad_norm": 0.5175420045852661,
+      "learning_rate": 1.9071693801770925e-05,
+      "loss": 0.3277,
+      "step": 10830
+    },
+    {
+      "epoch": 1.8577549271636675,
+      "grad_norm": 0.34820953011512756,
+      "learning_rate": 1.9043130534133104e-05,
+      "loss": 0.3458,
+      "step": 10840
+    },
+    {
+      "epoch": 1.8594687232219367,
+      "grad_norm": 0.46136412024497986,
+      "learning_rate": 1.9014567266495287e-05,
+      "loss": 0.3433,
+      "step": 10850
+    },
+    {
+      "epoch": 1.8611825192802056,
+      "grad_norm": 0.3980143666267395,
+      "learning_rate": 1.898600399885747e-05,
+      "loss": 0.3362,
+      "step": 10860
+    },
+    {
+      "epoch": 1.8628963153384748,
+      "grad_norm": 0.3734038472175598,
+      "learning_rate": 1.8957440731219652e-05,
+      "loss": 0.3279,
+      "step": 10870
+    },
+    {
+      "epoch": 1.8646101113967437,
+      "grad_norm": 0.38700205087661743,
+      "learning_rate": 1.8928877463581835e-05,
+      "loss": 0.3336,
+      "step": 10880
+    },
+    {
+      "epoch": 1.8663239074550129,
+      "grad_norm": 0.3614012897014618,
+      "learning_rate": 1.8900314195944017e-05,
+      "loss": 0.3314,
+      "step": 10890
+    },
+    {
+      "epoch": 1.868037703513282,
+      "grad_norm": 0.40407755970954895,
+      "learning_rate": 1.88717509283062e-05,
+      "loss": 0.326,
+      "step": 10900
+    },
+    {
+      "epoch": 1.869751499571551,
+      "grad_norm": 0.45859092473983765,
+      "learning_rate": 1.8843187660668383e-05,
+      "loss": 0.3307,
+      "step": 10910
+    },
+    {
+      "epoch": 1.87146529562982,
+      "grad_norm": 0.3943389058113098,
+      "learning_rate": 1.8814624393030565e-05,
+      "loss": 0.32,
+      "step": 10920
+    },
+    {
+      "epoch": 1.873179091688089,
+      "grad_norm": 0.4314815402030945,
+      "learning_rate": 1.8786061125392745e-05,
+      "loss": 0.321,
+      "step": 10930
+    },
+    {
+      "epoch": 1.8748928877463582,
+      "grad_norm": 0.4362752437591553,
+      "learning_rate": 1.8757497857754927e-05,
+      "loss": 0.3245,
+      "step": 10940
+    },
+    {
+      "epoch": 1.8766066838046274,
+      "grad_norm": 0.4172927439212799,
+      "learning_rate": 1.8728934590117113e-05,
+      "loss": 0.3297,
+      "step": 10950
+    },
+    {
+      "epoch": 1.8783204798628963,
+      "grad_norm": 0.350132554769516,
+      "learning_rate": 1.8700371322479292e-05,
+      "loss": 0.3232,
+      "step": 10960
+    },
+    {
+      "epoch": 1.8800342759211652,
+      "grad_norm": 0.38863199949264526,
+      "learning_rate": 1.8671808054841475e-05,
+      "loss": 0.3121,
+      "step": 10970
+    },
+    {
+      "epoch": 1.8817480719794344,
+      "grad_norm": 0.3894982933998108,
+      "learning_rate": 1.8643244787203658e-05,
+      "loss": 0.3203,
+      "step": 10980
+    },
+    {
+      "epoch": 1.8834618680377035,
+      "grad_norm": 0.405094712972641,
+      "learning_rate": 1.8614681519565837e-05,
+      "loss": 0.331,
+      "step": 10990
+    },
+    {
+      "epoch": 1.8851756640959727,
+      "grad_norm": 0.4262722134590149,
+      "learning_rate": 1.8586118251928023e-05,
+      "loss": 0.3411,
+      "step": 11000
+    },
+    {
+      "epoch": 1.8868894601542416,
+      "grad_norm": 0.37803834676742554,
+      "learning_rate": 1.8557554984290206e-05,
+      "loss": 0.327,
+      "step": 11010
+    },
+    {
+      "epoch": 1.8886032562125106,
+      "grad_norm": 0.44252264499664307,
+      "learning_rate": 1.8528991716652385e-05,
+      "loss": 0.322,
+      "step": 11020
+    },
+    {
+      "epoch": 1.8903170522707797,
+      "grad_norm": 0.41628533601760864,
+      "learning_rate": 1.8500428449014567e-05,
+      "loss": 0.3338,
+      "step": 11030
+    },
+    {
+      "epoch": 1.8920308483290489,
+      "grad_norm": 0.4093359112739563,
+      "learning_rate": 1.847186518137675e-05,
