diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4905 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9990309034133737,
+  "eval_steps": 500,
+  "global_step": 6963,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004307095940562076,
+      "grad_norm": 9.035698890686035,
+      "learning_rate": 5.730659025787966e-07,
+      "loss": 1.2789,
+      "step": 10
+    },
+    {
+      "epoch": 0.008614191881124151,
+      "grad_norm": 6.42362642288208,
+      "learning_rate": 1.1461318051575932e-06,
+      "loss": 1.1797,
+      "step": 20
+    },
+    {
+      "epoch": 0.012921287821686228,
+      "grad_norm": 1.6813161373138428,
+      "learning_rate": 1.7191977077363897e-06,
+      "loss": 0.9736,
+      "step": 30
+    },
+    {
+      "epoch": 0.017228383762248303,
+      "grad_norm": 0.9371042251586914,
+      "learning_rate": 2.2922636103151864e-06,
+      "loss": 0.871,
+      "step": 40
+    },
+    {
+      "epoch": 0.02153547970281038,
+      "grad_norm": 0.7695503234863281,
+      "learning_rate": 2.865329512893983e-06,
+      "loss": 0.8025,
+      "step": 50
+    },
+    {
+      "epoch": 0.025842575643372456,
+      "grad_norm": 0.716374397277832,
+      "learning_rate": 3.4383954154727795e-06,
+      "loss": 0.7698,
+      "step": 60
+    },
+    {
+      "epoch": 0.030149671583934532,
+      "grad_norm": 0.6864265203475952,
+      "learning_rate": 4.011461318051576e-06,
+      "loss": 0.7701,
+      "step": 70
+    },
+    {
+      "epoch": 0.034456767524496605,
+      "grad_norm": 0.7581704258918762,
+      "learning_rate": 4.584527220630373e-06,
+      "loss": 0.7619,
+      "step": 80
+    },
+    {
+      "epoch": 0.03876386346505868,
+      "grad_norm": 0.6616791486740112,
+      "learning_rate": 5.157593123209169e-06,
+      "loss": 0.7296,
+      "step": 90
+    },
+    {
+      "epoch": 0.04307095940562076,
+      "grad_norm": 0.6397051811218262,
+      "learning_rate": 5.730659025787966e-06,
+      "loss": 0.7453,
+      "step": 100
+    },
+    {
+      "epoch": 0.047378055346182835,
+      "grad_norm": 0.6572911143302917,
+      "learning_rate": 6.303724928366762e-06,
+      "loss": 0.767,
+      "step": 110
+    },
+    {
+      "epoch": 0.05168515128674491,
+      "grad_norm": 0.669222354888916,
+      "learning_rate": 6.876790830945559e-06,
+      "loss": 0.7369,
+      "step": 120
+    },
+    {
+      "epoch": 0.05599224722730699,
+      "grad_norm": 0.6517964601516724,
+      "learning_rate": 7.449856733524356e-06,
+      "loss": 0.7186,
+      "step": 130
+    },
+    {
+      "epoch": 0.060299343167869064,
+      "grad_norm": 0.6209223866462708,
+      "learning_rate": 8.022922636103152e-06,
+      "loss": 0.7155,
+      "step": 140
+    },
+    {
+      "epoch": 0.06460643910843114,
+      "grad_norm": 0.6591508388519287,
+      "learning_rate": 8.595988538681949e-06,
+      "loss": 0.7289,
+      "step": 150
+    },
+    {
+      "epoch": 0.06891353504899321,
+      "grad_norm": 0.5842370390892029,
+      "learning_rate": 9.169054441260746e-06,
+      "loss": 0.7183,
+      "step": 160
+    },
+    {
+      "epoch": 0.0732206309895553,
+      "grad_norm": 0.7117204070091248,
+      "learning_rate": 9.742120343839543e-06,
+      "loss": 0.7192,
+      "step": 170
+    },
+    {
+      "epoch": 0.07752772693011736,
+      "grad_norm": 0.6163178086280823,
+      "learning_rate": 1.0315186246418338e-05,
+      "loss": 0.7193,
+      "step": 180
+    },
+    {
+      "epoch": 0.08183482287067945,
+      "grad_norm": 0.5932906270027161,
+      "learning_rate": 1.0888252148997137e-05,
+      "loss": 0.714,
+      "step": 190
+    },
+    {
+      "epoch": 0.08614191881124152,
+      "grad_norm": 0.5982919335365295,
+      "learning_rate": 1.1461318051575932e-05,
+      "loss": 0.7058,
+      "step": 200
+    },
+    {
+      "epoch": 0.0904490147518036,
+      "grad_norm": 0.6208463907241821,
+      "learning_rate": 1.2034383954154729e-05,
+      "loss": 0.7189,
+      "step": 210
+    },
+    {
+      "epoch": 0.09475611069236567,
+      "grad_norm": 0.5887411236763,
+      "learning_rate": 1.2607449856733524e-05,
+      "loss": 0.7249,
+      "step": 220
+    },
+    {
+      "epoch": 0.09906320663292775,
+      "grad_norm": 0.5963988900184631,
+      "learning_rate": 1.3180515759312323e-05,
+      "loss": 0.7293,
+      "step": 230
+    },
+    {
+      "epoch": 0.10337030257348982,
+      "grad_norm": 0.5715692043304443,
+      "learning_rate": 1.3753581661891118e-05,
+      "loss": 0.6845,
+      "step": 240
+    },
+    {
+      "epoch": 0.1076773985140519,
+      "grad_norm": 0.639398455619812,
+      "learning_rate": 1.4326647564469915e-05,
+      "loss": 0.6994,
+      "step": 250
+    },
+    {
+      "epoch": 0.11198449445461398,
+      "grad_norm": 0.6884477734565735,
+      "learning_rate": 1.4899713467048712e-05,
+      "loss": 0.7126,
+      "step": 260
+    },
+    {
+      "epoch": 0.11629159039517606,
+      "grad_norm": 0.6021578907966614,
+      "learning_rate": 1.5472779369627507e-05,
+      "loss": 0.7215,
+      "step": 270
+    },
+    {
+      "epoch": 0.12059868633573813,
+      "grad_norm": 0.6716468930244446,
+      "learning_rate": 1.6045845272206304e-05,
+      "loss": 0.6969,
+      "step": 280
+    },
+    {
+      "epoch": 0.1249057822763002,
+      "grad_norm": 0.5783571600914001,
+      "learning_rate": 1.66189111747851e-05,
+      "loss": 0.7111,
+      "step": 290
+    },
+    {
+      "epoch": 0.12921287821686228,
+      "grad_norm": 0.5546681880950928,
+      "learning_rate": 1.7191977077363898e-05,
+      "loss": 0.7,
+      "step": 300
+    },
+    {
+      "epoch": 0.13351997415742436,
+      "grad_norm": 0.5409330129623413,
+      "learning_rate": 1.7765042979942695e-05,
+      "loss": 0.696,
+      "step": 310
+    },
+    {
+      "epoch": 0.13782707009798642,
+      "grad_norm": 0.5752865672111511,
+      "learning_rate": 1.833810888252149e-05,
+      "loss": 0.6883,
+      "step": 320
+    },
+    {
+      "epoch": 0.1421341660385485,
+      "grad_norm": 0.6340565085411072,
+      "learning_rate": 1.891117478510029e-05,
+      "loss": 0.6881,
+      "step": 330
+    },
+    {
+      "epoch": 0.1464412619791106,
+      "grad_norm": 0.5298891067504883,
+      "learning_rate": 1.9484240687679085e-05,
+      "loss": 0.6935,
+      "step": 340
+    },
+    {
+      "epoch": 0.15074835791967267,
+      "grad_norm": 0.5659753680229187,
+      "learning_rate": 1.9999998871916207e-05,
+      "loss": 0.7103,
+      "step": 350
+    },
+    {
+      "epoch": 0.15505545386023473,
+      "grad_norm": 0.6017744541168213,
+      "learning_rate": 1.999986350216883e-05,
+      "loss": 0.6855,
+      "step": 360
+    },
+    {
+      "epoch": 0.1593625498007968,
+      "grad_norm": 0.5426760911941528,
+      "learning_rate": 1.999950251916212e-05,
+      "loss": 0.6914,
+      "step": 370
+    },
+    {
+      "epoch": 0.1636696457413589,
+      "grad_norm": 0.5532637238502502,
+      "learning_rate": 1.999891593104044e-05,
+      "loss": 0.6895,
+      "step": 380
+    },
+    {
+      "epoch": 0.16797674168192098,
+      "grad_norm": 0.5581168532371521,
+      "learning_rate": 1.9998103751038177e-05,
+      "loss": 0.6897,
+      "step": 390
+    },
+    {
+      "epoch": 0.17228383762248303,
+      "grad_norm": 0.5208210945129395,
+      "learning_rate": 1.9997065997479442e-05,
+      "loss": 0.6889,
+      "step": 400
+    },
+    {
+      "epoch": 0.17659093356304512,
+      "grad_norm": 0.5863595604896545,
+      "learning_rate": 1.9995802693777644e-05,
+      "loss": 0.6905,
+      "step": 410
+    },
+    {
+      "epoch": 0.1808980295036072,
+      "grad_norm": 0.5605342984199524,
+      "learning_rate": 1.9994313868434988e-05,
+      "loss": 0.6815,
+      "step": 420
+    },
+    {
+      "epoch": 0.18520512544416926,
+      "grad_norm": 0.5580301880836487,
+      "learning_rate": 1.9992599555041798e-05,
+      "loss": 0.7067,
+      "step": 430
+    },
+    {
+      "epoch": 0.18951222138473134,
+      "grad_norm": 0.558312177658081,
+      "learning_rate": 1.999065979227579e-05,
+      "loss": 0.7061,
+      "step": 440
+    },
+    {
+      "epoch": 0.19381931732529342,
+      "grad_norm": 0.5273975133895874,
+      "learning_rate": 1.998849462390118e-05,
+      "loss": 0.6905,
+      "step": 450
+    },
+    {
+      "epoch": 0.1981264132658555,
+      "grad_norm": 0.4772217571735382,
+      "learning_rate": 1.9986104098767703e-05,
+      "loss": 0.686,
+      "step": 460
+    },
+    {
+      "epoch": 0.20243350920641756,
+      "grad_norm": 0.5336763858795166,
+      "learning_rate": 1.9983488270809515e-05,
+      "loss": 0.6861,
+      "step": 470
+    },
+    {
+      "epoch": 0.20674060514697964,
+      "grad_norm": 0.4961983859539032,
+      "learning_rate": 1.9980647199043966e-05,
+      "loss": 0.6882,
+      "step": 480
+    },
+    {
+      "epoch": 0.21104770108754173,
+      "grad_norm": 0.5408128499984741,
+      "learning_rate": 1.9977580947570275e-05,
+      "loss": 0.7001,
+      "step": 490
+    },
+    {
+      "epoch": 0.2153547970281038,
+      "grad_norm": 0.5350680351257324,
+      "learning_rate": 1.997428958556809e-05,
+      "loss": 0.6931,
+      "step": 500
+    },
+    {
+      "epoch": 0.21966189296866587,
+      "grad_norm": 0.5455281734466553,
+      "learning_rate": 1.9970773187295917e-05,
+      "loss": 0.6919,
+      "step": 510
+    },
+    {
+      "epoch": 0.22396898890922795,
+      "grad_norm": 0.524664580821991,
+      "learning_rate": 1.9967031832089438e-05,
+      "loss": 0.6738,
+      "step": 520
+    },
+    {
+      "epoch": 0.22827608484979003,
+      "grad_norm": 0.48598727583885193,
+      "learning_rate": 1.9963065604359746e-05,
+      "loss": 0.6678,
+      "step": 530
+    },
+    {
+      "epoch": 0.23258318079035212,
+      "grad_norm": 0.5560494065284729,
+      "learning_rate": 1.9958874593591418e-05,
+      "loss": 0.694,
+      "step": 540
+    },
+    {
+      "epoch": 0.23689027673091417,
+      "grad_norm": 0.5516777038574219,
+      "learning_rate": 1.99544588943405e-05,
+      "loss": 0.6715,
+      "step": 550
+    },
+    {
+      "epoch": 0.24119737267147626,
+      "grad_norm": 0.5097941756248474,
+      "learning_rate": 1.9949818606232393e-05,
+      "loss": 0.6782,
+      "step": 560
+    },
+    {
+      "epoch": 0.24550446861203834,
+      "grad_norm": 0.5353350639343262,
+      "learning_rate": 1.9944953833959567e-05,
+      "loss": 0.6904,
+      "step": 570
+    },
+    {
+      "epoch": 0.2498115645526004,
+      "grad_norm": 0.5160298943519592,
+      "learning_rate": 1.9939864687279237e-05,
+      "loss": 0.6756,
+      "step": 580
+    },
+    {
+      "epoch": 0.2541186604931625,
+      "grad_norm": 0.5377163887023926,
+      "learning_rate": 1.993455128101087e-05,
+      "loss": 0.712,
+      "step": 590
+    },
+    {
+      "epoch": 0.25842575643372456,
+      "grad_norm": 0.47318100929260254,
+      "learning_rate": 1.992901373503359e-05,
+      "loss": 0.6648,
+      "step": 600
+    },
+    {
+      "epoch": 0.2627328523742866,
+      "grad_norm": 0.4977729916572571,
+      "learning_rate": 1.992325217428348e-05,
+      "loss": 0.6893,
+      "step": 610
+    },
+    {
+      "epoch": 0.26703994831484873,
+      "grad_norm": 0.5569038391113281,
+      "learning_rate": 1.991726672875077e-05,
+      "loss": 0.6876,
+      "step": 620
+    },
+    {
+      "epoch": 0.2713470442554108,
+      "grad_norm": 0.544884443283081,
+      "learning_rate": 1.9911057533476884e-05,
+      "loss": 0.6736,
+      "step": 630
+    },
+    {
+      "epoch": 0.27565414019597284,
+      "grad_norm": 0.5159808993339539,
+      "learning_rate": 1.9904624728551417e-05,
+      "loss": 0.674,
+      "step": 640
+    },
+    {
+      "epoch": 0.27996123613653495,
+      "grad_norm": 0.48680537939071655,
+      "learning_rate": 1.989796845910896e-05,
+      "loss": 0.6903,
+      "step": 650
+    },
+    {
+      "epoch": 0.284268332077097,
+      "grad_norm": 0.527867317199707,
+      "learning_rate": 1.9891088875325827e-05,
+      "loss": 0.6693,
+      "step": 660
+    },
+    {
+      "epoch": 0.2885754280176591,
+      "grad_norm": 0.5441365838050842,
+      "learning_rate": 1.988398613241666e-05,
+      "loss": 0.6721,
+      "step": 670
+    },
+    {
+      "epoch": 0.2928825239582212,
+      "grad_norm": 0.5693966150283813,
+      "learning_rate": 1.9876660390630954e-05,
+      "loss": 0.6684,
+      "step": 680
+    },
+    {
+      "epoch": 0.29718961989878323,
+      "grad_norm": 0.5607503652572632,
+      "learning_rate": 1.986911181524941e-05,
+      "loss": 0.6783,
+      "step": 690
+    },
+    {
+      "epoch": 0.30149671583934534,
+      "grad_norm": 0.5421719551086426,
+      "learning_rate": 1.9861340576580225e-05,
+      "loss": 0.6658,
+      "step": 700
+    },
+    {
+      "epoch": 0.3058038117799074,
+      "grad_norm": 0.497612863779068,
+      "learning_rate": 1.9853346849955236e-05,
+      "loss": 0.6816,
+      "step": 710
+    },
+    {
+      "epoch": 0.31011090772046945,
+      "grad_norm": 0.5503632426261902,
+      "learning_rate": 1.984513081572598e-05,
+      "loss": 0.6663,
+      "step": 720
+    },
+    {
+      "epoch": 0.31441800366103156,
+      "grad_norm": 0.5319767594337463,
+      "learning_rate": 1.983669265925961e-05,
+      "loss": 0.6513,
+      "step": 730
+    },
+    {
+      "epoch": 0.3187250996015936,
+      "grad_norm": 0.5350950956344604,
+      "learning_rate": 1.9828032570934726e-05,
+      "loss": 0.6699,
+      "step": 740
+    },
+    {
+      "epoch": 0.3230321955421557,
+      "grad_norm": 0.5330127477645874,
+      "learning_rate": 1.9819150746137067e-05,
+      "loss": 0.6786,
+      "step": 750
+    },
+    {
+      "epoch": 0.3273392914827178,
+      "grad_norm": 0.4740910232067108,
+      "learning_rate": 1.981004738525512e-05,
+      "loss": 0.6867,
+      "step": 760
+    },
+    {
+      "epoch": 0.33164638742327984,
+      "grad_norm": 0.5131900906562805,
+      "learning_rate": 1.980072269367557e-05,
+      "loss": 0.6618,
+      "step": 770
+    },
+    {
+      "epoch": 0.33595348336384195,
+      "grad_norm": 0.4712623059749603,
+      "learning_rate": 1.97911768817787e-05,
+      "loss": 0.6863,
+      "step": 780
+    },
+    {
+      "epoch": 0.340260579304404,
+      "grad_norm": 0.5240254998207092,
+      "learning_rate": 1.9781410164933626e-05,
+      "loss": 0.6941,
+      "step": 790
+    },
+    {
+      "epoch": 0.34456767524496607,
+      "grad_norm": 0.5192612409591675,
+      "learning_rate": 1.9771422763493434e-05,
+      "loss": 0.6726,
+      "step": 800
+    },
+    {
+      "epoch": 0.3488747711855282,
+      "grad_norm": 0.4864448010921478,
+      "learning_rate": 1.9761214902790217e-05,
+      "loss": 0.6541,
+      "step": 810
+    },
+    {
+      "epoch": 0.35318186712609023,
+      "grad_norm": 0.5248873829841614,
+      "learning_rate": 1.9750786813129995e-05,
+      "loss": 0.6713,
+      "step": 820
+    },
+    {
+      "epoch": 0.3574889630666523,
+      "grad_norm": 0.5010212659835815,
+      "learning_rate": 1.9740138729787505e-05,
+      "loss": 0.6793,
+      "step": 830
+    },
+    {
+      "epoch": 0.3617960590072144,
+      "grad_norm": 0.4966225326061249,
+      "learning_rate": 1.9729270893000913e-05,
+      "loss": 0.6692,
+      "step": 840
+    },
+    {
+      "epoch": 0.36610315494777645,
+      "grad_norm": 0.48576685786247253,
+      "learning_rate": 1.9718183547966366e-05,
+      "loss": 0.6812,
+      "step": 850
+    },
+    {
+      "epoch": 0.3704102508883385,
+      "grad_norm": 0.5232109427452087,
+      "learning_rate": 1.9706876944832486e-05,
+      "loss": 0.6567,
+      "step": 860
+    },
+    {
+      "epoch": 0.3747173468289006,
