diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5327 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 19.997354497354497,
+  "eval_steps": 500,
+  "global_step": 7559,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.026455026455026454,
+      "grad_norm": 6.989788055419922,
+      "learning_rate": 5.291005291005291e-06,
+      "loss": 1.1645,
+      "step": 10
+    },
+    {
+      "epoch": 0.05291005291005291,
+      "grad_norm": 5.150631427764893,
+      "learning_rate": 1.0582010582010582e-05,
+      "loss": 0.9316,
+      "step": 20
+    },
+    {
+      "epoch": 0.07936507936507936,
+      "grad_norm": 3.7024333477020264,
+      "learning_rate": 1.5873015873015872e-05,
+      "loss": 0.5019,
+      "step": 30
+    },
+    {
+      "epoch": 0.10582010582010581,
+      "grad_norm": 0.8457810878753662,
+      "learning_rate": 2.1164021164021164e-05,
+      "loss": 0.3289,
+      "step": 40
+    },
+    {
+      "epoch": 0.13227513227513227,
+      "grad_norm": 1.1393985748291016,
+      "learning_rate": 2.6455026455026456e-05,
+      "loss": 0.2681,
+      "step": 50
+    },
+    {
+      "epoch": 0.15873015873015872,
+      "grad_norm": 1.797629714012146,
+      "learning_rate": 3.1746031746031745e-05,
+      "loss": 0.2378,
+      "step": 60
+    },
+    {
+      "epoch": 0.18518518518518517,
+      "grad_norm": 1.5710021257400513,
+      "learning_rate": 3.7037037037037037e-05,
+      "loss": 0.1902,
+      "step": 70
+    },
+    {
+      "epoch": 0.21164021164021163,
+      "grad_norm": 2.3919057846069336,
+      "learning_rate": 4.232804232804233e-05,
+      "loss": 0.1454,
+      "step": 80
+    },
+    {
+      "epoch": 0.23809523809523808,
+      "grad_norm": 1.9616649150848389,
+      "learning_rate": 4.761904761904762e-05,
+      "loss": 0.1157,
+      "step": 90
+    },
+    {
+      "epoch": 0.26455026455026454,
+      "grad_norm": 2.282738208770752,
+      "learning_rate": 5.291005291005291e-05,
+      "loss": 0.1006,
+      "step": 100
+    },
+    {
+      "epoch": 0.291005291005291,
+      "grad_norm": 0.6997108459472656,
+      "learning_rate": 5.82010582010582e-05,
+      "loss": 0.0969,
+      "step": 110
+    },
+    {
+      "epoch": 0.31746031746031744,
+      "grad_norm": 1.1011691093444824,
+      "learning_rate": 6.349206349206349e-05,
+      "loss": 0.0777,
+      "step": 120
+    },
+    {
+      "epoch": 0.3439153439153439,
+      "grad_norm": 1.2826348543167114,
+      "learning_rate": 6.878306878306878e-05,
+      "loss": 0.0701,
+      "step": 130
+    },
+    {
+      "epoch": 0.37037037037037035,
+      "grad_norm": 1.0125218629837036,
+      "learning_rate": 7.407407407407407e-05,
+      "loss": 0.0781,
+      "step": 140
+    },
+    {
+      "epoch": 0.3968253968253968,
+      "grad_norm": 1.344835877418518,
+      "learning_rate": 7.936507936507937e-05,
+      "loss": 0.0717,
+      "step": 150
+    },
+    {
+      "epoch": 0.42328042328042326,
+      "grad_norm": 0.7368573546409607,
+      "learning_rate": 8.465608465608466e-05,
+      "loss": 0.0681,
+      "step": 160
+    },
+    {
+      "epoch": 0.4497354497354497,
+      "grad_norm": 0.8334409594535828,
+      "learning_rate": 8.994708994708995e-05,
+      "loss": 0.0652,
+      "step": 170
+    },
+    {
+      "epoch": 0.47619047619047616,
+      "grad_norm": 0.9617886543273926,
+      "learning_rate": 9.523809523809524e-05,
+      "loss": 0.0591,
+      "step": 180
+    },
+    {
+      "epoch": 0.5026455026455027,
+      "grad_norm": 1.6967134475708008,
+      "learning_rate": 0.00010052910052910055,
+      "loss": 0.0603,
+      "step": 190
+    },
+    {
+      "epoch": 0.5291005291005291,
+      "grad_norm": 1.2042906284332275,
+      "learning_rate": 0.00010582010582010582,
+      "loss": 0.0569,
+      "step": 200
+    },
+    {
+      "epoch": 0.5555555555555556,
+      "grad_norm": 0.8804681897163391,
+      "learning_rate": 0.00011111111111111112,
+      "loss": 0.0572,
+      "step": 210
+    },
+    {
+      "epoch": 0.582010582010582,
+      "grad_norm": 0.5275452136993408,
+      "learning_rate": 0.0001164021164021164,
+      "loss": 0.0564,
+      "step": 220
+    },
+    {
+      "epoch": 0.6084656084656085,
+      "grad_norm": 0.8496862053871155,
+      "learning_rate": 0.0001216931216931217,
+      "loss": 0.0559,
+      "step": 230
+    },
+    {
+      "epoch": 0.6349206349206349,
+      "grad_norm": 0.9702103137969971,
+      "learning_rate": 0.00012698412698412698,
+      "loss": 0.0575,
+      "step": 240
+    },
+    {
+      "epoch": 0.6613756613756614,
+      "grad_norm": 0.5400204658508301,
+      "learning_rate": 0.00013227513227513228,
+      "loss": 0.0492,
+      "step": 250
+    },
+    {
+      "epoch": 0.6878306878306878,
+      "grad_norm": 0.7467755079269409,
+      "learning_rate": 0.00013756613756613756,
+      "loss": 0.0486,
+      "step": 260
+    },
+    {
+      "epoch": 0.7142857142857143,
+      "grad_norm": 0.7477474808692932,
+      "learning_rate": 0.00014285714285714287,
+      "loss": 0.0491,
+      "step": 270
+    },
+    {
+      "epoch": 0.7407407407407407,
+      "grad_norm": 0.9659186005592346,
+      "learning_rate": 0.00014814814814814815,
+      "loss": 0.0455,
+      "step": 280
+    },
+    {
+      "epoch": 0.7671957671957672,
+      "grad_norm": 0.8414769172668457,
+      "learning_rate": 0.00015343915343915345,
+      "loss": 0.0465,
+      "step": 290
+    },
+    {
+      "epoch": 0.7936507936507936,
+      "grad_norm": 0.690024733543396,
+      "learning_rate": 0.00015873015873015873,
+      "loss": 0.0506,
+      "step": 300
+    },
+    {
+      "epoch": 0.8201058201058201,
+      "grad_norm": 0.6750884056091309,
+      "learning_rate": 0.00016402116402116404,
+      "loss": 0.0525,
+      "step": 310
+    },
+    {
+      "epoch": 0.8465608465608465,
+      "grad_norm": 0.8909515142440796,
+      "learning_rate": 0.00016931216931216931,
+      "loss": 0.0417,
+      "step": 320
+    },
+    {
+      "epoch": 0.873015873015873,
+      "grad_norm": 0.6845270395278931,
+      "learning_rate": 0.00017460317460317462,
+      "loss": 0.0468,
+      "step": 330
+    },
+    {
+      "epoch": 0.8994708994708994,
+      "grad_norm": 0.5316764116287231,
+      "learning_rate": 0.0001798941798941799,
+      "loss": 0.0525,
+      "step": 340
+    },
+    {
+      "epoch": 0.9259259259259259,
+      "grad_norm": 0.7101507186889648,
+      "learning_rate": 0.0001851851851851852,
+      "loss": 0.0456,
+      "step": 350
+    },
+    {
+      "epoch": 0.9523809523809523,
+      "grad_norm": 0.6450228095054626,
+      "learning_rate": 0.00019047619047619048,
+      "loss": 0.0528,
+      "step": 360
+    },
+    {
+      "epoch": 0.9788359788359788,
+      "grad_norm": 0.5121487975120544,
+      "learning_rate": 0.0001957671957671958,
+      "loss": 0.0486,
+      "step": 370
+    },
+    {
+      "epoch": 1.0052910052910053,
+      "grad_norm": 0.8493865728378296,
+      "learning_rate": 0.0001999999617210638,
+      "loss": 0.0459,
+      "step": 380
+    },
+    {
+      "epoch": 1.0317460317460316,
+      "grad_norm": 0.6720256805419922,
+      "learning_rate": 0.00019999862196137396,
+      "loss": 0.0435,
+      "step": 390
+    },
+    {
+      "epoch": 1.0582010582010581,
+      "grad_norm": 0.43235063552856445,
+      "learning_rate": 0.00019999536828417959,
+      "loss": 0.0449,
+      "step": 400
+    },
+    {
+      "epoch": 1.0846560846560847,
+      "grad_norm": 0.5137812495231628,
+      "learning_rate": 0.00019999020075175426,
+      "loss": 0.0452,
+      "step": 410
+    },
+    {
+      "epoch": 1.1111111111111112,
+      "grad_norm": 0.6568519473075867,
+      "learning_rate": 0.00019998311946300173,
+      "loss": 0.0442,
+      "step": 420
+    },
+    {
+      "epoch": 1.1375661375661377,
+      "grad_norm": 1.1188125610351562,
+      "learning_rate": 0.0001999741245534538,
+      "loss": 0.0484,
+      "step": 430
+    },
+    {
+      "epoch": 1.164021164021164,
+      "grad_norm": 0.6816080808639526,
+      "learning_rate": 0.00019996321619526804,
+      "loss": 0.0443,
+      "step": 440
+    },
+    {
+      "epoch": 1.1904761904761905,
+      "grad_norm": 0.41056472063064575,
+      "learning_rate": 0.00019995039459722423,
+      "loss": 0.0462,
+      "step": 450
+    },
+    {
+      "epoch": 1.216931216931217,
+      "grad_norm": 0.528852641582489,
+      "learning_rate": 0.00019993566000472068,
+      "loss": 0.0416,
+      "step": 460
+    },
+    {
+      "epoch": 1.2433862433862433,
+      "grad_norm": 0.549899697303772,
+      "learning_rate": 0.00019991901269976911,
+      "loss": 0.0417,
+      "step": 470
+    },
+    {
+      "epoch": 1.2698412698412698,
+      "grad_norm": 0.48012372851371765,
+      "learning_rate": 0.00019990045300098967,
+      "loss": 0.0396,
+      "step": 480
+    },
+    {
+      "epoch": 1.2962962962962963,
+      "grad_norm": 0.5004888772964478,
+      "learning_rate": 0.00019987998126360453,
+      "loss": 0.0397,
+      "step": 490
+    },
+    {
+      "epoch": 1.3227513227513228,
+      "grad_norm": 0.5583204627037048,
+      "learning_rate": 0.00019985759787943127,
+      "loss": 0.0437,
+      "step": 500
+    },
+    {
+      "epoch": 1.3492063492063493,
+      "grad_norm": 0.7729330062866211,
+      "learning_rate": 0.00019983330327687533,
+      "loss": 0.0374,
+      "step": 510
+    },
+    {
+      "epoch": 1.3756613756613756,
+      "grad_norm": 0.5652401447296143,
+      "learning_rate": 0.0001998070979209217,
+      "loss": 0.0402,
+      "step": 520
+    },
+    {
+      "epoch": 1.402116402116402,
+      "grad_norm": 0.7112484574317932,
+      "learning_rate": 0.00019977898231312628,
+      "loss": 0.0428,
+      "step": 530
+    },
+    {
+      "epoch": 1.4285714285714286,
+      "grad_norm": 0.66225665807724,
+      "learning_rate": 0.00019974895699160597,
+      "loss": 0.0415,
+      "step": 540
+    },
+    {
+      "epoch": 1.455026455026455,
+      "grad_norm": 0.4615730047225952,
+      "learning_rate": 0.00019971702253102856,
+      "loss": 0.0351,
+      "step": 550
+    },
+    {
+      "epoch": 1.4814814814814814,
+      "grad_norm": 0.5178949236869812,
+      "learning_rate": 0.00019968317954260168,
+      "loss": 0.0409,
+      "step": 560
+    },
+    {
+      "epoch": 1.507936507936508,
+      "grad_norm": 0.5937110781669617,
+      "learning_rate": 0.00019964742867406126,
+      "loss": 0.0423,
+      "step": 570
+    },
+    {
+      "epoch": 1.5343915343915344,
+      "grad_norm": 0.45062029361724854,
+      "learning_rate": 0.00019960977060965872,
+      "loss": 0.0404,
+      "step": 580
+    },
+    {
+      "epoch": 1.560846560846561,
+      "grad_norm": 0.41606688499450684,
+      "learning_rate": 0.00019957020607014835,
+      "loss": 0.038,
+      "step": 590
+    },
+    {
+      "epoch": 1.5873015873015874,
+      "grad_norm": 0.36349207162857056,
+      "learning_rate": 0.00019952873581277324,
+      "loss": 0.0392,
+      "step": 600
+    },
+    {
+      "epoch": 1.6137566137566137,
+      "grad_norm": 0.7711144685745239,
+      "learning_rate": 0.0001994853606312508,
+      "loss": 0.0393,
+      "step": 610
+    },
+    {
+      "epoch": 1.6402116402116402,
+      "grad_norm": 0.5946980714797974,
+      "learning_rate": 0.00019944008135575772,
+      "loss": 0.0408,
+      "step": 620
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 0.21440181136131287,
+      "learning_rate": 0.00019939289885291384,
+      "loss": 0.0399,
+      "step": 630
+    },
+    {
+      "epoch": 1.693121693121693,
+      "grad_norm": 0.48613789677619934,
+      "learning_rate": 0.00019934381402576584,
+      "loss": 0.0406,
+      "step": 640
+    },
+    {
+      "epoch": 1.7195767195767195,
+      "grad_norm": 0.47162196040153503,
+      "learning_rate": 0.00019929282781376976,
+      "loss": 0.0433,
+      "step": 650
+    },
+    {
+      "epoch": 1.746031746031746,
+      "grad_norm": 0.5799857974052429,
+      "learning_rate": 0.00019923994119277309,
+      "loss": 0.0412,
+      "step": 660
+    },
+    {
+      "epoch": 1.7724867724867726,
+      "grad_norm": 0.6739380955696106,
+      "learning_rate": 0.00019918515517499606,
+      "loss": 0.0365,
+      "step": 670
+    },
+    {
+      "epoch": 1.798941798941799,
+      "grad_norm": 0.5352742671966553,
+      "learning_rate": 0.0001991284708090123,
+      "loss": 0.0367,
+      "step": 680
+    },
+    {
+      "epoch": 1.8253968253968254,
+      "grad_norm": 0.6227455735206604,
+      "learning_rate": 0.00019906988917972878,
+      "loss": 0.0357,
+      "step": 690
+    },
+    {
+      "epoch": 1.8518518518518519,
+      "grad_norm": 0.546350359916687,
+      "learning_rate": 0.00019900941140836497,
+      "loss": 0.0369,
+      "step": 700
+    },
+    {
+      "epoch": 1.8783068783068781,
+      "grad_norm": 0.5400755405426025,
+      "learning_rate": 0.00019894703865243156,
+      "loss": 0.0391,
+      "step": 710
+    },
+    {
+      "epoch": 1.9047619047619047,
+      "grad_norm": 0.6319063305854797,
+      "learning_rate": 0.000198882772105708,
+      "loss": 0.0382,
+      "step": 720
+    },
+    {
+      "epoch": 1.9312169312169312,
+      "grad_norm": 0.5921104550361633,
+      "learning_rate": 0.00019881661299822,
+      "loss": 0.0361,
+      "step": 730
+    },
+    {
+      "epoch": 1.9576719576719577,
+      "grad_norm": 0.5720839500427246,
+      "learning_rate": 0.00019874856259621568,
+      "loss": 0.0419,
+      "step": 740
+    },
+    {
+      "epoch": 1.9841269841269842,
+      "grad_norm": 0.47676175832748413,
+      "learning_rate": 0.0001986786222021416,
+      "loss": 0.0387,
+      "step": 750
+    },
+    {
+      "epoch": 2.0105820105820107,
+      "grad_norm": 0.4507744312286377,
+      "learning_rate": 0.0001986067931546176,
+      "loss": 0.0376,
+      "step": 760
+    },
+    {
+      "epoch": 2.037037037037037,
+      "grad_norm": 0.5624145269393921,
+      "learning_rate": 0.0001985330768284114,
+      "loss": 0.0402,
+      "step": 770
+    },
+    {
+      "epoch": 2.0634920634920633,
+      "grad_norm": 0.3807164430618286,
+      "learning_rate": 0.00019845747463441207,
+      "loss": 0.0375,
+      "step": 780
+    },
+    {
+      "epoch": 2.0899470899470898,
+      "grad_norm": 0.579364001750946,
+      "learning_rate": 0.0001983799880196032,
+      "loss": 0.0384,
+      "step": 790
+    },
+    {
+      "epoch": 2.1164021164021163,
+      "grad_norm": 0.5431824922561646,
+      "learning_rate": 0.00019830061846703507,
+      "loss": 0.0374,
+      "step": 800
+    },
+    {
+      "epoch": 2.142857142857143,
+      "grad_norm": 0.5862318873405457,
+      "learning_rate": 0.0001982193674957964,
+      "loss": 0.033,
+      "step": 810
+    },
+    {
+      "epoch": 2.1693121693121693,
+      "grad_norm": 0.6154080629348755,
+      "learning_rate": 0.00019813623666098523,
+      "loss": 0.0375,
+      "step": 820
+    },
+    {
+      "epoch": 2.195767195767196,
+      "grad_norm": 0.48721227049827576,
+      "learning_rate": 0.000198051227553679,
+      "loss": 0.0351,
+      "step": 830
+    },
+    {
+      "epoch": 2.2222222222222223,
+      "grad_norm": 0.4979737401008606,
+      "learning_rate": 0.00019796434180090436,
+      "loss": 0.0388,
+      "step": 840
+    },
+    {
+      "epoch": 2.248677248677249,
+      "grad_norm": 0.47598177194595337,
+      "learning_rate": 0.00019787558106560584,
+      "loss": 0.0362,
+      "step": 850
+    },
+    {
+      "epoch": 2.2751322751322753,
+      "grad_norm": 0.3844626247882843,
+      "learning_rate": 0.00019778494704661412,
+      "loss": 0.0356,
+      "step": 860
+    },
+    {
+      "epoch": 2.3015873015873014,
+      "grad_norm": 0.24894066154956818,
+      "learning_rate": 0.00019769244147861348,
+      "loss": 0.0331,
+      "step": 870
+    },
+    {
+      "epoch": 2.328042328042328,
+      "grad_norm": 0.5516997575759888,
+      "learning_rate": 0.00019759806613210853,
+      "loss": 0.0377,
+      "step": 880
+    },
+    {
+      "epoch": 2.3544973544973544,
+      "grad_norm": 0.23550976812839508,
+      "learning_rate": 0.0001975018228133904,
+      "loss": 0.0357,
+      "step": 890
+    },
+    {
+      "epoch": 2.380952380952381,
+      "grad_norm": 0.24155914783477783,
+      "learning_rate": 0.00019740371336450234,
+      "loss": 0.0378,
+      "step": 900
+    },
+    {
+      "epoch": 2.4074074074074074,
+      "grad_norm": 0.278036504983902,
+      "learning_rate": 0.00019730373966320402,
+      "loss": 0.0335,
+      "step": 910
+    },
+    {
+      "epoch": 2.433862433862434,
+      "grad_norm": 0.48762863874435425,
+      "learning_rate": 0.000197201903622936,
+      "loss": 0.0357,
+      "step": 920
+    },
+    {
+      "epoch": 2.4603174603174605,
+      "grad_norm": 0.37752997875213623,
+      "learning_rate": 0.000197098207192783,
+      "loss": 0.0358,
+      "step": 930
+    },
+    {
+      "epoch": 2.4867724867724865,
+      "grad_norm": 0.49835503101348877,
+      "learning_rate": 0.00019699265235743638,
+      "loss": 0.0359,
+      "step": 940
+    },
+    {
+      "epoch": 2.5132275132275135,
