diff --git "a/final/trainer_state.json" "b/final/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/final/trainer_state.json"
@@ -0,0 +1,4185 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.19014513183609394,
+  "eval_steps": 500,
+  "global_step": 5935,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0003203793291256848,
+      "grad_norm": 2697.9695951365356,
+      "learning_rate": 2.910112359550562e-06,
+      "loss": 2.1080955505371093,
+      "step": 10
+    },
+    {
+      "epoch": 0.0006407586582513696,
+      "grad_norm": 150.28792394404246,
+      "learning_rate": 3.921348314606742e-06,
+      "loss": 1.051735782623291,
+      "step": 20
+    },
+    {
+      "epoch": 0.0009611379873770544,
+      "grad_norm": 11.728004906904669,
+      "learning_rate": 4.932584269662922e-06,
+      "loss": 0.8452852249145508,
+      "step": 30
+    },
+    {
+      "epoch": 0.0012815173165027392,
+      "grad_norm": 79.82045143025138,
+      "learning_rate": 5.943820224719102e-06,
+      "loss": 0.8374807357788085,
+      "step": 40
+    },
+    {
+      "epoch": 0.001601896645628424,
+      "grad_norm": 15.487828868814473,
+      "learning_rate": 6.955056179775282e-06,
+      "loss": 0.8031275749206543,
+      "step": 50
+    },
+    {
+      "epoch": 0.0019222759747541088,
+      "grad_norm": 26.189373840069653,
+      "learning_rate": 7.966292134831462e-06,
+      "loss": 0.7619327545166016,
+      "step": 60
+    },
+    {
+      "epoch": 0.002242655303879794,
+      "grad_norm": 24.510054444181353,
+      "learning_rate": 8.977528089887641e-06,
+      "loss": 0.788119888305664,
+      "step": 70
+    },
+    {
+      "epoch": 0.0025630346330054784,
+      "grad_norm": 27.619677813912134,
+      "learning_rate": 9.988764044943822e-06,
+      "loss": 0.8074554443359375,
+      "step": 80
+    },
+    {
+      "epoch": 0.0028834139621311634,
+      "grad_norm": 53.27561070809039,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.7780509948730469,
+      "step": 90
+    },
+    {
+      "epoch": 0.003203793291256848,
+      "grad_norm": 20.22293994140261,
+      "learning_rate": 1.2011235955056182e-05,
+      "loss": 0.7639347076416015,
+      "step": 100
+    },
+    {
+      "epoch": 0.003524172620382533,
+      "grad_norm": 28.55291130912943,
+      "learning_rate": 1.3022471910112362e-05,
+      "loss": 0.7700469970703125,
+      "step": 110
+    },
+    {
+      "epoch": 0.0038445519495082176,
+      "grad_norm": 7.520305767933223,
+      "learning_rate": 1.403370786516854e-05,
+      "loss": 0.7953043937683105,
+      "step": 120
+    },
+    {
+      "epoch": 0.004164931278633903,
+      "grad_norm": 96.08078842951305,
+      "learning_rate": 1.504494382022472e-05,
+      "loss": 0.8506025314331055,
+      "step": 130
+    },
+    {
+      "epoch": 0.004485310607759588,
+      "grad_norm": 948.131373093307,
+      "learning_rate": 1.60561797752809e-05,
+      "loss": 0.8122509002685547,
+      "step": 140
+    },
+    {
+      "epoch": 0.004805689936885272,
+      "grad_norm": 128.55037969805156,
+      "learning_rate": 1.706741573033708e-05,
+      "loss": 0.7938695907592773,
+      "step": 150
+    },
+    {
+      "epoch": 0.005126069266010957,
+      "grad_norm": 25.93618158094234,
+      "learning_rate": 1.8078651685393256e-05,
+      "loss": 0.8438175201416016,
+      "step": 160
+    },
+    {
+      "epoch": 0.005446448595136642,
+      "grad_norm": 9.528266971779534,
+      "learning_rate": 1.908988764044944e-05,
+      "loss": 0.9011789321899414,
+      "step": 170
+    },
+    {
+      "epoch": 0.005766827924262327,
+      "grad_norm": 4.6353029582495315,
+      "learning_rate": 2e-05,
+      "loss": 0.8190475463867187,
+      "step": 180
+    },
+    {
+      "epoch": 0.006087207253388012,
+      "grad_norm": 7.804600561994388,
+      "learning_rate": 2e-05,
+      "loss": 0.7879319667816163,
+      "step": 190
+    },
+    {
+      "epoch": 0.006407586582513696,
+      "grad_norm": 8.759193053432988,
+      "learning_rate": 2e-05,
+      "loss": 0.8373113632202148,
+      "step": 200
+    },
+    {
+      "epoch": 0.006727965911639381,
+      "grad_norm": 5.7645630188571495,
+      "learning_rate": 2e-05,
+      "loss": 0.8225456237792969,
+      "step": 210
+    },
+    {
+      "epoch": 0.007048345240765066,
+      "grad_norm": 2.975476645382291,
+      "learning_rate": 2e-05,
+      "loss": 0.7898527145385742,
+      "step": 220
+    },
+    {
+      "epoch": 0.007368724569890751,
+      "grad_norm": 3.5660713690110923,
+      "learning_rate": 2e-05,
+      "loss": 0.7848415374755859,
+      "step": 230
+    },
+    {
+      "epoch": 0.007689103899016435,
+      "grad_norm": 3.776542520178389,
+      "learning_rate": 2e-05,
+      "loss": 0.7830493927001954,
+      "step": 240
+    },
+    {
+      "epoch": 0.008009483228142121,
+      "grad_norm": 5.652886706438173,
+      "learning_rate": 2e-05,
+      "loss": 0.7420679092407226,
+      "step": 250
+    },
+    {
+      "epoch": 0.008329862557267805,
+      "grad_norm": 2.7433625896377887,
+      "learning_rate": 2e-05,
+      "loss": 0.8062055587768555,
+      "step": 260
+    },
+    {
+      "epoch": 0.00865024188639349,
+      "grad_norm": 4.993469336501579,
+      "learning_rate": 2e-05,
+      "loss": 0.7772369384765625,
+      "step": 270
+    },
+    {
+      "epoch": 0.008970621215519175,
+      "grad_norm": 5.978072597304008,
+      "learning_rate": 2e-05,
+      "loss": 0.7567183494567871,
+      "step": 280
+    },
+    {
+      "epoch": 0.00929100054464486,
+      "grad_norm": 20.0019916771175,
+      "learning_rate": 2e-05,
+      "loss": 0.7632350921630859,
+      "step": 290
+    },
+    {
+      "epoch": 0.009611379873770544,
+      "grad_norm": 1813.803400221406,
+      "learning_rate": 2e-05,
+      "loss": 0.7820405006408692,
+      "step": 300
+    },
+    {
+      "epoch": 0.00993175920289623,
+      "grad_norm": 15.257478921571794,
+      "learning_rate": 2e-05,
+      "loss": 0.8594757080078125,
+      "step": 310
+    },
+    {
+      "epoch": 0.010252138532021914,
+      "grad_norm": 26.97548621262987,
+      "learning_rate": 2e-05,
+      "loss": 0.8610923767089844,
+      "step": 320
+    },
+    {
+      "epoch": 0.0105725178611476,
+      "grad_norm": 21.892790276243126,
+      "learning_rate": 2e-05,
+      "loss": 0.8686389923095703,
+      "step": 330
+    },
+    {
+      "epoch": 0.010892897190273284,
+      "grad_norm": 16.06260184223094,
+      "learning_rate": 2e-05,
+      "loss": 0.8639827728271484,
+      "step": 340
+    },
+    {
+      "epoch": 0.011213276519398968,
+      "grad_norm": 29.260614803206966,
+      "learning_rate": 2e-05,
+      "loss": 0.7904685497283935,
+      "step": 350
+    },
+    {
+      "epoch": 0.011533655848524654,
+      "grad_norm": 16.44688311222318,
+      "learning_rate": 2e-05,
+      "loss": 0.8587644577026368,
+      "step": 360
+    },
+    {
+      "epoch": 0.011854035177650338,
+      "grad_norm": 70.37237245987374,
+      "learning_rate": 2e-05,
+      "loss": 0.8287432670593262,
+      "step": 370
+    },
+    {
+      "epoch": 0.012174414506776024,
+      "grad_norm": 7.352519778805397,
+      "learning_rate": 2e-05,
+      "loss": 0.8243340492248535,
+      "step": 380
+    },
+    {
+      "epoch": 0.012494793835901708,
+      "grad_norm": 7.57180002114558,
+      "learning_rate": 2e-05,
+      "loss": 0.8488252639770508,
+      "step": 390
+    },
+    {
+      "epoch": 0.012815173165027392,
+      "grad_norm": 4.028963614155682,
+      "learning_rate": 2e-05,
+      "loss": 0.800967788696289,
+      "step": 400
+    },
+    {
+      "epoch": 0.013135552494153078,
+      "grad_norm": 7.604607607710578,
+      "learning_rate": 2e-05,
+      "loss": 0.8435401916503906,
+      "step": 410
+    },
+    {
+      "epoch": 0.013455931823278762,
+      "grad_norm": 5.686576942826079,
+      "learning_rate": 2e-05,
+      "loss": 0.8086394309997559,
+      "step": 420
+    },
+    {
+      "epoch": 0.013776311152404446,
+      "grad_norm": 29.16934098168494,
+      "learning_rate": 2e-05,
+      "loss": 0.8092558860778809,
+      "step": 430
+    },
+    {
+      "epoch": 0.014096690481530132,
+      "grad_norm": 7.838527647928418,
+      "learning_rate": 2e-05,
+      "loss": 0.7873313903808594,
+      "step": 440
+    },
+    {
+      "epoch": 0.014417069810655816,
+      "grad_norm": 4.360151373268049,
+      "learning_rate": 2e-05,
+      "loss": 0.8105262756347656,
+      "step": 450
+    },
+    {
+      "epoch": 0.014737449139781502,
+      "grad_norm": 3.093592672180748,
+      "learning_rate": 2e-05,
+      "loss": 0.7954371452331543,
+      "step": 460
+    },
+    {
+      "epoch": 0.015057828468907186,
+      "grad_norm": 2.8867780320165948,
+      "learning_rate": 2e-05,
+      "loss": 0.7867568969726563,
+      "step": 470
+    },
+    {
+      "epoch": 0.01537820779803287,