+      "grad_norm": 0.4847777485847473,
+      "learning_rate": 1.9695351338694713e-05,
+      "loss": 0.6638,
+      "step": 870
+    },
+    {
+      "epoch": 0.3790244427694627,
+      "grad_norm": 0.49412795901298523,
+      "learning_rate": 1.9683606989589553e-05,
+      "loss": 0.6731,
+      "step": 880
+    },
+    {
+      "epoch": 0.3833315387100248,
+      "grad_norm": 0.5143546462059021,
+      "learning_rate": 1.9671644162488716e-05,
+      "loss": 0.6779,
+      "step": 890
+    },
+    {
+      "epoch": 0.38763863465058684,
+      "grad_norm": 0.5516107082366943,
+      "learning_rate": 1.965946312729312e-05,
+      "loss": 0.6798,
+      "step": 900
+    },
+    {
+      "epoch": 0.3919457305911489,
+      "grad_norm": 0.5140990018844604,
+      "learning_rate": 1.9647064158826825e-05,
+      "loss": 0.6473,
+      "step": 910
+    },
+    {
+      "epoch": 0.396252826531711,
+      "grad_norm": 0.4911974370479584,
+      "learning_rate": 1.9634447536830815e-05,
+      "loss": 0.6565,
+      "step": 920
+    },
+    {
+      "epoch": 0.40055992247227307,
+      "grad_norm": 0.4995877742767334,
+      "learning_rate": 1.9621613545956703e-05,
+      "loss": 0.6514,
+      "step": 930
+    },
+    {
+      "epoch": 0.4048670184128351,
+      "grad_norm": 0.48752328753471375,
+      "learning_rate": 1.9608562475760287e-05,
+      "loss": 0.6751,
+      "step": 940
+    },
+    {
+      "epoch": 0.40917411435339723,
+      "grad_norm": 0.4956004321575165,
+      "learning_rate": 1.9595294620695036e-05,
+      "loss": 0.6492,
+      "step": 950
+    },
+    {
+      "epoch": 0.4134812102939593,
+      "grad_norm": 0.48215603828430176,
+      "learning_rate": 1.958181028010544e-05,
+      "loss": 0.6741,
+      "step": 960
+    },
+    {
+      "epoch": 0.4177883062345214,
+      "grad_norm": 0.48835939168930054,
+      "learning_rate": 1.9568109758220253e-05,
+      "loss": 0.6638,
+      "step": 970
+    },
+    {
+      "epoch": 0.42209540217508346,
+      "grad_norm": 0.47754788398742676,
+      "learning_rate": 1.9554193364145635e-05,
+      "loss": 0.6657,
+      "step": 980
+    },
+    {
+      "epoch": 0.4264024981156455,
+      "grad_norm": 0.5080917477607727,
+      "learning_rate": 1.9540061411858172e-05,
+      "loss": 0.6675,
+      "step": 990
+    },
+    {
+      "epoch": 0.4307095940562076,
+      "grad_norm": 0.4634297788143158,
+      "learning_rate": 1.9525714220197802e-05,
+      "loss": 0.6693,
+      "step": 1000
+    },
+    {
+      "epoch": 0.4350166899967697,
+      "grad_norm": 0.4760366678237915,
+      "learning_rate": 1.951115211286061e-05,
+      "loss": 0.6721,
+      "step": 1010
+    },
+    {
+      "epoch": 0.43932378593733173,
+      "grad_norm": 0.5227916836738586,
+      "learning_rate": 1.9496375418391525e-05,
+      "loss": 0.6691,
+      "step": 1020
+    },
+    {
+      "epoch": 0.44363088187789385,
+      "grad_norm": 0.5157990455627441,
+      "learning_rate": 1.948138447017692e-05,
+      "loss": 0.6774,
+      "step": 1030
+    },
+    {
+      "epoch": 0.4479379778184559,
+      "grad_norm": 0.49596408009529114,
+      "learning_rate": 1.9466179606437087e-05,
+      "loss": 0.6313,
+      "step": 1040
+    },
+    {
+      "epoch": 0.45224507375901796,
+      "grad_norm": 0.47041237354278564,
+      "learning_rate": 1.945076117021859e-05,
+      "loss": 0.6724,
+      "step": 1050
+    },
+    {
+      "epoch": 0.45655216969958007,
+      "grad_norm": 0.5206364989280701,
+      "learning_rate": 1.9435129509386538e-05,
+      "loss": 0.6843,
+      "step": 1060
+    },
+    {
+      "epoch": 0.4608592656401421,
+      "grad_norm": 0.5067657828330994,
+      "learning_rate": 1.9419284976616745e-05,
+      "loss": 0.6649,
+      "step": 1070
+    },
+    {
+      "epoch": 0.46516636158070424,
+      "grad_norm": 1.3445152044296265,
+      "learning_rate": 1.9403227929387756e-05,
+      "loss": 0.6548,
+      "step": 1080
+    },
+    {
+      "epoch": 0.4694734575212663,
+      "grad_norm": 0.5465224385261536,
+      "learning_rate": 1.93869587299728e-05,
+      "loss": 0.6427,
+      "step": 1090
+    },
+    {
+      "epoch": 0.47378055346182835,
+      "grad_norm": 0.49137911200523376,
+      "learning_rate": 1.9370477745431587e-05,
+      "loss": 0.6519,
+      "step": 1100
+    },
+    {
+      "epoch": 0.47808764940239046,
+      "grad_norm": 0.48190736770629883,
+      "learning_rate": 1.935378534760206e-05,
+      "loss": 0.6615,
+      "step": 1110
+    },
+    {
+      "epoch": 0.4823947453429525,
+      "grad_norm": 0.4869353771209717,
+      "learning_rate": 1.9336881913091992e-05,
+      "loss": 0.65,
+      "step": 1120
+    },
+    {
+      "epoch": 0.48670184128351457,
+      "grad_norm": 0.4473590552806854,
+      "learning_rate": 1.931976782327048e-05,
+      "loss": 0.6821,
+      "step": 1130
+    },
+    {
+      "epoch": 0.4910089372240767,
+      "grad_norm": 0.4703207314014435,
+      "learning_rate": 1.9302443464259352e-05,
+      "loss": 0.657,
+      "step": 1140
+    },
+    {
+      "epoch": 0.49531603316463874,
+      "grad_norm": 0.48172295093536377,
+      "learning_rate": 1.9284909226924457e-05,
+      "loss": 0.6581,
+      "step": 1150
+    },
+    {
+      "epoch": 0.4996231291052008,
+      "grad_norm": 0.4986841082572937,
+      "learning_rate": 1.9267165506866835e-05,
+      "loss": 0.664,
+      "step": 1160
+    },
+    {
+      "epoch": 0.5039302250457629,
+      "grad_norm": 0.4936910569667816,
+      "learning_rate": 1.9249212704413803e-05,
+      "loss": 0.6409,
+      "step": 1170
+    },
+    {
+      "epoch": 0.508237320986325,
+      "grad_norm": 0.48618724942207336,
+      "learning_rate": 1.9231051224609918e-05,
+      "loss": 0.6566,
+      "step": 1180
+    },
+    {
+      "epoch": 0.512544416926887,
+      "grad_norm": 0.5300356149673462,
+      "learning_rate": 1.921268147720784e-05,
+      "loss": 0.6533,
+      "step": 1190
+    },
+    {
+      "epoch": 0.5168515128674491,
+      "grad_norm": 0.4799743890762329,
+      "learning_rate": 1.919410387665908e-05,
+      "loss": 0.6677,
+      "step": 1200
+    },
+    {
+      "epoch": 0.5211586088080112,
+      "grad_norm": 0.5317394137382507,
+      "learning_rate": 1.9175318842104667e-05,
+      "loss": 0.6464,
+      "step": 1210
+    },
+    {
+      "epoch": 0.5254657047485732,
+      "grad_norm": 0.49199768900871277,
+      "learning_rate": 1.9156326797365665e-05,
+      "loss": 0.6655,
+      "step": 1220
+    },
+    {
+      "epoch": 0.5297728006891353,
+      "grad_norm": 0.4916874170303345,
+      "learning_rate": 1.913712817093364e-05,
+      "loss": 0.6372,
+      "step": 1230
+    },
+    {
+      "epoch": 0.5340798966296975,
+      "grad_norm": 0.48562970757484436,
+      "learning_rate": 1.9117723395960972e-05,
+      "loss": 0.6639,
+      "step": 1240
+    },
+    {
+      "epoch": 0.5383869925702595,
+      "grad_norm": 0.5152992010116577,
+      "learning_rate": 1.909811291025109e-05,
+      "loss": 0.6609,
+      "step": 1250
+    },
+    {
+      "epoch": 0.5426940885108216,
+      "grad_norm": 0.48352181911468506,
+      "learning_rate": 1.907829715624859e-05,
+      "loss": 0.6726,
+      "step": 1260
+    },
+    {
+      "epoch": 0.5470011844513837,
+      "grad_norm": 0.5064017176628113,
+      "learning_rate": 1.905827658102926e-05,
+      "loss": 0.6698,
+      "step": 1270
+    },
+    {
+      "epoch": 0.5513082803919457,
+      "grad_norm": 0.46494290232658386,
+      "learning_rate": 1.9038051636289997e-05,
+      "loss": 0.68,
+      "step": 1280
+    },
+    {
+      "epoch": 0.5556153763325078,
+      "grad_norm": 0.4788792133331299,
+      "learning_rate": 1.9017622778338585e-05,
+      "loss": 0.6501,
+      "step": 1290
+    },
+    {
+      "epoch": 0.5599224722730699,
+      "grad_norm": 0.4712987542152405,
+      "learning_rate": 1.8996990468083448e-05,
+      "loss": 0.6488,
+      "step": 1300
+    },
+    {
+      "epoch": 0.5642295682136319,
+      "grad_norm": 0.4997137784957886,
+      "learning_rate": 1.8976155171023216e-05,
+      "loss": 0.6518,
+      "step": 1310
+    },
+    {
+      "epoch": 0.568536664154194,
+      "grad_norm": 0.5003030896186829,
+      "learning_rate": 1.895511735723623e-05,
+      "loss": 0.6317,
+      "step": 1320
+    },
+    {
+      "epoch": 0.5728437600947561,
+      "grad_norm": 0.4551664888858795,
+      "learning_rate": 1.8933877501369944e-05,
+      "loss": 0.6634,
+      "step": 1330
+    },
+    {
+      "epoch": 0.5771508560353182,
+      "grad_norm": 0.532534122467041,
+      "learning_rate": 1.891243608263021e-05,
+      "loss": 0.6656,
+      "step": 1340
+    },
+    {
+      "epoch": 0.5814579519758802,
+      "grad_norm": 0.47166600823402405,
+      "learning_rate": 1.889079358477047e-05,
+      "loss": 0.657,
+      "step": 1350
+    },
+    {
+      "epoch": 0.5857650479164423,
+      "grad_norm": 0.45552805066108704,
+      "learning_rate": 1.8868950496080832e-05,
+      "loss": 0.6652,
+      "step": 1360
+    },
+    {
+      "epoch": 0.5900721438570045,
+      "grad_norm": 0.5267536044120789,
+      "learning_rate": 1.884690730937707e-05,
+      "loss": 0.6463,
+      "step": 1370
+    },
+    {
+      "epoch": 0.5943792397975665,
+      "grad_norm": 0.49093228578567505,
+      "learning_rate": 1.882466452198949e-05,
+      "loss": 0.6604,
+      "step": 1380
+    },
+    {
+      "epoch": 0.5986863357381286,
+      "grad_norm": 0.5105960369110107,
+      "learning_rate": 1.880222263575172e-05,
+      "loss": 0.6457,
+      "step": 1390
+    },
+    {
+      "epoch": 0.6029934316786907,
+      "grad_norm": 0.47326135635375977,
+      "learning_rate": 1.8779582156989384e-05,
+      "loss": 0.6464,
+      "step": 1400
+    },
+    {
+      "epoch": 0.6073005276192527,
+      "grad_norm": 0.4910115599632263,
+      "learning_rate": 1.875674359650867e-05,
+      "loss": 0.6547,
+      "step": 1410
+    },
+    {
+      "epoch": 0.6116076235598148,
+      "grad_norm": 0.48352956771850586,
+      "learning_rate": 1.873370746958482e-05,
+      "loss": 0.654,
+      "step": 1420
+    },
+    {
+      "epoch": 0.6159147195003769,
+      "grad_norm": 0.4722056984901428,
+      "learning_rate": 1.871047429595049e-05,
+      "loss": 0.6372,
+      "step": 1430
+    },
+    {
+      "epoch": 0.6202218154409389,
+      "grad_norm": 0.4340212345123291,
+      "learning_rate": 1.868704459978405e-05,
+      "loss": 0.6507,
+      "step": 1440
+    },
+    {
+      "epoch": 0.624528911381501,
+      "grad_norm": 0.48497867584228516,
+      "learning_rate": 1.8663418909697723e-05,
+      "loss": 0.6349,
+      "step": 1450
+    },
+    {
+      "epoch": 0.6288360073220631,
+      "grad_norm": 0.4707370102405548,
+      "learning_rate": 1.863959775872567e-05,
+      "loss": 0.6445,
+      "step": 1460
+    },
+    {
+      "epoch": 0.6331431032626251,
+      "grad_norm": 0.5151925683021545,
+      "learning_rate": 1.861558168431199e-05,
+      "loss": 0.6493,
+      "step": 1470
+    },
+    {
+      "epoch": 0.6374501992031872,
+      "grad_norm": 0.47226110100746155,
+      "learning_rate": 1.8591371228298554e-05,
+      "loss": 0.6211,
+      "step": 1480
+    },
+    {
+      "epoch": 0.6417572951437494,
+      "grad_norm": 0.48166829347610474,
+      "learning_rate": 1.856696693691281e-05,
+      "loss": 0.6476,
+      "step": 1490
+    },
+    {
+      "epoch": 0.6460643910843114,
+      "grad_norm": 0.5039719343185425,
+      "learning_rate": 1.8542369360755448e-05,
+      "loss": 0.636,
+      "step": 1500
+    },
+    {
+      "epoch": 0.6503714870248735,
+      "grad_norm": 0.45818519592285156,
+      "learning_rate": 1.8517579054787974e-05,
+      "loss": 0.658,
+      "step": 1510
+    },
+    {
+      "epoch": 0.6546785829654356,
+      "grad_norm": 0.4803057014942169,
+      "learning_rate": 1.8492596578320194e-05,
+      "loss": 0.6468,
+      "step": 1520
+    },
+    {
+      "epoch": 0.6589856789059977,
+      "grad_norm": 0.480227530002594,
+      "learning_rate": 1.8467422494997593e-05,
+      "loss": 0.641,
+      "step": 1530
+    },
+    {
+      "epoch": 0.6632927748465597,
+      "grad_norm": 0.49187588691711426,
+      "learning_rate": 1.844205737278863e-05,
+      "loss": 0.6572,
+      "step": 1540
+    },
+    {
+      "epoch": 0.6675998707871218,
+      "grad_norm": 0.49701517820358276,
+      "learning_rate": 1.84165017839719e-05,
+      "loss": 0.6567,
+      "step": 1550
+    },
+    {
+      "epoch": 0.6719069667276839,
+      "grad_norm": 0.48368483781814575,
+      "learning_rate": 1.8390756305123246e-05,
+      "loss": 0.669,
+      "step": 1560
+    },
+    {
+      "epoch": 0.6762140626682459,
+      "grad_norm": 0.5007254481315613,
+      "learning_rate": 1.836482151710273e-05,
+      "loss": 0.6448,
+      "step": 1570
+    },
+    {
+      "epoch": 0.680521158608808,
+      "grad_norm": 0.44526585936546326,
+      "learning_rate": 1.8338698005041556e-05,
+      "loss": 0.6386,
+      "step": 1580
+    },
+    {
+      "epoch": 0.6848282545493701,
+      "grad_norm": 0.4812663197517395,
+      "learning_rate": 1.8312386358328828e-05,
+      "loss": 0.6447,
+      "step": 1590
+    },
+    {
+      "epoch": 0.6891353504899321,
+      "grad_norm": 0.4910503029823303,
+      "learning_rate": 1.828588717059829e-05,
+      "loss": 0.6449,
+      "step": 1600
+    },
+    {
+      "epoch": 0.6934424464304942,
+      "grad_norm": 0.47431930899620056,
+      "learning_rate": 1.8259201039714914e-05,
+      "loss": 0.6372,
+      "step": 1610
+    },
+    {
+      "epoch": 0.6977495423710564,
+      "grad_norm": 0.5024338364601135,
+      "learning_rate": 1.8232328567761416e-05,
+      "loss": 0.6433,
+      "step": 1620
+    },
+    {
+      "epoch": 0.7020566383116184,
+      "grad_norm": 0.47510799765586853,
+      "learning_rate": 1.820527036102467e-05,
+      "loss": 0.6601,
+      "step": 1630
+    },
+    {
+      "epoch": 0.7063637342521805,
+      "grad_norm": 0.47990313172340393,
+      "learning_rate": 1.8178027029982027e-05,
+      "loss": 0.6463,
+      "step": 1640
+    },
+    {
+      "epoch": 0.7106708301927426,
+      "grad_norm": 0.5117030739784241,
+      "learning_rate": 1.8150599189287553e-05,
+      "loss": 0.6455,
+      "step": 1650
+    },
+    {
+      "epoch": 0.7149779261333046,
+      "grad_norm": 0.4917861819267273,
+      "learning_rate": 1.8122987457758147e-05,
+      "loss": 0.6688,
+      "step": 1660
+    },
+    {
+      "epoch": 0.7192850220738667,
+      "grad_norm": 0.49872297048568726,
+      "learning_rate": 1.8095192458359588e-05,
+      "loss": 0.6513,
+      "step": 1670
+    },
+    {
+      "epoch": 0.7235921180144288,
+      "grad_norm": 0.47510796785354614,
+      "learning_rate": 1.806721481819247e-05,
+      "loss": 0.649,
+      "step": 1680
+    },
+    {
+      "epoch": 0.7278992139549908,
+      "grad_norm": 0.4924173057079315,
+      "learning_rate": 1.8039055168478074e-05,
+      "loss": 0.6177,
+      "step": 1690
+    },
+    {
+      "epoch": 0.7322063098955529,
+      "grad_norm": 0.4918348789215088,
+      "learning_rate": 1.8010714144544104e-05,
+      "loss": 0.6543,
+      "step": 1700
+    },
+    {
+      "epoch": 0.736513405836115,
+      "grad_norm": 0.45298415422439575,
+      "learning_rate": 1.7982192385810372e-05,
+      "loss": 0.6367,
+      "step": 1710
+    },
+    {
+      "epoch": 0.740820501776677,
+      "grad_norm": 0.46879851818084717,
+      "learning_rate": 1.795349053577435e-05,
+      "loss": 0.6414,
+      "step": 1720
+    },
+    {
+      "epoch": 0.7451275977172391,
+      "grad_norm": 0.4573706388473511,
+      "learning_rate": 1.7924609241996672e-05,
+      "loss": 0.628,
+      "step": 1730
+    },
+    {
+      "epoch": 0.7494346936578012,