+      "grad_norm": 0.453590452671051,
+      "learning_rate": 0.00019688524113715657,
+      "loss": 0.0367,
+      "step": 950
+    },
+    {
+      "epoch": 2.5396825396825395,
+      "grad_norm": 0.375345379114151,
+      "learning_rate": 0.00019677597558773397,
+      "loss": 0.0339,
+      "step": 960
+    },
+    {
+      "epoch": 2.566137566137566,
+      "grad_norm": 0.5068901777267456,
+      "learning_rate": 0.0001966648578004499,
+      "loss": 0.0363,
+      "step": 970
+    },
+    {
+      "epoch": 2.5925925925925926,
+      "grad_norm": 0.3547157347202301,
+      "learning_rate": 0.00019655188990203647,
+      "loss": 0.0349,
+      "step": 980
+    },
+    {
+      "epoch": 2.619047619047619,
+      "grad_norm": 0.3495565354824066,
+      "learning_rate": 0.00019643707405463584,
+      "loss": 0.0416,
+      "step": 990
+    },
+    {
+      "epoch": 2.6455026455026456,
+      "grad_norm": 0.24774406850337982,
+      "learning_rate": 0.00019632041245575888,
+      "loss": 0.0339,
+      "step": 1000
+    },
+    {
+      "epoch": 2.671957671957672,
+      "grad_norm": 0.3105117082595825,
+      "learning_rate": 0.00019620190733824315,
+      "loss": 0.034,
+      "step": 1010
+    },
+    {
+      "epoch": 2.6984126984126986,
+      "grad_norm": 0.2921735644340515,
+      "learning_rate": 0.00019608156097021006,
+      "loss": 0.0433,
+      "step": 1020
+    },
+    {
+      "epoch": 2.7248677248677247,
+      "grad_norm": 0.4697803258895874,
+      "learning_rate": 0.00019595937565502157,
+      "loss": 0.0342,
+      "step": 1030
+    },
+    {
+      "epoch": 2.751322751322751,
+      "grad_norm": 0.2624008059501648,
+      "learning_rate": 0.00019583535373123606,
+      "loss": 0.0325,
+      "step": 1040
+    },
+    {
+      "epoch": 2.7777777777777777,
+      "grad_norm": 0.4016210734844208,
+      "learning_rate": 0.00019570949757256355,
+      "loss": 0.0332,
+      "step": 1050
+    },
+    {
+      "epoch": 2.804232804232804,
+      "grad_norm": 0.4004408121109009,
+      "learning_rate": 0.00019558180958782024,
+      "loss": 0.0302,
+      "step": 1060
+    },
+    {
+      "epoch": 2.8306878306878307,
+      "grad_norm": 0.17768095433712006,
+      "learning_rate": 0.0001954522922208825,
+      "loss": 0.0342,
+      "step": 1070
+    },
+    {
+      "epoch": 2.857142857142857,
+      "grad_norm": 0.443602591753006,
+      "learning_rate": 0.00019532094795064015,
+      "loss": 0.0345,
+      "step": 1080
+    },
+    {
+      "epoch": 2.8835978835978837,
+      "grad_norm": 0.3766569197177887,
+      "learning_rate": 0.00019518777929094868,
+      "loss": 0.0338,
+      "step": 1090
+    },
+    {
+      "epoch": 2.91005291005291,
+      "grad_norm": 0.3146336078643799,
+      "learning_rate": 0.00019505278879058157,
+      "loss": 0.0329,
+      "step": 1100
+    },
+    {
+      "epoch": 2.9365079365079367,
+      "grad_norm": 0.44845908880233765,
+      "learning_rate": 0.00019491597903318125,
+      "loss": 0.0318,
+      "step": 1110
+    },
+    {
+      "epoch": 2.962962962962963,
+      "grad_norm": 0.43101152777671814,
+      "learning_rate": 0.00019477735263720964,
+      "loss": 0.0344,
+      "step": 1120
+    },
+    {
+      "epoch": 2.9894179894179893,
+      "grad_norm": 0.3073725998401642,
+      "learning_rate": 0.0001946369122558982,
+      "loss": 0.0363,
+      "step": 1130
+    },
+    {
+      "epoch": 3.015873015873016,
+      "grad_norm": 0.5444226861000061,
+      "learning_rate": 0.00019449466057719703,
+      "loss": 0.0334,
+      "step": 1140
+    },
+    {
+      "epoch": 3.0423280423280423,
+      "grad_norm": 0.5659046173095703,
+      "learning_rate": 0.00019435060032372338,
+      "loss": 0.0319,
+      "step": 1150
+    },
+    {
+      "epoch": 3.068783068783069,
+      "grad_norm": 0.444764643907547,
+      "learning_rate": 0.0001942047342527097,
+      "loss": 0.0316,
+      "step": 1160
+    },
+    {
+      "epoch": 3.0952380952380953,
+      "grad_norm": 0.3891999125480652,
+      "learning_rate": 0.0001940570651559507,
+      "loss": 0.0361,
+      "step": 1170
+    },
+    {
+      "epoch": 3.121693121693122,
+      "grad_norm": 0.3401958644390106,
+      "learning_rate": 0.00019390759585975005,
+      "loss": 0.0336,
+      "step": 1180
+    },
+    {
+      "epoch": 3.148148148148148,
+      "grad_norm": 0.24082757532596588,
+      "learning_rate": 0.00019375632922486615,
+      "loss": 0.0319,
+      "step": 1190
+    },
+    {
+      "epoch": 3.1746031746031744,
+      "grad_norm": 0.34355929493904114,
+      "learning_rate": 0.00019360326814645752,
+      "loss": 0.035,
+      "step": 1200
+    },
+    {
+      "epoch": 3.201058201058201,
+      "grad_norm": 0.3685361444950104,
+      "learning_rate": 0.00019344841555402731,
+      "loss": 0.0307,
+      "step": 1210
+    },
+    {
+      "epoch": 3.2275132275132274,
+      "grad_norm": 0.30369487404823303,
+      "learning_rate": 0.00019329177441136723,
+      "loss": 0.0315,
+      "step": 1220
+    },
+    {
+      "epoch": 3.253968253968254,
+      "grad_norm": 0.27075985074043274,
+      "learning_rate": 0.00019313334771650084,
+      "loss": 0.0348,
+      "step": 1230
+    },
+    {
+      "epoch": 3.2804232804232805,
+      "grad_norm": 0.2865367829799652,
+      "learning_rate": 0.0001929731385016262,
+      "loss": 0.0322,
+      "step": 1240
+    },
+    {
+      "epoch": 3.306878306878307,
+      "grad_norm": 0.3830226957798004,
+      "learning_rate": 0.00019281114983305773,
+      "loss": 0.0312,
+      "step": 1250
+    },
+    {
+      "epoch": 3.3333333333333335,
+      "grad_norm": 0.41300126910209656,
+      "learning_rate": 0.00019264738481116763,
+      "loss": 0.0339,
+      "step": 1260
+    },
+    {
+      "epoch": 3.35978835978836,
+      "grad_norm": 0.2901158332824707,
+      "learning_rate": 0.00019248184657032655,
+      "loss": 0.0317,
+      "step": 1270
+    },
+    {
+      "epoch": 3.386243386243386,
+      "grad_norm": 0.29940736293792725,
+      "learning_rate": 0.00019231453827884352,
+      "loss": 0.035,
+      "step": 1280
+    },
+    {
+      "epoch": 3.4126984126984126,
+      "grad_norm": 0.36355215311050415,
+      "learning_rate": 0.0001921454631389053,
+      "loss": 0.0319,
+      "step": 1290
+    },
+    {
+      "epoch": 3.439153439153439,
+      "grad_norm": 0.34896090626716614,
+      "learning_rate": 0.00019197462438651523,
+      "loss": 0.0333,
+      "step": 1300
+    },
+    {
+      "epoch": 3.4656084656084656,
+      "grad_norm": 0.24769645929336548,
+      "learning_rate": 0.00019180202529143112,
+      "loss": 0.0305,
+      "step": 1310
+    },
+    {
+      "epoch": 3.492063492063492,
+      "grad_norm": 0.2804192006587982,
+      "learning_rate": 0.00019162766915710282,
+      "loss": 0.0266,
+      "step": 1320
+    },
+    {
+      "epoch": 3.5185185185185186,
+      "grad_norm": 0.1939479410648346,
+      "learning_rate": 0.00019145155932060885,
+      "loss": 0.0366,
+      "step": 1330
+    },
+    {
+      "epoch": 3.544973544973545,
+      "grad_norm": 0.3781014084815979,
+      "learning_rate": 0.0001912736991525927,
+      "loss": 0.0313,
+      "step": 1340
+    },
+    {
+      "epoch": 3.571428571428571,
+      "grad_norm": 0.28336915373802185,
+      "learning_rate": 0.00019109409205719812,
+      "loss": 0.0318,
+      "step": 1350
+    },
+    {
+      "epoch": 3.597883597883598,
+      "grad_norm": 0.32090702652931213,
+      "learning_rate": 0.00019091274147200413,
+      "loss": 0.0295,
+      "step": 1360
+    },
+    {
+      "epoch": 3.624338624338624,
+      "grad_norm": 0.4037337601184845,
+      "learning_rate": 0.0001907296508679592,
+      "loss": 0.0312,
+      "step": 1370
+    },
+    {
+      "epoch": 3.6507936507936507,
+      "grad_norm": 0.2988503575325012,
+      "learning_rate": 0.00019054482374931467,
+      "loss": 0.0285,
+      "step": 1380
+    },
+    {
+      "epoch": 3.677248677248677,
+      "grad_norm": 0.3322925567626953,
+      "learning_rate": 0.0001903582636535579,
+      "loss": 0.0307,
+      "step": 1390
+    },
+    {
+      "epoch": 3.7037037037037037,
+      "grad_norm": 0.5639650821685791,
+      "learning_rate": 0.00019016997415134443,
+      "loss": 0.0323,
+      "step": 1400
+    },
+    {
+      "epoch": 3.7301587301587302,
+      "grad_norm": 0.30782419443130493,
+      "learning_rate": 0.00018997995884642967,
+      "loss": 0.0305,
+      "step": 1410
+    },
+    {
+      "epoch": 3.7566137566137567,
+      "grad_norm": 0.36455053091049194,
+      "learning_rate": 0.00018978822137559988,
+      "loss": 0.0327,
+      "step": 1420
+    },
+    {
+      "epoch": 3.7830687830687832,
+      "grad_norm": 0.3742886185646057,
+      "learning_rate": 0.00018959476540860267,
+      "loss": 0.0287,
+      "step": 1430
+    },
+    {
+      "epoch": 3.8095238095238093,
+      "grad_norm": 0.36028748750686646,
+      "learning_rate": 0.00018939959464807672,
+      "loss": 0.0293,
+      "step": 1440
+    },
+    {
+      "epoch": 3.835978835978836,
+      "grad_norm": 0.34717637300491333,
+      "learning_rate": 0.0001892027128294808,
+      "loss": 0.0292,
+      "step": 1450
+    },
+    {
+      "epoch": 3.8624338624338623,
+      "grad_norm": 0.32690536975860596,
+      "learning_rate": 0.00018900412372102244,
+      "loss": 0.0349,
+      "step": 1460
+    },
+    {
+      "epoch": 3.888888888888889,
+      "grad_norm": 0.21768151223659515,
+      "learning_rate": 0.0001888038311235857,
+      "loss": 0.0295,
+      "step": 1470
+    },
+    {
+      "epoch": 3.9153439153439153,
+      "grad_norm": 0.2899485230445862,
+      "learning_rate": 0.00018860183887065856,
+      "loss": 0.0287,
+      "step": 1480
+    },
+    {
+      "epoch": 3.941798941798942,
+      "grad_norm": 0.2762286961078644,
+      "learning_rate": 0.00018839815082825933,
+      "loss": 0.0307,
+      "step": 1490
+    },
+    {
+      "epoch": 3.9682539682539684,
+      "grad_norm": 0.24080318212509155,
+      "learning_rate": 0.00018819277089486281,
+      "loss": 0.0317,
+      "step": 1500
+    },
+    {
+      "epoch": 3.9947089947089944,
+      "grad_norm": 0.25071361660957336,
+      "learning_rate": 0.00018798570300132571,
+      "loss": 0.0335,
+      "step": 1510
+    },
+    {
+      "epoch": 4.021164021164021,
+      "grad_norm": 0.47212567925453186,
+      "learning_rate": 0.00018777695111081132,
+      "loss": 0.0283,
+      "step": 1520
+    },
+    {
+      "epoch": 4.0476190476190474,
+      "grad_norm": 0.25096508860588074,
+      "learning_rate": 0.0001875665192187136,
+      "loss": 0.0339,
+      "step": 1530
+    },
+    {
+      "epoch": 4.074074074074074,
+      "grad_norm": 0.41704872250556946,
+      "learning_rate": 0.00018735441135258097,
+      "loss": 0.0305,
+      "step": 1540
+    },
+    {
+      "epoch": 4.1005291005291005,
+      "grad_norm": 0.3602231442928314,
+      "learning_rate": 0.00018714063157203886,
+      "loss": 0.0339,
+      "step": 1550
+    },
+    {
+      "epoch": 4.1269841269841265,
+      "grad_norm": 0.21357379853725433,
+      "learning_rate": 0.00018692518396871234,
+      "loss": 0.0336,
+      "step": 1560
+    },
+    {
+      "epoch": 4.1534391534391535,
+      "grad_norm": 0.37265297770500183,
+      "learning_rate": 0.00018670807266614757,
+      "loss": 0.0316,
+      "step": 1570
+    },
+    {
+      "epoch": 4.1798941798941796,
+      "grad_norm": 0.32641279697418213,
+      "learning_rate": 0.00018648930181973313,
+      "loss": 0.0298,
+      "step": 1580
+    },
+    {
+      "epoch": 4.2063492063492065,
+      "grad_norm": 0.28364816308021545,
+      "learning_rate": 0.00018626887561662016,
+      "loss": 0.0318,
+      "step": 1590
+    },
+    {
+      "epoch": 4.232804232804233,
+      "grad_norm": 0.38192281126976013,
+      "learning_rate": 0.00018604679827564254,
+      "loss": 0.0321,
+      "step": 1600
+    },
+    {
+      "epoch": 4.2592592592592595,
+      "grad_norm": 0.29732367396354675,
+      "learning_rate": 0.00018582307404723593,
+      "loss": 0.0302,
+      "step": 1610
+    },
+    {
+      "epoch": 4.285714285714286,
+      "grad_norm": 0.32292482256889343,
+      "learning_rate": 0.00018559770721335653,
+      "loss": 0.0296,
+      "step": 1620
+    },
+    {
+      "epoch": 4.3121693121693125,
+      "grad_norm": 0.36669158935546875,
+      "learning_rate": 0.00018537070208739898,
+      "loss": 0.0313,
+      "step": 1630
+    },
+    {
+      "epoch": 4.338624338624339,
+      "grad_norm": 0.32983410358428955,
+      "learning_rate": 0.00018514206301411407,
+      "loss": 0.0293,
+      "step": 1640
+    },
+    {
+      "epoch": 4.365079365079365,
+      "grad_norm": 0.38306760787963867,
+      "learning_rate": 0.00018491179436952532,
+      "loss": 0.0309,
+      "step": 1650
+    },
+    {
+      "epoch": 4.391534391534392,
+      "grad_norm": 0.31832560896873474,
+      "learning_rate": 0.00018467990056084536,
+      "loss": 0.03,
+      "step": 1660
+    },
+    {
+      "epoch": 4.417989417989418,
+      "grad_norm": 0.4297352731227875,
+      "learning_rate": 0.0001844463860263916,
+      "loss": 0.0302,
+      "step": 1670
+    },
+    {
+      "epoch": 4.444444444444445,
+      "grad_norm": 0.22931234538555145,
+      "learning_rate": 0.0001842112552355011,
+      "loss": 0.0318,
+      "step": 1680
+    },
+    {
+      "epoch": 4.470899470899471,
+      "grad_norm": 0.49042898416519165,
+      "learning_rate": 0.00018397451268844524,
+      "loss": 0.0353,
+      "step": 1690
+    },
+    {
+      "epoch": 4.497354497354498,
+      "grad_norm": 0.29628482460975647,
+      "learning_rate": 0.0001837361629163436,
+      "loss": 0.0364,
+      "step": 1700
+    },
+    {
+      "epoch": 4.523809523809524,
+      "grad_norm": 0.3375132977962494,
+      "learning_rate": 0.000183496210481077,
+      "loss": 0.0324,
+      "step": 1710
+    },
+    {
+      "epoch": 4.550264550264551,
+      "grad_norm": 0.2872624397277832,
+      "learning_rate": 0.00018325465997520035,
+      "loss": 0.0284,
+      "step": 1720
+    },
+    {
+      "epoch": 4.576719576719577,
+      "grad_norm": 0.3233286142349243,
+      "learning_rate": 0.0001830115160218548,
+      "loss": 0.0338,
+      "step": 1730
+    },
+    {
+      "epoch": 4.603174603174603,
+      "grad_norm": 0.25356996059417725,
+      "learning_rate": 0.00018276678327467923,
+      "loss": 0.0286,
+      "step": 1740
+    },
+    {
+      "epoch": 4.62962962962963,
+      "grad_norm": 0.21880115568637848,
+      "learning_rate": 0.00018252046641772097,
+      "loss": 0.0335,
+      "step": 1750
+    },
+    {
+      "epoch": 4.656084656084656,
+      "grad_norm": 0.30065304040908813,
+      "learning_rate": 0.0001822725701653465,
+      "loss": 0.0294,
+      "step": 1760
+    },
+    {
+      "epoch": 4.682539682539683,
+      "grad_norm": 0.44395706057548523,
+      "learning_rate": 0.00018202309926215092,
+      "loss": 0.0318,
+      "step": 1770
+    },
+    {
+      "epoch": 4.708994708994709,
+      "grad_norm": 0.4432872235774994,
+      "learning_rate": 0.00018177205848286736,
+      "loss": 0.0332,
+      "step": 1780
+    },
+    {
+      "epoch": 4.735449735449736,
+      "grad_norm": 0.3368462026119232,
+      "learning_rate": 0.00018151945263227543,
+      "loss": 0.0351,
+      "step": 1790
+    },
+    {
+      "epoch": 4.761904761904762,
+      "grad_norm": 0.4848134517669678,
+      "learning_rate": 0.00018126528654510934,
+      "loss": 0.0281,
+      "step": 1800
+    },
+    {
+      "epoch": 4.788359788359788,
+      "grad_norm": 0.339212566614151,
+      "learning_rate": 0.00018100956508596537,
+      "loss": 0.0295,
+      "step": 1810
+    },
+    {
+      "epoch": 4.814814814814815,
+      "grad_norm": 0.37739694118499756,
+      "learning_rate": 0.00018075229314920872,
+      "loss": 0.0293,
+      "step": 1820
+    },
+    {
+      "epoch": 4.841269841269841,
+      "grad_norm": 0.2847217619419098,
+      "learning_rate": 0.00018049347565887987,
+      "loss": 0.0291,
+      "step": 1830
+    },
+    {
+      "epoch": 4.867724867724868,
+      "grad_norm": 0.2565479278564453,
+      "learning_rate": 0.00018023311756860038,
+      "loss": 0.0318,
+      "step": 1840
+    },
+    {
+      "epoch": 4.894179894179894,
+      "grad_norm": 0.2911958396434784,
+      "learning_rate": 0.0001799712238614779,
+      "loss": 0.0293,
+      "step": 1850
+    },
+    {
+      "epoch": 4.920634920634921,
+      "grad_norm": 0.35511329770088196,
+      "learning_rate": 0.0001797077995500111,
+      "loss": 0.0325,
+      "step": 1860
+    },
+    {
+      "epoch": 4.947089947089947,
+      "grad_norm": 0.36546435952186584,
+      "learning_rate": 0.00017944284967599344,
+      "loss": 0.0288,
+      "step": 1870
+    },
+    {
+      "epoch": 4.973544973544973,
+      "grad_norm": 0.24980072677135468,
+      "learning_rate": 0.0001791763793104168,
+      "loss": 0.03,
+      "step": 1880
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.2070397436618805,
+      "learning_rate": 0.00017890839355337443,
+      "loss": 0.0334,
+      "step": 1890
+    },
+    {