+      "loss": 0.3329,
+      "step": 11040
+    },
+    {
+      "epoch": 1.893744644387318,
+      "grad_norm": 0.41291117668151855,
+      "learning_rate": 1.8443301913738933e-05,
+      "loss": 0.3206,
+      "step": 11050
+    },
+    {
+      "epoch": 1.895458440445587,
+      "grad_norm": 0.4110288918018341,
+      "learning_rate": 1.8414738646101115e-05,
+      "loss": 0.3396,
+      "step": 11060
+    },
+    {
+      "epoch": 1.897172236503856,
+      "grad_norm": 0.41824471950531006,
+      "learning_rate": 1.8386175378463298e-05,
+      "loss": 0.3272,
+      "step": 11070
+    },
+    {
+      "epoch": 1.898886032562125,
+      "grad_norm": 0.3943333029747009,
+      "learning_rate": 1.8357612110825477e-05,
+      "loss": 0.3448,
+      "step": 11080
+    },
+    {
+      "epoch": 1.9005998286203942,
+      "grad_norm": 0.4381183683872223,
+      "learning_rate": 1.8329048843187663e-05,
+      "loss": 0.3315,
+      "step": 11090
+    },
+    {
+      "epoch": 1.9023136246786634,
+      "grad_norm": 0.439394474029541,
+      "learning_rate": 1.8300485575549846e-05,
+      "loss": 0.3265,
+      "step": 11100
+    },
+    {
+      "epoch": 1.9040274207369323,
+      "grad_norm": 0.41325217485427856,
+      "learning_rate": 1.8271922307912025e-05,
+      "loss": 0.3234,
+      "step": 11110
+    },
+    {
+      "epoch": 1.9057412167952013,
+      "grad_norm": 0.44458985328674316,
+      "learning_rate": 1.8243359040274208e-05,
+      "loss": 0.3254,
+      "step": 11120
+    },
+    {
+      "epoch": 1.9074550128534704,
+      "grad_norm": 0.3895314335823059,
+      "learning_rate": 1.821479577263639e-05,
+      "loss": 0.3406,
+      "step": 11130
+    },
+    {
+      "epoch": 1.9091688089117396,
+      "grad_norm": 0.407094806432724,
+      "learning_rate": 1.8186232504998573e-05,
+      "loss": 0.3244,
+      "step": 11140
+    },
+    {
+      "epoch": 1.9108826049700087,
+      "grad_norm": 0.4123266935348511,
+      "learning_rate": 1.8157669237360755e-05,
+      "loss": 0.3346,
+      "step": 11150
+    },
+    {
+      "epoch": 1.9125964010282777,
+      "grad_norm": 0.3695683479309082,
+      "learning_rate": 1.8129105969722938e-05,
+      "loss": 0.3134,
+      "step": 11160
+    },
+    {
+      "epoch": 1.9143101970865466,
+      "grad_norm": 0.3902941942214966,
+      "learning_rate": 1.8100542702085117e-05,
+      "loss": 0.3358,
+      "step": 11170
+    },
+    {
+      "epoch": 1.9160239931448158,
+      "grad_norm": 0.3855278789997101,
+      "learning_rate": 1.80719794344473e-05,
+      "loss": 0.3261,
+      "step": 11180
+    },
+    {
+      "epoch": 1.917737789203085,
+      "grad_norm": 0.4039939045906067,
+      "learning_rate": 1.8043416166809486e-05,
+      "loss": 0.3453,
+      "step": 11190
+    },
+    {
+      "epoch": 1.919451585261354,
+      "grad_norm": 0.3597125709056854,
+      "learning_rate": 1.8014852899171665e-05,
+      "loss": 0.3347,
+      "step": 11200
+    },
+    {
+      "epoch": 1.921165381319623,
+      "grad_norm": 0.38155752420425415,
+      "learning_rate": 1.7986289631533848e-05,
+      "loss": 0.3383,
+      "step": 11210
+    },
+    {
+      "epoch": 1.922879177377892,
+      "grad_norm": 0.4478619396686554,
+      "learning_rate": 1.795772636389603e-05,
+      "loss": 0.3521,
+      "step": 11220
+    },
+    {
+      "epoch": 1.924592973436161,
+      "grad_norm": 0.36768054962158203,
+      "learning_rate": 1.7929163096258213e-05,
+      "loss": 0.3237,
+      "step": 11230
+    },
+    {
+      "epoch": 1.9263067694944302,
+      "grad_norm": 0.442877858877182,
+      "learning_rate": 1.7900599828620396e-05,
+      "loss": 0.3355,
+      "step": 11240
+    },
+    {
+      "epoch": 1.9280205655526992,
+      "grad_norm": 0.369087815284729,
+      "learning_rate": 1.787203656098258e-05,
+      "loss": 0.3292,
+      "step": 11250
+    },
+    {