+      "grad_norm": 0.46929094195365906,
+      "learning_rate": 1.7895549156086514e-05,
+      "loss": 0.6478,
+      "step": 1740
+    },
+    {
+      "epoch": 0.7537417895983634,
+      "grad_norm": 0.5428628325462341,
+      "learning_rate": 1.78663109336869e-05,
+      "loss": 0.6405,
+      "step": 1750
+    },
+    {
+      "epoch": 0.7580488855389254,
+      "grad_norm": 0.47853079438209534,
+      "learning_rate": 1.78368952344599e-05,
+      "loss": 0.6442,
+      "step": 1760
+    },
+    {
+      "epoch": 0.7623559814794875,
+      "grad_norm": 0.46747061610221863,
+      "learning_rate": 1.7807302722071742e-05,
+      "loss": 0.6369,
+      "step": 1770
+    },
+    {
+      "epoch": 0.7666630774200496,
+      "grad_norm": 0.5107671022415161,
+      "learning_rate": 1.7777534064177864e-05,
+      "loss": 0.6322,
+      "step": 1780
+    },
+    {
+      "epoch": 0.7709701733606116,
+      "grad_norm": 0.5013517141342163,
+      "learning_rate": 1.7747589932407826e-05,
+      "loss": 0.6384,
+      "step": 1790
+    },
+    {
+      "epoch": 0.7752772693011737,
+      "grad_norm": 0.5039073824882507,
+      "learning_rate": 1.7717471002350162e-05,
+      "loss": 0.6504,
+      "step": 1800
+    },
+    {
+      "epoch": 0.7795843652417358,
+      "grad_norm": 0.4767347276210785,
+      "learning_rate": 1.7687177953537148e-05,
+      "loss": 0.645,
+      "step": 1810
+    },
+    {
+      "epoch": 0.7838914611822978,
+      "grad_norm": 0.4766087532043457,
+      "learning_rate": 1.7656711469429464e-05,
+      "loss": 0.6249,
+      "step": 1820
+    },
+    {
+      "epoch": 0.7881985571228599,
+      "grad_norm": 0.5031486749649048,
+      "learning_rate": 1.7626072237400764e-05,
+      "loss": 0.6263,
+      "step": 1830
+    },
+    {
+      "epoch": 0.792505653063422,
+      "grad_norm": 0.444658488035202,
+      "learning_rate": 1.759526094872219e-05,
+      "loss": 0.6561,
+      "step": 1840
+    },
+    {
+      "epoch": 0.796812749003984,
+      "grad_norm": 0.5070600509643555,
+      "learning_rate": 1.7564278298546758e-05,
+      "loss": 0.6477,
+      "step": 1850
+    },
+    {
+      "epoch": 0.8011198449445461,
+      "grad_norm": 0.45487794280052185,
+      "learning_rate": 1.753312498589367e-05,
+      "loss": 0.6257,
+      "step": 1860
+    },
+    {
+      "epoch": 0.8054269408851082,
+      "grad_norm": 0.4745471477508545,
+      "learning_rate": 1.7501801713632568e-05,
+      "loss": 0.6586,
+      "step": 1870
+    },
+    {
+      "epoch": 0.8097340368256702,
+      "grad_norm": 0.4743909537792206,
+      "learning_rate": 1.7470309188467645e-05,
+      "loss": 0.6255,
+      "step": 1880
+    },
+    {
+      "epoch": 0.8140411327662324,
+      "grad_norm": 0.5165956020355225,
+      "learning_rate": 1.7438648120921736e-05,
+      "loss": 0.6592,
+      "step": 1890
+    },
+    {
+      "epoch": 0.8183482287067945,
+      "grad_norm": 0.455861359834671,
+      "learning_rate": 1.740681922532025e-05,
+      "loss": 0.6467,
+      "step": 1900
+    },
+    {
+      "epoch": 0.8226553246473565,
+      "grad_norm": 0.468013733625412,
+      "learning_rate": 1.7374823219775073e-05,
+      "loss": 0.6382,
+      "step": 1910
+    },
+    {
+      "epoch": 0.8269624205879186,
+      "grad_norm": 0.46119919419288635,
+      "learning_rate": 1.7342660826168374e-05,
+      "loss": 0.6437,
+      "step": 1920
+    },
+    {
+      "epoch": 0.8312695165284807,
+      "grad_norm": 0.4399983286857605,
+      "learning_rate": 1.73103327701363e-05,
+      "loss": 0.6379,
+      "step": 1930
+    },
+    {
+      "epoch": 0.8355766124690428,
+      "grad_norm": 0.46829739212989807,
+      "learning_rate": 1.7277839781052617e-05,
+      "loss": 0.6402,
+      "step": 1940
+    },
+    {
+      "epoch": 0.8398837084096048,
+      "grad_norm": 0.5193459987640381,
+      "learning_rate": 1.7245182592012248e-05,
+      "loss": 0.6348,
+      "step": 1950
+    },
+    {
+      "epoch": 0.8441908043501669,
+      "grad_norm": 0.5310715436935425,
+      "learning_rate": 1.7212361939814735e-05,
+      "loss": 0.6351,
+      "step": 1960
+    },
+    {
+      "epoch": 0.848497900290729,
+      "grad_norm": 0.4883059561252594,
+      "learning_rate": 1.7179378564947615e-05,
+      "loss": 0.6401,
+      "step": 1970
+    },
+    {
+      "epoch": 0.852804996231291,
+      "grad_norm": 0.5028474926948547,
+      "learning_rate": 1.7146233211569723e-05,
+      "loss": 0.6559,
+      "step": 1980
+    },
+    {
+      "epoch": 0.8571120921718531,
+      "grad_norm": 0.48668941855430603,
+      "learning_rate": 1.7112926627494385e-05,
+      "loss": 0.6572,
+      "step": 1990
+    },
+    {
+      "epoch": 0.8614191881124152,
+      "grad_norm": 0.4668605327606201,
+      "learning_rate": 1.7079459564172555e-05,
+      "loss": 0.6321,
+      "step": 2000
+    },
+    {
+      "epoch": 0.8657262840529772,
+      "grad_norm": 0.4556910991668701,
+      "learning_rate": 1.7045832776675863e-05,
+      "loss": 0.6268,
+      "step": 2010
+    },
+    {
+      "epoch": 0.8700333799935394,
+      "grad_norm": 0.45260846614837646,
+      "learning_rate": 1.701204702367958e-05,
+      "loss": 0.6271,
+      "step": 2020
+    },
+    {
+      "epoch": 0.8743404759341015,
+      "grad_norm": 0.4828309714794159,
+      "learning_rate": 1.6978103067445494e-05,
+      "loss": 0.6351,
+      "step": 2030
+    },
+    {
+      "epoch": 0.8786475718746635,
+      "grad_norm": 0.4691152274608612,
+      "learning_rate": 1.6944001673804723e-05,
+      "loss": 0.6512,
+      "step": 2040
+    },
+    {
+      "epoch": 0.8829546678152256,
+      "grad_norm": 0.4812765419483185,
+      "learning_rate": 1.6909743612140417e-05,
+      "loss": 0.6335,
+      "step": 2050
+    },
+    {
+      "epoch": 0.8872617637557877,
+      "grad_norm": 0.4415755867958069,
+      "learning_rate": 1.687532965537043e-05,
+      "loss": 0.6541,
+      "step": 2060
+    },
+    {
+      "epoch": 0.8915688596963497,
+      "grad_norm": 0.4993227422237396,
+      "learning_rate": 1.6840760579929846e-05,
+      "loss": 0.6318,
+      "step": 2070
+    },
+    {
+      "epoch": 0.8958759556369118,
+      "grad_norm": 0.4628779888153076,
+      "learning_rate": 1.6806037165753498e-05,
+      "loss": 0.6369,
+      "step": 2080
+    },
+    {
+      "epoch": 0.9001830515774739,
+      "grad_norm": 0.5235878229141235,
+      "learning_rate": 1.677116019625834e-05,
+      "loss": 0.6415,
+      "step": 2090
+    },
+    {
+      "epoch": 0.9044901475180359,
+      "grad_norm": 0.4750138819217682,
+      "learning_rate": 1.6736130458325793e-05,
+      "loss": 0.6101,
+      "step": 2100
+    },
+    {
+      "epoch": 0.908797243458598,
+      "grad_norm": 0.5292583107948303,
+      "learning_rate": 1.6700948742283977e-05,
+      "loss": 0.6248,
+      "step": 2110
+    },
+    {
+      "epoch": 0.9131043393991601,
+      "grad_norm": 0.45959070324897766,
+      "learning_rate": 1.6665615841889885e-05,
+      "loss": 0.6339,
+      "step": 2120
+    },
+    {
+      "epoch": 0.9174114353397222,
+      "grad_norm": 0.48287901282310486,
+      "learning_rate": 1.6630132554311486e-05,
+      "loss": 0.6161,
+      "step": 2130
+    },
+    {
+      "epoch": 0.9217185312802842,
+      "grad_norm": 0.4725618064403534,
+      "learning_rate": 1.6594499680109722e-05,
+      "loss": 0.627,
+      "step": 2140
+    },
+    {
+      "epoch": 0.9260256272208464,
+      "grad_norm": 0.4820912778377533,
+      "learning_rate": 1.6558718023220457e-05,
+      "loss": 0.6399,
+      "step": 2150
+    },
+    {
+      "epoch": 0.9303327231614085,
+      "grad_norm": 0.48815685510635376,
+      "learning_rate": 1.6522788390936328e-05,
+      "loss": 0.6437,
+      "step": 2160
+    },
+    {
+      "epoch": 0.9346398191019705,
+      "grad_norm": 0.4747340679168701,
+      "learning_rate": 1.648671159388855e-05,
+      "loss": 0.6455,
+      "step": 2170
+    },
+    {
+      "epoch": 0.9389469150425326,
+      "grad_norm": 0.4894673526287079,
+      "learning_rate": 1.6450488446028612e-05,
+      "loss": 0.6545,
+      "step": 2180
+    },
+    {
+      "epoch": 0.9432540109830947,
+      "grad_norm": 0.4756160080432892,
+      "learning_rate": 1.641411976460991e-05,
+      "loss": 0.6498,
+      "step": 2190
+    },
+    {
+      "epoch": 0.9475611069236567,
+      "grad_norm": 0.45228078961372375,
+      "learning_rate": 1.637760637016932e-05,
+      "loss": 0.6438,
+      "step": 2200
+    },
+    {
+      "epoch": 0.9518682028642188,
+      "grad_norm": 0.49898287653923035,
+      "learning_rate": 1.6340949086508676e-05,
+      "loss": 0.6518,
+      "step": 2210
+    },
+    {
+      "epoch": 0.9561752988047809,
+      "grad_norm": 0.4354493021965027,
+      "learning_rate": 1.6304148740676204e-05,
+      "loss": 0.6125,
+      "step": 2220
+    },
+    {
+      "epoch": 0.9604823947453429,
+      "grad_norm": 0.45118704438209534,
+      "learning_rate": 1.6267206162947823e-05,
+      "loss": 0.6146,
+      "step": 2230
+    },
+    {
+      "epoch": 0.964789490685905,
+      "grad_norm": 0.4822487533092499,
+      "learning_rate": 1.6230122186808443e-05,
+      "loss": 0.6425,
+      "step": 2240
+    },
+    {
+      "epoch": 0.9690965866264671,
+      "grad_norm": 0.490903377532959,
+      "learning_rate": 1.619289764893317e-05,
+      "loss": 0.6353,
+      "step": 2250
+    },
+    {
+      "epoch": 0.9734036825670291,
+      "grad_norm": 0.4738866686820984,
+      "learning_rate": 1.615553338916839e-05,
+      "loss": 0.6315,
+      "step": 2260
+    },
+    {
+      "epoch": 0.9777107785075912,
+      "grad_norm": 0.46285027265548706,
+      "learning_rate": 1.6118030250512863e-05,
+      "loss": 0.6501,
+      "step": 2270
+    },
+    {
+      "epoch": 0.9820178744481534,
+      "grad_norm": 0.46414172649383545,
+      "learning_rate": 1.6080389079098657e-05,
+      "loss": 0.6501,
+      "step": 2280
+    },
+    {
+      "epoch": 0.9863249703887154,
+      "grad_norm": 0.5042113661766052,
+      "learning_rate": 1.604261072417211e-05,
+      "loss": 0.6319,
+      "step": 2290
+    },
+    {
+      "epoch": 0.9906320663292775,
+      "grad_norm": 0.43653419613838196,
+      "learning_rate": 1.600469603807464e-05,
+      "loss": 0.6461,
+      "step": 2300
+    },
+    {
+      "epoch": 0.9949391622698396,
+      "grad_norm": 0.4572006165981293,
+      "learning_rate": 1.5966645876223505e-05,
+      "loss": 0.6477,
+      "step": 2310
+    },
+    {
+      "epoch": 0.9992462582104016,
+      "grad_norm": 0.43867436051368713,
+      "learning_rate": 1.5928461097092532e-05,
+      "loss": 0.6288,
+      "step": 2320
+    },
+    {
+      "epoch": 1.0035533541509638,
+      "grad_norm": 0.5620077848434448,
+      "learning_rate": 1.589014256219273e-05,
+      "loss": 0.5378,
+      "step": 2330
+    },
+    {
+      "epoch": 1.0078604500915258,
+      "grad_norm": 0.4836018681526184,
+      "learning_rate": 1.5851691136052842e-05,
+      "loss": 0.5421,
+      "step": 2340
+    },
+    {
+      "epoch": 1.0121675460320878,
+      "grad_norm": 0.49632197618484497,
+      "learning_rate": 1.581310768619988e-05,
+      "loss": 0.5237,
+      "step": 2350
+    },
+    {
+      "epoch": 1.01647464197265,
+      "grad_norm": 0.49445948004722595,
+      "learning_rate": 1.5774393083139513e-05,
+      "loss": 0.5313,
+      "step": 2360
+    },
+    {
+      "epoch": 1.020781737913212,
+      "grad_norm": 0.5299666523933411,
+      "learning_rate": 1.5735548200336435e-05,
+      "loss": 0.5326,
+      "step": 2370
+    },
+    {
+      "epoch": 1.025088833853774,
+      "grad_norm": 0.5012844204902649,
+      "learning_rate": 1.569657391419468e-05,
+      "loss": 0.5401,
+      "step": 2380
+    },
+    {
+      "epoch": 1.0293959297943363,
+      "grad_norm": 0.4741289019584656,
+      "learning_rate": 1.565747110403781e-05,
+      "loss": 0.5052,
+      "step": 2390
+    },
+    {
+      "epoch": 1.0337030257348983,
+      "grad_norm": 0.4950823485851288,
+      "learning_rate": 1.5618240652089123e-05,
+      "loss": 0.5294,
+      "step": 2400
+    },
+    {
+      "epoch": 1.0380101216754603,
+      "grad_norm": 0.4934958517551422,
+      "learning_rate": 1.557888344345171e-05,
+      "loss": 0.5278,
+      "step": 2410
+    },
+    {
+      "epoch": 1.0423172176160225,
+      "grad_norm": 0.467101514339447,
+      "learning_rate": 1.5539400366088503e-05,
+      "loss": 0.504,
+      "step": 2420
+    },
+    {
+      "epoch": 1.0466243135565845,
+      "grad_norm": 0.5479716062545776,
+      "learning_rate": 1.5499792310802238e-05,
+      "loss": 0.5256,
+      "step": 2430
+    },
+    {
+      "epoch": 1.0509314094971465,
+      "grad_norm": 0.4706737697124481,
+      "learning_rate": 1.5460060171215362e-05,
+      "loss": 0.5251,
+      "step": 2440
+    },
+    {
+      "epoch": 1.0552385054377087,
+      "grad_norm": 0.5142565965652466,
+      "learning_rate": 1.5420204843749857e-05,
+      "loss": 0.5333,
+      "step": 2450
+    },
+    {
+      "epoch": 1.0595456013782707,
+      "grad_norm": 0.5430694222450256,
+      "learning_rate": 1.5380227227607032e-05,
+      "loss": 0.5391,
+      "step": 2460
+    },
+    {
+      "epoch": 1.0638526973188327,
+      "grad_norm": 0.4780258536338806,
+      "learning_rate": 1.5340128224747225e-05,
+      "loss": 0.5338,
+      "step": 2470
+    },
+    {
+      "epoch": 1.068159793259395,
+      "grad_norm": 0.47647717595100403,
+      "learning_rate": 1.5299908739869464e-05,
+      "loss": 0.5178,
+      "step": 2480
+    },
+    {
+      "epoch": 1.072466889199957,
+      "grad_norm": 0.5330241918563843,
+      "learning_rate": 1.525956968039103e-05,
+      "loss": 0.5027,
+      "step": 2490
+    },
+    {
+      "epoch": 1.076773985140519,
+      "grad_norm": 0.4681854546070099,
+      "learning_rate": 1.5219111956427027e-05,
+      "loss": 0.5315,
+      "step": 2500
+    },
+    {
+      "epoch": 1.0810810810810811,
+      "grad_norm": 0.5060921311378479,
+      "learning_rate": 1.5178536480769803e-05,
+      "loss": 0.5103,
+      "step": 2510
+    },
+    {
+      "epoch": 1.0853881770216431,
+      "grad_norm": 0.497199147939682,
+      "learning_rate": 1.5137844168868391e-05,
+      "loss": 0.5302,
+      "step": 2520
+    },
+    {
+      "epoch": 1.0896952729622051,
+      "grad_norm": 0.4658927321434021,
+      "learning_rate": 1.5097035938807834e-05,
+      "loss": 0.5196,
+      "step": 2530
+    },
+    {
+      "epoch": 1.0940023689027674,
+      "grad_norm": 0.5109249353408813,
+      "learning_rate": 1.5056112711288475e-05,
+      "loss": 0.5099,
+      "step": 2540
+    },
+    {
+      "epoch": 1.0983094648433294,
+      "grad_norm": 0.5212246775627136,
+      "learning_rate": 1.5015075409605189e-05,
+      "loss": 0.4911,
+      "step": 2550
+    },
+    {
+      "epoch": 1.1026165607838914,
+      "grad_norm": 0.47850698232650757,
+      "learning_rate": 1.497392495962656e-05,
+      "loss": 0.5225,
+      "step": 2560
+    },
+    {
+      "epoch": 1.1069236567244536,
+      "grad_norm": 0.4982755184173584,
+      "learning_rate": 1.4932662289773969e-05,
+      "loss": 0.5278,
+      "step": 2570
+    },
+    {
+      "epoch": 1.1112307526650156,
+      "grad_norm": 0.49975791573524475,
+      "learning_rate": 1.4891288331000668e-05,
+      "loss": 0.5261,
+      "step": 2580
+    },
+    {
+      "epoch": 1.1155378486055776,
+      "grad_norm": 0.5002388954162598,
+      "learning_rate": 1.484980401677077e-05,
+      "loss": 0.5313,
+      "step": 2590
+    },
+    {
+      "epoch": 1.1198449445461398,
+      "grad_norm": 0.4950617253780365,
+      "learning_rate": 1.4808210283038183e-05,
+      "loss": 0.5286,
+      "step": 2600
+    },
+    {
+      "epoch": 1.1241520404867018,