+      "epoch": 5.026455026455026,
+      "grad_norm": 0.34367257356643677,
+      "learning_rate": 0.00017863889753396334,
+      "loss": 0.0287,
+      "step": 1900
+    },
+    {
+      "epoch": 5.052910052910053,
+      "grad_norm": 0.24305817484855652,
+      "learning_rate": 0.00017836789641018606,
+      "loss": 0.0297,
+      "step": 1910
+    },
+    {
+      "epoch": 5.079365079365079,
+      "grad_norm": 0.28586849570274353,
+      "learning_rate": 0.0001780953953688521,
+      "loss": 0.0324,
+      "step": 1920
+    },
+    {
+      "epoch": 5.105820105820106,
+      "grad_norm": 0.3378239572048187,
+      "learning_rate": 0.00017782139962547834,
+      "loss": 0.0308,
+      "step": 1930
+    },
+    {
+      "epoch": 5.132275132275132,
+      "grad_norm": 0.26583924889564514,
+      "learning_rate": 0.00017754591442418964,
+      "loss": 0.0298,
+      "step": 1940
+    },
+    {
+      "epoch": 5.158730158730159,
+      "grad_norm": 0.22006455063819885,
+      "learning_rate": 0.0001772689450376181,
+      "loss": 0.026,
+      "step": 1950
+    },
+    {
+      "epoch": 5.185185185185185,
+      "grad_norm": 0.26729926466941833,
+      "learning_rate": 0.00017699049676680236,
+      "loss": 0.0304,
+      "step": 1960
+    },
+    {
+      "epoch": 5.211640211640212,
+      "grad_norm": 0.34223079681396484,
+      "learning_rate": 0.00017671057494108604,
+      "loss": 0.0331,
+      "step": 1970
+    },
+    {
+      "epoch": 5.238095238095238,
+      "grad_norm": 0.18476738035678864,
+      "learning_rate": 0.00017642918491801582,
+      "loss": 0.0318,
+      "step": 1980
+    },
+    {
+      "epoch": 5.264550264550264,
+      "grad_norm": 0.2811238467693329,
+      "learning_rate": 0.00017614633208323877,
+      "loss": 0.029,
+      "step": 1990
+    },
+    {
+      "epoch": 5.291005291005291,
+      "grad_norm": 0.27177298069000244,
+      "learning_rate": 0.00017586202185039952,
+      "loss": 0.0304,
+      "step": 2000
+    },
+    {
+      "epoch": 5.317460317460317,
+      "grad_norm": 0.4007485508918762,
+      "learning_rate": 0.00017557625966103623,
+      "loss": 0.0285,
+      "step": 2010
+    },
+    {
+      "epoch": 5.343915343915344,
+      "grad_norm": 0.2058725655078888,
+      "learning_rate": 0.0001752890509844769,
+      "loss": 0.0296,
+      "step": 2020
+    },
+    {
+      "epoch": 5.37037037037037,
+      "grad_norm": 0.3283293843269348,
+      "learning_rate": 0.00017500040131773444,
+      "loss": 0.0304,
+      "step": 2030
+    },
+    {
+      "epoch": 5.396825396825397,
+      "grad_norm": 0.33090895414352417,
+      "learning_rate": 0.00017471031618540141,
+      "loss": 0.0289,
+      "step": 2040
+    },
+    {
+      "epoch": 5.423280423280423,
+      "grad_norm": 0.3411179780960083,
+      "learning_rate": 0.0001744188011395445,
+      "loss": 0.0306,
+      "step": 2050
+    },
+    {
+      "epoch": 5.449735449735449,
+      "grad_norm": 0.3357543349266052,
+      "learning_rate": 0.0001741258617595981,
+      "loss": 0.0294,
+      "step": 2060
+    },
+    {
+      "epoch": 5.476190476190476,
+      "grad_norm": 0.2584282457828522,
+      "learning_rate": 0.0001738315036522575,
+      "loss": 0.0301,
+      "step": 2070
+    },
+    {
+      "epoch": 5.502645502645502,
+      "grad_norm": 0.2944985628128052,
+      "learning_rate": 0.00017353573245137171,
+      "loss": 0.0256,
+      "step": 2080
+    },
+    {
+      "epoch": 5.529100529100529,
+      "grad_norm": 0.38644468784332275,
+      "learning_rate": 0.00017323855381783551,
+      "loss": 0.032,
+      "step": 2090
+    },
+    {
+      "epoch": 5.555555555555555,
+      "grad_norm": 0.2968820035457611,
+      "learning_rate": 0.00017293997343948117,
+      "loss": 0.0304,
+      "step": 2100
+    },
+    {
+      "epoch": 5.582010582010582,
+      "grad_norm": 0.428739458322525,
+      "learning_rate": 0.00017263999703096957,
+      "loss": 0.0284,
+      "step": 2110
+    },
+    {
+      "epoch": 5.608465608465608,
+      "grad_norm": 0.28147000074386597,
+      "learning_rate": 0.00017233863033368077,
+      "loss": 0.0315,
+      "step": 2120
+    },
+    {
+      "epoch": 5.634920634920634,
+      "grad_norm": 0.3461019992828369,
+      "learning_rate": 0.00017203587911560436,
+      "loss": 0.0313,
+      "step": 2130
+    },
+    {
+      "epoch": 5.661375661375661,
+      "grad_norm": 0.30748844146728516,
+      "learning_rate": 0.0001717317491712286,
+      "loss": 0.0276,
+      "step": 2140
+    },
+    {
+      "epoch": 5.6878306878306875,
+      "grad_norm": 0.3075932264328003,
+      "learning_rate": 0.00017142624632143002,
+      "loss": 0.0308,
+      "step": 2150
+    },
+    {
+      "epoch": 5.714285714285714,
+      "grad_norm": 0.2982342839241028,
+      "learning_rate": 0.0001711193764133617,
+      "loss": 0.0279,
+      "step": 2160
+    },
+    {
+      "epoch": 5.7407407407407405,
+      "grad_norm": 0.3656872510910034,
+      "learning_rate": 0.00017081114532034147,
+      "loss": 0.0278,
+      "step": 2170
+    },
+    {
+      "epoch": 5.767195767195767,
+      "grad_norm": 0.29646244645118713,
+      "learning_rate": 0.0001705015589417395,
+      "loss": 0.0283,
+      "step": 2180
+    },
+    {
+      "epoch": 5.7936507936507935,
+      "grad_norm": 0.2913481891155243,
+      "learning_rate": 0.00017019062320286526,
+      "loss": 0.0269,
+      "step": 2190
+    },
+    {
+      "epoch": 5.8201058201058204,
+      "grad_norm": 0.25158271193504333,
+      "learning_rate": 0.0001698783440548543,
+      "loss": 0.0324,
+      "step": 2200
+    },
+    {
+      "epoch": 5.8465608465608465,
+      "grad_norm": 0.34147316217422485,
+      "learning_rate": 0.0001695647274745544,
+      "loss": 0.0308,
+      "step": 2210
+    },
+    {
+      "epoch": 5.8730158730158735,
+      "grad_norm": 0.23324055969715118,
+      "learning_rate": 0.00016924977946441072,
+      "loss": 0.0309,
+      "step": 2220
+    },
+    {
+      "epoch": 5.8994708994708995,
+      "grad_norm": 0.375702440738678,
+      "learning_rate": 0.0001689335060523515,
+      "loss": 0.0298,
+      "step": 2230
+    },
+    {
+      "epoch": 5.925925925925926,
+      "grad_norm": 0.22085419297218323,
+      "learning_rate": 0.00016861591329167238,
+      "loss": 0.0297,
+      "step": 2240
+    },
+    {
+      "epoch": 5.9523809523809526,
+      "grad_norm": 0.29211702942848206,
+      "learning_rate": 0.0001682970072609205,
+      "loss": 0.0296,
+      "step": 2250
+    },
+    {
+      "epoch": 5.978835978835979,
+      "grad_norm": 0.2560863494873047,
+      "learning_rate": 0.00016797679406377837,
+      "loss": 0.0297,
+      "step": 2260
+    },
+    {
+      "epoch": 6.005291005291006,
+      "grad_norm": 0.22338557243347168,
+      "learning_rate": 0.00016765527982894689,
+      "loss": 0.0308,
+      "step": 2270
+    },
+    {
+      "epoch": 6.031746031746032,
+      "grad_norm": 0.29270821809768677,
+      "learning_rate": 0.00016733247071002802,
+      "loss": 0.0249,
+      "step": 2280
+    },
+    {
+      "epoch": 6.058201058201059,
+      "grad_norm": 0.33310195803642273,
+      "learning_rate": 0.00016700837288540716,
+      "loss": 0.0294,
+      "step": 2290
+    },
+    {
+      "epoch": 6.084656084656085,
+      "grad_norm": 0.28385937213897705,
+      "learning_rate": 0.0001666829925581348,
+      "loss": 0.0277,
+      "step": 2300
+    },
+    {
+      "epoch": 6.111111111111111,
+      "grad_norm": 0.2993493974208832,
+      "learning_rate": 0.0001663563359558078,
+      "loss": 0.0298,
+      "step": 2310
+    },
+    {
+      "epoch": 6.137566137566138,
+      "grad_norm": 0.308386892080307,
+      "learning_rate": 0.00016602840933045018,
+      "loss": 0.0295,
+      "step": 2320
+    },
+    {
+      "epoch": 6.164021164021164,
+      "grad_norm": 0.38717547059059143,
+      "learning_rate": 0.00016569921895839354,
+      "loss": 0.0342,
+      "step": 2330
+    },
+    {
+      "epoch": 6.190476190476191,
+      "grad_norm": 0.2936514914035797,
+      "learning_rate": 0.00016536877114015685,
+      "loss": 0.0278,
+      "step": 2340
+    },
+    {
+      "epoch": 6.216931216931217,
+      "grad_norm": 0.26422831416130066,
+      "learning_rate": 0.00016503707220032586,
+      "loss": 0.0304,
+      "step": 2350
+    },
+    {
+      "epoch": 6.243386243386244,
+      "grad_norm": 0.3753129839897156,
+      "learning_rate": 0.00016470412848743223,
+      "loss": 0.0278,
+      "step": 2360
+    },
+    {
+      "epoch": 6.26984126984127,
+      "grad_norm": 0.33631107211112976,
+      "learning_rate": 0.00016436994637383166,
+      "loss": 0.0279,
+      "step": 2370
+    },
+    {
+      "epoch": 6.296296296296296,
+      "grad_norm": 0.2561590075492859,
+      "learning_rate": 0.00016403453225558235,
+      "loss": 0.0266,
+      "step": 2380
+    },
+    {
+      "epoch": 6.322751322751323,
+      "grad_norm": 0.342010498046875,
+      "learning_rate": 0.00016369789255232223,
+      "loss": 0.0287,
+      "step": 2390
+    },
+    {
+      "epoch": 6.349206349206349,
+      "grad_norm": 0.2974427044391632,
+      "learning_rate": 0.0001633600337071463,
+      "loss": 0.0264,
+      "step": 2400
+    },
+    {
+      "epoch": 6.375661375661376,
+      "grad_norm": 0.24735130369663239,
+      "learning_rate": 0.0001630209621864833,
+      "loss": 0.025,
+      "step": 2410
+    },
+    {
+      "epoch": 6.402116402116402,
+      "grad_norm": 0.3415727913379669,
+      "learning_rate": 0.00016268068447997176,
+      "loss": 0.034,
+      "step": 2420
+    },
+    {
+      "epoch": 6.428571428571429,
+      "grad_norm": 0.21212635934352875,
+      "learning_rate": 0.00016233920710033607,
+      "loss": 0.0297,
+      "step": 2430
+    },
+    {
+      "epoch": 6.455026455026455,
+      "grad_norm": 0.4277040958404541,
+      "learning_rate": 0.00016199653658326168,
+      "loss": 0.0305,
+      "step": 2440
+    },
+    {
+      "epoch": 6.481481481481482,
+      "grad_norm": 0.377903014421463,
+      "learning_rate": 0.00016165267948726987,
+      "loss": 0.0292,
+      "step": 2450
+    },
+    {
+      "epoch": 6.507936507936508,
+      "grad_norm": 0.3249509632587433,
+      "learning_rate": 0.0001613076423935926,
+      "loss": 0.027,
+      "step": 2460
+    },
+    {
+      "epoch": 6.534391534391535,
+      "grad_norm": 0.2562958002090454,
+      "learning_rate": 0.0001609614319060461,
+      "loss": 0.0277,
+      "step": 2470
+    },
+    {
+      "epoch": 6.560846560846561,
+      "grad_norm": 0.24918289482593536,
+      "learning_rate": 0.0001606140546509049,
+      "loss": 0.0263,
+      "step": 2480
+    },
+    {
+      "epoch": 6.587301587301587,
+      "grad_norm": 0.24581879377365112,
+      "learning_rate": 0.00016026551727677465,
+      "loss": 0.0305,
+      "step": 2490
+    },
+    {
+      "epoch": 6.613756613756614,
+      "grad_norm": 0.20334988832473755,
+      "learning_rate": 0.0001599158264544651,
+      "loss": 0.0273,
+      "step": 2500
+    },
+    {
+      "epoch": 6.64021164021164,
+      "grad_norm": 0.20349211990833282,
+      "learning_rate": 0.00015956498887686232,
+      "loss": 0.0301,
+      "step": 2510
+    },
+    {
+      "epoch": 6.666666666666667,
+      "grad_norm": 0.27112945914268494,
+      "learning_rate": 0.0001592130112588007,
+      "loss": 0.027,
+      "step": 2520
+    },
+    {
+      "epoch": 6.693121693121693,
+      "grad_norm": 0.18712319433689117,
+      "learning_rate": 0.00015885990033693434,
+      "loss": 0.0262,
+      "step": 2530
+    },
+    {
+      "epoch": 6.71957671957672,
+      "grad_norm": 0.3410026729106903,
+      "learning_rate": 0.00015850566286960803,
+      "loss": 0.0266,
+      "step": 2540
+    },
+    {
+      "epoch": 6.746031746031746,
+      "grad_norm": 0.29726535081863403,
+      "learning_rate": 0.00015815030563672813,
+      "loss": 0.0286,
+      "step": 2550
+    },
+    {
+      "epoch": 6.772486772486772,
+      "grad_norm": 0.20028139650821686,
+      "learning_rate": 0.00015779383543963271,
+      "loss": 0.0298,
+      "step": 2560
+    },
+    {
+      "epoch": 6.798941798941799,
+      "grad_norm": 0.28673526644706726,
+      "learning_rate": 0.00015743625910096117,
+      "loss": 0.0258,
+      "step": 2570
+    },
+    {
+      "epoch": 6.825396825396825,
+      "grad_norm": 0.3958338499069214,
+      "learning_rate": 0.00015707758346452408,
+      "loss": 0.0286,
+      "step": 2580
+    },
+    {
+      "epoch": 6.851851851851852,
+      "grad_norm": 0.2631092369556427,
+      "learning_rate": 0.00015671781539517175,
+      "loss": 0.0269,
+      "step": 2590
+    },
+    {
+      "epoch": 6.878306878306878,
+      "grad_norm": 0.24416978657245636,
+      "learning_rate": 0.00015635696177866316,
+      "loss": 0.0314,
+      "step": 2600
+    },
+    {
+      "epoch": 6.904761904761905,
+      "grad_norm": 0.4131467640399933,
+      "learning_rate": 0.000155995029521534,
+      "loss": 0.0297,
+      "step": 2610
+    },
+    {
+      "epoch": 6.931216931216931,
+      "grad_norm": 0.42825278639793396,
+      "learning_rate": 0.00015563202555096453,
+      "loss": 0.0277,
+      "step": 2620
+    },
+    {
+      "epoch": 6.957671957671957,
+      "grad_norm": 0.2998770475387573,
+      "learning_rate": 0.00015526795681464713,
+      "loss": 0.027,
+      "step": 2630
+    },
+    {
+      "epoch": 6.984126984126984,
+      "grad_norm": 0.35698017477989197,
+      "learning_rate": 0.00015490283028065296,
+      "loss": 0.032,
+      "step": 2640
+    },
+    {
+      "epoch": 7.01058201058201,
+      "grad_norm": 0.3422069847583771,
+      "learning_rate": 0.00015453665293729905,
+      "loss": 0.029,
+      "step": 2650
+    },
+    {
+      "epoch": 7.037037037037037,
+      "grad_norm": 0.3283518850803375,
+      "learning_rate": 0.00015416943179301422,
+      "loss": 0.0285,
+      "step": 2660
+    },
+    {
+      "epoch": 7.063492063492063,
+      "grad_norm": 0.17980273067951202,
+      "learning_rate": 0.00015380117387620506,
+      "loss": 0.0262,
+      "step": 2670
+    },
+    {
+      "epoch": 7.08994708994709,
+      "grad_norm": 0.22317390143871307,
+      "learning_rate": 0.00015343188623512152,
+      "loss": 0.0292,
+      "step": 2680
+    },
+    {
+      "epoch": 7.116402116402116,
+      "grad_norm": 0.2997380793094635,
+      "learning_rate": 0.0001530615759377217,
+      "loss": 0.0271,
+      "step": 2690
+    },
+    {
+      "epoch": 7.142857142857143,
+      "grad_norm": 0.3536098301410675,
+      "learning_rate": 0.000152690250071537,
+      "loss": 0.0282,
+      "step": 2700
+    },
+    {
+      "epoch": 7.169312169312169,
+      "grad_norm": 0.23207667469978333,
+      "learning_rate": 0.00015231791574353594,
+      "loss": 0.0258,
+      "step": 2710
+    },
+    {
+      "epoch": 7.195767195767195,
+      "grad_norm": 0.4378697872161865,
+      "learning_rate": 0.00015194458007998876,
+      "loss": 0.0262,
+      "step": 2720
+    },
+    {
+      "epoch": 7.222222222222222,
+      "grad_norm": 0.28229495882987976,
+      "learning_rate": 0.00015157025022633042,
+      "loss": 0.0285,
+      "step": 2730
+    },
+    {
+      "epoch": 7.248677248677248,
+      "grad_norm": 0.3776451647281647,
+      "learning_rate": 0.00015119493334702432,
+      "loss": 0.0258,
+      "step": 2740
+    },
+    {
+      "epoch": 7.275132275132275,
+      "grad_norm": 0.37293151021003723,
+      "learning_rate": 0.00015081863662542487,
+      "loss": 0.0273,
+      "step": 2750
+    },
+    {
+      "epoch": 7.301587301587301,
+      "grad_norm": 0.19970954954624176,
+      "learning_rate": 0.00015044136726364015,
+      "loss": 0.0262,
+      "step": 2760
+    },
+    {
+      "epoch": 7.328042328042328,
+      "grad_norm": 0.3114972412586212,
+      "learning_rate": 0.000150063132482394,
+      "loss": 0.0275,
+      "step": 2770
+    },
+    {
+      "epoch": 7.354497354497354,
+      "grad_norm": 0.2646826207637787,
+      "learning_rate": 0.0001496839395208879,
+      "loss": 0.0282,
+      "step": 2780
+    },
+    {
+      "epoch": 7.380952380952381,
+      "grad_norm": 0.4848107397556305,
+      "learning_rate": 0.0001493037956366623,
+      "loss": 0.0276,
+      "step": 2790
+    },
+    {
+      "epoch": 7.407407407407407,
+      "grad_norm": 0.4195409119129181,
+      "learning_rate": 0.0001489227081054578,
+      "loss": 0.0286,
+      "step": 2800
+    },
+    {
+      "epoch": 7.4338624338624335,
+      "grad_norm": 0.2596321702003479,
+      "learning_rate": 0.00014854068422107589,
+      "loss": 0.0288,
+      "step": 2810
+    },
+    {
+      "epoch": 7.4603174603174605,
+      "grad_norm": 0.20885036885738373,
+      "learning_rate": 0.00014815773129523934,
+      "loss": 0.0262,
+      "step": 2820
+    },
+    {
+      "epoch": 7.4867724867724865,
+      "grad_norm": 0.32868385314941406,
+      "learning_rate": 0.0001477738566574522,
+      "loss": 0.0255,
+      "step": 2830
+    },
+    {
+      "epoch": 7.5132275132275135,
+      "grad_norm": 0.25176772475242615,
+      "learning_rate": 0.00014738906765485963,
+      "loss": 0.0263,
+      "step": 2840
+    },
+    {