+      "epoch": 1.9297343616109683,
+      "grad_norm": 0.4272878170013428,
+      "learning_rate": 1.7843473293344758e-05,
+      "loss": 0.3247,
+      "step": 11260
+    },
+    {
+      "epoch": 1.9314481576692373,
+      "grad_norm": 0.40085938572883606,
+      "learning_rate": 1.781491002570694e-05,
+      "loss": 0.3115,
+      "step": 11270
+    },
+    {
+      "epoch": 1.9331619537275064,
+      "grad_norm": 0.4040519595146179,
+      "learning_rate": 1.7786346758069126e-05,
+      "loss": 0.3345,
+      "step": 11280
+    },
+    {
+      "epoch": 1.9348757497857756,
+      "grad_norm": 0.38495534658432007,
+      "learning_rate": 1.7757783490431305e-05,
+      "loss": 0.3325,
+      "step": 11290
+    },
+    {
+      "epoch": 1.9365895458440445,
+      "grad_norm": 0.40545254945755005,
+      "learning_rate": 1.7729220222793488e-05,
+      "loss": 0.3401,
+      "step": 11300
+    },
+    {
+      "epoch": 1.9383033419023135,
+      "grad_norm": 0.40779033303260803,
+      "learning_rate": 1.770065695515567e-05,
+      "loss": 0.3293,
+      "step": 11310
+    },
+    {
+      "epoch": 1.9400171379605826,
+      "grad_norm": 0.4259167015552521,
+      "learning_rate": 1.7672093687517853e-05,
+      "loss": 0.3174,
+      "step": 11320
+    },
+    {
+      "epoch": 1.9417309340188518,
+      "grad_norm": 0.3741375207901001,
+      "learning_rate": 1.7643530419880036e-05,
+      "loss": 0.3279,
+      "step": 11330
+    },
+    {
+      "epoch": 1.943444730077121,
+      "grad_norm": 0.4534846842288971,
+      "learning_rate": 1.761496715224222e-05,
+      "loss": 0.348,
+      "step": 11340
+    },
+    {
+      "epoch": 1.9451585261353899,
+      "grad_norm": 0.41401243209838867,
+      "learning_rate": 1.7586403884604398e-05,
+      "loss": 0.3238,
+      "step": 11350
+    },
+    {
+      "epoch": 1.9468723221936588,
+      "grad_norm": 0.39744701981544495,
+      "learning_rate": 1.755784061696658e-05,
+      "loss": 0.3293,
+      "step": 11360
+    },
+    {
+      "epoch": 1.948586118251928,
+      "grad_norm": 0.46253296732902527,
+      "learning_rate": 1.7529277349328766e-05,
+      "loss": 0.3308,
+      "step": 11370
+    },
+    {
+      "epoch": 1.9502999143101971,
+      "grad_norm": 0.39185985922813416,
+      "learning_rate": 1.7500714081690946e-05,
+      "loss": 0.345,
+      "step": 11380
+    },
+    {
+      "epoch": 1.9520137103684663,
+      "grad_norm": 0.37337854504585266,
+      "learning_rate": 1.7472150814053128e-05,
+      "loss": 0.3404,
+      "step": 11390
+    },
+    {
+      "epoch": 1.9537275064267352,
+      "grad_norm": 0.47062331438064575,
+      "learning_rate": 1.744358754641531e-05,
+      "loss": 0.3361,
+      "step": 11400
+    },
+    {
+      "epoch": 1.9554413024850041,
+      "grad_norm": 0.3791089355945587,
+      "learning_rate": 1.7415024278777494e-05,
+      "loss": 0.3312,
+      "step": 11410
+    },
+    {
+      "epoch": 1.9571550985432733,
+      "grad_norm": 0.3983840048313141,
+      "learning_rate": 1.7386461011139676e-05,
+      "loss": 0.3433,
+      "step": 11420
+    },
+    {
+      "epoch": 1.9588688946015425,
+      "grad_norm": 0.4170704185962677,
+      "learning_rate": 1.735789774350186e-05,
+      "loss": 0.3179,
+      "step": 11430
+    },
+    {
+      "epoch": 1.9605826906598116,
+      "grad_norm": 0.38848522305488586,
+      "learning_rate": 1.7329334475864038e-05,
+      "loss": 0.3247,
+      "step": 11440
+    },
+    {
+      "epoch": 1.9622964867180805,
+      "grad_norm": 0.38156813383102417,
+      "learning_rate": 1.730077120822622e-05,
+      "loss": 0.3323,
+      "step": 11450
+    },
+    {
+      "epoch": 1.9640102827763495,
+      "grad_norm": 0.37670812010765076,
+      "learning_rate": 1.7272207940588403e-05,
+      "loss": 0.3362,
+      "step": 11460
+    },
+    {
+      "epoch": 1.9657240788346186,
+      "grad_norm": 0.38012853264808655,