+      "grad_norm": 0.49831753969192505,
+      "learning_rate": 1.47665080682255e-05,
+      "loss": 0.5133,
+      "step": 2610
+    },
+    {
+      "epoch": 1.128459136427264,
+      "grad_norm": 0.6730148792266846,
+      "learning_rate": 1.4724698313202825e-05,
+      "loss": 0.5224,
+      "step": 2620
+    },
+    {
+      "epoch": 1.132766232367826,
+      "grad_norm": 0.5355139374732971,
+      "learning_rate": 1.4682781961266546e-05,
+      "loss": 0.5188,
+      "step": 2630
+    },
+    {
+      "epoch": 1.137073328308388,
+      "grad_norm": 0.5199829936027527,
+      "learning_rate": 1.4640759958118045e-05,
+      "loss": 0.5121,
+      "step": 2640
+    },
+    {
+      "epoch": 1.14138042424895,
+      "grad_norm": 0.5292408466339111,
+      "learning_rate": 1.4598633251842373e-05,
+      "loss": 0.5267,
+      "step": 2650
+    },
+    {
+      "epoch": 1.1456875201895123,
+      "grad_norm": 0.5363121032714844,
+      "learning_rate": 1.4556402792886856e-05,
+      "loss": 0.5147,
+      "step": 2660
+    },
+    {
+      "epoch": 1.1499946161300743,
+      "grad_norm": 0.5359490513801575,
+      "learning_rate": 1.4514069534039649e-05,
+      "loss": 0.5155,
+      "step": 2670
+    },
+    {
+      "epoch": 1.1543017120706365,
+      "grad_norm": 0.4707220792770386,
+      "learning_rate": 1.4471634430408244e-05,
+      "loss": 0.5419,
+      "step": 2680
+    },
+    {
+      "epoch": 1.1586088080111985,
+      "grad_norm": 0.4798811376094818,
+      "learning_rate": 1.4429098439397901e-05,
+      "loss": 0.5152,
+      "step": 2690
+    },
+    {
+      "epoch": 1.1629159039517605,
+      "grad_norm": 0.4730081260204315,
+      "learning_rate": 1.4386462520690087e-05,
+      "loss": 0.5283,
+      "step": 2700
+    },
+    {
+      "epoch": 1.1672229998923225,
+      "grad_norm": 0.524276614189148,
+      "learning_rate": 1.4343727636220785e-05,
+      "loss": 0.5087,
+      "step": 2710
+    },
+    {
+      "epoch": 1.1715300958328847,
+      "grad_norm": 0.5093454122543335,
+      "learning_rate": 1.430089475015882e-05,
+      "loss": 0.5371,
+      "step": 2720
+    },
+    {
+      "epoch": 1.1758371917734467,
+      "grad_norm": 0.5228180289268494,
+      "learning_rate": 1.4257964828884077e-05,
+      "loss": 0.5121,
+      "step": 2730
+    },
+    {
+      "epoch": 1.180144287714009,
+      "grad_norm": 0.5263434052467346,
+      "learning_rate": 1.4214938840965729e-05,
+      "loss": 0.5104,
+      "step": 2740
+    },
+    {
+      "epoch": 1.184451383654571,
+      "grad_norm": 0.5519675612449646,
+      "learning_rate": 1.417181775714036e-05,
+      "loss": 0.5081,
+      "step": 2750
+    },
+    {
+      "epoch": 1.188758479595133,
+      "grad_norm": 0.48901626467704773,
+      "learning_rate": 1.4128602550290078e-05,
+      "loss": 0.5332,
+      "step": 2760
+    },
+    {
+      "epoch": 1.1930655755356951,
+      "grad_norm": 0.5022098422050476,
+      "learning_rate": 1.4085294195420563e-05,
+      "loss": 0.5267,
+      "step": 2770
+    },
+    {
+      "epoch": 1.1973726714762571,
+      "grad_norm": 0.5244942307472229,
+      "learning_rate": 1.4041893669639053e-05,
+      "loss": 0.5309,
+      "step": 2780
+    },
+    {
+      "epoch": 1.2016797674168191,
+      "grad_norm": 0.5060109496116638,
+      "learning_rate": 1.399840195213233e-05,
+      "loss": 0.509,
+      "step": 2790
+    },
+    {
+      "epoch": 1.2059868633573814,
+      "grad_norm": 0.48709142208099365,
+      "learning_rate": 1.3954820024144595e-05,
+      "loss": 0.5249,
+      "step": 2800
+    },
+    {
+      "epoch": 1.2102939592979434,
+      "grad_norm": 0.48755279183387756,
+      "learning_rate": 1.3911148868955357e-05,
+      "loss": 0.5216,
+      "step": 2810
+    },
+    {
+      "epoch": 1.2146010552385054,
+      "grad_norm": 0.4871668219566345,
+      "learning_rate": 1.3867389471857229e-05,
+      "loss": 0.5199,
+      "step": 2820
+    },
+    {
+      "epoch": 1.2189081511790676,
+      "grad_norm": 0.5313363671302795,
+      "learning_rate": 1.3823542820133706e-05,
+      "loss": 0.5146,
+      "step": 2830
+    },
+    {
+      "epoch": 1.2232152471196296,
+      "grad_norm": 0.48473960161209106,
+      "learning_rate": 1.3779609903036894e-05,
+      "loss": 0.5126,
+      "step": 2840
+    },
+    {
+      "epoch": 1.2275223430601916,
+      "grad_norm": 0.5411814451217651,
+      "learning_rate": 1.3735591711765189e-05,
+      "loss": 0.5186,
+      "step": 2850
+    },
+    {
+      "epoch": 1.2318294390007538,
+      "grad_norm": 0.5286210775375366,
+      "learning_rate": 1.3691489239440899e-05,
+      "loss": 0.513,
+      "step": 2860
+    },
+    {
+      "epoch": 1.2361365349413158,
+      "grad_norm": 0.47112423181533813,
+      "learning_rate": 1.3647303481087858e-05,
+      "loss": 0.5268,
+      "step": 2870
+    },
+    {
+      "epoch": 1.2404436308818778,
+      "grad_norm": 0.5465208888053894,
+      "learning_rate": 1.3603035433608977e-05,
+      "loss": 0.5109,
+      "step": 2880
+    },
+    {
+      "epoch": 1.24475072682244,
+      "grad_norm": 0.4758882522583008,
+      "learning_rate": 1.3558686095763732e-05,
+      "loss": 0.5307,
+      "step": 2890
+    },
+    {
+      "epoch": 1.249057822763002,
+      "grad_norm": 0.5721794962882996,
+      "learning_rate": 1.3514256468145645e-05,
+      "loss": 0.5104,
+      "step": 2900
+    },
+    {
+      "epoch": 1.2533649187035643,
+      "grad_norm": 0.5125982761383057,
+      "learning_rate": 1.3469747553159714e-05,
+      "loss": 0.5278,
+      "step": 2910
+    },
+    {
+      "epoch": 1.2576720146441263,
+      "grad_norm": 0.5272653698921204,
+      "learning_rate": 1.342516035499978e-05,
+      "loss": 0.5276,
+      "step": 2920
+    },
+    {
+      "epoch": 1.2619791105846883,
+      "grad_norm": 0.5423816442489624,
+      "learning_rate": 1.3380495879625884e-05,
+      "loss": 0.5408,
+      "step": 2930
+    },
+    {
+      "epoch": 1.2662862065252503,
+      "grad_norm": 0.4817509055137634,
+      "learning_rate": 1.333575513474157e-05,
+      "loss": 0.5152,
+      "step": 2940
+    },
+    {
+      "epoch": 1.2705933024658125,
+      "grad_norm": 0.5113592147827148,
+      "learning_rate": 1.3290939129771143e-05,
+      "loss": 0.5397,
+      "step": 2950
+    },
+    {
+      "epoch": 1.2749003984063745,
+      "grad_norm": 0.5106224417686462,
+      "learning_rate": 1.3246048875836898e-05,
+      "loss": 0.5269,
+      "step": 2960
+    },
+    {
+      "epoch": 1.2792074943469367,
+      "grad_norm": 0.5446826219558716,
+      "learning_rate": 1.3201085385736313e-05,
+      "loss": 0.5252,
+      "step": 2970
+    },
+    {
+      "epoch": 1.2835145902874987,
+      "grad_norm": 0.484943151473999,
+      "learning_rate": 1.3156049673919184e-05,
+      "loss": 0.525,
+      "step": 2980
+    },
+    {
+      "epoch": 1.2878216862280607,
+      "grad_norm": 0.5692194700241089,
+      "learning_rate": 1.3110942756464764e-05,
+      "loss": 0.5197,
+      "step": 2990
+    },
+    {
+      "epoch": 1.2921287821686227,
+      "grad_norm": 0.5009827017784119,
+      "learning_rate": 1.3065765651058802e-05,
+      "loss": 0.5325,
+      "step": 3000
+    },
+    {
+      "epoch": 1.296435878109185,
+      "grad_norm": 0.4953298568725586,
+      "learning_rate": 1.3020519376970613e-05,
+      "loss": 0.5095,
+      "step": 3010
+    },
+    {
+      "epoch": 1.300742974049747,
+      "grad_norm": 0.5116891264915466,
+      "learning_rate": 1.2975204955030068e-05,
+      "loss": 0.5263,
+      "step": 3020
+    },
+    {
+      "epoch": 1.3050500699903091,
+      "grad_norm": 0.4844088554382324,
+      "learning_rate": 1.2929823407604567e-05,
+      "loss": 0.5113,
+      "step": 3030
+    },
+    {
+      "epoch": 1.3093571659308711,
+      "grad_norm": 0.4732029438018799,
+      "learning_rate": 1.2884375758575967e-05,
+      "loss": 0.532,
+      "step": 3040
+    },
+    {
+      "epoch": 1.3136642618714331,
+      "grad_norm": 0.5469485521316528,
+      "learning_rate": 1.2838863033317484e-05,
+      "loss": 0.519,
+      "step": 3050
+    },
+    {
+      "epoch": 1.3179713578119951,
+      "grad_norm": 0.4888254702091217,
+      "learning_rate": 1.2793286258670565e-05,
+      "loss": 0.5097,
+      "step": 3060
+    },
+    {
+      "epoch": 1.3222784537525574,
+      "grad_norm": 0.5359517335891724,
+      "learning_rate": 1.2747646462921717e-05,
+      "loss": 0.5246,
+      "step": 3070
+    },
+    {
+      "epoch": 1.3265855496931194,
+      "grad_norm": 0.5013801455497742,
+      "learning_rate": 1.2701944675779299e-05,
+      "loss": 0.524,
+      "step": 3080
+    },
+    {
+      "epoch": 1.3308926456336816,
+      "grad_norm": 0.49307557940483093,
+      "learning_rate": 1.2656181928350301e-05,
+      "loss": 0.5403,
+      "step": 3090
+    },
+    {
+      "epoch": 1.3351997415742436,
+      "grad_norm": 0.47625210881233215,
+      "learning_rate": 1.2610359253117078e-05,
+      "loss": 0.5275,
+      "step": 3100
+    },
+    {
+      "epoch": 1.3395068375148056,
+      "grad_norm": 0.5096368789672852,
+      "learning_rate": 1.2564477683914053e-05,
+      "loss": 0.5231,
+      "step": 3110
+    },
+    {
+      "epoch": 1.3438139334553676,
+      "grad_norm": 0.4992668926715851,
+      "learning_rate": 1.2518538255904389e-05,
+      "loss": 0.5235,
+      "step": 3120
+    },
+    {
+      "epoch": 1.3481210293959298,
+      "grad_norm": 0.491062194108963,
+      "learning_rate": 1.2472542005556647e-05,
+      "loss": 0.5432,
+      "step": 3130
+    },
+    {
+      "epoch": 1.3524281253364918,
+      "grad_norm": 0.48666131496429443,
+      "learning_rate": 1.2426489970621385e-05,
+      "loss": 0.531,
+      "step": 3140
+    },
+    {
+      "epoch": 1.356735221277054,
+      "grad_norm": 0.4706876575946808,
+      "learning_rate": 1.2380383190107757e-05,
+      "loss": 0.5188,
+      "step": 3150
+    },
+    {
+      "epoch": 1.361042317217616,
+      "grad_norm": 0.4910385310649872,
+      "learning_rate": 1.2334222704260063e-05,
+      "loss": 0.5106,
+      "step": 3160
+    },
+    {
+      "epoch": 1.365349413158178,
+      "grad_norm": 0.506514847278595,
+      "learning_rate": 1.2288009554534291e-05,
+      "loss": 0.5292,
+      "step": 3170
+    },
+    {
+      "epoch": 1.36965650909874,
+      "grad_norm": 0.49671700596809387,
+      "learning_rate": 1.2241744783574596e-05,
+      "loss": 0.5284,
+      "step": 3180
+    },
+    {
+      "epoch": 1.3739636050393023,
+      "grad_norm": 0.4892718195915222,
+      "learning_rate": 1.219542943518981e-05,
+      "loss": 0.5215,
+      "step": 3190
+    },
+    {
+      "epoch": 1.3782707009798643,
+      "grad_norm": 0.5412102937698364,
+      "learning_rate": 1.2149064554329864e-05,
+      "loss": 0.5256,
+      "step": 3200
+    },
+    {
+      "epoch": 1.3825777969204265,
+      "grad_norm": 0.4869970679283142,
+      "learning_rate": 1.2102651187062227e-05,
+      "loss": 0.5218,
+      "step": 3210
+    },
+    {
+      "epoch": 1.3868848928609885,
+      "grad_norm": 0.5195066332817078,
+      "learning_rate": 1.2056190380548299e-05,
+      "loss": 0.5269,
+      "step": 3220
+    },
+    {
+      "epoch": 1.3911919888015505,
+      "grad_norm": 0.5343438982963562,
+      "learning_rate": 1.2009683183019788e-05,
+      "loss": 0.5301,
+      "step": 3230
+    },
+    {
+      "epoch": 1.3954990847421127,
+      "grad_norm": 0.522270679473877,
+      "learning_rate": 1.1963130643755055e-05,
+      "loss": 0.545,
+      "step": 3240
+    },
+    {
+      "epoch": 1.3998061806826747,
+      "grad_norm": 0.501485288143158,
+      "learning_rate": 1.191653381305545e-05,
+      "loss": 0.5253,
+      "step": 3250
+    },
+    {
+      "epoch": 1.4041132766232367,
+      "grad_norm": 0.5288712382316589,
+      "learning_rate": 1.186989374222161e-05,
+      "loss": 0.5181,
+      "step": 3260
+    },
+    {
+      "epoch": 1.408420372563799,
+      "grad_norm": 0.5131502151489258,
+      "learning_rate": 1.1823211483529733e-05,
+      "loss": 0.5138,
+      "step": 3270
+    },
+    {
+      "epoch": 1.412727468504361,
+      "grad_norm": 0.4853404462337494,
+      "learning_rate": 1.1776488090207852e-05,
+      "loss": 0.5319,
+      "step": 3280
+    },
+    {
+      "epoch": 1.417034564444923,
+      "grad_norm": 0.5093010663986206,
+      "learning_rate": 1.1729724616412062e-05,
+      "loss": 0.5155,
+      "step": 3290
+    },
+    {
+      "epoch": 1.4213416603854852,
+      "grad_norm": 0.5078168511390686,
+      "learning_rate": 1.1682922117202736e-05,
+      "loss": 0.5206,
+      "step": 3300
+    },
+    {
+      "epoch": 1.4256487563260472,
+      "grad_norm": 0.5315324664115906,
+      "learning_rate": 1.163608164852073e-05,
+      "loss": 0.5314,
+      "step": 3310
+    },
+    {
+      "epoch": 1.4299558522666094,
+      "grad_norm": 0.4705192446708679,
+      "learning_rate": 1.1589204267163545e-05,
+      "loss": 0.4966,
+      "step": 3320
+    },
+    {
+      "epoch": 1.4342629482071714,
+      "grad_norm": 0.48757535219192505,
+      "learning_rate": 1.15422910307615e-05,
+      "loss": 0.5299,
+      "step": 3330
+    },
+    {
+      "epoch": 1.4385700441477334,
+      "grad_norm": 0.5582148432731628,
+      "learning_rate": 1.1495342997753864e-05,
+      "loss": 0.5201,
+      "step": 3340
+    },
+    {
+      "epoch": 1.4428771400882954,
+      "grad_norm": 0.5134326219558716,
+      "learning_rate": 1.1448361227364963e-05,
+      "loss": 0.5061,
+      "step": 3350
+    },
+    {
+      "epoch": 1.4471842360288576,
+      "grad_norm": 0.5316387414932251,
+      "learning_rate": 1.1401346779580303e-05,
+      "loss": 0.5145,
+      "step": 3360
+    },
+    {
+      "epoch": 1.4514913319694196,
+      "grad_norm": 0.5328738689422607,
+      "learning_rate": 1.1354300715122637e-05,
+      "loss": 0.5288,
+      "step": 3370
+    },
+    {
+      "epoch": 1.4557984279099818,
+      "grad_norm": 0.5279168486595154,
+      "learning_rate": 1.1307224095428058e-05,
+      "loss": 0.5031,
+      "step": 3380
+    },
+    {
+      "epoch": 1.4601055238505438,
+      "grad_norm": 0.5049686431884766,
+      "learning_rate": 1.1260117982622021e-05,
+      "loss": 0.5004,
+      "step": 3390
+    },
+    {
+      "epoch": 1.4644126197911058,
+      "grad_norm": 0.47000184655189514,
+      "learning_rate": 1.1212983439495392e-05,
+      "loss": 0.5267,
+      "step": 3400
+    },
+    {
+      "epoch": 1.4687197157316678,
+      "grad_norm": 0.49505382776260376,
+      "learning_rate": 1.1165821529480483e-05,
+      "loss": 0.5278,
+      "step": 3410
+    },
+    {
+      "epoch": 1.47302681167223,
+      "grad_norm": 0.568454384803772,
+      "learning_rate": 1.1118633316627037e-05,
+      "loss": 0.5116,
+      "step": 3420
+    },
+    {
+      "epoch": 1.477333907612792,
+      "grad_norm": 0.5094279646873474,
+      "learning_rate": 1.1071419865578241e-05,
+      "loss": 0.5181,
+      "step": 3430
+    },
+    {
+      "epoch": 1.4816410035533543,
+      "grad_norm": 0.5605435371398926,
+      "learning_rate": 1.1024182241546686e-05,
+      "loss": 0.5191,
+      "step": 3440
+    },
+    {
+      "epoch": 1.4859480994939163,
+      "grad_norm": 0.49941274523735046,
+      "learning_rate": 1.097692151029036e-05,
+      "loss": 0.5036,
+      "step": 3450
+    },
+    {
+      "epoch": 1.4902551954344783,
+      "grad_norm": 0.5064433813095093,
+      "learning_rate": 1.0929638738088571e-05,
+      "loss": 0.5195,
+      "step": 3460
+    },
+    {
+      "epoch": 1.4945622913750403,
+      "grad_norm": 0.5021061301231384,
+      "learning_rate": 1.088233499171792e-05,
+      "loss": 0.522,
+      "step": 3470
+    },
+    {
+      "epoch": 1.4988693873156025,