+      "epoch": 7.5396825396825395,
+      "grad_norm": 0.265190988779068,
+      "learning_rate": 0.0001470033716521072,
+      "loss": 0.0268,
+      "step": 2850
+    },
+    {
+      "epoch": 7.5661375661375665,
+      "grad_norm": 0.3301750123500824,
+      "learning_rate": 0.00014661677603119993,
+      "loss": 0.0288,
+      "step": 2860
+    },
+    {
+      "epoch": 7.592592592592593,
+      "grad_norm": 0.2590659260749817,
+      "learning_rate": 0.00014622928819136105,
+      "loss": 0.0287,
+      "step": 2870
+    },
+    {
+      "epoch": 7.619047619047619,
+      "grad_norm": 0.2190064787864685,
+      "learning_rate": 0.00014584091554889034,
+      "loss": 0.025,
+      "step": 2880
+    },
+    {
+      "epoch": 7.645502645502646,
+      "grad_norm": 0.21431300044059753,
+      "learning_rate": 0.00014545166553702222,
+      "loss": 0.0267,
+      "step": 2890
+    },
+    {
+      "epoch": 7.671957671957672,
+      "grad_norm": 0.25257354974746704,
+      "learning_rate": 0.00014506154560578352,
+      "loss": 0.0251,
+      "step": 2900
+    },
+    {
+      "epoch": 7.698412698412699,
+      "grad_norm": 0.23436136543750763,
+      "learning_rate": 0.0001446705632218507,
+      "loss": 0.0259,
+      "step": 2910
+    },
+    {
+      "epoch": 7.724867724867725,
+      "grad_norm": 0.294412225484848,
+      "learning_rate": 0.00014427872586840725,
+      "loss": 0.0274,
+      "step": 2920
+    },
+    {
+      "epoch": 7.751322751322752,
+      "grad_norm": 0.29302552342414856,
+      "learning_rate": 0.00014388604104500016,
+      "loss": 0.0286,
+      "step": 2930
+    },
+    {
+      "epoch": 7.777777777777778,
+      "grad_norm": 0.3188011646270752,
+      "learning_rate": 0.00014349251626739656,
+      "loss": 0.0304,
+      "step": 2940
+    },
+    {
+      "epoch": 7.804232804232804,
+      "grad_norm": 0.28278741240501404,
+      "learning_rate": 0.00014309815906743994,
+      "loss": 0.0258,
+      "step": 2950
+    },
+    {
+      "epoch": 7.830687830687831,
+      "grad_norm": 0.311074823141098,
+      "learning_rate": 0.00014270297699290566,
+      "loss": 0.0276,
+      "step": 2960
+    },
+    {
+      "epoch": 7.857142857142857,
+      "grad_norm": 0.2844277322292328,
+      "learning_rate": 0.000142306977607357,
+      "loss": 0.0248,
+      "step": 2970
+    },
+    {
+      "epoch": 7.883597883597884,
+      "grad_norm": 0.3190961182117462,
+      "learning_rate": 0.0001419101684899998,
+      "loss": 0.027,
+      "step": 2980
+    },
+    {
+      "epoch": 7.91005291005291,
+      "grad_norm": 0.24902918934822083,
+      "learning_rate": 0.00014151255723553795,
+      "loss": 0.0262,
+      "step": 2990
+    },
+    {
+      "epoch": 7.936507936507937,
+      "grad_norm": 0.2748497724533081,
+      "learning_rate": 0.00014111415145402772,
+      "loss": 0.027,
+      "step": 3000
+    },
+    {
+      "epoch": 7.962962962962963,
+      "grad_norm": 0.24207575619220734,
+      "learning_rate": 0.00014071495877073222,
+      "loss": 0.0254,
+      "step": 3010
+    },
+    {
+      "epoch": 7.98941798941799,
+      "grad_norm": 0.19475680589675903,
+      "learning_rate": 0.00014031498682597532,
+      "loss": 0.0257,
+      "step": 3020
+    },
+    {
+      "epoch": 8.015873015873016,
+      "grad_norm": 0.2495572566986084,
+      "learning_rate": 0.00013991424327499557,
+      "loss": 0.023,
+      "step": 3030
+    },
+    {
+      "epoch": 8.042328042328043,
+      "grad_norm": 0.20358052849769592,
+      "learning_rate": 0.00013951273578779972,
+      "loss": 0.029,
+      "step": 3040
+    },
+    {
+      "epoch": 8.068783068783068,
+      "grad_norm": 0.27466002106666565,
+      "learning_rate": 0.0001391104720490156,
+      "loss": 0.0255,
+      "step": 3050
+    },
+    {
+      "epoch": 8.095238095238095,
+      "grad_norm": 0.22618280351161957,
+      "learning_rate": 0.00013870745975774559,
+      "loss": 0.0263,
+      "step": 3060
+    },
+    {
+      "epoch": 8.121693121693122,
+      "grad_norm": 0.23149581253528595,
+      "learning_rate": 0.0001383037066274186,
+      "loss": 0.0242,
+      "step": 3070
+    },
+    {
+      "epoch": 8.148148148148149,
+      "grad_norm": 0.32176461815834045,
+      "learning_rate": 0.0001378992203856431,
+      "loss": 0.0237,
+      "step": 3080
+    },
+    {
+      "epoch": 8.174603174603174,
+      "grad_norm": 0.2506023347377777,
+      "learning_rate": 0.00013749400877405865,
+      "loss": 0.0251,
+      "step": 3090
+    },
+    {
+      "epoch": 8.201058201058201,
+      "grad_norm": 0.18478137254714966,
+      "learning_rate": 0.00013708807954818819,
+      "loss": 0.0278,
+      "step": 3100
+    },
+    {
+      "epoch": 8.227513227513228,
+      "grad_norm": 0.27599525451660156,
+      "learning_rate": 0.0001366814404772892,
+      "loss": 0.0254,
+      "step": 3110
+    },
+    {
+      "epoch": 8.253968253968253,
+      "grad_norm": 0.2510387897491455,
+      "learning_rate": 0.00013627409934420534,
+      "loss": 0.0249,
+      "step": 3120
+    },
+    {
+      "epoch": 8.28042328042328,
+      "grad_norm": 0.2076123058795929,
+      "learning_rate": 0.0001358660639452173,
+      "loss": 0.0269,
+      "step": 3130
+    },
+    {
+      "epoch": 8.306878306878307,
+      "grad_norm": 0.23005664348602295,
+      "learning_rate": 0.0001354573420898936,
+      "loss": 0.0274,
+      "step": 3140
+    },
+    {
+      "epoch": 8.333333333333334,
+      "grad_norm": 0.1964070051908493,
+      "learning_rate": 0.00013504794160094116,
+      "loss": 0.0265,
+      "step": 3150
+    },
+    {
+      "epoch": 8.359788359788359,
+      "grad_norm": 0.22786502540111542,
+      "learning_rate": 0.0001346378703140556,
+      "loss": 0.0253,
+      "step": 3160
+    },
+    {
+      "epoch": 8.386243386243386,
+      "grad_norm": 0.32075852155685425,
+      "learning_rate": 0.00013422713607777123,
+      "loss": 0.0254,
+      "step": 3170
+    },
+    {
+      "epoch": 8.412698412698413,
+      "grad_norm": 0.40378686785697937,
+      "learning_rate": 0.00013381574675331082,
+      "loss": 0.0251,
+      "step": 3180
+    },
+    {
+      "epoch": 8.43915343915344,
+      "grad_norm": 0.29405686259269714,
+      "learning_rate": 0.00013340371021443515,
+      "loss": 0.0255,
+      "step": 3190
+    },
+    {
+      "epoch": 8.465608465608465,
+      "grad_norm": 0.2700364291667938,
+      "learning_rate": 0.0001329910343472924,
+      "loss": 0.0275,
+      "step": 3200
+    },
+    {
+      "epoch": 8.492063492063492,
+      "grad_norm": 0.29450562596321106,
+      "learning_rate": 0.00013257772705026693,
+      "loss": 0.026,
+      "step": 3210
+    },
+    {
+      "epoch": 8.518518518518519,
+      "grad_norm": 0.3529842793941498,
+      "learning_rate": 0.0001321637962338286,
+      "loss": 0.025,
+      "step": 3220
+    },
+    {
+      "epoch": 8.544973544973544,
+      "grad_norm": 0.35684412717819214,
+      "learning_rate": 0.00013174924982038087,
+      "loss": 0.0255,
+      "step": 3230
+    },
+    {
+      "epoch": 8.571428571428571,
+      "grad_norm": 0.2878738343715668,
+      "learning_rate": 0.00013133409574410943,
+      "loss": 0.0266,
+      "step": 3240
+    },
+    {
+      "epoch": 8.597883597883598,
+      "grad_norm": 0.2095337063074112,
+      "learning_rate": 0.00013091834195083036,
+      "loss": 0.0267,
+      "step": 3250
+    },
+    {
+      "epoch": 8.624338624338625,
+      "grad_norm": 0.337873637676239,
+      "learning_rate": 0.0001305019963978379,
+      "loss": 0.0252,
+      "step": 3260
+    },
+    {
+      "epoch": 8.65079365079365,
+      "grad_norm": 0.191969633102417,
+      "learning_rate": 0.0001300850670537523,
+      "loss": 0.028,
+      "step": 3270
+    },
+    {
+      "epoch": 8.677248677248677,
+      "grad_norm": 0.36041873693466187,
+      "learning_rate": 0.00012966756189836725,
+      "loss": 0.0254,
+      "step": 3280
+    },
+    {
+      "epoch": 8.703703703703704,
+      "grad_norm": 0.3025069832801819,
+      "learning_rate": 0.00012924948892249706,
+      "loss": 0.0309,
+      "step": 3290
+    },
+    {
+      "epoch": 8.73015873015873,
+      "grad_norm": 0.2568103075027466,
+      "learning_rate": 0.00012883085612782388,
+      "loss": 0.0251,
+      "step": 3300
+    },
+    {
+      "epoch": 8.756613756613756,
+      "grad_norm": 0.32407626509666443,
+      "learning_rate": 0.0001284116715267445,
+      "loss": 0.0248,
+      "step": 3310
+    },
+    {
+      "epoch": 8.783068783068783,
+      "grad_norm": 0.28700342774391174,
+      "learning_rate": 0.00012799194314221687,
+      "loss": 0.0281,
+      "step": 3320
+    },
+    {
+      "epoch": 8.80952380952381,
+      "grad_norm": 0.2833905816078186,
+      "learning_rate": 0.00012757167900760678,
+      "loss": 0.0234,
+      "step": 3330
+    },
+    {
+      "epoch": 8.835978835978835,
+      "grad_norm": 0.24850112199783325,
+      "learning_rate": 0.00012715088716653387,
+      "loss": 0.0264,
+      "step": 3340
+    },
+    {
+      "epoch": 8.862433862433862,
+      "grad_norm": 0.30938786268234253,
+      "learning_rate": 0.00012672957567271784,
+      "loss": 0.0246,
+      "step": 3350
+    },
+    {
+      "epoch": 8.88888888888889,
+      "grad_norm": 0.2675461769104004,
+      "learning_rate": 0.00012630775258982424,
+      "loss": 0.0244,
+      "step": 3360
+    },
+    {
+      "epoch": 8.915343915343914,
+      "grad_norm": 0.20268189907073975,
+      "learning_rate": 0.00012588542599131013,
+      "loss": 0.0279,
+      "step": 3370
+    },
+    {
+      "epoch": 8.941798941798941,
+      "grad_norm": 0.21915365755558014,
+      "learning_rate": 0.00012546260396026964,
+      "loss": 0.0225,
+      "step": 3380
+    },
+    {
+      "epoch": 8.968253968253968,
+      "grad_norm": 0.20165018737316132,
+      "learning_rate": 0.00012503929458927912,
+      "loss": 0.0229,
+      "step": 3390
+    },
+    {
+      "epoch": 8.994708994708995,
+      "grad_norm": 0.33086565136909485,
+      "learning_rate": 0.00012461550598024234,
+      "loss": 0.0264,
+      "step": 3400
+    },
+    {
+      "epoch": 9.02116402116402,
+      "grad_norm": 0.24809397757053375,
+      "learning_rate": 0.00012419124624423548,
+      "loss": 0.0242,
+      "step": 3410
+    },
+    {
+      "epoch": 9.047619047619047,
+      "grad_norm": 0.24256815016269684,
+      "learning_rate": 0.0001237665235013518,
+      "loss": 0.0215,
+      "step": 3420
+    },
+    {
+      "epoch": 9.074074074074074,
+      "grad_norm": 0.1913849264383316,
+      "learning_rate": 0.00012334134588054625,
+      "loss": 0.0237,
+      "step": 3430
+    },
+    {
+      "epoch": 9.100529100529101,
+      "grad_norm": 0.34869614243507385,
+      "learning_rate": 0.00012291572151947982,
+      "loss": 0.0266,
+      "step": 3440
+    },
+    {
+      "epoch": 9.126984126984127,
+      "grad_norm": 0.2579362690448761,
+      "learning_rate": 0.00012248965856436394,
+      "loss": 0.0239,
+      "step": 3450
+    },
+    {
+      "epoch": 9.153439153439153,
+      "grad_norm": 0.27626147866249084,
+      "learning_rate": 0.0001220631651698045,
+      "loss": 0.027,
+      "step": 3460
+    },
+    {
+      "epoch": 9.17989417989418,
+      "grad_norm": 0.2195921093225479,
+      "learning_rate": 0.00012163624949864569,
+      "loss": 0.0231,
+      "step": 3470
+    },
+    {
+      "epoch": 9.206349206349206,
+      "grad_norm": 0.2434544414281845,
+      "learning_rate": 0.00012120891972181387,
+      "loss": 0.0262,
+      "step": 3480
+    },
+    {
+      "epoch": 9.232804232804233,
+      "grad_norm": 0.31113576889038086,
+      "learning_rate": 0.00012078118401816119,
+      "loss": 0.0262,
+      "step": 3490
+    },
+    {
+      "epoch": 9.25925925925926,
+      "grad_norm": 0.24409563839435577,
+      "learning_rate": 0.00012035305057430888,
+      "loss": 0.0263,
+      "step": 3500
+    },
+    {
+      "epoch": 9.285714285714286,
+      "grad_norm": 0.2240586131811142,
+      "learning_rate": 0.00011992452758449084,
+      "loss": 0.0243,
+      "step": 3510
+    },
+    {
+      "epoch": 9.312169312169312,
+      "grad_norm": 0.306912362575531,
+      "learning_rate": 0.00011949562325039665,
+      "loss": 0.0259,
+      "step": 3520
+    },
+    {
+      "epoch": 9.338624338624339,
+      "grad_norm": 0.18710291385650635,
+      "learning_rate": 0.00011906634578101447,
+      "loss": 0.0252,
+      "step": 3530
+    },
+    {
+      "epoch": 9.365079365079366,
+      "grad_norm": 0.2894376218318939,
+      "learning_rate": 0.0001186367033924742,
+      "loss": 0.0239,
+      "step": 3540
+    },
+    {
+      "epoch": 9.39153439153439,
+      "grad_norm": 0.29621022939682007,
+      "learning_rate": 0.00011820670430789004,
+      "loss": 0.0237,
+      "step": 3550
+    },
+    {
+      "epoch": 9.417989417989418,
+      "grad_norm": 0.2515980303287506,
+      "learning_rate": 0.00011777635675720313,
+      "loss": 0.0243,
+      "step": 3560
+    },
+    {
+      "epoch": 9.444444444444445,
+      "grad_norm": 0.27100127935409546,
+      "learning_rate": 0.00011734566897702412,
+      "loss": 0.0239,
+      "step": 3570
+    },
+    {
+      "epoch": 9.470899470899472,
+      "grad_norm": 0.3107968270778656,
+      "learning_rate": 0.00011691464921047536,
+      "loss": 0.0259,
+      "step": 3580
+    },
+    {
+      "epoch": 9.497354497354497,
+      "grad_norm": 0.2154056280851364,
+      "learning_rate": 0.00011648330570703331,
+      "loss": 0.025,
+      "step": 3590
+    },
+    {
+      "epoch": 9.523809523809524,
+      "grad_norm": 0.25029587745666504,
+      "learning_rate": 0.00011605164672237059,
+      "loss": 0.0253,
+      "step": 3600
+    },
+    {
+      "epoch": 9.55026455026455,
+      "grad_norm": 0.3177552819252014,
+      "learning_rate": 0.0001156196805181979,
+      "loss": 0.0244,
+      "step": 3610
+    },
+    {
+      "epoch": 9.576719576719576,
+      "grad_norm": 0.2663916051387787,
+      "learning_rate": 0.00011518741536210594,
+      "loss": 0.0231,
+      "step": 3620
+    },
+    {
+      "epoch": 9.603174603174603,
+      "grad_norm": 0.2276095300912857,
+      "learning_rate": 0.00011475485952740718,
+      "loss": 0.0224,
+      "step": 3630
+    },
+    {
+      "epoch": 9.62962962962963,
+      "grad_norm": 0.30287131667137146,
+      "learning_rate": 0.0001143220212929776,
+      "loss": 0.0248,
+      "step": 3640
+    },
+    {
+      "epoch": 9.656084656084657,
+      "grad_norm": 0.2559273838996887,
+      "learning_rate": 0.00011388890894309812,
+      "loss": 0.0242,
+      "step": 3650
+    },
+    {
+      "epoch": 9.682539682539682,
+      "grad_norm": 0.3561594486236572,
+      "learning_rate": 0.000113455530767296,
+      "loss": 0.0258,
+      "step": 3660
+    },
+    {
+      "epoch": 9.708994708994709,
+      "grad_norm": 0.24624326825141907,
+      "learning_rate": 0.0001130218950601863,
+      "loss": 0.0258,
+      "step": 3670
+    },
+    {
+      "epoch": 9.735449735449736,
+      "grad_norm": 0.1778760552406311,
+      "learning_rate": 0.00011258801012131313,
+      "loss": 0.0224,
+      "step": 3680
+    },
+    {
+      "epoch": 9.761904761904763,
+      "grad_norm": 0.3034067153930664,
+      "learning_rate": 0.00011215388425499069,
+      "loss": 0.0247,
+      "step": 3690
+    },
+    {
+      "epoch": 9.788359788359788,
+      "grad_norm": 0.1993577629327774,
+      "learning_rate": 0.00011171952577014448,
+      "loss": 0.0253,
+      "step": 3700
+    },
+    {
+      "epoch": 9.814814814814815,
+      "grad_norm": 0.27741366624832153,
+      "learning_rate": 0.00011128494298015208,
+      "loss": 0.0222,
+      "step": 3710
+    },
+    {
+      "epoch": 9.841269841269842,
+      "grad_norm": 0.2273164540529251,
+      "learning_rate": 0.00011085014420268423,
+      "loss": 0.0239,
+      "step": 3720
+    },
+    {
+      "epoch": 9.867724867724867,
+      "grad_norm": 0.28204792737960815,
+      "learning_rate": 0.00011041513775954549,
+      "loss": 0.0229,
+      "step": 3730
+    },
+    {
+      "epoch": 9.894179894179894,
+      "grad_norm": 0.18121418356895447,
+      "learning_rate": 0.00010997993197651508,
+      "loss": 0.024,
+      "step": 3740
+    },
+    {
+      "epoch": 9.920634920634921,
+      "grad_norm": 0.20097821950912476,
+      "learning_rate": 0.00010954453518318746,
+      "loss": 0.0233,
+      "step": 3750
+    },
+    {
+      "epoch": 9.947089947089948,
+      "grad_norm": 0.22093629837036133,
+      "learning_rate": 0.00010910895571281286,
+      "loss": 0.0227,
+      "step": 3760
+    },
+    {
+      "epoch": 9.973544973544973,
+      "grad_norm": 0.19535768032073975,
+      "learning_rate": 0.00010867320190213792,
+      "loss": 0.0225,
+      "step": 3770
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 0.2424137443304062,
+      "learning_rate": 0.00010823728209124603,
+      "loss": 0.0212,
+      "step": 3780
+    },
+    {
+      "epoch": 10.026455026455027,
+      "grad_norm": 0.23939844965934753,
+      "learning_rate": 0.00010780120462339775,
+      "loss": 0.0249,
+      "step": 3790
+    },
+    {