+      "grad_norm": 2.5525945956595404,
+      "learning_rate": 2e-05,
+      "loss": 0.7605972290039062,
+      "step": 480
+    },
+    {
+      "epoch": 0.015698587127158554,
+      "grad_norm": 4.350350397516954,
+      "learning_rate": 2e-05,
+      "loss": 0.7946616172790527,
+      "step": 490
+    },
+    {
+      "epoch": 0.016018966456284242,
+      "grad_norm": 2.5601326184252624,
+      "learning_rate": 2e-05,
+      "loss": 0.7657347679138183,
+      "step": 500
+    },
+    {
+      "epoch": 0.016339345785409926,
+      "grad_norm": 3.0411596315285943,
+      "learning_rate": 2e-05,
+      "loss": 0.7469792366027832,
+      "step": 510
+    },
+    {
+      "epoch": 0.01665972511453561,
+      "grad_norm": 1.6677799044175043,
+      "learning_rate": 2e-05,
+      "loss": 0.737004280090332,
+      "step": 520
+    },
+    {
+      "epoch": 0.016980104443661295,
+      "grad_norm": 4.326231014259539,
+      "learning_rate": 2e-05,
+      "loss": 0.7799790859222412,
+      "step": 530
+    },
+    {
+      "epoch": 0.01730048377278698,
+      "grad_norm": 2.732975416289203,
+      "learning_rate": 2e-05,
+      "loss": 0.8164985656738282,
+      "step": 540
+    },
+    {
+      "epoch": 0.017620863101912666,
+      "grad_norm": 2.1962459439491036,
+      "learning_rate": 2e-05,
+      "loss": 0.7743217468261718,
+      "step": 550
+    },
+    {
+      "epoch": 0.01794124243103835,
+      "grad_norm": 4.329667095154606,
+      "learning_rate": 2e-05,
+      "loss": 0.8150390625,
+      "step": 560
+    },
+    {
+      "epoch": 0.018261621760164035,
+      "grad_norm": 1.7505089770107007,
+      "learning_rate": 2e-05,
+      "loss": 0.8161712646484375,
+      "step": 570
+    },
+    {
+      "epoch": 0.01858200108928972,
+      "grad_norm": 1.94080520422082,
+      "learning_rate": 2e-05,
+      "loss": 0.7340856075286866,
+      "step": 580
+    },
+    {
+      "epoch": 0.018902380418415403,
+      "grad_norm": 3.767043072114596,
+      "learning_rate": 2e-05,
+      "loss": 0.7465179443359375,
+      "step": 590
+    },
+    {
+      "epoch": 0.019222759747541087,
+      "grad_norm": 2.134739515696804,
+      "learning_rate": 2e-05,
+      "loss": 0.7511123657226563,
+      "step": 600
+    },
+    {
+      "epoch": 0.019543139076666775,
+      "grad_norm": 2.6553715328436724,
+      "learning_rate": 2e-05,
+      "loss": 0.7474520683288575,
+      "step": 610
+    },
+    {
+      "epoch": 0.01986351840579246,
+      "grad_norm": 23.554705629944248,
+      "learning_rate": 2e-05,
+      "loss": 0.7547467708587646,
+      "step": 620
+    },
+    {
+      "epoch": 0.020183897734918143,
+      "grad_norm": 9.663715178820514,
+      "learning_rate": 2e-05,
+      "loss": 0.7591798782348633,
+      "step": 630
+    },
+    {
+      "epoch": 0.020504277064043827,
+      "grad_norm": 2.857655614152204,
+      "learning_rate": 2e-05,
+      "loss": 0.7372615814208985,
+      "step": 640
+    },
+    {
+      "epoch": 0.02082465639316951,
+      "grad_norm": 19.794443617215876,
+      "learning_rate": 2e-05,
+      "loss": 0.7473020076751709,
+      "step": 650
+    },
+    {
+      "epoch": 0.0211450357222952,
+      "grad_norm": 3.4519708452708513,
+      "learning_rate": 2e-05,
+      "loss": 0.7669941902160644,
+      "step": 660
+    },
+    {
+      "epoch": 0.021465415051420883,
+      "grad_norm": 4.439475537657204,
+      "learning_rate": 2e-05,
+      "loss": 0.7708707809448242,
+      "step": 670
+    },
+    {
+      "epoch": 0.021785794380546567,
+      "grad_norm": 4.288236090232602,
+      "learning_rate": 2e-05,
+      "loss": 0.75105299949646,
+      "step": 680
+    },
+    {
+      "epoch": 0.02210617370967225,
+      "grad_norm": 19.247483032059645,
+      "learning_rate": 2e-05,
+      "loss": 0.7456221580505371,
+      "step": 690
+    },
+    {
+      "epoch": 0.022426553038797935,
+      "grad_norm": 6.277608344095431,
+      "learning_rate": 2e-05,
+      "loss": 0.772921371459961,
+      "step": 700
+    },
+    {
+      "epoch": 0.022746932367923623,
+      "grad_norm": 3.5768536610883284,
+      "learning_rate": 2e-05,
+      "loss": 0.7416125297546386,
+      "step": 710
+    },
+    {
+      "epoch": 0.023067311697049307,
+      "grad_norm": 15.05269132810213,
+      "learning_rate": 2e-05,
+      "loss": 0.7103744506835937,
+      "step": 720
+    },
+    {
+      "epoch": 0.02338769102617499,
+      "grad_norm": 4.468376174246059,
+      "learning_rate": 2e-05,
+      "loss": 0.727785873413086,
+      "step": 730
+    },
+    {
+      "epoch": 0.023708070355300676,
+      "grad_norm": 5.087071299517523,
+      "learning_rate": 2e-05,
+      "loss": 0.7700984477996826,
+      "step": 740
+    },
+    {
+      "epoch": 0.02402844968442636,
+      "grad_norm": 11.288253253171709,
+      "learning_rate": 2e-05,
+      "loss": 0.7581868648529053,
+      "step": 750
+    },
+    {
+      "epoch": 0.024348829013552047,
+      "grad_norm": 3.27468779155936,
+      "learning_rate": 2e-05,
+      "loss": 0.7562522888183594,
+      "step": 760
+    },
+    {
+      "epoch": 0.02466920834267773,
+      "grad_norm": 7.595749246101231,
+      "learning_rate": 2e-05,
+      "loss": 0.7546162605285645,
+      "step": 770
+    },
+    {
+      "epoch": 0.024989587671803416,
+      "grad_norm": 2.1701956042711728,
+      "learning_rate": 2e-05,
+      "loss": 0.7503818511962891,
+      "step": 780
+    },
+    {
+      "epoch": 0.0253099670009291,
+      "grad_norm": 7.551313750149006,
+      "learning_rate": 2e-05,
+      "loss": 0.7194967269897461,
+      "step": 790
+    },
+    {
+      "epoch": 0.025630346330054784,
+      "grad_norm": 3.0982279258014085,
+      "learning_rate": 2e-05,
+      "loss": 0.7183728218078613,
+      "step": 800
+    },
+    {
+      "epoch": 0.025950725659180468,
+      "grad_norm": 2.2319456779746507,
+      "learning_rate": 2e-05,
+      "loss": 0.741051197052002,
+      "step": 810
+    },
+    {
+      "epoch": 0.026271104988306156,
+      "grad_norm": 2.1272967612583904,
+      "learning_rate": 2e-05,
+      "loss": 0.7074191093444824,
+      "step": 820
+    },
+    {
+      "epoch": 0.02659148431743184,
+      "grad_norm": 2.3756745796722125,
+      "learning_rate": 2e-05,
+      "loss": 0.7466435909271241,
+      "step": 830
+    },
+    {
+      "epoch": 0.026911863646557524,
+      "grad_norm": 3.9008908753860925,
+      "learning_rate": 2e-05,
+      "loss": 0.7262076377868653,
+      "step": 840
+    },
+    {
+      "epoch": 0.027232242975683208,
+      "grad_norm": 2.0561831766929806,
+      "learning_rate": 2e-05,
+      "loss": 0.7282081604003906,
+      "step": 850
+    },
+    {
+      "epoch": 0.027552622304808892,
+      "grad_norm": 5.818198655017998,
+      "learning_rate": 2e-05,
+      "loss": 0.7566670894622802,
+      "step": 860
+    },
+    {
+      "epoch": 0.02787300163393458,
+      "grad_norm": 2.1494490230827337,
+      "learning_rate": 2e-05,
+      "loss": 0.7288908958435059,
+      "step": 870
+    },
+    {
+      "epoch": 0.028193380963060264,
+      "grad_norm": 1.409242496714709,
+      "learning_rate": 2e-05,
+      "loss": 0.7215320587158203,
+      "step": 880
+    },
+    {
+      "epoch": 0.028513760292185948,
+      "grad_norm": 7.0920590147047164,
+      "learning_rate": 2e-05,
+      "loss": 0.7246700286865234,
+      "step": 890
+    },
+    {
+      "epoch": 0.028834139621311632,
+      "grad_norm": 1.5684095793164903,
+      "learning_rate": 2e-05,
+      "loss": 0.7144109725952148,
+      "step": 900
+    },
+    {
+      "epoch": 0.029154518950437316,
+      "grad_norm": 1.3316395400018082,
+      "learning_rate": 2e-05,
+      "loss": 0.7477982997894287,
+      "step": 910
+    },
+    {
+      "epoch": 0.029474898279563004,
+      "grad_norm": 1.3602729023468434,
+      "learning_rate": 2e-05,
+      "loss": 0.7173037528991699,
+      "step": 920
+    },
+    {
+      "epoch": 0.029795277608688688,
+      "grad_norm": 49.53643391306756,
+      "learning_rate": 2e-05,
+      "loss": 0.7678704738616944,
+      "step": 930
+    },
+    {
+      "epoch": 0.030115656937814372,
+      "grad_norm": 4.340394359223178,
+      "learning_rate": 2e-05,
+      "loss": 0.7031218528747558,
+      "step": 940
+    },
+    {
+      "epoch": 0.030436036266940057,
+      "grad_norm": 2.1203555620583736,
+      "learning_rate": 2e-05,
+      "loss": 0.7084536552429199,
+      "step": 950
+    },
+    {
+      "epoch": 0.03075641559606574,
+      "grad_norm": 1.3371075573755178,
+      "learning_rate": 2e-05,
+      "loss": 0.7464202880859375,
+      "step": 960
+    },
+    {
+      "epoch": 0.03107679492519143,
+      "grad_norm": 1.6473009838115498,
+      "learning_rate": 2e-05,
+      "loss": 0.7116725444793701,
+      "step": 970
+    },
+    {
+      "epoch": 0.03139717425431711,
+      "grad_norm": 3.385121877804678,
+      "learning_rate": 2e-05,