+      "grad_norm": 0.5188096761703491,
+      "learning_rate": 1.0835011338428217e-05,
+      "loss": 0.5156,
+      "step": 3480
+    },
+    {
+      "epoch": 1.5031764832561645,
+      "grad_norm": 0.6124559640884399,
+      "learning_rate": 1.0787668845918393e-05,
+      "loss": 0.5145,
+      "step": 3490
+    },
+    {
+      "epoch": 1.5074835791967267,
+      "grad_norm": 0.48937344551086426,
+      "learning_rate": 1.074030858231244e-05,
+      "loss": 0.515,
+      "step": 3500
+    },
+    {
+      "epoch": 1.5117906751372887,
+      "grad_norm": 0.518526017665863,
+      "learning_rate": 1.0692931616135283e-05,
+      "loss": 0.505,
+      "step": 3510
+    },
+    {
+      "epoch": 1.5160977710778507,
+      "grad_norm": 0.5395667552947998,
+      "learning_rate": 1.0645539016288686e-05,
+      "loss": 0.5076,
+      "step": 3520
+    },
+    {
+      "epoch": 1.5204048670184127,
+      "grad_norm": 0.495190292596817,
+      "learning_rate": 1.059813185202714e-05,
+      "loss": 0.523,
+      "step": 3530
+    },
+    {
+      "epoch": 1.524711962958975,
+      "grad_norm": 0.49644342064857483,
+      "learning_rate": 1.055071119293373e-05,
+      "loss": 0.5038,
+      "step": 3540
+    },
+    {
+      "epoch": 1.5290190588995372,
+      "grad_norm": 0.483696848154068,
+      "learning_rate": 1.0503278108896e-05,
+      "loss": 0.5103,
+      "step": 3550
+    },
+    {
+      "epoch": 1.5333261548400992,
+      "grad_norm": 0.5149986147880554,
+      "learning_rate": 1.0455833670081831e-05,
+      "loss": 0.5402,
+      "step": 3560
+    },
+    {
+      "epoch": 1.5376332507806612,
+      "grad_norm": 0.4734952449798584,
+      "learning_rate": 1.0408378946915282e-05,
+      "loss": 0.5292,
+      "step": 3570
+    },
+    {
+      "epoch": 1.5419403467212232,
+      "grad_norm": 0.5490080118179321,
+      "learning_rate": 1.0360915010052443e-05,
+      "loss": 0.5155,
+      "step": 3580
+    },
+    {
+      "epoch": 1.5462474426617852,
+      "grad_norm": 0.5176838636398315,
+      "learning_rate": 1.0313442930357278e-05,
+      "loss": 0.5111,
+      "step": 3590
+    },
+    {
+      "epoch": 1.5505545386023474,
+      "grad_norm": 0.5659157633781433,
+      "learning_rate": 1.026596377887747e-05,
+      "loss": 0.5152,
+      "step": 3600
+    },
+    {
+      "epoch": 1.5548616345429096,
+      "grad_norm": 0.5195504426956177,
+      "learning_rate": 1.0218478626820256e-05,
+      "loss": 0.5178,
+      "step": 3610
+    },
+    {
+      "epoch": 1.5591687304834716,
+      "grad_norm": 0.533338189125061,
+      "learning_rate": 1.0170988545528248e-05,
+      "loss": 0.5138,
+      "step": 3620
+    },
+    {
+      "epoch": 1.5634758264240336,
+      "grad_norm": 0.5108840465545654,
+      "learning_rate": 1.0123494606455278e-05,
+      "loss": 0.5273,
+      "step": 3630
+    },
+    {
+      "epoch": 1.5677829223645956,
+      "grad_norm": 0.4785379469394684,
+      "learning_rate": 1.0075997881142208e-05,
+      "loss": 0.5071,
+      "step": 3640
+    },
+    {
+      "epoch": 1.5720900183051576,
+      "grad_norm": 0.49497827887535095,
+      "learning_rate": 1.0028499441192765e-05,
+      "loss": 0.5132,
+      "step": 3650
+    },
+    {
+      "epoch": 1.5763971142457198,
+      "grad_norm": 0.5214102864265442,
+      "learning_rate": 9.981000358249368e-06,
+      "loss": 0.5133,
+      "step": 3660
+    },
+    {
+      "epoch": 1.580704210186282,
+      "grad_norm": 0.47462400794029236,
+      "learning_rate": 9.933501703968928e-06,
+      "loss": 0.5226,
+      "step": 3670
+    },
+    {
+      "epoch": 1.585011306126844,
+      "grad_norm": 0.4743979275226593,
+      "learning_rate": 9.8860045499987e-06,
+      "loss": 0.5219,
+      "step": 3680
+    },
+    {
+      "epoch": 1.589318402067406,
+      "grad_norm": 0.5265910625457764,
+      "learning_rate": 9.838509967952076e-06,
+      "loss": 0.4945,
+      "step": 3690
+    },
+    {
+      "epoch": 1.593625498007968,
+      "grad_norm": 0.5075172185897827,
+      "learning_rate": 9.791019029384437e-06,
+      "loss": 0.5175,
+      "step": 3700
+    },
+    {
+      "epoch": 1.59793259394853,
+      "grad_norm": 0.5206677913665771,
+      "learning_rate": 9.743532805768948e-06,
+      "loss": 0.5188,
+      "step": 3710
+    },
+    {
+      "epoch": 1.6022396898890923,
+      "grad_norm": 0.4802674651145935,
+      "learning_rate": 9.696052368472406e-06,
+      "loss": 0.5064,
+      "step": 3720
+    },
+    {
+      "epoch": 1.6065467858296545,
+      "grad_norm": 0.5289535522460938,
+      "learning_rate": 9.648578788731044e-06,
+      "loss": 0.5281,
+      "step": 3730
+    },
+    {
+      "epoch": 1.6108538817702165,
+      "grad_norm": 0.47722700238227844,
+      "learning_rate": 9.601113137626394e-06,
+      "loss": 0.5151,
+      "step": 3740
+    },
+    {
+      "epoch": 1.6151609777107785,
+      "grad_norm": 0.4994152784347534,
+      "learning_rate": 9.553656486061098e-06,
+      "loss": 0.52,
+      "step": 3750
+    },
+    {
+      "epoch": 1.6194680736513405,
+      "grad_norm": 0.48130089044570923,
+      "learning_rate": 9.506209904734753e-06,
+      "loss": 0.5336,
+      "step": 3760
+    },
+    {
+      "epoch": 1.6237751695919027,
+      "grad_norm": 0.48449528217315674,
+      "learning_rate": 9.45877446411976e-06,
+      "loss": 0.5252,
+      "step": 3770
+    },
+    {
+      "epoch": 1.6280822655324647,
+      "grad_norm": 0.5411643981933594,
+      "learning_rate": 9.411351234437163e-06,
+      "loss": 0.5187,
+      "step": 3780
+    },
+    {
+      "epoch": 1.632389361473027,
+      "grad_norm": 0.5133873820304871,
+      "learning_rate": 9.363941285632507e-06,
+      "loss": 0.5217,
+      "step": 3790
+    },
+    {
+      "epoch": 1.636696457413589,
+      "grad_norm": 0.5814666748046875,
+      "learning_rate": 9.3165456873517e-06,
+      "loss": 0.5,
+      "step": 3800
+    },
+    {
+      "epoch": 1.641003553354151,
+      "grad_norm": 0.52715665102005,
+      "learning_rate": 9.269165508916883e-06,
+      "loss": 0.5184,
+      "step": 3810
+    },
+    {
+      "epoch": 1.645310649294713,
+      "grad_norm": 0.48196879029273987,
+      "learning_rate": 9.221801819302288e-06,
+      "loss": 0.5191,
+      "step": 3820
+    },
+    {
+      "epoch": 1.6496177452352752,
+      "grad_norm": 0.49397778511047363,
+      "learning_rate": 9.174455687110142e-06,
+      "loss": 0.5013,
+      "step": 3830
+    },
+    {
+      "epoch": 1.6539248411758372,
+      "grad_norm": 0.5037091970443726,
+      "learning_rate": 9.127128180546548e-06,
+      "loss": 0.5298,
+      "step": 3840
+    },
+    {
+      "epoch": 1.6582319371163994,
+      "grad_norm": 0.5031833052635193,
+      "learning_rate": 9.079820367397384e-06,
+      "loss": 0.4929,
+      "step": 3850
+    },
+    {
+      "epoch": 1.6625390330569614,
+      "grad_norm": 0.5380353927612305,
+      "learning_rate": 9.032533315004207e-06,
+      "loss": 0.4968,
+      "step": 3860
+    },
+    {
+      "epoch": 1.6668461289975234,
+      "grad_norm": 0.5191226005554199,
+      "learning_rate": 8.98526809024018e-06,
+      "loss": 0.5267,
+      "step": 3870
+    },
+    {
+      "epoch": 1.6711532249380854,
+      "grad_norm": 0.5179468393325806,
+      "learning_rate": 8.938025759486007e-06,
+      "loss": 0.5159,
+      "step": 3880
+    },
+    {
+      "epoch": 1.6754603208786476,
+      "grad_norm": 0.4779166579246521,
+      "learning_rate": 8.89080738860585e-06,
+      "loss": 0.5211,
+      "step": 3890
+    },
+    {
+      "epoch": 1.6797674168192096,
+      "grad_norm": 0.5136571526527405,
+      "learning_rate": 8.843614042923318e-06,
+      "loss": 0.5003,
+      "step": 3900
+    },
+    {
+      "epoch": 1.6840745127597718,
+      "grad_norm": 0.540773332118988,
+      "learning_rate": 8.796446787197383e-06,
+      "loss": 0.5131,
+      "step": 3910
+    },
+    {
+      "epoch": 1.6883816087003338,
+      "grad_norm": 0.5126665234565735,
+      "learning_rate": 8.749306685598409e-06,
+      "loss": 0.5093,
+      "step": 3920
+    },
+    {
+      "epoch": 1.6926887046408958,
+      "grad_norm": 0.47659188508987427,
+      "learning_rate": 8.702194801684112e-06,
+      "loss": 0.5158,
+      "step": 3930
+    },
+    {
+      "epoch": 1.6969958005814578,
+      "grad_norm": 0.47945475578308105,
+      "learning_rate": 8.655112198375564e-06,
+      "loss": 0.5026,
+      "step": 3940
+    },
+    {
+      "epoch": 1.70130289652202,
+      "grad_norm": 0.4939498007297516,
+      "learning_rate": 8.60805993793323e-06,
+      "loss": 0.5099,
+      "step": 3950
+    },
+    {
+      "epoch": 1.7056099924625823,
+      "grad_norm": 0.5328351259231567,
+      "learning_rate": 8.561039081932975e-06,
+      "loss": 0.52,
+      "step": 3960
+    },
+    {
+      "epoch": 1.7099170884031443,
+      "grad_norm": 0.49865198135375977,
+      "learning_rate": 8.514050691242145e-06,
+      "loss": 0.5077,
+      "step": 3970
+    },
+    {
+      "epoch": 1.7142241843437063,
+      "grad_norm": 0.49807870388031006,
+      "learning_rate": 8.467095825995605e-06,
+      "loss": 0.4976,
+      "step": 3980
+    },
+    {
+      "epoch": 1.7185312802842683,
+      "grad_norm": 0.5023031234741211,
+      "learning_rate": 8.420175545571837e-06,
+      "loss": 0.5233,
+      "step": 3990
+    },
+    {
+      "epoch": 1.7228383762248303,
+      "grad_norm": 0.49054110050201416,
+      "learning_rate": 8.373290908569026e-06,
+      "loss": 0.5115,
+      "step": 4000
+    },
+    {
+      "epoch": 1.7271454721653925,
+      "grad_norm": 0.47637811303138733,
+      "learning_rate": 8.32644297278119e-06,
+      "loss": 0.5103,
+      "step": 4010
+    },
+    {
+      "epoch": 1.7314525681059547,
+      "grad_norm": 0.5239661931991577,
+      "learning_rate": 8.279632795174304e-06,
+      "loss": 0.5161,
+      "step": 4020
+    },
+    {
+      "epoch": 1.7357596640465167,
+      "grad_norm": 0.5000544190406799,
+      "learning_rate": 8.232861431862457e-06,
+      "loss": 0.5113,
+      "step": 4030
+    },
+    {
+      "epoch": 1.7400667599870787,
+      "grad_norm": 0.5361005067825317,
+      "learning_rate": 8.186129938084028e-06,
+      "loss": 0.5137,
+      "step": 4040
+    },
+    {
+      "epoch": 1.7443738559276407,
+      "grad_norm": 0.48270535469055176,
+      "learning_rate": 8.139439368177868e-06,
+      "loss": 0.5116,
+      "step": 4050
+    },
+    {
+      "epoch": 1.7486809518682027,
+      "grad_norm": 0.48645904660224915,
+      "learning_rate": 8.092790775559522e-06,
+      "loss": 0.517,
+      "step": 4060
+    },
+    {
+      "epoch": 1.752988047808765,
+      "grad_norm": 0.4865799844264984,
+      "learning_rate": 8.046185212697459e-06,
+      "loss": 0.5202,
+      "step": 4070
+    },
+    {
+      "epoch": 1.7572951437493272,
+      "grad_norm": 0.5095897912979126,
+      "learning_rate": 7.999623731089327e-06,
+      "loss": 0.5186,
+      "step": 4080
+    },
+    {
+      "epoch": 1.7616022396898892,
+      "grad_norm": 0.49918055534362793,
+      "learning_rate": 7.953107381238226e-06,
+      "loss": 0.5091,
+      "step": 4090
+    },
+    {
+      "epoch": 1.7659093356304512,
+      "grad_norm": 0.5209227204322815,
+      "learning_rate": 7.906637212629011e-06,
+      "loss": 0.5098,
+      "step": 4100
+    },
+    {
+      "epoch": 1.7702164315710132,
+      "grad_norm": 0.5320930480957031,
+      "learning_rate": 7.860214273704614e-06,
+      "loss": 0.5172,
+      "step": 4110
+    },
+    {
+      "epoch": 1.7745235275115752,
+      "grad_norm": 0.4841155707836151,
+      "learning_rate": 7.813839611842387e-06,
+      "loss": 0.4851,
+      "step": 4120
+    },
+    {
+      "epoch": 1.7788306234521374,
+      "grad_norm": 0.5300472378730774,
+      "learning_rate": 7.767514273330473e-06,
+      "loss": 0.4953,
+      "step": 4130
+    },
+    {
+      "epoch": 1.7831377193926996,
+      "grad_norm": 0.5021957159042358,
+      "learning_rate": 7.721239303344201e-06,
+      "loss": 0.5112,
+      "step": 4140
+    },
+    {
+      "epoch": 1.7874448153332616,
+      "grad_norm": 0.498737096786499,
+      "learning_rate": 7.675015745922499e-06,
+      "loss": 0.5045,
+      "step": 4150
+    },
+    {
+      "epoch": 1.7917519112738236,
+      "grad_norm": 0.4690532684326172,
+      "learning_rate": 7.628844643944349e-06,
+      "loss": 0.5102,
+      "step": 4160
+    },
+    {
+      "epoch": 1.7960590072143856,
+      "grad_norm": 0.5077162384986877,
+      "learning_rate": 7.582727039105255e-06,
+      "loss": 0.5105,
+      "step": 4170
+    },
+    {
+      "epoch": 1.8003661031549478,
+      "grad_norm": 0.47492554783821106,
+      "learning_rate": 7.536663971893724e-06,
+      "loss": 0.5008,
+      "step": 4180
+    },
+    {
+      "epoch": 1.8046731990955098,
+      "grad_norm": 0.5036799907684326,
+      "learning_rate": 7.4906564815678205e-06,
+      "loss": 0.5179,
+      "step": 4190
+    },
+    {
+      "epoch": 1.808980295036072,
+      "grad_norm": 0.5044455528259277,
+      "learning_rate": 7.444705606131697e-06,
+      "loss": 0.5171,
+      "step": 4200
+    },
+    {
+      "epoch": 1.813287390976634,
+      "grad_norm": 0.5645790696144104,
+      "learning_rate": 7.39881238231218e-06,
+      "loss": 0.5111,
+      "step": 4210
+    },
+    {
+      "epoch": 1.817594486917196,
+      "grad_norm": 0.4966265857219696,
+      "learning_rate": 7.352977845535387e-06,
+      "loss": 0.5144,
+      "step": 4220
+    },
+    {
+      "epoch": 1.821901582857758,
+      "grad_norm": 0.5225628614425659,
+      "learning_rate": 7.307203029903354e-06,
+      "loss": 0.5115,
+      "step": 4230
+    },
+    {
+      "epoch": 1.8262086787983203,
+      "grad_norm": 0.5282090902328491,
+      "learning_rate": 7.261488968170713e-06,
+      "loss": 0.5251,
+      "step": 4240
+    },
+    {
+      "epoch": 1.8305157747388823,
+      "grad_norm": 0.5346629023551941,
+      "learning_rate": 7.21583669172139e-06,
+      "loss": 0.5042,
+      "step": 4250
+    },
+    {
+      "epoch": 1.8348228706794445,
+      "grad_norm": 0.5141210556030273,
+      "learning_rate": 7.170247230545335e-06,
+      "loss": 0.5199,
+      "step": 4260
+    },
+    {
+      "epoch": 1.8391299666200065,
+      "grad_norm": 0.5251668691635132,
+      "learning_rate": 7.124721613215275e-06,
+      "loss": 0.4936,
+      "step": 4270
+    },
+    {
+      "epoch": 1.8434370625605685,
+      "grad_norm": 0.5125293731689453,
+      "learning_rate": 7.079260866863523e-06,
+      "loss": 0.5161,
+      "step": 4280
+    },
+    {
+      "epoch": 1.8477441585011305,
+      "grad_norm": 0.4881208837032318,
+      "learning_rate": 7.033866017158797e-06,
+      "loss": 0.5142,
+      "step": 4290
+    },
+    {
+      "epoch": 1.8520512544416927,
+      "grad_norm": 0.5215027928352356,
+      "learning_rate": 6.9885380882830735e-06,
+      "loss": 0.5097,
+      "step": 4300
+    },
+    {
+      "epoch": 1.8563583503822547,
+      "grad_norm": 0.4931368827819824,
+      "learning_rate": 6.943278102908491e-06,
+      "loss": 0.5123,
+      "step": 4310
+    },
+    {
+      "epoch": 1.860665446322817,
+      "grad_norm": 0.5080362558364868,
+      "learning_rate": 6.898087082174267e-06,
+      "loss": 0.5093,
+      "step": 4320
+    },
+    {
+      "epoch": 1.864972542263379,
+      "grad_norm": 0.537807285785675,
+      "learning_rate": 6.852966045663671e-06,
+      "loss": 0.5245,
+      "step": 4330
+    },
+    {
+      "epoch": 1.869279638203941,
+      "grad_norm": 0.5395597815513611,
+      "learning_rate": 6.807916011381008e-06,
+      "loss": 0.5016,
+      "step": 4340
+    },
+    {
+      "epoch": 1.873586734144503,
+      "grad_norm": 0.48623430728912354,
+      "learning_rate": 6.762937995728663e-06,