+      "epoch": 10.052910052910052,
+      "grad_norm": 0.2456304281949997,
+      "learning_rate": 0.00010736497784487104,
+      "loss": 0.0228,
+      "step": 3800
+    },
+    {
+      "epoch": 10.079365079365079,
+      "grad_norm": 0.26618778705596924,
+      "learning_rate": 0.00010692861010480166,
+      "loss": 0.0202,
+      "step": 3810
+    },
+    {
+      "epoch": 10.105820105820106,
+      "grad_norm": 0.29209059476852417,
+      "learning_rate": 0.00010649210975502324,
+      "loss": 0.0217,
+      "step": 3820
+    },
+    {
+      "epoch": 10.132275132275133,
+      "grad_norm": 0.21608638763427734,
+      "learning_rate": 0.00010605548514990753,
+      "loss": 0.022,
+      "step": 3830
+    },
+    {
+      "epoch": 10.158730158730158,
+      "grad_norm": 0.21823786199092865,
+      "learning_rate": 0.00010561874464620447,
+      "loss": 0.0222,
+      "step": 3840
+    },
+    {
+      "epoch": 10.185185185185185,
+      "grad_norm": 0.21151448786258698,
+      "learning_rate": 0.00010518189660288209,
+      "loss": 0.0232,
+      "step": 3850
+    },
+    {
+      "epoch": 10.211640211640212,
+      "grad_norm": 0.20828621089458466,
+      "learning_rate": 0.00010474494938096687,
+      "loss": 0.0236,
+      "step": 3860
+    },
+    {
+      "epoch": 10.238095238095237,
+      "grad_norm": 0.29258930683135986,
+      "learning_rate": 0.0001043079113433834,
+      "loss": 0.0242,
+      "step": 3870
+    },
+    {
+      "epoch": 10.264550264550264,
+      "grad_norm": 0.23913761973381042,
+      "learning_rate": 0.00010387079085479443,
+      "loss": 0.0222,
+      "step": 3880
+    },
+    {
+      "epoch": 10.291005291005291,
+      "grad_norm": 0.3311878740787506,
+      "learning_rate": 0.00010343359628144082,
+      "loss": 0.0235,
+      "step": 3890
+    },
+    {
+      "epoch": 10.317460317460318,
+      "grad_norm": 0.1927531361579895,
+      "learning_rate": 0.00010299633599098135,
+      "loss": 0.0229,
+      "step": 3900
+    },
+    {
+      "epoch": 10.343915343915343,
+      "grad_norm": 0.18032406270503998,
+      "learning_rate": 0.00010255901835233254,
+      "loss": 0.0233,
+      "step": 3910
+    },
+    {
+      "epoch": 10.37037037037037,
+      "grad_norm": 0.2483786642551422,
+      "learning_rate": 0.00010212165173550863,
+      "loss": 0.0238,
+      "step": 3920
+    },
+    {
+      "epoch": 10.396825396825397,
+      "grad_norm": 0.20934239029884338,
+      "learning_rate": 0.00010168424451146119,
+      "loss": 0.0222,
+      "step": 3930
+    },
+    {
+      "epoch": 10.423280423280424,
+      "grad_norm": 0.2310916781425476,
+      "learning_rate": 0.00010124680505191901,
+      "loss": 0.0233,
+      "step": 3940
+    },
+    {
+      "epoch": 10.44973544973545,
+      "grad_norm": 0.2244623601436615,
+      "learning_rate": 0.00010080934172922784,
+      "loss": 0.0242,
+      "step": 3950
+    },
+    {
+      "epoch": 10.476190476190476,
+      "grad_norm": 0.19429625570774078,
+      "learning_rate": 0.00010037186291619017,
+      "loss": 0.0244,
+      "step": 3960
+    },
+    {
+      "epoch": 10.502645502645503,
+      "grad_norm": 0.3608129918575287,
+      "learning_rate": 9.9934376985905e-05,
+      "loss": 0.0258,
+      "step": 3970
+    },
+    {
+      "epoch": 10.529100529100528,
+      "grad_norm": 0.25932079553604126,
+      "learning_rate": 9.94968923116075e-05,
+      "loss": 0.0279,
+      "step": 3980
+    },
+    {
+      "epoch": 10.555555555555555,
+      "grad_norm": 0.2552861273288727,
+      "learning_rate": 9.90594172665088e-05,
+      "loss": 0.0257,
+      "step": 3990
+    },
+    {
+      "epoch": 10.582010582010582,
+      "grad_norm": 0.24272707104682922,
+      "learning_rate": 9.86219602236357e-05,
+      "loss": 0.0238,
+      "step": 4000
+    },
+    {
+      "epoch": 10.60846560846561,
+      "grad_norm": 0.33407488465309143,
+      "learning_rate": 9.818452955567063e-05,
+      "loss": 0.0243,
+      "step": 4010
+    },
+    {
+      "epoch": 10.634920634920634,
+      "grad_norm": 0.25362735986709595,
+      "learning_rate": 9.774713363479099e-05,
+      "loss": 0.0228,
+      "step": 4020
+    },
+    {
+      "epoch": 10.661375661375661,
+      "grad_norm": 0.2759397029876709,
+      "learning_rate": 9.730978083250923e-05,
+      "loss": 0.021,
+      "step": 4030
+    },
+    {
+      "epoch": 10.687830687830688,
+      "grad_norm": 0.21675390005111694,
+      "learning_rate": 9.68724795195126e-05,
+      "loss": 0.0213,
+      "step": 4040
+    },
+    {
+      "epoch": 10.714285714285714,
+      "grad_norm": 0.23203110694885254,
+      "learning_rate": 9.643523806550278e-05,
+      "loss": 0.0206,
+      "step": 4050
+    },
+    {
+      "epoch": 10.74074074074074,
+      "grad_norm": 0.20020541548728943,
+      "learning_rate": 9.59980648390358e-05,
+      "loss": 0.0233,
+      "step": 4060
+    },
+    {
+      "epoch": 10.767195767195767,
+      "grad_norm": 0.26615190505981445,
+      "learning_rate": 9.556096820736193e-05,
+      "loss": 0.0214,
+      "step": 4070
+    },
+    {
+      "epoch": 10.793650793650794,
+      "grad_norm": 0.22349591553211212,
+      "learning_rate": 9.512395653626533e-05,
+      "loss": 0.0241,
+      "step": 4080
+    },
+    {
+      "epoch": 10.82010582010582,
+      "grad_norm": 0.2637469172477722,
+      "learning_rate": 9.468703818990412e-05,
+      "loss": 0.0221,
+      "step": 4090
+    },
+    {
+      "epoch": 10.846560846560847,
+      "grad_norm": 0.21997188031673431,
+      "learning_rate": 9.425022153065034e-05,
+      "loss": 0.0217,
+      "step": 4100
+    },
+    {
+      "epoch": 10.873015873015873,
+      "grad_norm": 0.2709382474422455,
+      "learning_rate": 9.381351491892966e-05,
+      "loss": 0.0238,
+      "step": 4110
+    },
+    {
+      "epoch": 10.899470899470899,
+      "grad_norm": 0.29111942648887634,
+      "learning_rate": 9.33769267130615e-05,
+      "loss": 0.0222,
+      "step": 4120
+    },
+    {
+      "epoch": 10.925925925925926,
+      "grad_norm": 0.2430836260318756,
+      "learning_rate": 9.294046526909917e-05,
+      "loss": 0.0203,
+      "step": 4130
+    },
+    {
+      "epoch": 10.952380952380953,
+      "grad_norm": 0.29842931032180786,
+      "learning_rate": 9.250413894066974e-05,
+      "loss": 0.0263,
+      "step": 4140
+    },
+    {
+      "epoch": 10.97883597883598,
+      "grad_norm": 0.21202361583709717,
+      "learning_rate": 9.206795607881419e-05,
+      "loss": 0.0206,
+      "step": 4150
+    },
+    {
+      "epoch": 11.005291005291005,
+      "grad_norm": 0.2802582085132599,
+      "learning_rate": 9.163192503182789e-05,
+      "loss": 0.0239,
+      "step": 4160
+    },
+    {
+      "epoch": 11.031746031746032,
+      "grad_norm": 0.176875039935112,
+      "learning_rate": 9.119605414510021e-05,
+      "loss": 0.0228,
+      "step": 4170
+    },
+    {
+      "epoch": 11.058201058201059,
+      "grad_norm": 0.24896326661109924,
+      "learning_rate": 9.076035176095533e-05,
+      "loss": 0.0232,
+      "step": 4180
+    },
+    {
+      "epoch": 11.084656084656086,
+      "grad_norm": 0.17039614915847778,
+      "learning_rate": 9.032482621849244e-05,
+      "loss": 0.0202,
+      "step": 4190
+    },
+    {
+      "epoch": 11.11111111111111,
+      "grad_norm": 0.17327331006526947,
+      "learning_rate": 8.988948585342597e-05,
+      "loss": 0.0198,
+      "step": 4200
+    },
+    {
+      "epoch": 11.137566137566138,
+      "grad_norm": 0.22939947247505188,
+      "learning_rate": 8.945433899792614e-05,
+      "loss": 0.0207,
+      "step": 4210
+    },
+    {
+      "epoch": 11.164021164021165,
+      "grad_norm": 0.1739424169063568,
+      "learning_rate": 8.901939398045963e-05,
+      "loss": 0.0183,
+      "step": 4220
+    },
+    {
+      "epoch": 11.19047619047619,
+      "grad_norm": 0.20427298545837402,
+      "learning_rate": 8.858465912562991e-05,
+      "loss": 0.0205,
+      "step": 4230
+    },
+    {
+      "epoch": 11.216931216931217,
+      "grad_norm": 0.30859696865081787,
+      "learning_rate": 8.81501427540182e-05,
+      "loss": 0.0223,
+      "step": 4240
+    },
+    {
+      "epoch": 11.243386243386244,
+      "grad_norm": 0.27141624689102173,
+      "learning_rate": 8.771585318202397e-05,
+      "loss": 0.0213,
+      "step": 4250
+    },
+    {
+      "epoch": 11.26984126984127,
+      "grad_norm": 0.31934279203414917,
+      "learning_rate": 8.728179872170587e-05,
+      "loss": 0.0232,
+      "step": 4260
+    },
+    {
+      "epoch": 11.296296296296296,
+      "grad_norm": 0.35330238938331604,
+      "learning_rate": 8.684798768062273e-05,
+      "loss": 0.0212,
+      "step": 4270
+    },
+    {
+      "epoch": 11.322751322751323,
+      "grad_norm": 0.27975142002105713,
+      "learning_rate": 8.64144283616744e-05,
+      "loss": 0.0234,
+      "step": 4280
+    },
+    {
+      "epoch": 11.34920634920635,
+      "grad_norm": 0.3430609107017517,
+      "learning_rate": 8.598112906294287e-05,
+      "loss": 0.0217,
+      "step": 4290
+    },
+    {
+      "epoch": 11.375661375661375,
+      "grad_norm": 0.17652027308940887,
+      "learning_rate": 8.554809807753364e-05,
+      "loss": 0.023,
+      "step": 4300
+    },
+    {
+      "epoch": 11.402116402116402,
+      "grad_norm": 0.24938417971134186,
+      "learning_rate": 8.51153436934167e-05,
+      "loss": 0.0236,
+      "step": 4310
+    },
+    {
+      "epoch": 11.428571428571429,
+      "grad_norm": 0.3108935058116913,
+      "learning_rate": 8.468287419326808e-05,
+      "loss": 0.0196,
+      "step": 4320
+    },
+    {
+      "epoch": 11.455026455026456,
+      "grad_norm": 0.17053309082984924,
+      "learning_rate": 8.425069785431137e-05,
+      "loss": 0.0202,
+      "step": 4330
+    },
+    {
+      "epoch": 11.481481481481481,
+      "grad_norm": 0.21392136812210083,
+      "learning_rate": 8.38188229481591e-05,
+      "loss": 0.0247,
+      "step": 4340
+    },
+    {
+      "epoch": 11.507936507936508,
+      "grad_norm": 0.27775999903678894,
+      "learning_rate": 8.338725774065462e-05,
+      "loss": 0.0209,
+      "step": 4350
+    },
+    {
+      "epoch": 11.534391534391535,
+      "grad_norm": 0.277752161026001,
+      "learning_rate": 8.295601049171384e-05,
+      "loss": 0.0212,
+      "step": 4360
+    },
+    {
+      "epoch": 11.56084656084656,
+      "grad_norm": 0.23016278445720673,
+      "learning_rate": 8.252508945516704e-05,
+      "loss": 0.0217,
+      "step": 4370
+    },
+    {
+      "epoch": 11.587301587301587,
+      "grad_norm": 0.22993695735931396,
+      "learning_rate": 8.209450287860103e-05,
+      "loss": 0.0232,
+      "step": 4380
+    },
+    {
+      "epoch": 11.613756613756614,
+      "grad_norm": 0.26726651191711426,
+      "learning_rate": 8.166425900320126e-05,
+      "loss": 0.0235,
+      "step": 4390
+    },
+    {
+      "epoch": 11.640211640211641,
+      "grad_norm": 0.24999497830867767,
+      "learning_rate": 8.123436606359403e-05,
+      "loss": 0.0236,
+      "step": 4400
+    },
+    {
+      "epoch": 11.666666666666666,
+      "grad_norm": 0.20490936934947968,
+      "learning_rate": 8.080483228768892e-05,
+      "loss": 0.0224,
+      "step": 4410
+    },
+    {
+      "epoch": 11.693121693121693,
+      "grad_norm": 0.1914476752281189,
+      "learning_rate": 8.037566589652141e-05,
+      "loss": 0.021,
+      "step": 4420
+    },
+    {
+      "epoch": 11.71957671957672,
+      "grad_norm": 0.19754275679588318,
+      "learning_rate": 7.994687510409536e-05,
+      "loss": 0.0212,
+      "step": 4430
+    },
+    {
+      "epoch": 11.746031746031747,
+      "grad_norm": 0.19561535120010376,
+      "learning_rate": 7.951846811722583e-05,
+      "loss": 0.021,
+      "step": 4440
+    },
+    {
+      "epoch": 11.772486772486772,
+      "grad_norm": 0.28997474908828735,
+      "learning_rate": 7.909045313538222e-05,
+      "loss": 0.0204,
+      "step": 4450
+    },
+    {
+      "epoch": 11.798941798941799,
+      "grad_norm": 0.18335215747356415,
+      "learning_rate": 7.8662838350531e-05,
+      "loss": 0.0205,
+      "step": 4460
+    },
+    {
+      "epoch": 11.825396825396826,
+      "grad_norm": 0.2745349109172821,
+      "learning_rate": 7.823563194697919e-05,
+      "loss": 0.0206,
+      "step": 4470
+    },
+    {
+      "epoch": 11.851851851851851,
+      "grad_norm": 0.2598635256290436,
+      "learning_rate": 7.780884210121767e-05,
+      "loss": 0.0238,
+      "step": 4480
+    },
+    {
+      "epoch": 11.878306878306878,
+      "grad_norm": 0.24171608686447144,
+      "learning_rate": 7.73824769817645e-05,
+      "loss": 0.0223,
+      "step": 4490
+    },
+    {
+      "epoch": 11.904761904761905,
+      "grad_norm": 0.19994641840457916,
+      "learning_rate": 7.695654474900875e-05,
+      "loss": 0.0214,
+      "step": 4500
+    },
+    {
+      "epoch": 11.93121693121693,
+      "grad_norm": 0.22974646091461182,
+      "learning_rate": 7.653105355505442e-05,
+      "loss": 0.0201,
+      "step": 4510
+    },
+    {
+      "epoch": 11.957671957671957,
+      "grad_norm": 0.24392633140087128,
+      "learning_rate": 7.610601154356413e-05,
+      "loss": 0.0189,
+      "step": 4520
+    },
+    {
+      "epoch": 11.984126984126984,
+      "grad_norm": 0.22622373700141907,
+      "learning_rate": 7.568142684960342e-05,
+      "loss": 0.0205,
+      "step": 4530
+    },
+    {
+      "epoch": 12.010582010582011,
+      "grad_norm": 0.21608489751815796,
+      "learning_rate": 7.525730759948508e-05,
+      "loss": 0.0213,
+      "step": 4540
+    },
+    {
+      "epoch": 12.037037037037036,
+      "grad_norm": 0.20682625472545624,
+      "learning_rate": 7.483366191061354e-05,
+      "loss": 0.0202,
+      "step": 4550
+    },
+    {
+      "epoch": 12.063492063492063,
+      "grad_norm": 0.22226907312870026,
+      "learning_rate": 7.441049789132948e-05,
+      "loss": 0.0222,
+      "step": 4560
+    },
+    {
+      "epoch": 12.08994708994709,
+      "grad_norm": 0.3071605861186981,
+      "learning_rate": 7.398782364075485e-05,
+      "loss": 0.019,
+      "step": 4570
+    },
+    {
+      "epoch": 12.116402116402117,
+      "grad_norm": 0.22628101706504822,
+      "learning_rate": 7.35656472486375e-05,
+      "loss": 0.0229,
+      "step": 4580
+    },
+    {
+      "epoch": 12.142857142857142,
+      "grad_norm": 0.24755476415157318,
+      "learning_rate": 7.314397679519673e-05,
+      "loss": 0.0231,
+      "step": 4590
+    },
+    {
+      "epoch": 12.16931216931217,
+      "grad_norm": 0.3300286531448364,
+      "learning_rate": 7.272282035096831e-05,
+      "loss": 0.0212,
+      "step": 4600
+    },
+    {
+      "epoch": 12.195767195767196,
+      "grad_norm": 0.24590083956718445,
+      "learning_rate": 7.230218597665024e-05,
+      "loss": 0.0212,
+      "step": 4610
+    },
+    {
+      "epoch": 12.222222222222221,
+      "grad_norm": 0.21087048947811127,
+      "learning_rate": 7.188208172294841e-05,
+      "loss": 0.0202,
+      "step": 4620
+    },
+    {
+      "epoch": 12.248677248677248,
+      "grad_norm": 0.18404772877693176,
+      "learning_rate": 7.146251563042246e-05,
+      "loss": 0.0225,
+      "step": 4630
+    },
+    {
+      "epoch": 12.275132275132275,
+      "grad_norm": 0.22472181916236877,
+      "learning_rate": 7.104349572933187e-05,
+      "loss": 0.0207,
+      "step": 4640
+    },
+    {
+      "epoch": 12.301587301587302,
+      "grad_norm": 0.17883093655109406,
+      "learning_rate": 7.06250300394825e-05,
+      "loss": 0.0203,
+      "step": 4650
+    },
+    {
+      "epoch": 12.328042328042327,
+      "grad_norm": 0.23257587850093842,
+      "learning_rate": 7.020712657007276e-05,
+      "loss": 0.0199,
+      "step": 4660
+    },
+    {
+      "epoch": 12.354497354497354,
+      "grad_norm": 0.30548185110092163,
+      "learning_rate": 6.97897933195405e-05,
+      "loss": 0.022,
+      "step": 4670
+    },
+    {
+      "epoch": 12.380952380952381,
+      "grad_norm": 0.2183784395456314,
+      "learning_rate": 6.937303827540996e-05,
+      "loss": 0.0211,
+      "step": 4680
+    },
+    {
+      "epoch": 12.407407407407407,
+      "grad_norm": 0.22535958886146545,
+      "learning_rate": 6.89568694141388e-05,
+      "loss": 0.0233,
+      "step": 4690
+    },
+    {
+      "epoch": 12.433862433862434,
+      "grad_norm": 0.2497279942035675,
+      "learning_rate": 6.854129470096539e-05,
+      "loss": 0.019,
+      "step": 4700
+    },
+    {
+      "epoch": 12.46031746031746,
+      "grad_norm": 0.2837906777858734,
+      "learning_rate": 6.812632208975667e-05,
+      "loss": 0.0197,
+      "step": 4710
+    },
+    {
+      "epoch": 12.486772486772487,
+      "grad_norm": 0.1884712129831314,
+      "learning_rate": 6.77119595228554e-05,
+      "loss": 0.022,
+      "step": 4720
+    },
+    {
+      "epoch": 12.513227513227513,
+      "grad_norm": 0.23100075125694275,
+      "learning_rate": 6.72982149309286e-05,
+      "loss": 0.0206,
+      "step": 4730
+    },
+    {
+      "epoch": 12.53968253968254,
+      "grad_norm": 0.2137850522994995,
+      "learning_rate": 6.688509623281559e-05,
+      "loss": 0.0208,
+      "step": 4740