+      "loss": 0.738076114654541,
+      "step": 980
+    },
+    {
+      "epoch": 0.0317175535834428,
+      "grad_norm": 1.7147230699245448,
+      "learning_rate": 2e-05,
+      "loss": 0.7158563137054443,
+      "step": 990
+    },
+    {
+      "epoch": 0.032037932912568484,
+      "grad_norm": 3.3445220753393468,
+      "learning_rate": 2e-05,
+      "loss": 0.8030566215515137,
+      "step": 1000
+    },
+    {
+      "epoch": 0.032358312241694165,
+      "grad_norm": 2.077502109986157,
+      "learning_rate": 2e-05,
+      "loss": 0.7472703456878662,
+      "step": 1010
+    },
+    {
+      "epoch": 0.03267869157081985,
+      "grad_norm": 17.329143506946714,
+      "learning_rate": 2e-05,
+      "loss": 0.7366230964660645,
+      "step": 1020
+    },
+    {
+      "epoch": 0.03299907089994553,
+      "grad_norm": 2.6268238025254735,
+      "learning_rate": 2e-05,
+      "loss": 0.7404189109802246,
+      "step": 1030
+    },
+    {
+      "epoch": 0.03331945022907122,
+      "grad_norm": 1.977866371730136,
+      "learning_rate": 2e-05,
+      "loss": 0.7236545085906982,
+      "step": 1040
+    },
+    {
+      "epoch": 0.03363982955819691,
+      "grad_norm": 1.498659413315871,
+      "learning_rate": 2e-05,
+      "loss": 0.7335420608520508,
+      "step": 1050
+    },
+    {
+      "epoch": 0.03396020888732259,
+      "grad_norm": 1.7062214656423396,
+      "learning_rate": 2e-05,
+      "loss": 0.7051788330078125,
+      "step": 1060
+    },
+    {
+      "epoch": 0.03428058821644828,
+      "grad_norm": 1.3397922877558162,
+      "learning_rate": 2e-05,
+      "loss": 0.7626142024993896,
+      "step": 1070
+    },
+    {
+      "epoch": 0.03460096754557396,
+      "grad_norm": 3.93050291862445,
+      "learning_rate": 2e-05,
+      "loss": 0.7514806270599366,
+      "step": 1080
+    },
+    {
+      "epoch": 0.034921346874699645,
+      "grad_norm": 3.398990137450838,
+      "learning_rate": 2e-05,
+      "loss": 0.7132882595062255,
+      "step": 1090
+    },
+    {
+      "epoch": 0.03524172620382533,
+      "grad_norm": 1.4600798546048035,
+      "learning_rate": 2e-05,
+      "loss": 0.716385793685913,
+      "step": 1100
+    },
+    {
+      "epoch": 0.03556210553295101,
+      "grad_norm": 1.3160301135986112,
+      "learning_rate": 2e-05,
+      "loss": 0.7191272735595703,
+      "step": 1110
+    },
+    {
+      "epoch": 0.0358824848620767,
+      "grad_norm": 1.2481731524679547,
+      "learning_rate": 2e-05,
+      "loss": 0.7158429145812988,
+      "step": 1120
+    },
+    {
+      "epoch": 0.03620286419120238,
+      "grad_norm": 2.8179317711592504,
+      "learning_rate": 2e-05,
+      "loss": 0.7038905143737793,
+      "step": 1130
+    },
+    {
+      "epoch": 0.03652324352032807,
+      "grad_norm": 1.8284267025580432,
+      "learning_rate": 2e-05,
+      "loss": 0.7065417289733886,
+      "step": 1140
+    },
+    {
+      "epoch": 0.03684362284945375,
+      "grad_norm": 2.212892645914091,
+      "learning_rate": 2e-05,
+      "loss": 0.7265023708343505,
+      "step": 1150
+    },
+    {
+      "epoch": 0.03716400217857944,
+      "grad_norm": 2.2597923858522404,
+      "learning_rate": 2e-05,
+      "loss": 0.7199366569519043,
+      "step": 1160
+    },
+    {
+      "epoch": 0.037484381507705125,
+      "grad_norm": 2.0160695548330936,
+      "learning_rate": 2e-05,
+      "loss": 0.7658578395843506,
+      "step": 1170
+    },
+    {
+      "epoch": 0.037804760836830806,
+      "grad_norm": 2.5572488896871843,
+      "learning_rate": 2e-05,
+      "loss": 0.7220034599304199,
+      "step": 1180
+    },
+    {
+      "epoch": 0.03812514016595649,
+      "grad_norm": 2.459122478774182,
+      "learning_rate": 2e-05,
+      "loss": 0.6983201026916503,
+      "step": 1190
+    },
+    {
+      "epoch": 0.038445519495082174,
+      "grad_norm": 5.318818273924634,
+      "learning_rate": 2e-05,
+      "loss": 0.7152614116668701,
+      "step": 1200
+    },
+    {
+      "epoch": 0.03876589882420786,
+      "grad_norm": 4.348661337562832,
+      "learning_rate": 2e-05,
+      "loss": 0.7531236171722412,
+      "step": 1210
+    },
+    {
+      "epoch": 0.03908627815333355,
+      "grad_norm": 3.5073096880649666,
+      "learning_rate": 2e-05,
+      "loss": 0.7070003509521484,
+      "step": 1220
+    },
+    {
+      "epoch": 0.03940665748245923,
+      "grad_norm": 6.250765626260255,
+      "learning_rate": 2e-05,
+      "loss": 0.7422648429870605,
+      "step": 1230
+    },
+    {
+      "epoch": 0.03972703681158492,
+      "grad_norm": 2.1134272476858693,
+      "learning_rate": 2e-05,
+      "loss": 0.7167387962341308,
+      "step": 1240
+    },
+    {
+      "epoch": 0.0400474161407106,
+      "grad_norm": 2.728613864385018,
+      "learning_rate": 2e-05,
+      "loss": 0.7613626480102539,
+      "step": 1250
+    },
+    {
+      "epoch": 0.040367795469836286,
+      "grad_norm": 2.8289243765180094,
+      "learning_rate": 2e-05,
+      "loss": 0.7406154155731202,
+      "step": 1260
+    },
+    {
+      "epoch": 0.040688174798961974,
+      "grad_norm": 1.460978022101602,
+      "learning_rate": 2e-05,
+      "loss": 0.7171460628509522,
+      "step": 1270
+    },
+    {
+      "epoch": 0.041008554128087654,
+      "grad_norm": 11.187948658326427,
+      "learning_rate": 2e-05,
+      "loss": 0.7119218349456787,
+      "step": 1280
+    },
+    {
+      "epoch": 0.04132893345721334,
+      "grad_norm": 1.5366780655953838,
+      "learning_rate": 2e-05,
+      "loss": 0.7250003337860107,
+      "step": 1290
+    },
+    {
+      "epoch": 0.04164931278633902,
+      "grad_norm": 2.6451401695263392,
+      "learning_rate": 2e-05,
+      "loss": 0.7535340785980225,
+      "step": 1300
+    },
+    {
+      "epoch": 0.04196969211546471,
+      "grad_norm": 5.247231775991024,
+      "learning_rate": 2e-05,
+      "loss": 0.7120182991027832,
+      "step": 1310
+    },
+    {
+      "epoch": 0.0422900714445904,
+      "grad_norm": 1.265414487137248,
+      "learning_rate": 2e-05,
+      "loss": 0.7114714622497559,
+      "step": 1320
+    },
+    {
+      "epoch": 0.04261045077371608,
+      "grad_norm": 32.52328999465322,
+      "learning_rate": 2e-05,
+      "loss": 0.6845267295837403,
+      "step": 1330
+    },
+    {
+      "epoch": 0.042930830102841766,
+      "grad_norm": 6.1768247607509394,
+      "learning_rate": 2e-05,
+      "loss": 0.702857780456543,
+      "step": 1340
+    },
+    {
+      "epoch": 0.04325120943196745,
+      "grad_norm": 7.1933410968196725,
+      "learning_rate": 2e-05,
+      "loss": 0.72010817527771,
+      "step": 1350
+    },
+    {
+      "epoch": 0.043571588761093134,
+      "grad_norm": 1.1755562476246728,
+      "learning_rate": 2e-05,
+      "loss": 0.6978877067565918,
+      "step": 1360
+    },
+    {
+      "epoch": 0.04389196809021882,
+      "grad_norm": 2.8540830213502844,
+      "learning_rate": 2e-05,
+      "loss": 0.7204657554626465,
+      "step": 1370
+    },
+    {
+      "epoch": 0.0442123474193445,
+      "grad_norm": 1.1871852286932287,
+      "learning_rate": 2e-05,
+      "loss": 0.7182430267333985,
+      "step": 1380
+    },
+    {
+      "epoch": 0.04453272674847019,
+      "grad_norm": 1.0566209393736636,
+      "learning_rate": 2e-05,
+      "loss": 0.7123696804046631,
+      "step": 1390
+    },
+    {
+      "epoch": 0.04485310607759587,
+      "grad_norm": 4.237024261846737,
+      "learning_rate": 2e-05,
+      "loss": 0.6963154792785644,
+      "step": 1400
+    },
+    {
+      "epoch": 0.04517348540672156,
+      "grad_norm": 2.008976090493564,
+      "learning_rate": 2e-05,
+      "loss": 0.6849660396575927,
+      "step": 1410
+    },
+    {
+      "epoch": 0.045493864735847246,
+      "grad_norm": 1.3493113476458192,
+      "learning_rate": 2e-05,
+      "loss": 0.7087863922119141,
+      "step": 1420
+    },
+    {
+      "epoch": 0.04581424406497293,
+      "grad_norm": 1.0453265260241618,
+      "learning_rate": 2e-05,
+      "loss": 0.6820076942443848,
+      "step": 1430
+    },
+    {
+      "epoch": 0.046134623394098614,
+      "grad_norm": 1.367769743280532,
+      "learning_rate": 2e-05,
+      "loss": 0.7149313926696778,
+      "step": 1440
+    },
+    {
+      "epoch": 0.046455002723224295,
+      "grad_norm": 1.5424097504581287,
+      "learning_rate": 2e-05,
+      "loss": 0.7253753185272217,
+      "step": 1450
+    },
+    {
+      "epoch": 0.04677538205234998,
+      "grad_norm": 2.3372729347553025,
+      "learning_rate": 2e-05,
+      "loss": 0.7671535015106201,
+      "step": 1460
+    },
+    {
+      "epoch": 0.04709576138147567,
+      "grad_norm": 1.8742828809078558,
+      "learning_rate": 2e-05,
+      "loss": 0.7182416915893555,
+      "step": 1470
+    },
+    {
+      "epoch": 0.04741614071060135,
+      "grad_norm": 6.87172572472952,
+      "learning_rate": 2e-05,
+      "loss": 0.7154425621032715,
+      "step": 1480
+    },