+      "loss": 0.4962,
+      "step": 4350
+    },
+    {
+      "epoch": 1.8778938300850652,
+      "grad_norm": 0.5058403611183167,
+      "learning_rate": 6.718033013484147e-06,
+      "loss": 0.5401,
+      "step": 4360
+    },
+    {
+      "epoch": 1.8822009260256274,
+      "grad_norm": 0.5220633149147034,
+      "learning_rate": 6.673202077777239e-06,
+      "loss": 0.5112,
+      "step": 4370
+    },
+    {
+      "epoch": 1.8865080219661894,
+      "grad_norm": 0.5163370966911316,
+      "learning_rate": 6.6284462000670924e-06,
+      "loss": 0.5231,
+      "step": 4380
+    },
+    {
+      "epoch": 1.8908151179067514,
+      "grad_norm": 0.508660614490509,
+      "learning_rate": 6.583766390119437e-06,
+      "loss": 0.5304,
+      "step": 4390
+    },
+    {
+      "epoch": 1.8951222138473134,
+      "grad_norm": 0.568144679069519,
+      "learning_rate": 6.539163655983786e-06,
+      "loss": 0.5086,
+      "step": 4400
+    },
+    {
+      "epoch": 1.8994293097878754,
+      "grad_norm": 0.5001341700553894,
+      "learning_rate": 6.494639003970701e-06,
+      "loss": 0.5084,
+      "step": 4410
+    },
+    {
+      "epoch": 1.9037364057284376,
+      "grad_norm": 0.5228297710418701,
+      "learning_rate": 6.450193438629078e-06,
+      "loss": 0.504,
+      "step": 4420
+    },
+    {
+      "epoch": 1.9080435016689998,
+      "grad_norm": 0.4816001057624817,
+      "learning_rate": 6.40582796272349e-06,
+      "loss": 0.5102,
+      "step": 4430
+    },
+    {
+      "epoch": 1.9123505976095618,
+      "grad_norm": 0.5058324933052063,
+      "learning_rate": 6.361543577211566e-06,
+      "loss": 0.524,
+      "step": 4440
+    },
+    {
+      "epoch": 1.9166576935501238,
+      "grad_norm": 0.5428106188774109,
+      "learning_rate": 6.317341281221392e-06,
+      "loss": 0.5082,
+      "step": 4450
+    },
+    {
+      "epoch": 1.9209647894906858,
+      "grad_norm": 0.5131290555000305,
+      "learning_rate": 6.273222072028991e-06,
+      "loss": 0.5316,
+      "step": 4460
+    },
+    {
+      "epoch": 1.9252718854312478,
+      "grad_norm": 0.5238609910011292,
+      "learning_rate": 6.2291869450358074e-06,
+      "loss": 0.5021,
+      "step": 4470
+    },
+    {
+      "epoch": 1.92957898137181,
+      "grad_norm": 0.4843258261680603,
+      "learning_rate": 6.1852368937462585e-06,
+      "loss": 0.5048,
+      "step": 4480
+    },
+    {
+      "epoch": 1.9338860773123723,
+      "grad_norm": 0.5138316750526428,
+      "learning_rate": 6.141372909745307e-06,
+      "loss": 0.5352,
+      "step": 4490
+    },
+    {
+      "epoch": 1.9381931732529343,
+      "grad_norm": 0.49319642782211304,
+      "learning_rate": 6.097595982676103e-06,
+      "loss": 0.5065,
+      "step": 4500
+    },
+    {
+      "epoch": 1.9425002691934963,
+      "grad_norm": 0.5176106095314026,
+      "learning_rate": 6.053907100217648e-06,
+      "loss": 0.5155,
+      "step": 4510
+    },
+    {
+      "epoch": 1.9468073651340583,
+      "grad_norm": 0.4772352874279022,
+      "learning_rate": 6.010307248062514e-06,
+      "loss": 0.5056,
+      "step": 4520
+    },
+    {
+      "epoch": 1.9511144610746203,
+      "grad_norm": 0.5366437435150146,
+      "learning_rate": 5.966797409894607e-06,
+      "loss": 0.4888,
+      "step": 4530
+    },
+    {
+      "epoch": 1.9554215570151825,
+      "grad_norm": 0.4917809069156647,
+      "learning_rate": 5.923378567366956e-06,
+      "loss": 0.5221,
+      "step": 4540
+    },
+    {
+      "epoch": 1.9597286529557447,
+      "grad_norm": 0.5597509741783142,
+      "learning_rate": 5.880051700079596e-06,
+      "loss": 0.5225,
+      "step": 4550
+    },
+    {
+      "epoch": 1.9640357488963067,
+      "grad_norm": 0.5258151888847351,
+      "learning_rate": 5.836817785557448e-06,
+      "loss": 0.5031,
+      "step": 4560
+    },
+    {
+      "epoch": 1.9683428448368687,
+      "grad_norm": 0.5679864287376404,
+      "learning_rate": 5.7936777992282565e-06,
+      "loss": 0.5074,
+      "step": 4570
+    },
+    {
+      "epoch": 1.9726499407774307,
+      "grad_norm": 0.5309889912605286,
+      "learning_rate": 5.750632714400607e-06,
+      "loss": 0.521,
+      "step": 4580
+    },
+    {
+      "epoch": 1.976957036717993,
+      "grad_norm": 0.5293132662773132,
+      "learning_rate": 5.707683502241936e-06,
+      "loss": 0.5133,
+      "step": 4590
+    },
+    {
+      "epoch": 1.981264132658555,
+      "grad_norm": 0.5223381519317627,
+      "learning_rate": 5.664831131756652e-06,
+      "loss": 0.5129,
+      "step": 4600
+    },
+    {
+      "epoch": 1.9855712285991172,
+      "grad_norm": 0.5365522503852844,
+      "learning_rate": 5.622076569764247e-06,
+      "loss": 0.504,
+      "step": 4610
+    },
+    {
+      "epoch": 1.9898783245396792,
+      "grad_norm": 0.5084212422370911,
+      "learning_rate": 5.5794207808774904e-06,
+      "loss": 0.488,
+      "step": 4620
+    },
+    {
+      "epoch": 1.9941854204802412,
+      "grad_norm": 0.4913804531097412,
+      "learning_rate": 5.536864727480683e-06,
+      "loss": 0.5098,
+      "step": 4630
+    },
+    {
+      "epoch": 1.9984925164208032,
+      "grad_norm": 0.5197212100028992,
+      "learning_rate": 5.4944093697079136e-06,
+      "loss": 0.5066,
+      "step": 4640
+    },
+    {
+      "epoch": 2.002799612361365,
+      "grad_norm": 0.51143479347229,
+      "learning_rate": 5.45205566542143e-06,
+      "loss": 0.4521,
+      "step": 4650
+    },
+    {
+      "epoch": 2.0071067083019276,
+      "grad_norm": 0.5107315182685852,
+      "learning_rate": 5.4098045701899934e-06,
+      "loss": 0.3968,
+      "step": 4660
+    },
+    {
+      "epoch": 2.0114138042424896,
+      "grad_norm": 0.5407351851463318,
+      "learning_rate": 5.367657037267354e-06,
+      "loss": 0.3933,
+      "step": 4670
+    },
+    {
+      "epoch": 2.0157209001830516,
+      "grad_norm": 0.5835046172142029,
+      "learning_rate": 5.325614017570712e-06,
+      "loss": 0.3897,
+      "step": 4680
+    },
+    {
+      "epoch": 2.0200279961236136,
+      "grad_norm": 0.5047739744186401,
+      "learning_rate": 5.283676459659288e-06,
+      "loss": 0.3992,
+      "step": 4690
+    },
+    {
+      "epoch": 2.0243350920641756,
+      "grad_norm": 0.5422953963279724,
+      "learning_rate": 5.241845309712921e-06,
+      "loss": 0.4131,
+      "step": 4700
+    },
+    {
+      "epoch": 2.0286421880047376,
+      "grad_norm": 0.5471384525299072,
+      "learning_rate": 5.2001215115106814e-06,
+      "loss": 0.3955,
+      "step": 4710
+    },
+    {
+      "epoch": 2.0329492839453,
+      "grad_norm": 0.5800908803939819,
+      "learning_rate": 5.158506006409644e-06,
+      "loss": 0.397,
+      "step": 4720
+    },
+    {
+      "epoch": 2.037256379885862,
+      "grad_norm": 0.5329377055168152,
+      "learning_rate": 5.116999733323591e-06,
+      "loss": 0.4017,
+      "step": 4730
+    },
+    {
+      "epoch": 2.041563475826424,
+      "grad_norm": 0.556845486164093,
+      "learning_rate": 5.075603628701869e-06,
+      "loss": 0.4009,
+      "step": 4740
+    },
+    {
+      "epoch": 2.045870571766986,
+      "grad_norm": 0.5501790642738342,
+      "learning_rate": 5.034318626508223e-06,
+      "loss": 0.3969,
+      "step": 4750
+    },
+    {
+      "epoch": 2.050177667707548,
+      "grad_norm": 0.5467825531959534,
+      "learning_rate": 4.993145658199766e-06,
+      "loss": 0.3996,
+      "step": 4760
+    },
+    {
+      "epoch": 2.05448476364811,
+      "grad_norm": 0.5644121766090393,
+      "learning_rate": 4.952085652705938e-06,
+      "loss": 0.3926,
+      "step": 4770
+    },
+    {
+      "epoch": 2.0587918595886725,
+      "grad_norm": 0.5279033780097961,
+      "learning_rate": 4.911139536407542e-06,
+      "loss": 0.3742,
+      "step": 4780
+    },
+    {
+      "epoch": 2.0630989555292345,
+      "grad_norm": 0.5283676981925964,
+      "learning_rate": 4.870308233115876e-06,
+      "loss": 0.3893,
+      "step": 4790
+    },
+    {
+      "epoch": 2.0674060514697965,
+      "grad_norm": 0.5302291512489319,
+      "learning_rate": 4.82959266405184e-06,
+      "loss": 0.3956,
+      "step": 4800
+    },
+    {
+      "epoch": 2.0717131474103585,
+      "grad_norm": 0.5381713509559631,
+      "learning_rate": 4.788993747825209e-06,
+      "loss": 0.4124,
+      "step": 4810
+    },
+    {
+      "epoch": 2.0760202433509205,
+      "grad_norm": 0.5772622227668762,
+      "learning_rate": 4.748512400413861e-06,
+      "loss": 0.405,
+      "step": 4820
+    },
+    {
+      "epoch": 2.0803273392914825,
+      "grad_norm": 0.5383191704750061,
+      "learning_rate": 4.708149535143138e-06,
+      "loss": 0.3874,
+      "step": 4830
+    },
+    {
+      "epoch": 2.084634435232045,
+      "grad_norm": 0.5546970963478088,
+      "learning_rate": 4.667906062665234e-06,
+      "loss": 0.3994,
+      "step": 4840
+    },
+    {
+      "epoch": 2.088941531172607,
+      "grad_norm": 0.5541481375694275,
+      "learning_rate": 4.627782890938632e-06,
+      "loss": 0.4073,
+      "step": 4850
+    },
+    {
+      "epoch": 2.093248627113169,
+      "grad_norm": 0.5656886100769043,
+      "learning_rate": 4.587780925207654e-06,
+      "loss": 0.3986,
+      "step": 4860
+    },
+    {
+      "epoch": 2.097555723053731,
+      "grad_norm": 0.5167860984802246,
+      "learning_rate": 4.5479010679819965e-06,
+      "loss": 0.3994,
+      "step": 4870
+    },
+    {
+      "epoch": 2.101862818994293,
+      "grad_norm": 0.585415780544281,
+      "learning_rate": 4.50814421901641e-06,
+      "loss": 0.3959,
+      "step": 4880
+    },
+    {
+      "epoch": 2.1061699149348554,
+      "grad_norm": 0.5390037894248962,
+      "learning_rate": 4.46851127529035e-06,
+      "loss": 0.393,
+      "step": 4890
+    },
+    {
+      "epoch": 2.1104770108754174,
+      "grad_norm": 0.5685362815856934,
+      "learning_rate": 4.42900313098779e-06,
+      "loss": 0.4031,
+      "step": 4900
+    },
+    {
+      "epoch": 2.1147841068159794,
+      "grad_norm": 0.5294394493103027,
+      "learning_rate": 4.389620677477023e-06,
+      "loss": 0.3926,
+      "step": 4910
+    },
+    {
+      "epoch": 2.1190912027565414,
+      "grad_norm": 0.5693227648735046,
+      "learning_rate": 4.3503648032905384e-06,
+      "loss": 0.3909,
+      "step": 4920
+    },
+    {
+      "epoch": 2.1233982986971034,
+      "grad_norm": 0.6294069886207581,
+      "learning_rate": 4.311236394105006e-06,
+      "loss": 0.3908,
+      "step": 4930
+    },
+    {
+      "epoch": 2.1277053946376654,
+      "grad_norm": 0.566862165927887,
+      "learning_rate": 4.27223633272126e-06,
+      "loss": 0.4019,
+      "step": 4940
+    },
+    {
+      "epoch": 2.132012490578228,
+      "grad_norm": 0.5680539608001709,
+      "learning_rate": 4.233365499044416e-06,
+      "loss": 0.3957,
+      "step": 4950
+    },
+    {
+      "epoch": 2.13631958651879,
+      "grad_norm": 0.5697780251502991,
+      "learning_rate": 4.194624770063985e-06,
+      "loss": 0.3876,
+      "step": 4960
+    },
+    {
+      "epoch": 2.140626682459352,
+      "grad_norm": 0.5857852697372437,
+      "learning_rate": 4.1560150198341174e-06,
+      "loss": 0.3986,
+      "step": 4970
+    },
+    {
+      "epoch": 2.144933778399914,
+      "grad_norm": 0.5707722306251526,
+      "learning_rate": 4.11753711945386e-06,
+      "loss": 0.4165,
+      "step": 4980
+    },
+    {
+      "epoch": 2.149240874340476,
+      "grad_norm": 0.5498836040496826,
+      "learning_rate": 4.079191937047511e-06,
+      "loss": 0.4236,
+      "step": 4990
+    },
+    {
+      "epoch": 2.153547970281038,
+      "grad_norm": 0.6008414626121521,
+      "learning_rate": 4.040980337745044e-06,
+      "loss": 0.3955,
+      "step": 5000
+    },
+    {
+      "epoch": 2.1578550662216003,
+      "grad_norm": 0.5871570110321045,
+      "learning_rate": 4.002903183662566e-06,
+      "loss": 0.3939,
+      "step": 5010
+    },
+    {
+      "epoch": 2.1621621621621623,
+      "grad_norm": 0.5556260347366333,
+      "learning_rate": 3.964961333882893e-06,
+      "loss": 0.4005,
+      "step": 5020
+    },
+    {
+      "epoch": 2.1664692581027243,
+      "grad_norm": 0.5592585206031799,
+      "learning_rate": 3.927155644436144e-06,
+      "loss": 0.4035,
+      "step": 5030
+    },
+    {
+      "epoch": 2.1707763540432863,
+      "grad_norm": 0.5638931393623352,
+      "learning_rate": 3.889486968280448e-06,
+      "loss": 0.3961,
+      "step": 5040
+    },
+    {
+      "epoch": 2.1750834499838483,
+      "grad_norm": 0.5473156571388245,
+      "learning_rate": 3.851956155282682e-06,
+      "loss": 0.3999,
+      "step": 5050
+    },
+    {
+      "epoch": 2.1793905459244103,
+      "grad_norm": 0.7088154554367065,
+      "learning_rate": 3.814564052199313e-06,
+      "loss": 0.3919,
+      "step": 5060
+    },
+    {
+      "epoch": 2.1836976418649727,
+      "grad_norm": 0.569315493106842,
+      "learning_rate": 3.777311502657279e-06,
+      "loss": 0.3924,
+      "step": 5070
+    },
+    {
+      "epoch": 2.1880047378055347,
+      "grad_norm": 0.6128218770027161,
+      "learning_rate": 3.7401993471349616e-06,
+      "loss": 0.4094,
+      "step": 5080
+    },
+    {
+      "epoch": 2.1923118337460967,
+      "grad_norm": 0.5971004962921143,
+      "learning_rate": 3.7032284229432325e-06,
+      "loss": 0.3786,
+      "step": 5090
+    },
+    {
+      "epoch": 2.1966189296866587,
+      "grad_norm": 0.5701526999473572,
+      "learning_rate": 3.666399564206541e-06,
+      "loss": 0.3912,
+      "step": 5100
+    },
+    {
+      "epoch": 2.2009260256272207,
+      "grad_norm": 0.5547009706497192,
+      "learning_rate": 3.6297136018441215e-06,
+      "loss": 0.3866,
+      "step": 5110
+    },
+    {
+      "epoch": 2.2052331215677827,
+      "grad_norm": 0.5613463521003723,
+      "learning_rate": 3.59317136355122e-06,
+      "loss": 0.3926,
+      "step": 5120
+    },
+    {
+      "epoch": 2.209540217508345,
+      "grad_norm": 0.6126610040664673,
+      "learning_rate": 3.556773673780446e-06,
+      "loss": 0.389,
+      "step": 5130
+    },
+    {
+      "epoch": 2.213847313448907,
+      "grad_norm": 0.5699272751808167,
+      "learning_rate": 3.520521353723142e-06,
+      "loss": 0.3982,
+      "step": 5140
+    },
+    {
+      "epoch": 2.218154409389469,
+      "grad_norm": 0.593333899974823,
+      "learning_rate": 3.484415221290889e-06,
+      "loss": 0.3826,
+      "step": 5150
+    },
+    {
+      "epoch": 2.222461505330031,
+      "grad_norm": 0.6188777685165405,
+      "learning_rate": 3.448456091097023e-06,
+      "loss": 0.4,
+      "step": 5160
+    },
+    {
+      "epoch": 2.226768601270593,
+      "grad_norm": 0.5949888825416565,
+      "learning_rate": 3.4126447744382753e-06,
+      "loss": 0.4062,
+      "step": 5170
+    },
+    {
+      "epoch": 2.231075697211155,
+      "grad_norm": 0.5788257718086243,
+      "learning_rate": 3.376982079276464e-06,
+      "loss": 0.3881,
+      "step": 5180
+    },
+    {
+      "epoch": 2.2353827931517176,
+      "grad_norm": 0.5726456642150879,
+      "learning_rate": 3.3414688102202564e-06,
+      "loss": 0.3968,
+      "step": 5190
+    },
+    {
+      "epoch": 2.2396898890922796,
+      "grad_norm": 0.5855600833892822,
+      "learning_rate": 3.3061057685070354e-06,
+      "loss": 0.3925,
+      "step": 5200
+    },
+    {
+      "epoch": 2.2439969850328416,
+      "grad_norm": 0.5823237299919128,
+      "learning_rate": 3.2708937519847916e-06,
+      "loss": 0.3875,
+      "step": 5210
+    },
+    {
+      "epoch": 2.2483040809734036,
+      "grad_norm": 0.5852989554405212,
+      "learning_rate": 3.23583355509416e-06,
+      "loss": 0.3985,
+      "step": 5220
+    },
+    {
+      "epoch": 2.2526111769139656,