+    },
+    {
+      "epoch": 12.566137566137566,
+      "grad_norm": 0.2913530766963959,
+      "learning_rate": 6.647261133537642e-05,
+      "loss": 0.0212,
+      "step": 4750
+    },
+    {
+      "epoch": 12.592592592592592,
+      "grad_norm": 0.21118688583374023,
+      "learning_rate": 6.606076813334046e-05,
+      "loss": 0.0199,
+      "step": 4760
+    },
+    {
+      "epoch": 12.619047619047619,
+      "grad_norm": 0.20391783118247986,
+      "learning_rate": 6.564957450915556e-05,
+      "loss": 0.0217,
+      "step": 4770
+    },
+    {
+      "epoch": 12.645502645502646,
+      "grad_norm": 0.3074709177017212,
+      "learning_rate": 6.523903833283688e-05,
+      "loss": 0.023,
+      "step": 4780
+    },
+    {
+      "epoch": 12.671957671957673,
+      "grad_norm": 0.22795026004314423,
+      "learning_rate": 6.48291674618164e-05,
+      "loss": 0.022,
+      "step": 4790
+    },
+    {
+      "epoch": 12.698412698412698,
+      "grad_norm": 0.26107490062713623,
+      "learning_rate": 6.441996974079264e-05,
+      "loss": 0.0204,
+      "step": 4800
+    },
+    {
+      "epoch": 12.724867724867725,
+      "grad_norm": 0.33507269620895386,
+      "learning_rate": 6.401145300158033e-05,
+      "loss": 0.0238,
+      "step": 4810
+    },
+    {
+      "epoch": 12.751322751322752,
+      "grad_norm": 0.27531009912490845,
+      "learning_rate": 6.360362506296052e-05,
+      "loss": 0.0214,
+      "step": 4820
+    },
+    {
+      "epoch": 12.777777777777779,
+      "grad_norm": 0.18488064408302307,
+      "learning_rate": 6.319649373053116e-05,
+      "loss": 0.0216,
+      "step": 4830
+    },
+    {
+      "epoch": 12.804232804232804,
+      "grad_norm": 0.23100340366363525,
+      "learning_rate": 6.279006679655746e-05,
+      "loss": 0.0203,
+      "step": 4840
+    },
+    {
+      "epoch": 12.83068783068783,
+      "grad_norm": 0.19215638935565948,
+      "learning_rate": 6.238435203982278e-05,
+      "loss": 0.0202,
+      "step": 4850
+    },
+    {
+      "epoch": 12.857142857142858,
+      "grad_norm": 0.20781166851520538,
+      "learning_rate": 6.197935722547996e-05,
+      "loss": 0.0214,
+      "step": 4860
+    },
+    {
+      "epoch": 12.883597883597883,
+      "grad_norm": 0.27218201756477356,
+      "learning_rate": 6.157509010490243e-05,
+      "loss": 0.022,
+      "step": 4870
+    },
+    {
+      "epoch": 12.91005291005291,
+      "grad_norm": 0.2349989116191864,
+      "learning_rate": 6.117155841553594e-05,
+      "loss": 0.0194,
+      "step": 4880
+    },
+    {
+      "epoch": 12.936507936507937,
+      "grad_norm": 0.22253037989139557,
+      "learning_rate": 6.076876988075064e-05,
+      "loss": 0.0199,
+      "step": 4890
+    },
+    {
+      "epoch": 12.962962962962964,
+      "grad_norm": 0.2949942350387573,
+      "learning_rate": 6.0366732209693035e-05,
+      "loss": 0.0206,
+      "step": 4900
+    },
+    {
+      "epoch": 12.989417989417989,
+      "grad_norm": 0.23691920936107635,
+      "learning_rate": 5.9965453097138455e-05,
+      "loss": 0.0192,
+      "step": 4910
+    },
+    {
+      "epoch": 13.015873015873016,
+      "grad_norm": 0.22252142429351807,
+      "learning_rate": 5.9564940223344e-05,
+      "loss": 0.0241,
+      "step": 4920
+    },
+    {
+      "epoch": 13.042328042328043,
+      "grad_norm": 0.31423982977867126,
+      "learning_rate": 5.916520125390124e-05,
+      "loss": 0.0219,
+      "step": 4930
+    },
+    {
+      "epoch": 13.068783068783068,
+      "grad_norm": 0.29873350262641907,
+      "learning_rate": 5.8766243839589755e-05,
+      "loss": 0.0193,
+      "step": 4940
+    },
+    {
+      "epoch": 13.095238095238095,
+      "grad_norm": 0.16471320390701294,
+      "learning_rate": 5.83680756162305e-05,
+      "loss": 0.0213,
+      "step": 4950
+    },
+    {
+      "epoch": 13.121693121693122,
+      "grad_norm": 0.22079427540302277,
+      "learning_rate": 5.797070420453989e-05,
+      "loss": 0.0216,
+      "step": 4960
+    },
+    {
+      "epoch": 13.148148148148149,
+      "grad_norm": 0.34240761399269104,
+      "learning_rate": 5.7574137209983706e-05,
+      "loss": 0.0207,
+      "step": 4970
+    },
+    {
+      "epoch": 13.174603174603174,
+      "grad_norm": 0.24463868141174316,
+      "learning_rate": 5.717838222263165e-05,
+      "loss": 0.022,
+      "step": 4980
+    },
+    {
+      "epoch": 13.201058201058201,
+      "grad_norm": 0.27385976910591125,
+      "learning_rate": 5.678344681701206e-05,
+      "loss": 0.0184,
+      "step": 4990
+    },
+    {
+      "epoch": 13.227513227513228,
+      "grad_norm": 0.18755285441875458,
+      "learning_rate": 5.6389338551967044e-05,
+      "loss": 0.021,
+      "step": 5000
+    },
+    {
+      "epoch": 13.253968253968253,
+      "grad_norm": 0.2744396924972534,
+      "learning_rate": 5.5996064970507555e-05,
+      "loss": 0.0208,
+      "step": 5010
+    },
+    {
+      "epoch": 13.28042328042328,
+      "grad_norm": 0.24906226992607117,
+      "learning_rate": 5.56036335996692e-05,
+      "loss": 0.0206,
+      "step": 5020
+    },
+    {
+      "epoch": 13.306878306878307,
+      "grad_norm": 0.2173585295677185,
+      "learning_rate": 5.5212051950368296e-05,
+      "loss": 0.0192,
+      "step": 5030
+    },
+    {
+      "epoch": 13.333333333333334,
+      "grad_norm": 0.2265741527080536,
+      "learning_rate": 5.4821327517257756e-05,
+      "loss": 0.0188,
+      "step": 5040
+    },
+    {
+      "epoch": 13.359788359788359,
+      "grad_norm": 0.1848832368850708,
+      "learning_rate": 5.4431467778584e-05,
+      "loss": 0.0186,
+      "step": 5050
+    },
+    {
+      "epoch": 13.386243386243386,
+      "grad_norm": 0.18450133502483368,
+      "learning_rate": 5.404248019604361e-05,
+      "loss": 0.0193,
+      "step": 5060
+    },
+    {
+      "epoch": 13.412698412698413,
+      "grad_norm": 0.1744503378868103,
+      "learning_rate": 5.3654372214640625e-05,
+      "loss": 0.018,
+      "step": 5070
+    },
+    {
+      "epoch": 13.43915343915344,
+      "grad_norm": 0.23882748186588287,
+      "learning_rate": 5.326715126254396e-05,
+      "loss": 0.02,
+      "step": 5080
+    },
+    {
+      "epoch": 13.465608465608465,
+      "grad_norm": 0.23424524068832397,
+      "learning_rate": 5.2880824750945404e-05,
+      "loss": 0.0207,
+      "step": 5090
+    },
+    {
+      "epoch": 13.492063492063492,
+      "grad_norm": 0.3038886487483978,
+      "learning_rate": 5.249540007391752e-05,
+      "loss": 0.0198,
+      "step": 5100
+    },
+    {
+      "epoch": 13.518518518518519,
+      "grad_norm": 0.180315300822258,
+      "learning_rate": 5.2110884608272315e-05,
+      "loss": 0.0184,
+      "step": 5110
+    },
+    {
+      "epoch": 13.544973544973544,
+      "grad_norm": 0.2620660662651062,
+      "learning_rate": 5.172728571342008e-05,
+      "loss": 0.0197,
+      "step": 5120
+    },
+    {
+      "epoch": 13.571428571428571,
+      "grad_norm": 0.24521809816360474,
+      "learning_rate": 5.1344610731228334e-05,
+      "loss": 0.0198,
+      "step": 5130
+    },
+    {
+      "epoch": 13.597883597883598,
+      "grad_norm": 0.2860581874847412,
+      "learning_rate": 5.0962866985881495e-05,
+      "loss": 0.0205,
+      "step": 5140
+    },
+    {
+      "epoch": 13.624338624338625,
+      "grad_norm": 0.20572273433208466,
+      "learning_rate": 5.05820617837406e-05,
+      "loss": 0.0199,
+      "step": 5150
+    },
+    {
+      "epoch": 13.65079365079365,
+      "grad_norm": 0.27270907163619995,
+      "learning_rate": 5.0202202413203506e-05,
+      "loss": 0.0213,
+      "step": 5160
+    },
+    {
+      "epoch": 13.677248677248677,
+      "grad_norm": 0.16998854279518127,
+      "learning_rate": 4.9823296144565346e-05,
+      "loss": 0.0169,
+      "step": 5170
+    },
+    {
+      "epoch": 13.703703703703704,
+      "grad_norm": 0.25765907764434814,
+      "learning_rate": 4.944535022987954e-05,
+      "loss": 0.0186,
+      "step": 5180
+    },
+    {
+      "epoch": 13.73015873015873,
+      "grad_norm": 0.2524023652076721,
+      "learning_rate": 4.906837190281875e-05,
+      "loss": 0.0199,
+      "step": 5190
+    },
+    {
+      "epoch": 13.756613756613756,
+      "grad_norm": 0.20784050226211548,
+      "learning_rate": 4.8692368378536545e-05,
+      "loss": 0.0211,
+      "step": 5200
+    },
+    {
+      "epoch": 13.783068783068783,
+      "grad_norm": 0.2766719162464142,
+      "learning_rate": 4.8317346853529465e-05,
+      "loss": 0.019,
+      "step": 5210
+    },
+    {
+      "epoch": 13.80952380952381,
+      "grad_norm": 0.1722410023212433,
+      "learning_rate": 4.7943314505498974e-05,
+      "loss": 0.0199,
+      "step": 5220
+    },
+    {
+      "epoch": 13.835978835978835,
+      "grad_norm": 0.3369690179824829,
+      "learning_rate": 4.757027849321427e-05,
+      "loss": 0.0184,
+      "step": 5230
+    },
+    {
+      "epoch": 13.862433862433862,
+      "grad_norm": 0.20445048809051514,
+      "learning_rate": 4.7198245956375395e-05,
+      "loss": 0.0192,
+      "step": 5240
+    },
+    {
+      "epoch": 13.88888888888889,
+      "grad_norm": 0.2391098439693451,
+      "learning_rate": 4.682722401547617e-05,
+      "loss": 0.0175,
+      "step": 5250
+    },
+    {
+      "epoch": 13.915343915343914,
+      "grad_norm": 0.23185895383358002,
+      "learning_rate": 4.6457219771668426e-05,
+      "loss": 0.0204,
+      "step": 5260
+    },
+    {
+      "epoch": 13.941798941798941,
+      "grad_norm": 0.19505229592323303,
+      "learning_rate": 4.608824030662571e-05,
+      "loss": 0.0192,
+      "step": 5270
+    },
+    {
+      "epoch": 13.968253968253968,
+      "grad_norm": 0.3211570382118225,
+      "learning_rate": 4.5720292682407874e-05,
+      "loss": 0.0179,
+      "step": 5280
+    },
+    {
+      "epoch": 13.994708994708995,
+      "grad_norm": 0.24055220186710358,
+      "learning_rate": 4.535338394132602e-05,
+      "loss": 0.0205,
+      "step": 5290
+    },
+    {
+      "epoch": 14.02116402116402,
+      "grad_norm": 0.24234339594841003,
+      "learning_rate": 4.498752110580752e-05,
+      "loss": 0.0192,
+      "step": 5300
+    },
+    {
+      "epoch": 14.047619047619047,
+      "grad_norm": 0.317220002412796,
+      "learning_rate": 4.4622711178261654e-05,
+      "loss": 0.019,
+      "step": 5310
+    },
+    {
+      "epoch": 14.074074074074074,
+      "grad_norm": 0.24808919429779053,
+      "learning_rate": 4.425896114094581e-05,
+      "loss": 0.0181,
+      "step": 5320
+    },
+    {
+      "epoch": 14.100529100529101,
+      "grad_norm": 0.21401309967041016,
+      "learning_rate": 4.38962779558315e-05,
+      "loss": 0.0191,
+      "step": 5330
+    },
+    {
+      "epoch": 14.126984126984127,
+      "grad_norm": 0.1712522953748703,
+      "learning_rate": 4.353466856447135e-05,
+      "loss": 0.0176,
+      "step": 5340
+    },
+    {
+      "epoch": 14.153439153439153,
+      "grad_norm": 0.1504759043455124,
+      "learning_rate": 4.317413988786618e-05,
+      "loss": 0.0182,
+      "step": 5350
+    },
+    {
+      "epoch": 14.17989417989418,
+      "grad_norm": 0.2544901669025421,
+      "learning_rate": 4.281469882633254e-05,
+      "loss": 0.0201,
+      "step": 5360
+    },
+    {
+      "epoch": 14.206349206349206,
+      "grad_norm": 0.17256955802440643,
+      "learning_rate": 4.2456352259370566e-05,
+      "loss": 0.0176,
+      "step": 5370
+    },
+    {
+      "epoch": 14.232804232804233,
+      "grad_norm": 0.2197188436985016,
+      "learning_rate": 4.209910704553254e-05,
+      "loss": 0.0173,
+      "step": 5380
+    },
+    {
+      "epoch": 14.25925925925926,
+      "grad_norm": 0.18674366176128387,
+      "learning_rate": 4.174297002229132e-05,
+      "loss": 0.0191,
+      "step": 5390
+    },
+    {
+      "epoch": 14.285714285714286,
+      "grad_norm": 0.19098690152168274,
+      "learning_rate": 4.138794800590963e-05,
+      "loss": 0.0186,
+      "step": 5400
+    },
+    {
+      "epoch": 14.312169312169312,
+      "grad_norm": 0.3053850531578064,
+      "learning_rate": 4.103404779130968e-05,
+      "loss": 0.0195,
+      "step": 5410
+    },
+    {
+      "epoch": 14.338624338624339,
+      "grad_norm": 0.2690741717815399,
+      "learning_rate": 4.0681276151942946e-05,
+      "loss": 0.0196,
+      "step": 5420
+    },
+    {
+      "epoch": 14.365079365079366,
+      "grad_norm": 0.18734146654605865,
+      "learning_rate": 4.032963983966064e-05,
+      "loss": 0.0171,
+      "step": 5430
+    },
+    {
+      "epoch": 14.39153439153439,
+      "grad_norm": 0.19453099370002747,
+      "learning_rate": 3.997914558458441e-05,
+      "loss": 0.0182,
+      "step": 5440
+    },
+    {
+      "epoch": 14.417989417989418,
+      "grad_norm": 0.14661167562007904,
+      "learning_rate": 3.962980009497762e-05,
+      "loss": 0.0166,
+      "step": 5450
+    },
+    {
+      "epoch": 14.444444444444445,
+      "grad_norm": 0.23112380504608154,
+      "learning_rate": 3.928161005711685e-05,
+      "loss": 0.0186,
+      "step": 5460
+    },
+    {
+      "epoch": 14.470899470899472,
+      "grad_norm": 0.19269657135009766,
+      "learning_rate": 3.8934582135164065e-05,
+      "loss": 0.02,
+      "step": 5470
+    },
+    {
+      "epoch": 14.497354497354497,
+      "grad_norm": 0.2662561237812042,
+      "learning_rate": 3.858872297103893e-05,
+      "loss": 0.0177,
+      "step": 5480
+    },
+    {
+      "epoch": 14.523809523809524,
+      "grad_norm": 0.16409893333911896,
+      "learning_rate": 3.8244039184291705e-05,
+      "loss": 0.0193,
+      "step": 5490
+    },
+    {
+      "epoch": 14.55026455026455,
+      "grad_norm": 0.2753513753414154,
+      "learning_rate": 3.790053737197668e-05,
+      "loss": 0.0202,
+      "step": 5500
+    },
+    {
+      "epoch": 14.576719576719576,
+      "grad_norm": 0.2297661155462265,
+      "learning_rate": 3.7558224108525766e-05,
+      "loss": 0.0172,
+      "step": 5510
+    },
+    {
+      "epoch": 14.603174603174603,
+      "grad_norm": 0.17383983731269836,
+      "learning_rate": 3.72171059456227e-05,
+      "loss": 0.0195,
+      "step": 5520
+    },
+    {
+      "epoch": 14.62962962962963,
+      "grad_norm": 0.16098396480083466,
+      "learning_rate": 3.687718941207767e-05,
+      "loss": 0.0167,
+      "step": 5530
+    },
+    {
+      "epoch": 14.656084656084657,
+      "grad_norm": 0.26572316884994507,
+      "learning_rate": 3.653848101370239e-05,
+      "loss": 0.0189,
+      "step": 5540
+    },
+    {
+      "epoch": 14.682539682539682,
+      "grad_norm": 0.19508256018161774,
+      "learning_rate": 3.6200987233185466e-05,
+      "loss": 0.02,
+      "step": 5550
+    },
+    {
+      "epoch": 14.708994708994709,
+      "grad_norm": 0.2050306499004364,
+      "learning_rate": 3.586471452996853e-05,
+      "loss": 0.0184,
+      "step": 5560
+    },
+    {
+      "epoch": 14.735449735449736,
+      "grad_norm": 0.2767617702484131,
+      "learning_rate": 3.5529669340122375e-05,
+      "loss": 0.0197,
+      "step": 5570
+    },
+    {
+      "epoch": 14.761904761904763,
+      "grad_norm": 0.21244259178638458,
+      "learning_rate": 3.5195858076223856e-05,
+      "loss": 0.0222,
+      "step": 5580
+    },
+    {
+      "epoch": 14.788359788359788,
+      "grad_norm": 0.20567195117473602,
+      "learning_rate": 3.48632871272333e-05,
+      "loss": 0.0177,
+      "step": 5590
+    },
+    {
+      "epoch": 14.814814814814815,
+      "grad_norm": 0.1751723289489746,
+      "learning_rate": 3.4531962858372005e-05,
+      "loss": 0.0207,
+      "step": 5600
+    },
+    {
+      "epoch": 14.841269841269842,
+      "grad_norm": 0.23504501581192017,
+      "learning_rate": 3.4201891611000526e-05,
+      "loss": 0.0198,
+      "step": 5610
+    },
+    {
+      "epoch": 14.867724867724867,
+      "grad_norm": 0.1576506644487381,
+      "learning_rate": 3.38730797024973e-05,
+      "loss": 0.0186,
+      "step": 5620
+    },
+    {
+      "epoch": 14.894179894179894,
+      "grad_norm": 0.2593693435192108,
+      "learning_rate": 3.35455334261377e-05,
+      "loss": 0.0198,
+      "step": 5630
+    },
+    {
+      "epoch": 14.920634920634921,
+      "grad_norm": 0.17774498462677002,
+      "learning_rate": 3.3219259050973694e-05,
+      "loss": 0.0172,
+      "step": 5640
+    },
+    {
+      "epoch": 14.947089947089948,
+      "grad_norm": 0.186201810836792,
+      "learning_rate": 3.2894262821713686e-05,
+      "loss": 0.0173,
+      "step": 5650
+    },
+    {
+      "epoch": 14.973544973544973,
+      "grad_norm": 0.1857055425643921,
+      "learning_rate": 3.257055095860306e-05,
+      "loss": 0.0162,
+      "step": 5660
+    },
+    {
+      "epoch": 15.0,
+      "grad_norm": 0.1835918426513672,
+      "learning_rate": 3.224812965730531e-05,
+      "loss": 0.0178,
+      "step": 5670
+    },
+    {
+      "epoch": 15.026455026455027,
+      "grad_norm": 0.22426722943782806,
+      "learning_rate": 3.192700508878315e-05,
+      "loss": 0.0184,
+      "step": 5680
+    },
+    {
+      "epoch": 15.052910052910052,
+      "grad_norm": 0.19990061223506927,
+      "learning_rate": 3.16071833991806e-05,