+    {
+      "epoch": 0.04773652003972704,
+      "grad_norm": 2.4729840403642545,
+      "learning_rate": 2e-05,
+      "loss": 0.7049578666687012,
+      "step": 1490
+    },
+    {
+      "epoch": 0.04805689936885272,
+      "grad_norm": 1.3804743410158802,
+      "learning_rate": 2e-05,
+      "loss": 0.7238261222839355,
+      "step": 1500
+    },
+    {
+      "epoch": 0.04837727869797841,
+      "grad_norm": 1.2556616325656202,
+      "learning_rate": 2e-05,
+      "loss": 0.7582721710205078,
+      "step": 1510
+    },
+    {
+      "epoch": 0.048697658027104095,
+      "grad_norm": 1.4068662819395166,
+      "learning_rate": 2e-05,
+      "loss": 0.6943454742431641,
+      "step": 1520
+    },
+    {
+      "epoch": 0.049018037356229775,
+      "grad_norm": 1.5191574570193485,
+      "learning_rate": 2e-05,
+      "loss": 0.7089879512786865,
+      "step": 1530
+    },
+    {
+      "epoch": 0.04933841668535546,
+      "grad_norm": 1.1100647663489502,
+      "learning_rate": 2e-05,
+      "loss": 0.6814855098724365,
+      "step": 1540
+    },
+    {
+      "epoch": 0.049658796014481144,
+      "grad_norm": 1.2368089337986163,
+      "learning_rate": 2e-05,
+      "loss": 0.7276053428649902,
+      "step": 1550
+    },
+    {
+      "epoch": 0.04997917534360683,
+      "grad_norm": 1.596887157213148,
+      "learning_rate": 2e-05,
+      "loss": 0.7387615203857422,
+      "step": 1560
+    },
+    {
+      "epoch": 0.05029955467273251,
+      "grad_norm": 1.3857783026385584,
+      "learning_rate": 2e-05,
+      "loss": 0.6814603805541992,
+      "step": 1570
+    },
+    {
+      "epoch": 0.0506199340018582,
+      "grad_norm": 1.5717792680161595,
+      "learning_rate": 2e-05,
+      "loss": 0.6634132385253906,
+      "step": 1580
+    },
+    {
+      "epoch": 0.05094031333098389,
+      "grad_norm": 5.076622025769138,
+      "learning_rate": 2e-05,
+      "loss": 0.7170121669769287,
+      "step": 1590
+    },
+    {
+      "epoch": 0.05126069266010957,
+      "grad_norm": 3.075120261257635,
+      "learning_rate": 2e-05,
+      "loss": 0.6901824951171875,
+      "step": 1600
+    },
+    {
+      "epoch": 0.051581071989235255,
+      "grad_norm": 3.1443073556392838,
+      "learning_rate": 2e-05,
+      "loss": 0.7126097679138184,
+      "step": 1610
+    },
+    {
+      "epoch": 0.051901451318360936,
+      "grad_norm": 1.3279065854265553,
+      "learning_rate": 2e-05,
+      "loss": 0.696572208404541,
+      "step": 1620
+    },
+    {
+      "epoch": 0.052221830647486624,
+      "grad_norm": 2.2922495579972213,
+      "learning_rate": 2e-05,
+      "loss": 0.7236587047576905,
+      "step": 1630
+    },
+    {
+      "epoch": 0.05254220997661231,
+      "grad_norm": 1.853599288786833,
+      "learning_rate": 2e-05,
+      "loss": 0.6959040641784668,
+      "step": 1640
+    },
+    {
+      "epoch": 0.05286258930573799,
+      "grad_norm": 1.6356938422959257,
+      "learning_rate": 2e-05,
+      "loss": 0.6883045196533203,
+      "step": 1650
+    },
+    {
+      "epoch": 0.05318296863486368,
+      "grad_norm": 1.4258158618988481,
+      "learning_rate": 2e-05,
+      "loss": 0.7095949172973632,
+      "step": 1660
+    },
+    {
+      "epoch": 0.05350334796398936,
+      "grad_norm": 1.3818626048983265,
+      "learning_rate": 2e-05,
+      "loss": 0.7279641628265381,
+      "step": 1670
+    },
+    {
+      "epoch": 0.05382372729311505,
+      "grad_norm": 1.3148725391903133,
+      "learning_rate": 2e-05,
+      "loss": 0.7071706295013428,
+      "step": 1680
+    },
+    {
+      "epoch": 0.054144106622240736,
+      "grad_norm": 1.3205843385485543,
+      "learning_rate": 2e-05,
+      "loss": 0.7102875709533691,
+      "step": 1690
+    },
+    {
+      "epoch": 0.054464485951366416,
+      "grad_norm": 1.0245118537391376,
+      "learning_rate": 2e-05,
+      "loss": 0.7355536460876465,
+      "step": 1700
+    },
+    {
+      "epoch": 0.054784865280492104,
+      "grad_norm": 0.9830319826014532,
+      "learning_rate": 2e-05,
+      "loss": 0.711072301864624,
+      "step": 1710
+    },
+    {
+      "epoch": 0.055105244609617784,
+      "grad_norm": 2.2427073223135,
+      "learning_rate": 2e-05,
+      "loss": 0.7257272720336914,
+      "step": 1720
+    },
+    {
+      "epoch": 0.05542562393874347,
+      "grad_norm": 1.7432652030049554,
+      "learning_rate": 2e-05,
+      "loss": 0.7266391754150391,
+      "step": 1730
+    },
+    {
+      "epoch": 0.05574600326786916,
+      "grad_norm": 7.3855085380936965,
+      "learning_rate": 2e-05,
+      "loss": 0.7229394435882568,
+      "step": 1740
+    },
+    {
+      "epoch": 0.05606638259699484,
+      "grad_norm": 2.1417039515536542,
+      "learning_rate": 2e-05,
+      "loss": 0.6833599090576172,
+      "step": 1750
+    },
+    {
+      "epoch": 0.05638676192612053,
+      "grad_norm": 2.4468519837946165,
+      "learning_rate": 2e-05,
+      "loss": 0.7663132667541503,
+      "step": 1760
+    },
+    {
+      "epoch": 0.05670714125524621,
+      "grad_norm": 3.0137174378050293,
+      "learning_rate": 2e-05,
+      "loss": 0.7224421977996827,
+      "step": 1770
+    },
+    {
+      "epoch": 0.057027520584371896,
+      "grad_norm": 2.2523438896707706,
+      "learning_rate": 2e-05,
+      "loss": 0.7207423210144043,
+      "step": 1780
+    },
+    {
+      "epoch": 0.057347899913497584,
+      "grad_norm": 2.919705865322085,
+      "learning_rate": 2e-05,
+      "loss": 0.6854898452758789,
+      "step": 1790
+    },
+    {
+      "epoch": 0.057668279242623265,
+      "grad_norm": 3.984396788478948,
+      "learning_rate": 2e-05,
+      "loss": 0.7234033107757568,
+      "step": 1800
+    },
+    {
+      "epoch": 0.05798865857174895,
+      "grad_norm": 6.67851094941151,
+      "learning_rate": 2e-05,
+      "loss": 0.7022351264953614,
+      "step": 1810
+    },
+    {
+      "epoch": 0.05830903790087463,
+      "grad_norm": 1.6866126836675677,
+      "learning_rate": 2e-05,
+      "loss": 0.7415154457092286,
+      "step": 1820
+    },
+    {
+      "epoch": 0.05862941723000032,
+      "grad_norm": 3.218820869797509,
+      "learning_rate": 2e-05,
+      "loss": 0.7005694389343262,
+      "step": 1830
+    },
+    {
+      "epoch": 0.05894979655912601,
+      "grad_norm": 3.0737908739781106,
+      "learning_rate": 2e-05,
+      "loss": 0.6981851100921631,
+      "step": 1840
+    },
+    {
+      "epoch": 0.05927017588825169,
+      "grad_norm": 2.2327032848527795,
+      "learning_rate": 2e-05,
+      "loss": 0.7604454040527344,
+      "step": 1850
+    },
+    {
+      "epoch": 0.059590555217377376,
+      "grad_norm": 2.8761311961580973,
+      "learning_rate": 2e-05,
+      "loss": 0.6888795852661133,
+      "step": 1860
+    },
+    {
+      "epoch": 0.05991093454650306,
+      "grad_norm": 2.2317201691382533,
+      "learning_rate": 2e-05,
+      "loss": 0.708838415145874,
+      "step": 1870
+    },
+    {
+      "epoch": 0.060231313875628745,
+      "grad_norm": 2.0355186191234806,
+      "learning_rate": 2e-05,
+      "loss": 0.7040927886962891,
+      "step": 1880
+    },
+    {
+      "epoch": 0.06055169320475443,
+      "grad_norm": 2.014047361446366,
+      "learning_rate": 2e-05,
+      "loss": 0.7197917461395263,
+      "step": 1890
+    },
+    {
+      "epoch": 0.06087207253388011,
+      "grad_norm": 1.977104681851146,
+      "learning_rate": 2e-05,
+      "loss": 0.6776611328125,
+      "step": 1900
+    },
+    {
+      "epoch": 0.0611924518630058,
+      "grad_norm": 13.901447338457938,
+      "learning_rate": 2e-05,
+      "loss": 0.6899932861328125,
+      "step": 1910
+    },
+    {
+      "epoch": 0.06151283119213148,
+      "grad_norm": 2.1123993495266435,
+      "learning_rate": 2e-05,
+      "loss": 0.6683700084686279,
+      "step": 1920
+    },
+    {
+      "epoch": 0.06183321052125717,
+      "grad_norm": 2.3306798040564676,
+      "learning_rate": 2e-05,
+      "loss": 0.691620922088623,
+      "step": 1930
+    },
+    {
+      "epoch": 0.06215358985038286,
+      "grad_norm": 13.395393364218384,
+      "learning_rate": 2e-05,
+      "loss": 0.7234976768493653,
+      "step": 1940
+    },
+    {
+      "epoch": 0.06247396917950854,
+      "grad_norm": 7.087449907589583,
+      "learning_rate": 2e-05,
+      "loss": 0.7050728797912598,
+      "step": 1950
+    },
+    {
+      "epoch": 0.06279434850863422,
+      "grad_norm": 2.912496650205425,
+      "learning_rate": 2e-05,
+      "loss": 0.6966752529144287,
+      "step": 1960
+    },
+    {
+      "epoch": 0.06311472783775991,
+      "grad_norm": 8.623415554665941,
+      "learning_rate": 2e-05,
+      "loss": 0.67595534324646,
+      "step": 1970
+    },
+    {
+      "epoch": 0.0634351071668856,
+      "grad_norm": 3.239938193275378,
+      "learning_rate": 2e-05,
+      "loss": 0.6972695350646972,
+      "step": 1980
+    },
+    {
+      "epoch": 0.06375548649601127,