+      "learning_rate": 1.7243644672950586e-05,
+      "loss": 0.3289,
+      "step": 11470
+    },
+    {
+      "epoch": 1.9674378748928878,
+      "grad_norm": 0.38730666041374207,
+      "learning_rate": 1.721508140531277e-05,
+      "loss": 0.3247,
+      "step": 11480
+    },
+    {
+      "epoch": 1.969151670951157,
+      "grad_norm": 0.3926190733909607,
+      "learning_rate": 1.718651813767495e-05,
+      "loss": 0.3211,
+      "step": 11490
+    },
+    {
+      "epoch": 1.9708654670094259,
+      "grad_norm": 0.39375394582748413,
+      "learning_rate": 1.7157954870037134e-05,
+      "loss": 0.3248,
+      "step": 11500
+    },
+    {
+      "epoch": 1.9725792630676948,
+      "grad_norm": 0.3916712701320648,
+      "learning_rate": 1.7129391602399313e-05,
+      "loss": 0.3269,
+      "step": 11510
+    },
+    {
+      "epoch": 1.974293059125964,
+      "grad_norm": 0.41252121329307556,
+      "learning_rate": 1.71008283347615e-05,
+      "loss": 0.3331,
+      "step": 11520
+    },
+    {
+      "epoch": 1.9760068551842331,
+      "grad_norm": 0.4166218042373657,
+      "learning_rate": 1.707226506712368e-05,
+      "loss": 0.3457,
+      "step": 11530
+    },
+    {
+      "epoch": 1.9777206512425023,
+      "grad_norm": 0.4022178649902344,
+      "learning_rate": 1.704370179948586e-05,
+      "loss": 0.3446,
+      "step": 11540
+    },
+    {
+      "epoch": 1.9794344473007712,
+      "grad_norm": 0.43955641984939575,
+      "learning_rate": 1.7015138531848044e-05,
+      "loss": 0.3377,
+      "step": 11550
+    },
+    {
+      "epoch": 1.9811482433590402,
+      "grad_norm": 0.4006490707397461,
+      "learning_rate": 1.6986575264210226e-05,
+      "loss": 0.3356,
+      "step": 11560
+    },
+    {
+      "epoch": 1.9828620394173093,
+      "grad_norm": 0.4111030101776123,
+      "learning_rate": 1.695801199657241e-05,
+      "loss": 0.3344,
+      "step": 11570
+    },
+    {
+      "epoch": 1.9845758354755785,
+      "grad_norm": 0.36849358677864075,
+      "learning_rate": 1.692944872893459e-05,
+      "loss": 0.3427,
+      "step": 11580
+    },
+    {
+      "epoch": 1.9862896315338476,
+      "grad_norm": 0.40527480840682983,
+      "learning_rate": 1.6900885461296774e-05,
+      "loss": 0.3309,
+      "step": 11590
+    },
+    {
+      "epoch": 1.9880034275921166,
+      "grad_norm": 0.3989039361476898,
+      "learning_rate": 1.6872322193658953e-05,
+      "loss": 0.3303,
+      "step": 11600
+    },
+    {
+      "epoch": 1.9897172236503855,
+      "grad_norm": 0.39155837893486023,
+      "learning_rate": 1.684375892602114e-05,
+      "loss": 0.3364,
+      "step": 11610
+    },
+    {
+      "epoch": 1.9914310197086547,
+      "grad_norm": 0.4368874132633209,
+      "learning_rate": 1.6815195658383322e-05,
+      "loss": 0.3434,
+      "step": 11620
+    },
+    {
+      "epoch": 1.9931448157669238,
+      "grad_norm": 0.40511763095855713,
+      "learning_rate": 1.67866323907455e-05,
+      "loss": 0.3221,
+      "step": 11630
+    },
+    {
+      "epoch": 1.9948586118251928,
+      "grad_norm": 0.3981911540031433,
+      "learning_rate": 1.6758069123107684e-05,
+      "loss": 0.3376,
+      "step": 11640
+    },
+    {
+      "epoch": 1.996572407883462,
+      "grad_norm": 0.3849552571773529,
+      "learning_rate": 1.6729505855469866e-05,
+      "loss": 0.3236,
+      "step": 11650
+    },
+    {
+      "epoch": 1.9982862039417308,
+      "grad_norm": 0.4233885109424591,
+      "learning_rate": 1.670094258783205e-05,
+      "loss": 0.3181,
+      "step": 11660
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.8736078143119812,
+      "learning_rate": 1.667237932019423e-05,
+      "loss": 0.3204,
+      "step": 11670
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 17505,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 9.7565399580672e+16,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}