+      "loss": 0.0187,
+      "step": 5690
+    },
+    {
+      "epoch": 15.079365079365079,
+      "grad_norm": 0.2128632515668869,
+      "learning_rate": 3.12886707097053e-05,
+      "loss": 0.0152,
+      "step": 5700
+    },
+    {
+      "epoch": 15.105820105820106,
+      "grad_norm": 0.18279944360256195,
+      "learning_rate": 3.0971473116511394e-05,
+      "loss": 0.0196,
+      "step": 5710
+    },
+    {
+      "epoch": 15.132275132275133,
+      "grad_norm": 0.21548713743686676,
+      "learning_rate": 3.0655596690582764e-05,
+      "loss": 0.018,
+      "step": 5720
+    },
+    {
+      "epoch": 15.158730158730158,
+      "grad_norm": 0.3082822263240814,
+      "learning_rate": 3.0341047477617012e-05,
+      "loss": 0.0209,
+      "step": 5730
+    },
+    {
+      "epoch": 15.185185185185185,
+      "grad_norm": 0.15852923691272736,
+      "learning_rate": 3.0027831497909486e-05,
+      "loss": 0.0183,
+      "step": 5740
+    },
+    {
+      "epoch": 15.211640211640212,
+      "grad_norm": 0.17777982354164124,
+      "learning_rate": 2.9715954746238238e-05,
+      "loss": 0.0171,
+      "step": 5750
+    },
+    {
+      "epoch": 15.238095238095237,
+      "grad_norm": 0.14747212827205658,
+      "learning_rate": 2.9405423191749338e-05,
+      "loss": 0.0185,
+      "step": 5760
+    },
+    {
+      "epoch": 15.264550264550264,
+      "grad_norm": 0.2689704895019531,
+      "learning_rate": 2.909624277784243e-05,
+      "loss": 0.018,
+      "step": 5770
+    },
+    {
+      "epoch": 15.291005291005291,
+      "grad_norm": 0.18944290280342102,
+      "learning_rate": 2.878841942205709e-05,
+      "loss": 0.0172,
+      "step": 5780
+    },
+    {
+      "epoch": 15.317460317460318,
+      "grad_norm": 0.14912503957748413,
+      "learning_rate": 2.848195901595969e-05,
+      "loss": 0.0165,
+      "step": 5790
+    },
+    {
+      "epoch": 15.343915343915343,
+      "grad_norm": 0.1962699443101883,
+      "learning_rate": 2.817686742503033e-05,
+      "loss": 0.018,
+      "step": 5800
+    },
+    {
+      "epoch": 15.37037037037037,
+      "grad_norm": 0.17675501108169556,
+      "learning_rate": 2.7873150488550826e-05,
+      "loss": 0.0173,
+      "step": 5810
+    },
+    {
+      "epoch": 15.396825396825397,
+      "grad_norm": 0.22969458997249603,
+      "learning_rate": 2.7570814019492964e-05,
+      "loss": 0.0176,
+      "step": 5820
+    },
+    {
+      "epoch": 15.423280423280424,
+      "grad_norm": 0.15065893530845642,
+      "learning_rate": 2.7269863804407058e-05,
+      "loss": 0.0162,
+      "step": 5830
+    },
+    {
+      "epoch": 15.44973544973545,
+      "grad_norm": 0.19528082013130188,
+      "learning_rate": 2.6970305603311297e-05,
+      "loss": 0.0166,
+      "step": 5840
+    },
+    {
+      "epoch": 15.476190476190476,
+      "grad_norm": 0.236960306763649,
+      "learning_rate": 2.6672145149581605e-05,
+      "loss": 0.018,
+      "step": 5850
+    },
+    {
+      "epoch": 15.502645502645503,
+      "grad_norm": 0.22975318133831024,
+      "learning_rate": 2.6375388149841696e-05,
+      "loss": 0.0192,
+      "step": 5860
+    },
+    {
+      "epoch": 15.529100529100528,
+      "grad_norm": 0.17225511372089386,
+      "learning_rate": 2.608004028385398e-05,
+      "loss": 0.0162,
+      "step": 5870
+    },
+    {
+      "epoch": 15.555555555555555,
+      "grad_norm": 0.1448196917772293,
+      "learning_rate": 2.5786107204410913e-05,
+      "loss": 0.0167,
+      "step": 5880
+    },
+    {
+      "epoch": 15.582010582010582,
+      "grad_norm": 0.13869361579418182,
+      "learning_rate": 2.5493594537226682e-05,
+      "loss": 0.016,
+      "step": 5890
+    },
+    {
+      "epoch": 15.60846560846561,
+      "grad_norm": 0.1881704032421112,
+      "learning_rate": 2.5202507880829484e-05,
+      "loss": 0.0188,
+      "step": 5900
+    },
+    {
+      "epoch": 15.634920634920634,
+      "grad_norm": 0.2088797241449356,
+      "learning_rate": 2.4912852806454667e-05,
+      "loss": 0.0182,
+      "step": 5910
+    },
+    {
+      "epoch": 15.661375661375661,
+      "grad_norm": 0.18913866579532623,
+      "learning_rate": 2.4624634857937755e-05,
+      "loss": 0.0173,
+      "step": 5920
+    },
+    {
+      "epoch": 15.687830687830688,
+      "grad_norm": 0.13622598350048065,
+      "learning_rate": 2.433785955160851e-05,
+      "loss": 0.0188,
+      "step": 5930
+    },
+    {
+      "epoch": 15.714285714285714,
+      "grad_norm": 0.16862928867340088,
+      "learning_rate": 2.405253237618541e-05,
+      "loss": 0.0179,
+      "step": 5940
+    },
+    {
+      "epoch": 15.74074074074074,
+      "grad_norm": 0.23106427490711212,
+      "learning_rate": 2.3768658792670407e-05,
+      "loss": 0.0195,
+      "step": 5950
+    },
+    {
+      "epoch": 15.767195767195767,
+      "grad_norm": 0.2322971224784851,
+      "learning_rate": 2.3486244234244637e-05,
+      "loss": 0.0177,
+      "step": 5960
+    },
+    {
+      "epoch": 15.793650793650794,
+      "grad_norm": 0.18966665863990784,
+      "learning_rate": 2.3205294106164223e-05,
+      "loss": 0.0176,
+      "step": 5970
+    },
+    {
+      "epoch": 15.82010582010582,
+      "grad_norm": 0.20752005279064178,
+      "learning_rate": 2.2925813785656945e-05,
+      "loss": 0.0173,
+      "step": 5980
+    },
+    {
+      "epoch": 15.846560846560847,
+      "grad_norm": 0.14546965062618256,
+      "learning_rate": 2.264780862181931e-05,
+      "loss": 0.0182,
+      "step": 5990
+    },
+    {
+      "epoch": 15.873015873015873,
+      "grad_norm": 0.27061018347740173,
+      "learning_rate": 2.2371283935514108e-05,
+      "loss": 0.017,
+      "step": 6000
+    },
+    {
+      "epoch": 15.899470899470899,
+      "grad_norm": 0.2078131139278412,
+      "learning_rate": 2.2096245019268634e-05,
+      "loss": 0.0171,
+      "step": 6010
+    },
+    {
+      "epoch": 15.925925925925926,
+      "grad_norm": 0.21537968516349792,
+      "learning_rate": 2.1822697137173444e-05,
+      "loss": 0.0179,
+      "step": 6020
+    },
+    {
+      "epoch": 15.952380952380953,
+      "grad_norm": 0.2436002641916275,
+      "learning_rate": 2.1550645524781464e-05,
+      "loss": 0.0174,
+      "step": 6030
+    },
+    {
+      "epoch": 15.97883597883598,
+      "grad_norm": 0.23792731761932373,
+      "learning_rate": 2.1280095389007836e-05,
+      "loss": 0.0184,
+      "step": 6040
+    },
+    {
+      "epoch": 16.005291005291006,
+      "grad_norm": 0.19010236859321594,
+      "learning_rate": 2.1011051908030387e-05,
+      "loss": 0.0167,
+      "step": 6050
+    },
+    {
+      "epoch": 16.03174603174603,
+      "grad_norm": 0.325145959854126,
+      "learning_rate": 2.074352023119034e-05,
+      "loss": 0.0168,
+      "step": 6060
+    },
+    {
+      "epoch": 16.058201058201057,
+      "grad_norm": 0.20572441816329956,
+      "learning_rate": 2.0477505478893823e-05,
+      "loss": 0.0162,
+      "step": 6070
+    },
+    {
+      "epoch": 16.084656084656086,
+      "grad_norm": 0.15513984858989716,
+      "learning_rate": 2.0213012742513927e-05,
+      "loss": 0.0211,
+      "step": 6080
+    },
+    {
+      "epoch": 16.11111111111111,
+      "grad_norm": 0.18062983453273773,
+      "learning_rate": 1.9950047084293187e-05,
+      "loss": 0.0171,
+      "step": 6090
+    },
+    {
+      "epoch": 16.137566137566136,
+      "grad_norm": 0.20299988985061646,
+      "learning_rate": 1.968861353724668e-05,
+      "loss": 0.0173,
+      "step": 6100
+    },
+    {
+      "epoch": 16.164021164021165,
+      "grad_norm": 0.23924869298934937,
+      "learning_rate": 1.9428717105065864e-05,
+      "loss": 0.0176,
+      "step": 6110
+    },
+    {
+      "epoch": 16.19047619047619,
+      "grad_norm": 0.24165087938308716,
+      "learning_rate": 1.917036276202253e-05,
+      "loss": 0.0195,
+      "step": 6120
+    },
+    {
+      "epoch": 16.21693121693122,
+      "grad_norm": 0.1869451254606247,
+      "learning_rate": 1.891355545287382e-05,
+      "loss": 0.0185,
+      "step": 6130
+    },
+    {
+      "epoch": 16.243386243386244,
+      "grad_norm": 0.19750215113162994,
+      "learning_rate": 1.8658300092767544e-05,
+      "loss": 0.0174,
+      "step": 6140
+    },
+    {
+      "epoch": 16.26984126984127,
+      "grad_norm": 0.2032746970653534,
+      "learning_rate": 1.8404601567148026e-05,
+      "loss": 0.0177,
+      "step": 6150
+    },
+    {
+      "epoch": 16.296296296296298,
+      "grad_norm": 0.18800148367881775,
+      "learning_rate": 1.8152464731662666e-05,
+      "loss": 0.0166,
+      "step": 6160
+    },
+    {
+      "epoch": 16.322751322751323,
+      "grad_norm": 0.16922627389431,
+      "learning_rate": 1.790189441206899e-05,
+      "loss": 0.0158,
+      "step": 6170
+    },
+    {
+      "epoch": 16.349206349206348,
+      "grad_norm": 0.21169210970401764,
+      "learning_rate": 1.76528954041423e-05,
+      "loss": 0.0167,
+      "step": 6180
+    },
+    {
+      "epoch": 16.375661375661377,
+      "grad_norm": 0.24106602370738983,
+      "learning_rate": 1.740547247358384e-05,
+      "loss": 0.0157,
+      "step": 6190
+    },
+    {
+      "epoch": 16.402116402116402,
+      "grad_norm": 0.20507925748825073,
+      "learning_rate": 1.715963035592969e-05,
+      "loss": 0.0179,
+      "step": 6200
+    },
+    {
+      "epoch": 16.428571428571427,
+      "grad_norm": 0.18841566145420074,
+      "learning_rate": 1.6915373756460006e-05,
+      "loss": 0.0179,
+      "step": 6210
+    },
+    {
+      "epoch": 16.455026455026456,
+      "grad_norm": 0.20179235935211182,
+      "learning_rate": 1.6672707350108973e-05,
+      "loss": 0.0168,
+      "step": 6220
+    },
+    {
+      "epoch": 16.48148148148148,
+      "grad_norm": 0.1482539176940918,
+      "learning_rate": 1.64316357813755e-05,
+      "loss": 0.0149,
+      "step": 6230
+    },
+    {
+      "epoch": 16.507936507936506,
+      "grad_norm": 0.23648443818092346,
+      "learning_rate": 1.6192163664234082e-05,
+      "loss": 0.0193,
+      "step": 6240
+    },
+    {
+      "epoch": 16.534391534391535,
+      "grad_norm": 0.1912665218114853,
+      "learning_rate": 1.5954295582046642e-05,
+      "loss": 0.0159,
+      "step": 6250
+    },
+    {
+      "epoch": 16.56084656084656,
+      "grad_norm": 0.20422996580600739,
+      "learning_rate": 1.5718036087474796e-05,
+      "loss": 0.0171,
+      "step": 6260
+    },
+    {
+      "epoch": 16.58730158730159,
+      "grad_norm": 0.13860765099525452,
+      "learning_rate": 1.5483389702392657e-05,
+      "loss": 0.0164,
+      "step": 6270
+    },
+    {
+      "epoch": 16.613756613756614,
+      "grad_norm": 0.20141619443893433,
+      "learning_rate": 1.5250360917800356e-05,
+      "loss": 0.0163,
+      "step": 6280
+    },
+    {
+      "epoch": 16.64021164021164,
+      "grad_norm": 0.2113092839717865,
+      "learning_rate": 1.5018954193738077e-05,
+      "loss": 0.0176,
+      "step": 6290
+    },
+    {
+      "epoch": 16.666666666666668,
+      "grad_norm": 0.23735524713993073,
+      "learning_rate": 1.478917395920062e-05,
+      "loss": 0.0157,
+      "step": 6300
+    },
+    {
+      "epoch": 16.693121693121693,
+      "grad_norm": 0.1875062733888626,
+      "learning_rate": 1.4561024612052754e-05,
+      "loss": 0.0161,
+      "step": 6310
+    },
+    {
+      "epoch": 16.719576719576718,
+      "grad_norm": 0.18014168739318848,
+      "learning_rate": 1.4334510518944932e-05,
+      "loss": 0.0162,
+      "step": 6320
+    },
+    {
+      "epoch": 16.746031746031747,
+      "grad_norm": 0.23076355457305908,
+      "learning_rate": 1.4109636015229766e-05,
+      "loss": 0.0177,
+      "step": 6330
+    },
+    {
+      "epoch": 16.772486772486772,
+      "grad_norm": 0.21106301248073578,
+      "learning_rate": 1.3886405404879055e-05,
+      "loss": 0.0165,
+      "step": 6340
+    },
+    {
+      "epoch": 16.798941798941797,
+      "grad_norm": 0.24004612863063812,
+      "learning_rate": 1.3664822960401413e-05,
+      "loss": 0.0174,
+      "step": 6350
+    },
+    {
+      "epoch": 16.825396825396826,
+      "grad_norm": 0.17413417994976044,
+      "learning_rate": 1.3444892922760444e-05,
+      "loss": 0.0143,
+      "step": 6360
+    },
+    {
+      "epoch": 16.85185185185185,
+      "grad_norm": 0.1467863917350769,
+      "learning_rate": 1.3226619501293692e-05,
+      "loss": 0.0166,
+      "step": 6370
+    },
+    {
+      "epoch": 16.87830687830688,
+      "grad_norm": 0.17535103857517242,
+      "learning_rate": 1.3010006873631919e-05,
+      "loss": 0.0162,
+      "step": 6380
+    },
+    {
+      "epoch": 16.904761904761905,
+      "grad_norm": 0.16379229724407196,
+      "learning_rate": 1.2795059185619229e-05,
+      "loss": 0.0167,
+      "step": 6390
+    },
+    {
+      "epoch": 16.93121693121693,
+      "grad_norm": 0.17413221299648285,
+      "learning_rate": 1.2581780551233801e-05,
+      "loss": 0.0179,
+      "step": 6400
+    },
+    {
+      "epoch": 16.95767195767196,
+      "grad_norm": 0.2470989227294922,
+      "learning_rate": 1.237017505250897e-05,
+      "loss": 0.0179,
+      "step": 6410
+    },
+    {
+      "epoch": 16.984126984126984,
+      "grad_norm": 0.16806478798389435,
+      "learning_rate": 1.2160246739455206e-05,
+      "loss": 0.0193,
+      "step": 6420
+    },
+    {
+      "epoch": 17.01058201058201,
+      "grad_norm": 0.19091880321502686,
+      "learning_rate": 1.195199962998268e-05,
+      "loss": 0.0158,
+      "step": 6430
+    },
+    {
+      "epoch": 17.037037037037038,
+      "grad_norm": 0.20383872091770172,
+      "learning_rate": 1.1745437709824114e-05,
+      "loss": 0.0161,
+      "step": 6440
+    },
+    {
+      "epoch": 17.063492063492063,
+      "grad_norm": 0.15551620721817017,
+      "learning_rate": 1.1540564932458753e-05,
+      "loss": 0.0155,
+      "step": 6450
+    },
+    {
+      "epoch": 17.08994708994709,
+      "grad_norm": 0.2658255100250244,
+      "learning_rate": 1.1337385219036623e-05,
+      "loss": 0.0176,
+      "step": 6460
+    },
+    {
+      "epoch": 17.116402116402117,
+      "grad_norm": 0.2648050785064697,
+      "learning_rate": 1.1135902458303393e-05,
+      "loss": 0.0152,
+      "step": 6470
+    },
+    {
+      "epoch": 17.142857142857142,
+      "grad_norm": 0.17353011667728424,
+      "learning_rate": 1.0936120506525994e-05,
+      "loss": 0.0151,
+      "step": 6480
+    },
+    {
+      "epoch": 17.16931216931217,
+      "grad_norm": 0.1721569001674652,
+      "learning_rate": 1.0738043187418922e-05,
+      "loss": 0.0193,
+      "step": 6490
+    },
+    {
+      "epoch": 17.195767195767196,
+      "grad_norm": 0.2267681360244751,
+      "learning_rate": 1.0541674292070868e-05,
+      "loss": 0.0173,
+      "step": 6500
+    },
+    {
+      "epoch": 17.22222222222222,
+      "grad_norm": 0.28807544708251953,
+      "learning_rate": 1.0347017578872276e-05,
+      "loss": 0.0181,
+      "step": 6510
+    },
+    {
+      "epoch": 17.24867724867725,
+      "grad_norm": 0.1608646959066391,
+      "learning_rate": 1.0154076773443432e-05,
+      "loss": 0.0166,
+      "step": 6520
+    },
+    {
+      "epoch": 17.275132275132275,
+      "grad_norm": 0.26136720180511475,
+      "learning_rate": 9.962855568563067e-06,
+      "loss": 0.0166,
+      "step": 6530
+    },
+    {
+      "epoch": 17.3015873015873,
+      "grad_norm": 0.17855539917945862,
+      "learning_rate": 9.773357624097678e-06,
+      "loss": 0.0152,
+      "step": 6540
+    },
+    {
+      "epoch": 17.32804232804233,
+      "grad_norm": 0.14467965066432953,
+      "learning_rate": 9.585586566931625e-06,
+      "loss": 0.0161,
+      "step": 6550
+    },
+    {
+      "epoch": 17.354497354497354,
+      "grad_norm": 0.2321748286485672,
+      "learning_rate": 9.39954599089754e-06,
+      "loss": 0.0161,
+      "step": 6560
+    },
+    {
+      "epoch": 17.38095238095238,
+      "grad_norm": 0.18389014899730682,
+      "learning_rate": 9.215239456707635e-06,
+      "loss": 0.0163,
+      "step": 6570
+    },
+    {
+      "epoch": 17.40740740740741,
+      "grad_norm": 0.13891808688640594,
+      "learning_rate": 9.032670491885575e-06,
+      "loss": 0.0138,
+      "step": 6580
+    },
+    {
+      "epoch": 17.433862433862434,
+      "grad_norm": 0.21685755252838135,
+      "learning_rate": 8.851842590698877e-06,
+      "loss": 0.0169,
+      "step": 6590
+    },
+    {
+      "epoch": 17.46031746031746,
+      "grad_norm": 0.1880199909210205,
+      "learning_rate": 8.672759214092042e-06,
+      "loss": 0.0146,
+      "step": 6600
+    },
+    {
+      "epoch": 17.486772486772487,
+      "grad_norm": 0.14832906424999237,
+      "learning_rate": 8.49542378962046e-06,
+      "loss": 0.0177,
+      "step": 6610
+    },
+    {
+      "epoch": 17.513227513227513,
+      "grad_norm": 0.1760573536157608,
+      "learning_rate": 8.319839711384603e-06,
+      "loss": 0.0181,
+      "step": 6620
+    },
+    {
+      "epoch": 17.53968253968254,
+      "grad_norm": 0.19962358474731445,
+      "learning_rate": 8.146010339965193e-06,
+      "loss": 0.0173,
+      "step": 6630
+    },
+    {
+      "epoch": 17.566137566137566,