+      "grad_norm": 4.165992500546785,
+      "learning_rate": 2e-05,
+      "loss": 0.7464128017425538,
+      "step": 1990
+    },
+    {
+      "epoch": 0.06407586582513697,
+      "grad_norm": 3.308146075178026,
+      "learning_rate": 2e-05,
+      "loss": 0.7509006023406982,
+      "step": 2000
+    },
+    {
+      "epoch": 0.06439624515426265,
+      "grad_norm": 3.4969526861271802,
+      "learning_rate": 2e-05,
+      "loss": 0.7300586700439453,
+      "step": 2010
+    },
+    {
+      "epoch": 0.06471662448338833,
+      "grad_norm": 4.743378784447764,
+      "learning_rate": 2e-05,
+      "loss": 0.6886860847473144,
+      "step": 2020
+    },
+    {
+      "epoch": 0.06503700381251401,
+      "grad_norm": 1.9438372463888733,
+      "learning_rate": 2e-05,
+      "loss": 0.705808448791504,
+      "step": 2030
+    },
+    {
+      "epoch": 0.0653573831416397,
+      "grad_norm": 11.001500428867326,
+      "learning_rate": 2e-05,
+      "loss": 0.698797607421875,
+      "step": 2040
+    },
+    {
+      "epoch": 0.06567776247076539,
+      "grad_norm": 1.7716709742858978,
+      "learning_rate": 2e-05,
+      "loss": 0.7007397651672364,
+      "step": 2050
+    },
+    {
+      "epoch": 0.06599814179989107,
+      "grad_norm": 2.0422284531425343,
+      "learning_rate": 2e-05,
+      "loss": 0.7145570755004883,
+      "step": 2060
+    },
+    {
+      "epoch": 0.06631852112901676,
+      "grad_norm": 5.491456733860842,
+      "learning_rate": 2e-05,
+      "loss": 0.669373893737793,
+      "step": 2070
+    },
+    {
+      "epoch": 0.06663890045814244,
+      "grad_norm": 17.321502365857935,
+      "learning_rate": 2e-05,
+      "loss": 0.7140804290771484,
+      "step": 2080
+    },
+    {
+      "epoch": 0.06695927978726812,
+      "grad_norm": 1.6192456244078586,
+      "learning_rate": 2e-05,
+      "loss": 0.7083226680755615,
+      "step": 2090
+    },
+    {
+      "epoch": 0.06727965911639382,
+      "grad_norm": 1.70128491897758,
+      "learning_rate": 2e-05,
+      "loss": 0.696595573425293,
+      "step": 2100
+    },
+    {
+      "epoch": 0.0676000384455195,
+      "grad_norm": 1.6411121971094411,
+      "learning_rate": 2e-05,
+      "loss": 0.6904148101806641,
+      "step": 2110
+    },
+    {
+      "epoch": 0.06792041777464518,
+      "grad_norm": 1.2162810019964114,
+      "learning_rate": 2e-05,
+      "loss": 0.7027925014495849,
+      "step": 2120
+    },
+    {
+      "epoch": 0.06824079710377086,
+      "grad_norm": 4.319041953230064,
+      "learning_rate": 2e-05,
+      "loss": 0.8154801368713379,
+      "step": 2130
+    },
+    {
+      "epoch": 0.06856117643289655,
+      "grad_norm": 6.8100351134727335,
+      "learning_rate": 2e-05,
+      "loss": 0.7859745025634766,
+      "step": 2140
+    },
+    {
+      "epoch": 0.06888155576202223,
+      "grad_norm": 41.32429680397114,
+      "learning_rate": 2e-05,
+      "loss": 0.8241888046264648,
+      "step": 2150
+    },
+    {
+      "epoch": 0.06920193509114791,
+      "grad_norm": 7.447130807285079,
+      "learning_rate": 2e-05,
+      "loss": 0.738440465927124,
+      "step": 2160
+    },
+    {
+      "epoch": 0.06952231442027361,
+      "grad_norm": 2.4311779687133095,
+      "learning_rate": 2e-05,
+      "loss": 0.7003934860229493,
+      "step": 2170
+    },
+    {
+      "epoch": 0.06984269374939929,
+      "grad_norm": 1.888452194069137,
+      "learning_rate": 2e-05,
+      "loss": 0.6981853485107422,
+      "step": 2180
+    },
+    {
+      "epoch": 0.07016307307852497,
+      "grad_norm": 1.3884978841961744,
+      "learning_rate": 2e-05,
+      "loss": 0.6570146560668946,
+      "step": 2190
+    },
+    {
+      "epoch": 0.07048345240765067,
+      "grad_norm": 1.90260673711977,
+      "learning_rate": 2e-05,
+      "loss": 0.6660157680511475,
+      "step": 2200
+    },
+    {
+      "epoch": 0.07080383173677635,
+      "grad_norm": 2.313870497589669,
+      "learning_rate": 2e-05,
+      "loss": 0.7752554893493653,
+      "step": 2210
+    },
+    {
+      "epoch": 0.07112421106590203,
+      "grad_norm": 17.868103729169547,
+      "learning_rate": 2e-05,
+      "loss": 0.745037841796875,
+      "step": 2220
+    },
+    {
+      "epoch": 0.07144459039502771,
+      "grad_norm": 2.3214674515548737,
+      "learning_rate": 2e-05,
+      "loss": 0.7210993766784668,
+      "step": 2230
+    },
+    {
+      "epoch": 0.0717649697241534,
+      "grad_norm": 3.1963888308689614,
+      "learning_rate": 2e-05,
+      "loss": 0.6749308586120606,
+      "step": 2240
+    },
+    {
+      "epoch": 0.07208534905327908,
+      "grad_norm": 3.3505276313125614,
+      "learning_rate": 2e-05,
+      "loss": 0.6825098991394043,
+      "step": 2250
+    },
+    {
+      "epoch": 0.07240572838240476,
+      "grad_norm": 1.7104131993752278,
+      "learning_rate": 2e-05,
+      "loss": 0.7080960750579834,
+      "step": 2260
+    },
+    {
+      "epoch": 0.07272610771153046,
+      "grad_norm": 1.4805583634540915,
+      "learning_rate": 2e-05,
+      "loss": 0.6782462120056152,
+      "step": 2270
+    },
+    {
+      "epoch": 0.07304648704065614,
+      "grad_norm": 1.3197929147416767,
+      "learning_rate": 2e-05,
+      "loss": 0.6485519409179688,
+      "step": 2280
+    },
+    {
+      "epoch": 0.07336686636978182,
+      "grad_norm": 1.221515239216845,
+      "learning_rate": 2e-05,
+      "loss": 0.6853775024414063,
+      "step": 2290
+    },
+    {
+      "epoch": 0.0736872456989075,
+      "grad_norm": 30.204078348726764,
+      "learning_rate": 2e-05,
+      "loss": 0.6834779262542725,
+      "step": 2300
+    },
+    {
+      "epoch": 0.0740076250280332,
+      "grad_norm": 1.4517575091324884,
+      "learning_rate": 2e-05,
+      "loss": 0.6975203514099121,
+      "step": 2310
+    },
+    {
+      "epoch": 0.07432800435715887,
+      "grad_norm": 1.3422521684581996,
+      "learning_rate": 2e-05,
+      "loss": 0.759161901473999,
+      "step": 2320
+    },
+    {
+      "epoch": 0.07464838368628456,
+      "grad_norm": 2.3245979856998127,
+      "learning_rate": 2e-05,
+      "loss": 0.7400575637817383,
+      "step": 2330
+    },
+    {
+      "epoch": 0.07496876301541025,
+      "grad_norm": 2.4522055207504607,
+      "learning_rate": 2e-05,
+      "loss": 0.6582594871520996,
+      "step": 2340
+    },
+    {
+      "epoch": 0.07528914234453593,
+      "grad_norm": 3.9583378563876668,
+      "learning_rate": 2e-05,
+      "loss": 0.7118147850036621,
+      "step": 2350
+    },
+    {
+      "epoch": 0.07560952167366161,
+      "grad_norm": 1.3241532087375179,
+      "learning_rate": 2e-05,
+      "loss": 0.750539493560791,
+      "step": 2360
+    },
+    {
+      "epoch": 0.0759299010027873,
+      "grad_norm": 1.6871548656996915,
+      "learning_rate": 2e-05,
+      "loss": 0.7293619155883789,
+      "step": 2370
+    },
+    {
+      "epoch": 0.07625028033191299,
+      "grad_norm": 1.9502293710817282,
+      "learning_rate": 2e-05,
+      "loss": 0.7126898765563965,
+      "step": 2380
+    },
+    {
+      "epoch": 0.07657065966103867,
+      "grad_norm": 2.5257667368767187,
+      "learning_rate": 2e-05,
+      "loss": 0.6581045150756836,
+      "step": 2390
+    },
+    {
+      "epoch": 0.07689103899016435,
+      "grad_norm": 2.4638908736573693,
+      "learning_rate": 2e-05,
+      "loss": 0.7101143836975098,
+      "step": 2400
+    },
+    {
+      "epoch": 0.07721141831929004,
+      "grad_norm": 2.047332045907796,
+      "learning_rate": 2e-05,
+      "loss": 0.6816272735595703,
+      "step": 2410
+    },
+    {
+      "epoch": 0.07753179764841572,
+      "grad_norm": 1.0750576692726481,
+      "learning_rate": 2e-05,
+      "loss": 0.7101527214050293,
+      "step": 2420
+    },
+    {
+      "epoch": 0.0778521769775414,
+      "grad_norm": 1.3667366477929122,
+      "learning_rate": 2e-05,
+      "loss": 0.680915641784668,
+      "step": 2430
+    },
+    {
+      "epoch": 0.0781725563066671,
+      "grad_norm": 1.9499465487529435,
+      "learning_rate": 2e-05,
+      "loss": 0.67269287109375,
+      "step": 2440
+    },
+    {
+      "epoch": 0.07849293563579278,
+      "grad_norm": 9.520772501016044,
+      "learning_rate": 2e-05,
+      "loss": 0.7161269187927246,
+      "step": 2450
+    },
+    {
+      "epoch": 0.07881331496491846,
+      "grad_norm": 1.2730224537721924,
+      "learning_rate": 2e-05,
+      "loss": 0.6880199432373046,
+      "step": 2460
+    },
+    {
+      "epoch": 0.07913369429404415,
+      "grad_norm": 4.122823755612335,
+      "learning_rate": 2e-05,
+      "loss": 0.7019095420837402,
+      "step": 2470
+    },
+    {
+      "epoch": 0.07945407362316984,
+      "grad_norm": 1.2350902700761086,
+      "learning_rate": 2e-05,
+      "loss": 0.7402851104736328,
+      "step": 2480
+    },
+    {
+      "epoch": 0.07977445295229552,
+      "grad_norm": 1.6557513951573315,