+      "grad_norm": 0.5461825728416443,
+      "learning_rate": 3.200925968850459e-06,
+      "loss": 0.3917,
+      "step": 5230
+    },
+    {
+      "epoch": 2.256918272854528,
+      "grad_norm": 0.5536659359931946,
+      "learning_rate": 3.166171780825876e-06,
+      "loss": 0.3963,
+      "step": 5240
+    },
+    {
+      "epoch": 2.26122536879509,
+      "grad_norm": 0.5736192464828491,
+      "learning_rate": 3.1315717751316755e-06,
+      "loss": 0.4114,
+      "step": 5250
+    },
+    {
+      "epoch": 2.265532464735652,
+      "grad_norm": 0.5808764100074768,
+      "learning_rate": 3.097126732400515e-06,
+      "loss": 0.3795,
+      "step": 5260
+    },
+    {
+      "epoch": 2.269839560676214,
+      "grad_norm": 0.5790621042251587,
+      "learning_rate": 3.0628374297688436e-06,
+      "loss": 0.3991,
+      "step": 5270
+    },
+    {
+      "epoch": 2.274146656616776,
+      "grad_norm": 0.5211635231971741,
+      "learning_rate": 3.0287046408593478e-06,
+      "loss": 0.3796,
+      "step": 5280
+    },
+    {
+      "epoch": 2.278453752557338,
+      "grad_norm": 0.6152241230010986,
+      "learning_rate": 2.994729135763522e-06,
+      "loss": 0.3976,
+      "step": 5290
+    },
+    {
+      "epoch": 2.2827608484979,
+      "grad_norm": 0.6017261147499084,
+      "learning_rate": 2.9609116810242677e-06,
+      "loss": 0.4031,
+      "step": 5300
+    },
+    {
+      "epoch": 2.2870679444384625,
+      "grad_norm": 0.5612776279449463,
+      "learning_rate": 2.9272530396186194e-06,
+      "loss": 0.3985,
+      "step": 5310
+    },
+    {
+      "epoch": 2.2913750403790245,
+      "grad_norm": 0.6065710186958313,
+      "learning_rate": 2.893753970940525e-06,
+      "loss": 0.3975,
+      "step": 5320
+    },
+    {
+      "epoch": 2.2956821363195865,
+      "grad_norm": 0.5793972611427307,
+      "learning_rate": 2.8604152307837064e-06,
+      "loss": 0.3889,
+      "step": 5330
+    },
+    {
+      "epoch": 2.2999892322601485,
+      "grad_norm": 0.5591062307357788,
+      "learning_rate": 2.8272375713246125e-06,
+      "loss": 0.3903,
+      "step": 5340
+    },
+    {
+      "epoch": 2.3042963282007105,
+      "grad_norm": 0.5505937337875366,
+      "learning_rate": 2.794221741105446e-06,
+      "loss": 0.397,
+      "step": 5350
+    },
+    {
+      "epoch": 2.308603424141273,
+      "grad_norm": 0.6174246668815613,
+      "learning_rate": 2.7613684850172882e-06,
+      "loss": 0.3966,
+      "step": 5360
+    },
+    {
+      "epoch": 2.312910520081835,
+      "grad_norm": 0.6093124747276306,
+      "learning_rate": 2.7286785442832685e-06,
+      "loss": 0.3902,
+      "step": 5370
+    },
+    {
+      "epoch": 2.317217616022397,
+      "grad_norm": 0.5350244045257568,
+      "learning_rate": 2.696152656441868e-06,
+      "loss": 0.3935,
+      "step": 5380
+    },
+    {
+      "epoch": 2.321524711962959,
+      "grad_norm": 0.5422816276550293,
+      "learning_rate": 2.663791555330255e-06,
+      "loss": 0.3924,
+      "step": 5390
+    },
+    {
+      "epoch": 2.325831807903521,
+      "grad_norm": 0.5582048892974854,
+      "learning_rate": 2.6315959710677464e-06,
+      "loss": 0.397,
+      "step": 5400
+    },
+    {
+      "epoch": 2.330138903844083,
+      "grad_norm": 0.5601301789283752,
+      "learning_rate": 2.599566630039332e-06,
+      "loss": 0.3813,
+      "step": 5410
+    },
+    {
+      "epoch": 2.334445999784645,
+      "grad_norm": 0.5601345896720886,
+      "learning_rate": 2.567704254879274e-06,
+      "loss": 0.3974,
+      "step": 5420
+    },
+    {
+      "epoch": 2.3387530957252074,
+      "grad_norm": 0.614778459072113,
+      "learning_rate": 2.536009564454817e-06,
+      "loss": 0.3836,
+      "step": 5430
+    },
+    {
+      "epoch": 2.3430601916657694,
+      "grad_norm": 0.5759994983673096,
+      "learning_rate": 2.504483273849958e-06,
+      "loss": 0.3949,
+      "step": 5440
+    },
+    {
+      "epoch": 2.3473672876063314,
+      "grad_norm": 0.586625874042511,
+      "learning_rate": 2.473126094349331e-06,
+      "loss": 0.3829,
+      "step": 5450
+    },
+    {
+      "epoch": 2.3516743835468934,
+      "grad_norm": 0.5470960736274719,
+      "learning_rate": 2.4419387334221333e-06,
+      "loss": 0.3881,
+      "step": 5460
+    },
+    {
+      "epoch": 2.3559814794874554,
+      "grad_norm": 0.5486071705818176,
+      "learning_rate": 2.4109218947061884e-06,
+      "loss": 0.399,
+      "step": 5470
+    },
+    {
+      "epoch": 2.360288575428018,
+      "grad_norm": 0.5942230820655823,
+      "learning_rate": 2.3800762779920574e-06,
+      "loss": 0.3921,
+      "step": 5480
+    },
+    {
+      "epoch": 2.36459567136858,
+      "grad_norm": 0.5786502957344055,
+      "learning_rate": 2.3494025792072474e-06,
+      "loss": 0.3901,
+      "step": 5490
+    },
+    {
+      "epoch": 2.368902767309142,
+      "grad_norm": 0.6082814931869507,
+      "learning_rate": 2.3189014904005247e-06,
+      "loss": 0.391,
+      "step": 5500
+    },
+    {
+      "epoch": 2.373209863249704,
+      "grad_norm": 0.612694501876831,
+      "learning_rate": 2.2885736997262863e-06,
+      "loss": 0.3981,
+      "step": 5510
+    },
+    {
+      "epoch": 2.377516959190266,
+      "grad_norm": 0.5050374865531921,
+      "learning_rate": 2.2584198914290435e-06,
+      "loss": 0.3951,
+      "step": 5520
+    },
+    {
+      "epoch": 2.381824055130828,
+      "grad_norm": 0.5465214848518372,
+      "learning_rate": 2.2284407458279743e-06,
+      "loss": 0.4,
+      "step": 5530
+    },
+    {
+      "epoch": 2.3861311510713903,
+      "grad_norm": 0.5544529557228088,
+      "learning_rate": 2.1986369393015914e-06,
+      "loss": 0.3836,
+      "step": 5540
+    },
+    {
+      "epoch": 2.3904382470119523,
+      "grad_norm": 0.586337149143219,
+      "learning_rate": 2.169009144272467e-06,
+      "loss": 0.4139,
+      "step": 5550
+    },
+    {
+      "epoch": 2.3947453429525143,
+      "grad_norm": 0.6219981908798218,
+      "learning_rate": 2.1395580291920625e-06,
+      "loss": 0.4011,
+      "step": 5560
+    },
+    {
+      "epoch": 2.3990524388930763,
+      "grad_norm": 0.6941688060760498,
+      "learning_rate": 2.110284258525658e-06,
+      "loss": 0.405,
+      "step": 5570
+    },
+    {
+      "epoch": 2.4033595348336383,
+      "grad_norm": 0.5210332274436951,
+      "learning_rate": 2.081188492737345e-06,
+      "loss": 0.4017,
+      "step": 5580
+    },
+    {
+      "epoch": 2.4076666307742007,
+      "grad_norm": 0.5930879712104797,
+      "learning_rate": 2.0522713882751445e-06,
+      "loss": 0.3918,
+      "step": 5590
+    },
+    {
+      "epoch": 2.4119737267147627,
+      "grad_norm": 0.5910641551017761,
+      "learning_rate": 2.0235335975561775e-06,
+      "loss": 0.3996,
+      "step": 5600
+    },
+    {
+      "epoch": 2.4162808226553247,
+      "grad_norm": 0.5827698111534119,
+      "learning_rate": 1.9949757689519555e-06,
+      "loss": 0.3854,
+      "step": 5610
+    },
+    {
+      "epoch": 2.4205879185958867,
+      "grad_norm": 0.5518185496330261,
+      "learning_rate": 1.966598546773757e-06,
+      "loss": 0.4077,
+      "step": 5620
+    },
+    {
+      "epoch": 2.4248950145364487,
+      "grad_norm": 0.6005439162254333,
+      "learning_rate": 1.938402571258073e-06,
+      "loss": 0.4095,
+      "step": 5630
+    },
+    {
+      "epoch": 2.4292021104770107,
+      "grad_norm": 0.5761522054672241,
+      "learning_rate": 1.9103884785521887e-06,
+      "loss": 0.3966,
+      "step": 5640
+    },
+    {
+      "epoch": 2.4335092064175727,
+      "grad_norm": 0.5546764135360718,
+      "learning_rate": 1.8825569006998012e-06,
+      "loss": 0.395,
+      "step": 5650
+    },
+    {
+      "epoch": 2.437816302358135,
+      "grad_norm": 0.5639533996582031,
+      "learning_rate": 1.8549084656267846e-06,
+      "loss": 0.3938,
+      "step": 5660
+    },
+    {
+      "epoch": 2.442123398298697,
+      "grad_norm": 0.5662581324577332,
+      "learning_rate": 1.8274437971270044e-06,
+      "loss": 0.4004,
+      "step": 5670
+    },
+    {
+      "epoch": 2.446430494239259,
+      "grad_norm": 0.5856819748878479,
+      "learning_rate": 1.8001635148482621e-06,
+      "loss": 0.3946,
+      "step": 5680
+    },
+    {
+      "epoch": 2.450737590179821,
+      "grad_norm": 0.5766512751579285,
+      "learning_rate": 1.7730682342782967e-06,
+      "loss": 0.3931,
+      "step": 5690
+    },
+    {
+      "epoch": 2.455044686120383,
+      "grad_norm": 0.6373909711837769,
+      "learning_rate": 1.7461585667309045e-06,
+      "loss": 0.4006,
+      "step": 5700
+    },
+    {
+      "epoch": 2.4593517820609456,
+      "grad_norm": 0.5694748759269714,
+      "learning_rate": 1.719435119332159e-06,
+      "loss": 0.3989,
+      "step": 5710
+    },
+    {
+      "epoch": 2.4636588780015076,
+      "grad_norm": 0.5339934229850769,
+      "learning_rate": 1.6928984950066918e-06,
+      "loss": 0.3966,
+      "step": 5720
+    },
+    {
+      "epoch": 2.4679659739420696,
+      "grad_norm": 0.5888383388519287,
+      "learning_rate": 1.6665492924641113e-06,
+      "loss": 0.3833,
+      "step": 5730
+    },
+    {
+      "epoch": 2.4722730698826316,
+      "grad_norm": 0.5573282241821289,
+      "learning_rate": 1.6403881061854732e-06,
+      "loss": 0.4,
+      "step": 5740
+    },
+    {
+      "epoch": 2.4765801658231936,
+      "grad_norm": 0.5756634473800659,
+      "learning_rate": 1.6144155264098883e-06,
+      "loss": 0.3964,
+      "step": 5750
+    },
+    {
+      "epoch": 2.4808872617637556,
+      "grad_norm": 0.5784355401992798,
+      "learning_rate": 1.58863213912119e-06,
+      "loss": 0.3762,
+      "step": 5760
+    },
+    {
+      "epoch": 2.4851943577043176,
+      "grad_norm": 0.6090006828308105,
+      "learning_rate": 1.563038526034727e-06,
+      "loss": 0.3986,
+      "step": 5770
+    },
+    {
+      "epoch": 2.48950145364488,
+      "grad_norm": 0.5565779209136963,
+      "learning_rate": 1.5376352645842242e-06,
+      "loss": 0.3916,
+      "step": 5780
+    },
+    {
+      "epoch": 2.493808549585442,
+      "grad_norm": 0.6107103228569031,
+      "learning_rate": 1.5124229279087655e-06,
+      "loss": 0.4093,
+      "step": 5790
+    },
+    {
+      "epoch": 2.498115645526004,
+      "grad_norm": 0.5300205945968628,
+      "learning_rate": 1.487402084839864e-06,
+      "loss": 0.4047,
+      "step": 5800
+    },
+    {
+      "epoch": 2.502422741466566,
+      "grad_norm": 0.6008495688438416,
+      "learning_rate": 1.4625732998886178e-06,
+      "loss": 0.4023,
+      "step": 5810
+    },
+    {
+      "epoch": 2.5067298374071285,
+      "grad_norm": 0.5560673475265503,
+      "learning_rate": 1.437937133232985e-06,
+      "loss": 0.3968,
+      "step": 5820
+    },
+    {
+      "epoch": 2.5110369333476905,
+      "grad_norm": 0.5503118634223938,
+      "learning_rate": 1.413494140705136e-06,
+      "loss": 0.3876,
+      "step": 5830
+    },
+    {
+      "epoch": 2.5153440292882525,
+      "grad_norm": 0.5559957027435303,
+      "learning_rate": 1.3892448737789243e-06,
+      "loss": 0.392,
+      "step": 5840
+    },
+    {
+      "epoch": 2.5196511252288145,
+      "grad_norm": 0.5354902148246765,
+      "learning_rate": 1.365189879557426e-06,
+      "loss": 0.3988,
+      "step": 5850
+    },
+    {
+      "epoch": 2.5239582211693765,
+      "grad_norm": 0.577046275138855,
+      "learning_rate": 1.3413297007606196e-06,
+      "loss": 0.3948,
+      "step": 5860
+    },
+    {
+      "epoch": 2.5282653171099385,
+      "grad_norm": 0.5745800733566284,
+      "learning_rate": 1.3176648757131205e-06,
+      "loss": 0.395,
+      "step": 5870
+    },
+    {
+      "epoch": 2.5325724130505005,
+      "grad_norm": 0.5721185207366943,
+      "learning_rate": 1.2941959383320478e-06,
+      "loss": 0.3918,
+      "step": 5880
+    },
+    {
+      "epoch": 2.5368795089910625,
+      "grad_norm": 0.5935482978820801,
+      "learning_rate": 1.2709234181149765e-06,
+      "loss": 0.376,
+      "step": 5890
+    },
+    {
+      "epoch": 2.541186604931625,
+      "grad_norm": 0.5709375143051147,
+      "learning_rate": 1.2478478401279848e-06,
+      "loss": 0.3881,
+      "step": 5900
+    },
+    {
+      "epoch": 2.545493700872187,
+      "grad_norm": 0.5233684182167053,
+      "learning_rate": 1.2249697249938197e-06,
+      "loss": 0.3945,
+      "step": 5910
+    },
+    {
+      "epoch": 2.549800796812749,
+      "grad_norm": 0.5812388062477112,
+      "learning_rate": 1.2022895888801333e-06,
+      "loss": 0.3984,
+      "step": 5920
+    },
+    {
+      "epoch": 2.554107892753311,
+      "grad_norm": 0.560550332069397,
+      "learning_rate": 1.1798079434878584e-06,
+      "loss": 0.3942,
+      "step": 5930
+    },
+    {
+      "epoch": 2.5584149886938734,
+      "grad_norm": 0.6010858416557312,
+      "learning_rate": 1.1575252960396422e-06,
+      "loss": 0.3851,
+      "step": 5940
+    },
+    {
+      "epoch": 2.5627220846344354,
+      "grad_norm": 0.5857875347137451,
+      "learning_rate": 1.1354421492684252e-06,
+      "loss": 0.3993,
+      "step": 5950
+    },
+    {
+      "epoch": 2.5670291805749974,
+      "grad_norm": 0.604179859161377,
+      "learning_rate": 1.1135590014060772e-06,
+      "loss": 0.388,
+      "step": 5960
+    },
+    {
+      "epoch": 2.5713362765155594,
+      "grad_norm": 0.569106936454773,
+      "learning_rate": 1.0918763461721648e-06,
+      "loss": 0.4014,
+      "step": 5970
+    },
+    {
+      "epoch": 2.5756433724561214,
+      "grad_norm": 0.5742547512054443,
+      "learning_rate": 1.0703946727628234e-06,
+      "loss": 0.3839,
+      "step": 5980
+    },
+    {
+      "epoch": 2.5799504683966834,
+      "grad_norm": 0.5561407208442688,
+      "learning_rate": 1.0491144658397e-06,
+      "loss": 0.3853,
+      "step": 5990
+    },
+    {
+      "epoch": 2.5842575643372454,
+      "grad_norm": 0.5482295155525208,
+      "learning_rate": 1.0280362055190341e-06,
+      "loss": 0.3876,
+      "step": 6000
+    },
+    {
+      "epoch": 2.588564660277808,
+      "grad_norm": 0.5737982392311096,
+      "learning_rate": 1.0071603673608176e-06,
+      "loss": 0.4059,
+      "step": 6010
+    },
+    {
+      "epoch": 2.59287175621837,
+      "grad_norm": 0.547715961933136,
+      "learning_rate": 9.864874223580668e-07,
+      "loss": 0.3837,
+      "step": 6020
+    },
+    {
+      "epoch": 2.597178852158932,
+      "grad_norm": 0.607851505279541,
+      "learning_rate": 9.66017836926203e-07,
+      "loss": 0.3779,
+      "step": 6030
+    },
+    {
+      "epoch": 2.601485948099494,
+      "grad_norm": 0.5557613968849182,
+      "learning_rate": 9.457520728925151e-07,
+      "loss": 0.3995,
+      "step": 6040
+    },
+    {
+      "epoch": 2.605793044040056,
+      "grad_norm": 0.5470052361488342,
+      "learning_rate": 9.256905874857535e-07,
+      "loss": 0.3916,
+      "step": 6050
+    },
+    {
+      "epoch": 2.6101001399806183,
+      "grad_norm": 0.5718830227851868,
+      "learning_rate": 9.058338333258032e-07,
+      "loss": 0.3997,
+      "step": 6060
+    },
+    {
+      "epoch": 2.6144072359211803,
+      "grad_norm": 0.5838637948036194,
+      "learning_rate": 8.861822584134882e-07,
+      "loss": 0.39,
+      "step": 6070
+    },
+    {
+      "epoch": 2.6187143318617423,
+      "grad_norm": 0.5819488763809204,
+      "learning_rate": 8.667363061204415e-07,
+      "loss": 0.4028,
+      "step": 6080
+    },
+    {
+      "epoch": 2.6230214278023043,
+      "grad_norm": 0.5477743744850159,
+      "learning_rate": 8.474964151791232e-07,
+      "loss": 0.3979,
+      "step": 6090
+    },
+    {
+      "epoch": 2.6273285237428663,
+      "grad_norm": 0.6217262744903564,
+      "learning_rate": 8.284630196729059e-07,