+      "grad_norm": 0.16452929377555847,
+      "learning_rate": 7.973939002358833e-06,
+      "loss": 0.0198,
+      "step": 6640
+    },
+    {
+      "epoch": 17.59259259259259,
+      "grad_norm": 0.15749408304691315,
+      "learning_rate": 7.803628991914358e-06,
+      "loss": 0.0191,
+      "step": 6650
+    },
+    {
+      "epoch": 17.61904761904762,
+      "grad_norm": 0.25544941425323486,
+      "learning_rate": 7.635083568269818e-06,
+      "loss": 0.015,
+      "step": 6660
+    },
+    {
+      "epoch": 17.645502645502646,
+      "grad_norm": 0.19979047775268555,
+      "learning_rate": 7.468305957290012e-06,
+      "loss": 0.0173,
+      "step": 6670
+    },
+    {
+      "epoch": 17.67195767195767,
+      "grad_norm": 0.20462428033351898,
+      "learning_rate": 7.3032993510048e-06,
+      "loss": 0.0168,
+      "step": 6680
+    },
+    {
+      "epoch": 17.6984126984127,
+      "grad_norm": 0.16061115264892578,
+      "learning_rate": 7.14006690754806e-06,
+      "loss": 0.0153,
+      "step": 6690
+    },
+    {
+      "epoch": 17.724867724867725,
+      "grad_norm": 0.2491457462310791,
+      "learning_rate": 6.9786117510971415e-06,
+      "loss": 0.017,
+      "step": 6700
+    },
+    {
+      "epoch": 17.75132275132275,
+      "grad_norm": 0.21707351505756378,
+      "learning_rate": 6.8189369718131165e-06,
+      "loss": 0.0162,
+      "step": 6710
+    },
+    {
+      "epoch": 17.77777777777778,
+      "grad_norm": 0.1476307213306427,
+      "learning_rate": 6.661045625781659e-06,
+      "loss": 0.0177,
+      "step": 6720
+    },
+    {
+      "epoch": 17.804232804232804,
+      "grad_norm": 0.1483127921819687,
+      "learning_rate": 6.504940734954512e-06,
+      "loss": 0.0176,
+      "step": 6730
+    },
+    {
+      "epoch": 17.83068783068783,
+      "grad_norm": 0.2029716670513153,
+      "learning_rate": 6.350625287091683e-06,
+      "loss": 0.0163,
+      "step": 6740
+    },
+    {
+      "epoch": 17.857142857142858,
+      "grad_norm": 0.17894791066646576,
+      "learning_rate": 6.198102235704251e-06,
+      "loss": 0.0184,
+      "step": 6750
+    },
+    {
+      "epoch": 17.883597883597883,
+      "grad_norm": 0.1477387249469757,
+      "learning_rate": 6.047374499997827e-06,
+      "loss": 0.0159,
+      "step": 6760
+    },
+    {
+      "epoch": 17.91005291005291,
+      "grad_norm": 0.2098730206489563,
+      "learning_rate": 5.898444964816652e-06,
+      "loss": 0.0168,
+      "step": 6770
+    },
+    {
+      "epoch": 17.936507936507937,
+      "grad_norm": 0.15353702008724213,
+      "learning_rate": 5.751316480588475e-06,
+      "loss": 0.0164,
+      "step": 6780
+    },
+    {
+      "epoch": 17.962962962962962,
+      "grad_norm": 0.15158168971538544,
+      "learning_rate": 5.6059918632699215e-06,
+      "loss": 0.0176,
+      "step": 6790
+    },
+    {
+      "epoch": 17.98941798941799,
+      "grad_norm": 0.19479572772979736,
+      "learning_rate": 5.462473894292597e-06,
+      "loss": 0.017,
+      "step": 6800
+    },
+    {
+      "epoch": 18.015873015873016,
+      "grad_norm": 0.1755516678094864,
+      "learning_rate": 5.3207653205098905e-06,
+      "loss": 0.0165,
+      "step": 6810
+    },
+    {
+      "epoch": 18.04232804232804,
+      "grad_norm": 0.2217196822166443,
+      "learning_rate": 5.18086885414436e-06,
+      "loss": 0.0185,
+      "step": 6820
+    },
+    {
+      "epoch": 18.06878306878307,
+      "grad_norm": 0.1340661495923996,
+      "learning_rate": 5.042787172735863e-06,
+      "loss": 0.0157,
+      "step": 6830
+    },
+    {
+      "epoch": 18.095238095238095,
+      "grad_norm": 0.2034100890159607,
+      "learning_rate": 4.906522919090318e-06,
+      "loss": 0.0144,
+      "step": 6840
+    },
+    {
+      "epoch": 18.12169312169312,
+      "grad_norm": 0.21613432466983795,
+      "learning_rate": 4.772078701229044e-06,
+      "loss": 0.0174,
+      "step": 6850
+    },
+    {
+      "epoch": 18.14814814814815,
+      "grad_norm": 0.20390327274799347,
+      "learning_rate": 4.6394570923389055e-06,
+      "loss": 0.0142,
+      "step": 6860
+    },
+    {
+      "epoch": 18.174603174603174,
+      "grad_norm": 0.2700379192829132,
+      "learning_rate": 4.508660630723116e-06,
+      "loss": 0.0161,
+      "step": 6870
+    },
+    {
+      "epoch": 18.201058201058203,
+      "grad_norm": 0.21042710542678833,
+      "learning_rate": 4.379691819752507e-06,
+      "loss": 0.0171,
+      "step": 6880
+    },
+    {
+      "epoch": 18.227513227513228,
+      "grad_norm": 0.20383331179618835,
+      "learning_rate": 4.252553127817749e-06,
+      "loss": 0.0165,
+      "step": 6890
+    },
+    {
+      "epoch": 18.253968253968253,
+      "grad_norm": 0.21164458990097046,
+      "learning_rate": 4.1272469882820745e-06,
+      "loss": 0.0163,
+      "step": 6900
+    },
+    {
+      "epoch": 18.280423280423282,
+      "grad_norm": 0.1821017563343048,
+      "learning_rate": 4.00377579943465e-06,
+      "loss": 0.0187,
+      "step": 6910
+    },
+    {
+      "epoch": 18.306878306878307,
+      "grad_norm": 0.1857336163520813,
+      "learning_rate": 3.882141924444727e-06,
+      "loss": 0.0175,
+      "step": 6920
+    },
+    {
+      "epoch": 18.333333333333332,
+      "grad_norm": 0.20543745160102844,
+      "learning_rate": 3.762347691316437e-06,
+      "loss": 0.0157,
+      "step": 6930
+    },
+    {
+      "epoch": 18.35978835978836,
+      "grad_norm": 0.202567458152771,
+      "learning_rate": 3.6443953928441687e-06,
+      "loss": 0.0141,
+      "step": 6940
+    },
+    {
+      "epoch": 18.386243386243386,
+      "grad_norm": 0.19636744260787964,
+      "learning_rate": 3.52828728656871e-06,
+      "loss": 0.017,
+      "step": 6950
+    },
+    {
+      "epoch": 18.41269841269841,
+      "grad_norm": 0.22278830409049988,
+      "learning_rate": 3.4140255947340605e-06,
+      "loss": 0.0173,
+      "step": 6960
+    },
+    {
+      "epoch": 18.43915343915344,
+      "grad_norm": 0.19748513400554657,
+      "learning_rate": 3.3016125042448776e-06,
+      "loss": 0.0192,
+      "step": 6970
+    },
+    {
+      "epoch": 18.465608465608465,
+      "grad_norm": 0.18418549001216888,
+      "learning_rate": 3.191050166624632e-06,
+      "loss": 0.0161,
+      "step": 6980
+    },
+    {
+      "epoch": 18.49206349206349,
+      "grad_norm": 0.21692697703838348,
+      "learning_rate": 3.082340697974395e-06,
+      "loss": 0.0156,
+      "step": 6990
+    },
+    {
+      "epoch": 18.51851851851852,
+      "grad_norm": 0.14934659004211426,
+      "learning_rate": 2.9754861789324073e-06,
+      "loss": 0.0161,
+      "step": 7000
+    },
+    {
+      "epoch": 18.544973544973544,
+      "grad_norm": 0.20961934328079224,
+      "learning_rate": 2.8704886546341956e-06,
+      "loss": 0.0164,
+      "step": 7010
+    },
+    {
+      "epoch": 18.571428571428573,
+      "grad_norm": 0.16422855854034424,
+      "learning_rate": 2.767350134673441e-06,
+      "loss": 0.0153,
+      "step": 7020
+    },
+    {
+      "epoch": 18.597883597883598,
+      "grad_norm": 0.17347002029418945,
+      "learning_rate": 2.66607259306354e-06,
+      "loss": 0.0157,
+      "step": 7030
+    },
+    {
+      "epoch": 18.624338624338623,
+      "grad_norm": 0.10744515806436539,
+      "learning_rate": 2.5666579681998036e-06,
+      "loss": 0.0163,
+      "step": 7040
+    },
+    {
+      "epoch": 18.650793650793652,
+      "grad_norm": 0.16614548861980438,
+      "learning_rate": 2.4691081628223533e-06,
+      "loss": 0.0143,
+      "step": 7050
+    },
+    {
+      "epoch": 18.677248677248677,
+      "grad_norm": 0.23060919344425201,
+      "learning_rate": 2.3734250439797155e-06,
+      "loss": 0.0156,
+      "step": 7060
+    },
+    {
+      "epoch": 18.703703703703702,
+      "grad_norm": 0.21790482103824615,
+      "learning_rate": 2.2796104429930963e-06,
+      "loss": 0.017,
+      "step": 7070
+    },
+    {
+      "epoch": 18.73015873015873,
+      "grad_norm": 0.19607238471508026,
+      "learning_rate": 2.1876661554213197e-06,
+      "loss": 0.0147,
+      "step": 7080
+    },
+    {
+      "epoch": 18.756613756613756,
+      "grad_norm": 0.14693410694599152,
+      "learning_rate": 2.097593941026421e-06,
+      "loss": 0.0171,
+      "step": 7090
+    },
+    {
+      "epoch": 18.78306878306878,
+      "grad_norm": 0.15378044545650482,
+      "learning_rate": 2.009395523740054e-06,
+      "loss": 0.0153,
+      "step": 7100
+    },
+    {
+      "epoch": 18.80952380952381,
+      "grad_norm": 0.1617782860994339,
+      "learning_rate": 1.923072591630415e-06,
+      "loss": 0.0169,
+      "step": 7110
+    },
+    {
+      "epoch": 18.835978835978835,
+      "grad_norm": 0.16385476291179657,
+      "learning_rate": 1.838626796869991e-06,
+      "loss": 0.0164,
+      "step": 7120
+    },
+    {
+      "epoch": 18.862433862433864,
+      "grad_norm": 0.16427811980247498,
+      "learning_rate": 1.7560597557038982e-06,
+      "loss": 0.0172,
+      "step": 7130
+    },
+    {
+      "epoch": 18.88888888888889,
+      "grad_norm": 0.15692010521888733,
+      "learning_rate": 1.6753730484189822e-06,
+      "loss": 0.0158,
+      "step": 7140
+    },
+    {
+      "epoch": 18.915343915343914,
+      "grad_norm": 0.17159394919872284,
+      "learning_rate": 1.59656821931351e-06,
+      "loss": 0.0144,
+      "step": 7150
+    },
+    {
+      "epoch": 18.941798941798943,
+      "grad_norm": 0.13447363674640656,
+      "learning_rate": 1.5196467766677047e-06,
+      "loss": 0.0162,
+      "step": 7160
+    },
+    {
+      "epoch": 18.96825396825397,
+      "grad_norm": 0.16987329721450806,
+      "learning_rate": 1.4446101927148014e-06,
+      "loss": 0.0166,
+      "step": 7170
+    },
+    {
+      "epoch": 18.994708994708994,
+      "grad_norm": 0.1536870151758194,
+      "learning_rate": 1.371459903612915e-06,
+      "loss": 0.0168,
+      "step": 7180
+    },
+    {
+      "epoch": 19.021164021164022,
+      "grad_norm": 0.21800269186496735,
+      "learning_rate": 1.3001973094175169e-06,
+      "loss": 0.0172,
+      "step": 7190
+    },
+    {
+      "epoch": 19.047619047619047,
+      "grad_norm": 0.17673099040985107,
+      "learning_rate": 1.2308237740546791e-06,
+      "loss": 0.0163,
+      "step": 7200
+    },
+    {
+      "epoch": 19.074074074074073,
+      "grad_norm": 0.15784895420074463,
+      "learning_rate": 1.1633406252949174e-06,
+      "loss": 0.0155,
+      "step": 7210
+    },
+    {
+      "epoch": 19.1005291005291,
+      "grad_norm": 0.1905474215745926,
+      "learning_rate": 1.097749154727834e-06,
+      "loss": 0.0139,
+      "step": 7220
+    },
+    {
+      "epoch": 19.126984126984127,
+      "grad_norm": 0.1780664622783661,
+      "learning_rate": 1.034050617737381e-06,
+      "loss": 0.0172,
+      "step": 7230
+    },
+    {
+      "epoch": 19.15343915343915,
+      "grad_norm": 0.16737215220928192,
+      "learning_rate": 9.722462334777805e-07,
+      "loss": 0.0149,
+      "step": 7240
+    },
+    {
+      "epoch": 19.17989417989418,
+      "grad_norm": 0.17369748651981354,
+      "learning_rate": 9.123371848502871e-07,
+      "loss": 0.0169,
+      "step": 7250
+    },
+    {
+      "epoch": 19.206349206349206,
+      "grad_norm": 0.17061293125152588,
+      "learning_rate": 8.54324618480462e-07,
+      "loss": 0.0148,
+      "step": 7260
+    },
+    {
+      "epoch": 19.232804232804234,
+      "grad_norm": 0.2849694490432739,
+      "learning_rate": 7.982096446962684e-07,
+      "loss": 0.0172,
+      "step": 7270
+    },
+    {
+      "epoch": 19.25925925925926,
+      "grad_norm": 0.16373994946479797,
+      "learning_rate": 7.439933375068097e-07,
+      "loss": 0.0155,
+      "step": 7280
+    },
+    {
+      "epoch": 19.285714285714285,
+      "grad_norm": 0.1858544796705246,
+      "learning_rate": 6.91676734581781e-07,
+      "loss": 0.0178,
+      "step": 7290
+    },
+    {
+      "epoch": 19.312169312169313,
+      "grad_norm": 0.22137916088104248,
+      "learning_rate": 6.412608372315943e-07,
+      "loss": 0.0158,
+      "step": 7300
+    },
+    {
+      "epoch": 19.33862433862434,
+      "grad_norm": 0.21592257916927338,
+      "learning_rate": 5.927466103882395e-07,
+      "loss": 0.0147,
+      "step": 7310
+    },
+    {
+      "epoch": 19.365079365079364,
+      "grad_norm": 0.16293910145759583,
+      "learning_rate": 5.461349825867879e-07,
+      "loss": 0.0164,
+      "step": 7320
+    },
+    {
+      "epoch": 19.391534391534393,
+      "grad_norm": 0.19730526208877563,
+      "learning_rate": 5.014268459476501e-07,
+      "loss": 0.0149,
+      "step": 7330
+    },
+    {
+      "epoch": 19.417989417989418,
+      "grad_norm": 0.32044050097465515,
+      "learning_rate": 4.586230561594462e-07,
+      "loss": 0.015,
+      "step": 7340
+    },
+    {
+      "epoch": 19.444444444444443,
+      "grad_norm": 0.1815277338027954,
+      "learning_rate": 4.177244324627183e-07,
+      "loss": 0.0148,
+      "step": 7350
+    },
+    {
+      "epoch": 19.47089947089947,
+      "grad_norm": 0.19404444098472595,
+      "learning_rate": 3.7873175763415427e-07,
+      "loss": 0.0198,
+      "step": 7360
+    },
+    {
+      "epoch": 19.497354497354497,
+      "grad_norm": 0.24406063556671143,
+      "learning_rate": 3.416457779716664e-07,
+      "loss": 0.0179,
+      "step": 7370
+    },
+    {
+      "epoch": 19.523809523809526,
+      "grad_norm": 0.19478613138198853,
+      "learning_rate": 3.0646720328010303e-07,
+      "loss": 0.0168,
+      "step": 7380
+    },
+    {
+      "epoch": 19.55026455026455,
+      "grad_norm": 0.1256127655506134,
+      "learning_rate": 2.731967068576369e-07,
+      "loss": 0.0178,
+      "step": 7390
+    },
+    {
+      "epoch": 19.576719576719576,
+      "grad_norm": 0.17166411876678467,
+      "learning_rate": 2.4183492548288667e-07,
+      "loss": 0.0165,
+      "step": 7400
+    },
+    {
+      "epoch": 19.603174603174605,
+      "grad_norm": 0.1664625108242035,
+      "learning_rate": 2.1238245940276014e-07,
+      "loss": 0.0148,
+      "step": 7410
+    },
+    {
+      "epoch": 19.62962962962963,
+      "grad_norm": 0.1637825071811676,
+      "learning_rate": 1.8483987232094101e-07,
+      "loss": 0.0165,
+      "step": 7420
+    },
+    {
+      "epoch": 19.656084656084655,
+      "grad_norm": 0.17463766038417816,
+      "learning_rate": 1.5920769138706438e-07,
+      "loss": 0.0154,
+      "step": 7430
+    },
+    {
+      "epoch": 19.682539682539684,
+      "grad_norm": 0.1705995351076126,
+      "learning_rate": 1.3548640718669124e-07,
+      "loss": 0.0151,
+      "step": 7440
+    },
+    {
+      "epoch": 19.70899470899471,
+      "grad_norm": 0.14153502881526947,
+      "learning_rate": 1.1367647373190516e-07,
+      "loss": 0.016,
+      "step": 7450
+    },
+    {
+      "epoch": 19.735449735449734,
+      "grad_norm": 0.2553439438343048,
+      "learning_rate": 9.377830845258561e-08,
+      "loss": 0.0162,
+      "step": 7460
+    },
+    {
+      "epoch": 19.761904761904763,
+      "grad_norm": 0.1255766898393631,
+      "learning_rate": 7.579229218843686e-08,
+      "loss": 0.0158,
+      "step": 7470
+    },
+    {
+      "epoch": 19.788359788359788,
+      "grad_norm": 0.12066392600536346,
+      "learning_rate": 5.971876918172692e-08,
+      "loss": 0.016,
+      "step": 7480
+    },
+    {
+      "epoch": 19.814814814814813,
+      "grad_norm": 0.14524872601032257,
+      "learning_rate": 4.555804707067069e-08,
+      "loss": 0.0133,
+      "step": 7490
+    },
+    {
+      "epoch": 19.841269841269842,
+      "grad_norm": 0.19802507758140564,
+      "learning_rate": 3.331039688353465e-08,
+      "loss": 0.0157,
+      "step": 7500
+    },
+    {
+      "epoch": 19.867724867724867,
+      "grad_norm": 0.16935652494430542,
+      "learning_rate": 2.297605303347439e-08,
+      "loss": 0.0162,
+      "step": 7510
+    },
+    {
+      "epoch": 19.894179894179896,
+      "grad_norm": 0.18010082840919495,
+      "learning_rate": 1.4555213314026984e-08,
+      "loss": 0.0155,
+      "step": 7520
+    },
+    {
+      "epoch": 19.92063492063492,
+      "grad_norm": 0.14799325168132782,
+      "learning_rate": 8.04803889533634e-09,
+      "loss": 0.0161,
+      "step": 7530
+    },
+    {
+      "epoch": 19.947089947089946,
+      "grad_norm": 0.1539669781923294,
+      "learning_rate": 3.4546543210778325e-09,
+      "loss": 0.0154,
+      "step": 7540
+    },
+    {
+      "epoch": 19.973544973544975,
+      "grad_norm": 0.1273125857114792,
+      "learning_rate": 7.751475060491231e-10,
+      "loss": 0.0144,
+      "step": 7550
+    },
+    {
+      "epoch": 19.997354497354497,
+      "step": 7559,
+      "total_flos": 5.331242062105344e+17,
+      "train_loss": 0.030525887037320242,
+      "train_runtime": 4964.64,
+      "train_samples_per_second": 97.444,
+      "train_steps_per_second": 1.523
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7559,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 20,
+  "save_steps": 10000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 5.331242062105344e+17,
+  "train_batch_size": 64,
+  "trial_name": null,
+  "trial_params": null
+}