+      "learning_rate": 2e-05,
+      "loss": 0.6856629371643066,
+      "step": 2490
+    },
+    {
+      "epoch": 0.0800948322814212,
+      "grad_norm": 2.76250457287939,
+      "learning_rate": 2e-05,
+      "loss": 0.7310905456542969,
+      "step": 2500
+    },
+    {
+      "epoch": 0.08041521161054689,
+      "grad_norm": 10.355727739658676,
+      "learning_rate": 2e-05,
+      "loss": 0.769098424911499,
+      "step": 2510
+    },
+    {
+      "epoch": 0.08073559093967257,
+      "grad_norm": 1.505013422609845,
+      "learning_rate": 2e-05,
+      "loss": 0.699889850616455,
+      "step": 2520
+    },
+    {
+      "epoch": 0.08105597026879825,
+      "grad_norm": 1.1570475749103184,
+      "learning_rate": 2e-05,
+      "loss": 0.6950805187225342,
+      "step": 2530
+    },
+    {
+      "epoch": 0.08137634959792395,
+      "grad_norm": 2.1024924495057995,
+      "learning_rate": 2e-05,
+      "loss": 0.701189374923706,
+      "step": 2540
+    },
+    {
+      "epoch": 0.08169672892704963,
+      "grad_norm": 1.7444425666233216,
+      "learning_rate": 2e-05,
+      "loss": 0.7059882164001465,
+      "step": 2550
+    },
+    {
+      "epoch": 0.08201710825617531,
+      "grad_norm": 1.0322806923298704,
+      "learning_rate": 2e-05,
+      "loss": 0.8018281936645508,
+      "step": 2560
+    },
+    {
+      "epoch": 0.082337487585301,
+      "grad_norm": 2.6526418970250307,
+      "learning_rate": 2e-05,
+      "loss": 0.6831948757171631,
+      "step": 2570
+    },
+    {
+      "epoch": 0.08265786691442668,
+      "grad_norm": 2.8515124039952076,
+      "learning_rate": 2e-05,
+      "loss": 0.7252332687377929,
+      "step": 2580
+    },
+    {
+      "epoch": 0.08297824624355236,
+      "grad_norm": 1.5075998819752536,
+      "learning_rate": 2e-05,
+      "loss": 0.6708373069763184,
+      "step": 2590
+    },
+    {
+      "epoch": 0.08329862557267805,
+      "grad_norm": 2.4918434345399083,
+      "learning_rate": 2e-05,
+      "loss": 0.6856244087219239,
+      "step": 2600
+    },
+    {
+      "epoch": 0.08361900490180374,
+      "grad_norm": 2.9895550752304025,
+      "learning_rate": 2e-05,
+      "loss": 0.6946100234985352,
+      "step": 2610
+    },
+    {
+      "epoch": 0.08393938423092942,
+      "grad_norm": 1.2195098935112956,
+      "learning_rate": 2e-05,
+      "loss": 0.6571969509124755,
+      "step": 2620
+    },
+    {
+      "epoch": 0.0842597635600551,
+      "grad_norm": 1.5659331888464,
+      "learning_rate": 2e-05,
+      "loss": 0.6633186340332031,
+      "step": 2630
+    },
+    {
+      "epoch": 0.0845801428891808,
+      "grad_norm": 1.2623772678903336,
+      "learning_rate": 2e-05,
+      "loss": 0.7006426811218261,
+      "step": 2640
+    },
+    {
+      "epoch": 0.08490052221830648,
+      "grad_norm": 53.51607876951121,
+      "learning_rate": 2e-05,
+      "loss": 0.7465255260467529,
+      "step": 2650
+    },
+    {
+      "epoch": 0.08522090154743216,
+      "grad_norm": 7.141894169140511,
+      "learning_rate": 2e-05,
+      "loss": 0.6905827999114991,
+      "step": 2660
+    },
+    {
+      "epoch": 0.08554128087655785,
+      "grad_norm": 12.021818224070394,
+      "learning_rate": 2e-05,
+      "loss": 0.7053852081298828,
+      "step": 2670
+    },
+    {
+      "epoch": 0.08586166020568353,
+      "grad_norm": 1.9077255089530207,
+      "learning_rate": 2e-05,
+      "loss": 0.6775362968444825,
+      "step": 2680
+    },
+    {
+      "epoch": 0.08618203953480921,
+      "grad_norm": 3.124384881028629,
+      "learning_rate": 2e-05,
+      "loss": 0.6948652267456055,
+      "step": 2690
+    },
+    {
+      "epoch": 0.0865024188639349,
+      "grad_norm": 1.3550553388372282,
+      "learning_rate": 2e-05,
+      "loss": 0.6418863296508789,
+      "step": 2700
+    },
+    {
+      "epoch": 0.08682279819306059,
+      "grad_norm": 1.6996387488097904,
+      "learning_rate": 2e-05,
+      "loss": 0.7217937469482422,
+      "step": 2710
+    },
+    {
+      "epoch": 0.08714317752218627,
+      "grad_norm": 4.721987876312749,
+      "learning_rate": 2e-05,
+      "loss": 0.678973388671875,
+      "step": 2720
+    },
+    {
+      "epoch": 0.08746355685131195,
+      "grad_norm": 207.96463050114306,
+      "learning_rate": 2e-05,
+      "loss": 0.7110597610473632,
+      "step": 2730
+    },
+    {
+      "epoch": 0.08778393618043764,
+      "grad_norm": 3.3362197281306725,
+      "learning_rate": 2e-05,
+      "loss": 0.7547786235809326,
+      "step": 2740
+    },
+    {
+      "epoch": 0.08810431550956332,
+      "grad_norm": 1.7625372241440926,
+      "learning_rate": 2e-05,
+      "loss": 0.678510046005249,
+      "step": 2750
+    },
+    {
+      "epoch": 0.088424694838689,
+      "grad_norm": 1.7993132875303999,
+      "learning_rate": 2e-05,
+      "loss": 0.6995362758636474,
+      "step": 2760
+    },
+    {
+      "epoch": 0.08874507416781469,
+      "grad_norm": 7.076835439735431,
+      "learning_rate": 2e-05,
+      "loss": 0.6843991279602051,
+      "step": 2770
+    },
+    {
+      "epoch": 0.08906545349694038,
+      "grad_norm": 1.6043720642575496,
+      "learning_rate": 2e-05,
+      "loss": 0.6938441276550293,
+      "step": 2780
+    },
+    {
+      "epoch": 0.08938583282606606,
+      "grad_norm": 1.3640725070465638,
+      "learning_rate": 2e-05,
+      "loss": 0.7047198295593262,
+      "step": 2790
+    },
+    {
+      "epoch": 0.08970621215519174,
+      "grad_norm": 2.562596485102053,
+      "learning_rate": 2e-05,
+      "loss": 0.6708857536315918,
+      "step": 2800
+    },
+    {
+      "epoch": 0.09002659148431744,
+      "grad_norm": 1.9416658533084818,
+      "learning_rate": 2e-05,
+      "loss": 0.6964911937713623,
+      "step": 2810
+    },
+    {
+      "epoch": 0.09034697081344312,
+      "grad_norm": 1.5822864532309853,
+      "learning_rate": 2e-05,
+      "loss": 0.6741170406341552,
+      "step": 2820
+    },
+    {
+      "epoch": 0.0906673501425688,
+      "grad_norm": 2.492955276963355,
+      "learning_rate": 2e-05,
+      "loss": 0.6612090110778809,
+      "step": 2830
+    },
+    {
+      "epoch": 0.09098772947169449,
+      "grad_norm": 12.577880739333416,
+      "learning_rate": 2e-05,
+      "loss": 0.6880275249481201,
+      "step": 2840
+    },
+    {
+      "epoch": 0.09130810880082017,
+      "grad_norm": 2.498526802339318,
+      "learning_rate": 2e-05,
+      "loss": 0.6688164710998535,
+      "step": 2850
+    },
+    {
+      "epoch": 0.09162848812994585,
+      "grad_norm": 2.480207709776554,
+      "learning_rate": 2e-05,
+      "loss": 0.8187369346618653,
+      "step": 2860
+    },
+    {
+      "epoch": 0.09194886745907153,
+      "grad_norm": 1.7387835479887794,
+      "learning_rate": 2e-05,
+      "loss": 0.6645703315734863,
+      "step": 2870
+    },
+    {
+      "epoch": 0.09226924678819723,
+      "grad_norm": 2.463777428019542,
+      "learning_rate": 2e-05,
+      "loss": 0.680314016342163,
+      "step": 2880
+    },
+    {
+      "epoch": 0.09258962611732291,
+      "grad_norm": 2.1125181898998457,
+      "learning_rate": 2e-05,
+      "loss": 0.7115361213684082,
+      "step": 2890
+    },
+    {
+      "epoch": 0.09291000544644859,
+      "grad_norm": 3.6819788201167962,
+      "learning_rate": 2e-05,
+      "loss": 0.7376339912414551,
+      "step": 2900
+    },
+    {
+      "epoch": 0.09323038477557428,
+      "grad_norm": 5.364519716326931,
+      "learning_rate": 2e-05,
+      "loss": 0.7281315803527832,
+      "step": 2910
+    },
+    {
+      "epoch": 0.09355076410469997,
+      "grad_norm": 5.2868820878632246,
+      "learning_rate": 2e-05,
+      "loss": 0.6780884742736817,
+      "step": 2920
+    },
+    {
+      "epoch": 0.09387114343382565,
+      "grad_norm": 2.569546482429642,
+      "learning_rate": 2e-05,
+      "loss": 0.6810887813568115,
+      "step": 2930
+    },
+    {
+      "epoch": 0.09419152276295134,
+      "grad_norm": 3.1590297185599887,
+      "learning_rate": 2e-05,
+      "loss": 0.6913040161132813,
+      "step": 2940
+    },
+    {
+      "epoch": 0.09451190209207702,
+      "grad_norm": 2.2404038565045,
+      "learning_rate": 2e-05,
+      "loss": 0.639086103439331,
+      "step": 2950
+    },
+    {
+      "epoch": 0.0948322814212027,
+      "grad_norm": 1.6593714513437243,
+      "learning_rate": 2e-05,
+      "loss": 0.6881752014160156,
+      "step": 2960
+    },
+    {
+      "epoch": 0.09515266075032838,
+      "grad_norm": 1.9761898426256463,
+      "learning_rate": 2e-05,
+      "loss": 0.6969388484954834,
+      "step": 2970
+    },
+    {
+      "epoch": 0.09547304007945408,
+      "grad_norm": 1.5176347302651876,
+      "learning_rate": 2e-05,
+      "loss": 0.6573243141174316,
+      "step": 2980
+    },
+    {
+      "epoch": 0.09579341940857976,
+      "grad_norm": 1.5490982461158973,
+      "learning_rate": 2e-05,
+      "loss": 0.6932000160217285,
+      "step": 2990