+      "loss": 0.3993,
+      "step": 6100
+    },
+    {
+      "epoch": 2.6316356196834283,
+      "grad_norm": 0.5514227747917175,
+      "learning_rate": 8.096365490262925e-07,
+      "loss": 0.4058,
+      "step": 6110
+    },
+    {
+      "epoch": 2.6359427156239903,
+      "grad_norm": 0.645946204662323,
+      "learning_rate": 7.910174279952232e-07,
+      "loss": 0.3992,
+      "step": 6120
+    },
+    {
+      "epoch": 2.6402498115645527,
+      "grad_norm": 0.5741420984268188,
+      "learning_rate": 7.726060766574883e-07,
+      "loss": 0.3938,
+      "step": 6130
+    },
+    {
+      "epoch": 2.6445569075051147,
+      "grad_norm": 0.5910946726799011,
+      "learning_rate": 7.544029104032558e-07,
+      "loss": 0.3898,
+      "step": 6140
+    },
+    {
+      "epoch": 2.6488640034456767,
+      "grad_norm": 0.5803595185279846,
+      "learning_rate": 7.364083399256971e-07,
+      "loss": 0.388,
+      "step": 6150
+    },
+    {
+      "epoch": 2.6531710993862387,
+      "grad_norm": 0.596809446811676,
+      "learning_rate": 7.186227712117266e-07,
+      "loss": 0.388,
+      "step": 6160
+    },
+    {
+      "epoch": 2.6574781953268007,
+      "grad_norm": 0.6213387250900269,
+      "learning_rate": 7.010466055328313e-07,
+      "loss": 0.3839,
+      "step": 6170
+    },
+    {
+      "epoch": 2.661785291267363,
+      "grad_norm": 0.5913180112838745,
+      "learning_rate": 6.836802394360276e-07,
+      "loss": 0.3989,
+      "step": 6180
+    },
+    {
+      "epoch": 2.666092387207925,
+      "grad_norm": 0.6089721322059631,
+      "learning_rate": 6.665240647349125e-07,
+      "loss": 0.4039,
+      "step": 6190
+    },
+    {
+      "epoch": 2.670399483148487,
+      "grad_norm": 0.5730729103088379,
+      "learning_rate": 6.495784685008133e-07,
+      "loss": 0.3951,
+      "step": 6200
+    },
+    {
+      "epoch": 2.674706579089049,
+      "grad_norm": 0.5562758445739746,
+      "learning_rate": 6.32843833054072e-07,
+      "loss": 0.3837,
+      "step": 6210
+    },
+    {
+      "epoch": 2.679013675029611,
+      "grad_norm": 0.5627213716506958,
+      "learning_rate": 6.16320535955407e-07,
+      "loss": 0.3712,
+      "step": 6220
+    },
+    {
+      "epoch": 2.683320770970173,
+      "grad_norm": 0.559660017490387,
+      "learning_rate": 6.000089499973971e-07,
+      "loss": 0.3901,
+      "step": 6230
+    },
+    {
+      "epoch": 2.687627866910735,
+      "grad_norm": 0.6018761992454529,
+      "learning_rate": 5.839094431960713e-07,
+      "loss": 0.383,
+      "step": 6240
+    },
+    {
+      "epoch": 2.6919349628512976,
+      "grad_norm": 0.5534284710884094,
+      "learning_rate": 5.680223787826089e-07,
+      "loss": 0.3925,
+      "step": 6250
+    },
+    {
+      "epoch": 2.6962420587918596,
+      "grad_norm": 0.5682888031005859,
+      "learning_rate": 5.523481151951427e-07,
+      "loss": 0.3929,
+      "step": 6260
+    },
+    {
+      "epoch": 2.7005491547324216,
+      "grad_norm": 0.6271238923072815,
+      "learning_rate": 5.368870060706677e-07,
+      "loss": 0.3942,
+      "step": 6270
+    },
+    {
+      "epoch": 2.7048562506729836,
+      "grad_norm": 0.5881267786026001,
+      "learning_rate": 5.216394002370695e-07,
+      "loss": 0.3876,
+      "step": 6280
+    },
+    {
+      "epoch": 2.709163346613546,
+      "grad_norm": 0.6085900068283081,
+      "learning_rate": 5.066056417052445e-07,
+      "loss": 0.3958,
+      "step": 6290
+    },
+    {
+      "epoch": 2.713470442554108,
+      "grad_norm": 0.5912172198295593,
+      "learning_rate": 4.917860696613541e-07,
+      "loss": 0.3887,
+      "step": 6300
+    },
+    {
+      "epoch": 2.71777753849467,
+      "grad_norm": 0.6698789596557617,
+      "learning_rate": 4.771810184591541e-07,
+      "loss": 0.3899,
+      "step": 6310
+    },
+    {
+      "epoch": 2.722084634435232,
+      "grad_norm": 0.5682712197303772,
+      "learning_rate": 4.627908176124618e-07,
+      "loss": 0.3826,
+      "step": 6320
+    },
+    {
+      "epoch": 2.726391730375794,
+      "grad_norm": 0.5702280402183533,
+      "learning_rate": 4.486157917877232e-07,
+      "loss": 0.3908,
+      "step": 6330
+    },
+    {
+      "epoch": 2.730698826316356,
+      "grad_norm": 0.5540564060211182,
+      "learning_rate": 4.346562607966787e-07,
+      "loss": 0.3962,
+      "step": 6340
+    },
+    {
+      "epoch": 2.735005922256918,
+      "grad_norm": 0.6031074523925781,
+      "learning_rate": 4.209125395891589e-07,
+      "loss": 0.3791,
+      "step": 6350
+    },
+    {
+      "epoch": 2.73931301819748,
+      "grad_norm": 0.5727553963661194,
+      "learning_rate": 4.0738493824596715e-07,
+      "loss": 0.4023,
+      "step": 6360
+    },
+    {
+      "epoch": 2.7436201141380425,
+      "grad_norm": 0.5374717116355896,
+      "learning_rate": 3.940737619718937e-07,
+      "loss": 0.38,
+      "step": 6370
+    },
+    {
+      "epoch": 2.7479272100786045,
+      "grad_norm": 0.5720168352127075,
+      "learning_rate": 3.809793110888249e-07,
+      "loss": 0.4011,
+      "step": 6380
+    },
+    {
+      "epoch": 2.7522343060191665,
+      "grad_norm": 0.5751203894615173,
+      "learning_rate": 3.6810188102896605e-07,
+      "loss": 0.3941,
+      "step": 6390
+    },
+    {
+      "epoch": 2.7565414019597285,
+      "grad_norm": 0.5838513970375061,
+      "learning_rate": 3.554417623281825e-07,
+      "loss": 0.3834,
+      "step": 6400
+    },
+    {
+      "epoch": 2.760848497900291,
+      "grad_norm": 0.6204310059547424,
+      "learning_rate": 3.429992406194338e-07,
+      "loss": 0.3933,
+      "step": 6410
+    },
+    {
+      "epoch": 2.765155593840853,
+      "grad_norm": 0.6237754225730896,
+      "learning_rate": 3.3077459662634205e-07,
+      "loss": 0.3911,
+      "step": 6420
+    },
+    {
+      "epoch": 2.769462689781415,
+      "grad_norm": 0.561553418636322,
+      "learning_rate": 3.1876810615684705e-07,
+      "loss": 0.3847,
+      "step": 6430
+    },
+    {
+      "epoch": 2.773769785721977,
+      "grad_norm": 0.568580150604248,
+      "learning_rate": 3.069800400969947e-07,
+      "loss": 0.3967,
+      "step": 6440
+    },
+    {
+      "epoch": 2.778076881662539,
+      "grad_norm": 0.6103531122207642,
+      "learning_rate": 2.954106644048127e-07,
+      "loss": 0.3731,
+      "step": 6450
+    },
+    {
+      "epoch": 2.782383977603101,
+      "grad_norm": 0.560199499130249,
+      "learning_rate": 2.840602401043213e-07,
+      "loss": 0.3889,
+      "step": 6460
+    },
+    {
+      "epoch": 2.786691073543663,
+      "grad_norm": 0.5612174868583679,
+      "learning_rate": 2.7292902327963776e-07,
+      "loss": 0.3915,
+      "step": 6470
+    },
+    {
+      "epoch": 2.7909981694842254,
+      "grad_norm": 0.5860500335693359,
+      "learning_rate": 2.620172650692021e-07,
+      "loss": 0.4063,
+      "step": 6480
+    },
+    {
+      "epoch": 2.7953052654247874,
+      "grad_norm": 0.6044652462005615,
+      "learning_rate": 2.513252116601062e-07,
+      "loss": 0.39,
+      "step": 6490
+    },
+    {
+      "epoch": 2.7996123613653494,
+      "grad_norm": 0.5966377258300781,
+      "learning_rate": 2.408531042825446e-07,
+      "loss": 0.3965,
+      "step": 6500
+    },
+    {
+      "epoch": 2.8039194573059114,
+      "grad_norm": 0.5729289650917053,
+      "learning_rate": 2.3060117920437164e-07,
+      "loss": 0.3798,
+      "step": 6510
+    },
+    {
+      "epoch": 2.8082265532464734,
+      "grad_norm": 0.6403810977935791,
+      "learning_rate": 2.2056966772576626e-07,
+      "loss": 0.4096,
+      "step": 6520
+    },
+    {
+      "epoch": 2.812533649187036,
+      "grad_norm": 0.5852852463722229,
+      "learning_rate": 2.1075879617401984e-07,
+      "loss": 0.383,
+      "step": 6530
+    },
+    {
+      "epoch": 2.816840745127598,
+      "grad_norm": 0.6858223080635071,
+      "learning_rate": 2.0116878589842236e-07,
+      "loss": 0.3763,
+      "step": 6540
+    },
+    {
+      "epoch": 2.82114784106816,
+      "grad_norm": 0.5583459138870239,
+      "learning_rate": 1.917998532652765e-07,
+      "loss": 0.4007,
+      "step": 6550
+    },
+    {
+      "epoch": 2.825454937008722,
+      "grad_norm": 0.6212313175201416,
+      "learning_rate": 1.8265220965300812e-07,
+      "loss": 0.3946,
+      "step": 6560
+    },
+    {
+      "epoch": 2.829762032949284,
+      "grad_norm": 0.5777102112770081,
+      "learning_rate": 1.7372606144740567e-07,
+      "loss": 0.3908,
+      "step": 6570
+    },
+    {
+      "epoch": 2.834069128889846,
+      "grad_norm": 0.5885289311408997,
+      "learning_rate": 1.6502161003695615e-07,
+      "loss": 0.4051,
+      "step": 6580
+    },
+    {
+      "epoch": 2.838376224830408,
+      "grad_norm": 0.6133362054824829,
+      "learning_rate": 1.5653905180830432e-07,
+      "loss": 0.3909,
+      "step": 6590
+    },
+    {
+      "epoch": 2.8426833207709703,
+      "grad_norm": 0.5662548542022705,
+      "learning_rate": 1.48278578141825e-07,
+      "loss": 0.3689,
+      "step": 6600
+    },
+    {
+      "epoch": 2.8469904167115323,
+      "grad_norm": 0.5703479647636414,
+      "learning_rate": 1.4024037540730006e-07,
+      "loss": 0.3812,
+      "step": 6610
+    },
+    {
+      "epoch": 2.8512975126520943,
+      "grad_norm": 0.5604844689369202,
+      "learning_rate": 1.324246249597183e-07,
+      "loss": 0.3992,
+      "step": 6620
+    },
+    {
+      "epoch": 2.8556046085926563,
+      "grad_norm": 0.6033147573471069,
+      "learning_rate": 1.2483150313517766e-07,
+      "loss": 0.3937,
+      "step": 6630
+    },
+    {
+      "epoch": 2.8599117045332187,
+      "grad_norm": 0.5846080780029297,
+      "learning_rate": 1.1746118124691508e-07,
+      "loss": 0.4123,
+      "step": 6640
+    },
+    {
+      "epoch": 2.8642188004737807,
+      "grad_norm": 0.63025963306427,
+      "learning_rate": 1.103138255814329e-07,
+      "loss": 0.3998,
+      "step": 6650
+    },
+    {
+      "epoch": 2.8685258964143427,
+      "grad_norm": 0.5580465197563171,
+      "learning_rate": 1.0338959739475296e-07,
+      "loss": 0.4007,
+      "step": 6660
+    },
+    {
+      "epoch": 2.8728329923549047,
+      "grad_norm": 0.5767059326171875,
+      "learning_rate": 9.66886529087785e-08,
+      "loss": 0.4008,
+      "step": 6670
+    },
+    {
+      "epoch": 2.8771400882954667,
+      "grad_norm": 0.583044707775116,
+      "learning_rate": 9.021114330776348e-08,
+      "loss": 0.403,
+      "step": 6680
+    },
+    {
+      "epoch": 2.8814471842360287,
+      "grad_norm": 0.5440847873687744,
+      "learning_rate": 8.395721473490992e-08,
+      "loss": 0.3839,
+      "step": 6690
+    },
+    {
+      "epoch": 2.8857542801765907,
+      "grad_norm": 0.55162513256073,
+      "learning_rate": 7.792700828906374e-08,
+      "loss": 0.4017,
+      "step": 6700
+    },
+    {
+      "epoch": 2.8900613761171527,
+      "grad_norm": 0.5817933082580566,
+      "learning_rate": 7.212066002153518e-08,
+      "loss": 0.4009,
+      "step": 6710
+    },
+    {
+      "epoch": 2.894368472057715,
+      "grad_norm": 0.6080750226974487,
+      "learning_rate": 6.653830093302782e-08,
+      "loss": 0.3964,
+      "step": 6720
+    },
+    {
+      "epoch": 2.898675567998277,
+      "grad_norm": 0.5681482553482056,
+      "learning_rate": 6.11800569706833e-08,
+      "loss": 0.4003,
+      "step": 6730
+    },
+    {
+      "epoch": 2.902982663938839,
+      "grad_norm": 0.5769705176353455,
+      "learning_rate": 5.604604902524235e-08,
+      "loss": 0.4017,
+      "step": 6740
+    },
+    {
+      "epoch": 2.907289759879401,
+      "grad_norm": 0.546116828918457,
+      "learning_rate": 5.113639292831152e-08,
+      "loss": 0.3828,
+      "step": 6750
+    },
+    {
+      "epoch": 2.9115968558199636,
+      "grad_norm": 0.590798020362854,
+      "learning_rate": 4.645119944975296e-08,
+      "loss": 0.3853,
+      "step": 6760
+    },
+    {
+      "epoch": 2.9159039517605256,
+      "grad_norm": 0.5748469233512878,
+      "learning_rate": 4.1990574295187606e-08,
+      "loss": 0.4107,
+      "step": 6770
+    },
+    {
+      "epoch": 2.9202110477010876,
+      "grad_norm": 0.5733410716056824,
+      "learning_rate": 3.7754618103608144e-08,
+      "loss": 0.4052,
+      "step": 6780
+    },
+    {
+      "epoch": 2.9245181436416496,
+      "grad_norm": 0.5576743483543396,
+      "learning_rate": 3.374342644510531e-08,
+      "loss": 0.3846,
+      "step": 6790
+    },
+    {
+      "epoch": 2.9288252395822116,
+      "grad_norm": 0.596834123134613,
+      "learning_rate": 2.9957089818718476e-08,
+      "loss": 0.4029,
+      "step": 6800
+    },
+    {
+      "epoch": 2.9331323355227736,
+      "grad_norm": 0.5680873990058899,
+      "learning_rate": 2.639569365038841e-08,
+      "loss": 0.381,
+      "step": 6810
+    },
+    {
+      "epoch": 2.9374394314633356,
+      "grad_norm": 0.5597060918807983,
+      "learning_rate": 2.305931829102992e-08,
+      "loss": 0.3974,
+      "step": 6820
+    },
+    {
+      "epoch": 2.941746527403898,
+      "grad_norm": 0.5827191472053528,
+      "learning_rate": 1.9948039014724417e-08,
+      "loss": 0.3973,
+      "step": 6830
+    },
+    {
+      "epoch": 2.94605362334446,
+      "grad_norm": 0.6119829416275024,
+      "learning_rate": 1.706192601701462e-08,
+      "loss": 0.3984,
+      "step": 6840
+    },
+    {
+      "epoch": 2.950360719285022,
+      "grad_norm": 0.602497935295105,
+      "learning_rate": 1.4401044413324682e-08,
+      "loss": 0.4086,
+      "step": 6850
+    },
+    {
+      "epoch": 2.954667815225584,
+      "grad_norm": 0.5783790349960327,
+      "learning_rate": 1.1965454237493623e-08,
+      "loss": 0.3945,
+      "step": 6860
+    },
+    {
+      "epoch": 2.958974911166146,
+      "grad_norm": 0.5653091073036194,
+      "learning_rate": 9.755210440413055e-09,
+      "loss": 0.3938,
+      "step": 6870
+    },
+    {
+      "epoch": 2.9632820071067085,
+      "grad_norm": 0.5716846585273743,
+      "learning_rate": 7.770362888795957e-09,
+      "loss": 0.3935,
+      "step": 6880
+    },
+    {
+      "epoch": 2.9675891030472705,
+      "grad_norm": 0.6015262603759766,
+      "learning_rate": 6.0109563640442515e-09,
+      "loss": 0.3955,
+      "step": 6890
+    },
+    {
+      "epoch": 2.9718961989878325,
+      "grad_norm": 0.5763514041900635,
+      "learning_rate": 4.477030561246265e-09,
+      "loss": 0.4069,
+      "step": 6900
+    },
+    {
+      "epoch": 2.9762032949283945,
+      "grad_norm": 0.5644577741622925,
+      "learning_rate": 3.168620088271901e-09,
+      "loss": 0.3921,
+      "step": 6910
+    },
+    {
+      "epoch": 2.9805103908689565,
+      "grad_norm": 0.5302848219871521,
+      "learning_rate": 2.0857544650010332e-09,
+      "loss": 0.404,
+      "step": 6920
+    },
+    {
+      "epoch": 2.9848174868095185,
+      "grad_norm": 0.6025976538658142,
+      "learning_rate": 1.2284581226507108e-09,
+      "loss": 0.4037,
+      "step": 6930
+    },
+    {
+      "epoch": 2.9891245827500805,
+      "grad_norm": 0.5681896805763245,
+      "learning_rate": 5.967504032267091e-10,
+      "loss": 0.4031,
+      "step": 6940
+    },
+    {
+      "epoch": 2.993431678690643,
+      "grad_norm": 0.5708478093147278,
+      "learning_rate": 1.906455590883205e-10,
+      "loss": 0.4206,
+      "step": 6950
+    },
+    {
+      "epoch": 2.997738774631205,
+      "grad_norm": 0.5966918468475342,
+      "learning_rate": 1.015275262306048e-11,
+      "loss": 0.4014,
+      "step": 6960
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 6963,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 300,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 7523782707118080.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}