+    },
+    {
+      "epoch": 0.09611379873770544,
+      "grad_norm": 3.9352475270631273,
+      "learning_rate": 2e-05,
+      "loss": 0.649666690826416,
+      "step": 3000
+    },
+    {
+      "epoch": 0.09643417806683113,
+      "grad_norm": 3.0813498769346834,
+      "learning_rate": 2e-05,
+      "loss": 0.6733899593353272,
+      "step": 3010
+    },
+    {
+      "epoch": 0.09675455739595681,
+      "grad_norm": 1.4582486886298482,
+      "learning_rate": 2e-05,
+      "loss": 0.6860736846923828,
+      "step": 3020
+    },
+    {
+      "epoch": 0.0970749367250825,
+      "grad_norm": 2.1316688325298223,
+      "learning_rate": 2e-05,
+      "loss": 0.6471640586853027,
+      "step": 3030
+    },
+    {
+      "epoch": 0.09739531605420819,
+      "grad_norm": 1.624813158041112,
+      "learning_rate": 2e-05,
+      "loss": 0.6769994258880615,
+      "step": 3040
+    },
+    {
+      "epoch": 0.09771569538333387,
+      "grad_norm": 1.1991400569586823,
+      "learning_rate": 2e-05,
+      "loss": 0.7090652465820313,
+      "step": 3050
+    },
+    {
+      "epoch": 0.09803607471245955,
+      "grad_norm": 1.8553248908588058,
+      "learning_rate": 2e-05,
+      "loss": 0.6960935115814209,
+      "step": 3060
+    },
+    {
+      "epoch": 0.09835645404158523,
+      "grad_norm": 1.0947674863297467,
+      "learning_rate": 2e-05,
+      "loss": 0.7006158828735352,
+      "step": 3070
+    },
+    {
+      "epoch": 0.09867683337071093,
+      "grad_norm": 1.9345154791117662,
+      "learning_rate": 2e-05,
+      "loss": 0.6712536811828613,
+      "step": 3080
+    },
+    {
+      "epoch": 0.0989972126998366,
+      "grad_norm": 1.206110040611478,
+      "learning_rate": 2e-05,
+      "loss": 0.6646307945251465,
+      "step": 3090
+    },
+    {
+      "epoch": 0.09931759202896229,
+      "grad_norm": 1.5706700337393924,
+      "learning_rate": 2e-05,
+      "loss": 0.721156644821167,
+      "step": 3100
+    },
+    {
+      "epoch": 0.09963797135808798,
+      "grad_norm": 1.8852007896443763,
+      "learning_rate": 2e-05,
+      "loss": 0.6588726997375488,
+      "step": 3110
+    },
+    {
+      "epoch": 0.09995835068721366,
+      "grad_norm": 1.9309188732059843,
+      "learning_rate": 2e-05,
+      "loss": 0.6430281639099121,
+      "step": 3120
+    },
+    {
+      "epoch": 0.10027873001633934,
+      "grad_norm": 1.8247916600609566,
+      "learning_rate": 2e-05,
+      "loss": 0.662308645248413,
+      "step": 3130
+    },
+    {
+      "epoch": 0.10059910934546502,
+      "grad_norm": 20.234568515276454,
+      "learning_rate": 2e-05,
+      "loss": 0.6764077186584473,
+      "step": 3140
+    },
+    {
+      "epoch": 0.10091948867459072,
+      "grad_norm": 2.270401562255825,
+      "learning_rate": 2e-05,
+      "loss": 0.6858000755310059,
+      "step": 3150
+    },
+    {
+      "epoch": 0.1012398680037164,
+      "grad_norm": 3.3124667080997154,
+      "learning_rate": 2e-05,
+      "loss": 0.7174823760986329,
+      "step": 3160
+    },
+    {
+      "epoch": 0.10156024733284208,
+      "grad_norm": 1.21628975030768,
+      "learning_rate": 2e-05,
+      "loss": 0.7047437191009521,
+      "step": 3170
+    },
+    {
+      "epoch": 0.10188062666196777,
+      "grad_norm": 1.8787293361467063,
+      "learning_rate": 2e-05,
+      "loss": 0.6899389266967774,
+      "step": 3180
+    },
+    {
+      "epoch": 0.10220100599109345,
+      "grad_norm": 3.435196857519197,
+      "learning_rate": 2e-05,
+      "loss": 0.678690242767334,
+      "step": 3190
+    },
+    {
+      "epoch": 0.10252138532021914,
+      "grad_norm": 2.4128135351683486,
+      "learning_rate": 2e-05,
+      "loss": 0.6727965354919434,
+      "step": 3200
+    },
+    {
+      "epoch": 0.10284176464934483,
+      "grad_norm": 5.463858532009493,
+      "learning_rate": 2e-05,
+      "loss": 0.7546576499938965,
+      "step": 3210
+    },
+    {
+      "epoch": 0.10316214397847051,
+      "grad_norm": 1.1458273204458618,
+      "learning_rate": 2e-05,
+      "loss": 0.6437710762023926,
+      "step": 3220
+    },
+    {
+      "epoch": 0.10348252330759619,
+      "grad_norm": 1.3601974957658387,
+      "learning_rate": 2e-05,
+      "loss": 0.6784330368041992,
+      "step": 3230
+    },
+    {
+      "epoch": 0.10380290263672187,
+      "grad_norm": 1.4840886563319795,
+      "learning_rate": 2e-05,
+      "loss": 0.6686971664428711,
+      "step": 3240
+    },
+    {
+      "epoch": 0.10412328196584757,
+      "grad_norm": 1.3597430285112961,
+      "learning_rate": 2e-05,
+      "loss": 0.6590595245361328,
+      "step": 3250
+    },
+    {
+      "epoch": 0.10444366129497325,
+      "grad_norm": 2.7391179546585467,
+      "learning_rate": 2e-05,
+      "loss": 0.6681632995605469,
+      "step": 3260
+    },
+    {
+      "epoch": 0.10476404062409893,
+      "grad_norm": 2.9881385443242867,
+      "learning_rate": 2e-05,
+      "loss": 0.6919444084167481,
+      "step": 3270
+    },
+    {
+      "epoch": 0.10508441995322462,
+      "grad_norm": 1.7848898733758627,
+      "learning_rate": 2e-05,
+      "loss": 0.678253173828125,
+      "step": 3280
+    },
+    {
+      "epoch": 0.1054047992823503,
+      "grad_norm": 1.0941482258819155,
+      "learning_rate": 2e-05,
+      "loss": 0.6677752494812011,
+      "step": 3290
+    },
+    {
+      "epoch": 0.10572517861147598,
+      "grad_norm": 1.1439491512926567,
+      "learning_rate": 2e-05,
+      "loss": 0.6404883384704589,
+      "step": 3300
+    },
+    {
+      "epoch": 0.10604555794060168,
+      "grad_norm": 1.6084438657234088,
+      "learning_rate": 2e-05,
+      "loss": 0.6679693222045898,
+      "step": 3310
+    },
+    {
+      "epoch": 0.10636593726972736,
+      "grad_norm": 2.3221962830811265,
+      "learning_rate": 2e-05,
+      "loss": 0.6701303482055664,
+      "step": 3320
+    },
+    {
+      "epoch": 0.10668631659885304,
+      "grad_norm": 1.1626176984318963,
+      "learning_rate": 2e-05,
+      "loss": 0.6760411262512207,
+      "step": 3330
+    },
+    {
+      "epoch": 0.10700669592797872,
+      "grad_norm": 9.60833542052224,
+      "learning_rate": 2e-05,
+      "loss": 0.6958446502685547,
+      "step": 3340
+    },
+    {
+      "epoch": 0.10732707525710442,
+      "grad_norm": 1.4570489896161745,
+      "learning_rate": 2e-05,
+      "loss": 0.7142772674560547,
+      "step": 3350
+    },
+    {
+      "epoch": 0.1076474545862301,
+      "grad_norm": 1.589527243320569,
+      "learning_rate": 2e-05,
+      "loss": 0.6480647087097168,
+      "step": 3360
+    },
+    {
+      "epoch": 0.10796783391535578,
+      "grad_norm": 3.7539533800821854,
+      "learning_rate": 2e-05,
+      "loss": 0.6926058769226074,
+      "step": 3370
+    },
+    {
+      "epoch": 0.10828821324448147,
+      "grad_norm": 1.4343813974850903,
+      "learning_rate": 2e-05,
+      "loss": 0.6597871303558349,
+      "step": 3380
+    },
+    {
+      "epoch": 0.10860859257360715,
+      "grad_norm": 1.700489983652249,
+      "learning_rate": 2e-05,
+      "loss": 0.6854336738586426,
+      "step": 3390
+    },
+    {
+      "epoch": 0.10892897190273283,
+      "grad_norm": 2.262978455318886,
+      "learning_rate": 2e-05,
+      "loss": 0.6611383438110352,
+      "step": 3400
+    },
+    {
+      "epoch": 0.10924935123185853,
+      "grad_norm": 3.7384144861916924,
+      "learning_rate": 2e-05,
+      "loss": 0.6642798900604248,
+      "step": 3410
+    },
+    {
+      "epoch": 0.10956973056098421,
+      "grad_norm": 5.828313398251216,
+      "learning_rate": 2e-05,
+      "loss": 0.6781678676605225,
+      "step": 3420
+    },
+    {
+      "epoch": 0.10989010989010989,
+      "grad_norm": 5.487121962637965,
+      "learning_rate": 2e-05,
+      "loss": 0.7087061882019043,
+      "step": 3430
+    },
+    {
+      "epoch": 0.11021048921923557,
+      "grad_norm": 1.6357218209966338,
+      "learning_rate": 2e-05,
+      "loss": 0.6441525936126709,
+      "step": 3440
+    },
+    {
+      "epoch": 0.11053086854836126,
+      "grad_norm": 1.9006368500272885,
+      "learning_rate": 2e-05,
+      "loss": 0.6848069667816162,
+      "step": 3450
+    },
+    {
+      "epoch": 0.11085124787748694,
+      "grad_norm": 1.4766899104425295,
+      "learning_rate": 2e-05,
+      "loss": 0.6893895149230957,
+      "step": 3460
+    },
+    {
+      "epoch": 0.11117162720661262,
+      "grad_norm": 2.769884408658176,
+      "learning_rate": 2e-05,
+      "loss": 0.6294010162353516,
+      "step": 3470
+    },
+    {
+      "epoch": 0.11149200653573832,
+      "grad_norm": 4.65526982930436,
+      "learning_rate": 2e-05,
+      "loss": 0.6821585655212402,
+      "step": 3480
+    },
+    {
+      "epoch": 0.111812385864864,
+      "grad_norm": 1.7647620795155232,
+      "learning_rate": 2e-05,
+      "loss": 0.6744832515716552,
+      "step": 3490
+    },
+    {
+      "epoch": 0.11213276519398968,
+      "grad_norm": 1.2028872660134693,
+      "learning_rate": 2e-05,
+      "loss": 0.663267707824707,
+      "step": 3500
+    },
+    {
+      "epoch": 0.11245314452311538,
+      "grad_norm": 4.612458331540539,
+      "learning_rate": 2e-05,
+      "loss": 0.6622488021850585,
+      "step": 3510
+    },
+    {
+      "epoch": 0.11277352385224106,
+      "grad_norm": 3.060996296931568,
+      "learning_rate": 2e-05,
+      "loss": 0.6657146453857422,
+      "step": 3520
+    },
+    {
+      "epoch": 0.11309390318136674,
+      "grad_norm": 1.3030075500669922,
+      "learning_rate": 2e-05,
+      "loss": 0.6406706809997559,
+      "step": 3530
+    },
+    {
+      "epoch": 0.11341428251049242,
+      "grad_norm": 2.019294690660488,
+      "learning_rate": 2e-05,
+      "loss": 0.6651981353759766,
+      "step": 3540
+    },
+    {
+      "epoch": 0.11373466183961811,
+      "grad_norm": 1.2086238864300596,
+      "learning_rate": 2e-05,
+      "loss": 0.6543137550354003,
+      "step": 3550
+    },
+    {
+      "epoch": 0.11405504116874379,
+      "grad_norm": 1.2655664277456855,
+      "learning_rate": 2e-05,
+      "loss": 0.6376056671142578,
+      "step": 3560
+    },
+    {
+      "epoch": 0.11437542049786947,
+      "grad_norm": 2.0908487857785176,
+      "learning_rate": 2e-05,
+      "loss": 0.6656252861022949,
+      "step": 3570
+    },
+    {
+      "epoch": 0.11469579982699517,
+      "grad_norm": 5.261221133161816,
+      "learning_rate": 2e-05,
+      "loss": 0.6583863258361816,
+      "step": 3580
+    },
+    {
+      "epoch": 0.11501617915612085,
+      "grad_norm": 2.3910426503344664,
+      "learning_rate": 2e-05,
+      "loss": 0.6792299270629882,
+      "step": 3590
+    },
+    {
+      "epoch": 0.11533655848524653,
+      "grad_norm": 1.211765741906558,
+      "learning_rate": 2e-05,
+      "loss": 0.7030490398406982,
+      "step": 3600
+    },
+    {
+      "epoch": 0.11565693781437221,
+      "grad_norm": 1.9832630555170212,
+      "learning_rate": 2e-05,
+      "loss": 0.6824533462524414,
+      "step": 3610
+    },
+    {
+      "epoch": 0.1159773171434979,
+      "grad_norm": 2.252162260913484,
+      "learning_rate": 2e-05,
+      "loss": 0.6910503387451172,
+      "step": 3620
+    },
+    {
+      "epoch": 0.11629769647262359,
+      "grad_norm": 1.2137738587533493,
+      "learning_rate": 2e-05,
+      "loss": 0.6761364936828613,
+      "step": 3630
+    },
+    {
+      "epoch": 0.11661807580174927,
+      "grad_norm": 1.6480847954089972,
+      "learning_rate": 2e-05,
+      "loss": 0.6949734687805176,
+      "step": 3640
+    },
+    {
+      "epoch": 0.11693845513087496,
+      "grad_norm": 2.773368722100306,
+      "learning_rate": 2e-05,
+      "loss": 0.6389493942260742,
+      "step": 3650
+    },
+    {
+      "epoch": 0.11725883446000064,
+      "grad_norm": 2.5465741003121365,
+      "learning_rate": 2e-05,
+      "loss": 0.6798902034759522,
+      "step": 3660
+    },
+    {
+      "epoch": 0.11757921378912632,
+      "grad_norm": 3.003062238735499,
+      "learning_rate": 2e-05,
+      "loss": 0.6937604427337647,
+      "step": 3670
+    },
+    {
+      "epoch": 0.11789959311825202,
+      "grad_norm": 1.0771834972401186,
+      "learning_rate": 2e-05,
+      "loss": 0.6856004238128662,
+      "step": 3680
+    },
+    {
+      "epoch": 0.1182199724473777,
+      "grad_norm": 3.2223166693736465,
+      "learning_rate": 2e-05,
+      "loss": 0.7183493614196778,
+      "step": 3690
+    },
+    {
+      "epoch": 0.11854035177650338,
+      "grad_norm": 1.90983436518728,
+      "learning_rate": 2e-05,
+      "loss": 0.6716906547546386,
+      "step": 3700
+    },
+    {
+      "epoch": 0.11886073110562906,
+      "grad_norm": 2.0873401588953433,
+      "learning_rate": 2e-05,
+      "loss": 0.666513729095459,
+      "step": 3710
+    },
+    {
+      "epoch": 0.11918111043475475,
+      "grad_norm": 1.1975205039758825,
+      "learning_rate": 2e-05,
+      "loss": 0.6365289211273193,
+      "step": 3720
+    },
+    {
+      "epoch": 0.11950148976388043,
+      "grad_norm": 41.55973254495974,
+      "learning_rate": 2e-05,
+      "loss": 0.7242575645446777,
+      "step": 3730
+    },
+    {
+      "epoch": 0.11982186909300611,
+      "grad_norm": 1.7969703921280566,
+      "learning_rate": 2e-05,
+      "loss": 0.6333506584167481,
+      "step": 3740
+    },
+    {
+      "epoch": 0.12014224842213181,
+      "grad_norm": 1.1484967465503775,
+      "learning_rate": 2e-05,
+      "loss": 0.681650447845459,
+      "step": 3750
+    },
+    {
+      "epoch": 0.12046262775125749,
+      "grad_norm": 1.491743333469875,
+      "learning_rate": 2e-05,
+      "loss": 0.683896541595459,
+      "step": 3760
+    },
+    {
+      "epoch": 0.12078300708038317,
+      "grad_norm": 1.9896310599000153,
+      "learning_rate": 2e-05,
+      "loss": 0.6723162651062011,
+      "step": 3770
+    },
+    {
+      "epoch": 0.12110338640950886,
+      "grad_norm": 4.355204703369626,
+      "learning_rate": 2e-05,
+      "loss": 0.6825237274169922,
+      "step": 3780
+    },
+    {
+      "epoch": 0.12142376573863455,
+      "grad_norm": 2.6337758244589446,
+      "learning_rate": 2e-05,
+      "loss": 0.6493152618408203,
+      "step": 3790
+    },
+    {
+      "epoch": 0.12174414506776023,
+      "grad_norm": 1.4311522270790977,
+      "learning_rate": 2e-05,
+      "loss": 0.6465260982513428,
+      "step": 3800
+    },
+    {
+      "epoch": 0.1220645243968859,
+      "grad_norm": 1.2584234760746995,
+      "learning_rate": 2e-05,
+      "loss": 0.6119470119476318,
+      "step": 3810
+    },
+    {
+      "epoch": 0.1223849037260116,
+      "grad_norm": 1.8362919442209469,
+      "learning_rate": 2e-05,
+      "loss": 0.6505841732025146,
+      "step": 3820
+    },
+    {
+      "epoch": 0.12270528305513728,
+      "grad_norm": 3.2911678224166363,
+      "learning_rate": 2e-05,
+      "loss": 0.7023921012878418,
+      "step": 3830
+    },
+    {
+      "epoch": 0.12302566238426296,
+      "grad_norm": 1.8619496153981334,
+      "learning_rate": 2e-05,
+      "loss": 0.676816749572754,
+      "step": 3840
+    },
+    {
+      "epoch": 0.12334604171338866,
+      "grad_norm": 1.2607793375322882,
+      "learning_rate": 2e-05,
+      "loss": 0.6600100994110107,
+      "step": 3850
+    },
+    {
+      "epoch": 0.12366642104251434,
+      "grad_norm": 1.3500576986027626,
+      "learning_rate": 2e-05,
+      "loss": 0.6765620708465576,
+      "step": 3860
+    },
+    {
+      "epoch": 0.12398680037164002,
+      "grad_norm": 1.3401379911166065,
+      "learning_rate": 2e-05,
+      "loss": 0.6644556045532226,
+      "step": 3870
+    },
+    {
+      "epoch": 0.12430717970076571,
+      "grad_norm": 1.3436995912683865,
+      "learning_rate": 2e-05,
+      "loss": 0.6688062191009522,
+      "step": 3880
+    },
+    {
+      "epoch": 0.1246275590298914,
+      "grad_norm": 2.1554623740628323,
+      "learning_rate": 2e-05,
+      "loss": 0.6625113487243652,
+      "step": 3890
+    },
+    {
+      "epoch": 0.12494793835901707,
+      "grad_norm": 2.5490411787256693,
+      "learning_rate": 2e-05,
+      "loss": 0.6453998565673829,
+      "step": 3900
+    },
+    {
+      "epoch": 0.12526831768814276,
+      "grad_norm": 2.1422672897977475,
+      "learning_rate": 2e-05,
+      "loss": 0.6868985652923584,
+      "step": 3910
+    },
+    {
+      "epoch": 0.12558869701726844,
+      "grad_norm": 8.647039761010506,
+      "learning_rate": 2e-05,
+      "loss": 0.6461306571960449,
+      "step": 3920
+    },
+    {
+      "epoch": 0.12590907634639414,
+      "grad_norm": 4.353384006867587,
+      "learning_rate": 2e-05,
+      "loss": 0.6968627452850342,
+      "step": 3930
+    },
+    {
+      "epoch": 0.12622945567551982,
+      "grad_norm": 5.312771003446082,
+      "learning_rate": 2e-05,
+      "loss": 0.6628563880920411,
+      "step": 3940
+    },
+    {
+      "epoch": 0.1265498350046455,
+      "grad_norm": 1.2483064513173903,
+      "learning_rate": 2e-05,
+      "loss": 0.6820049285888672,
+      "step": 3950
+    },
+    {
+      "epoch": 0.1268702143337712,
+      "grad_norm": 1.7052048413362177,
+      "learning_rate": 2e-05,
+      "loss": 0.6507192134857178,
+      "step": 3960
+    },
+    {
+      "epoch": 0.12719059366289687,
+      "grad_norm": 1.442220593830901,
+      "learning_rate": 2e-05,
+      "loss": 0.6692718505859375,
+      "step": 3970
+    },
+    {
+      "epoch": 0.12751097299202255,
+      "grad_norm": 21.674983962515157,
+      "learning_rate": 2e-05,
+      "loss": 0.6392804622650147,
+      "step": 3980
+    },
+    {
+      "epoch": 0.12783135232114823,
+      "grad_norm": 9.728263472908212,
+      "learning_rate": 2e-05,
+      "loss": 0.656775426864624,
+      "step": 3990
+    },
+    {
+      "epoch": 0.12815173165027394,
+      "grad_norm": 2.039616161760123,
+      "learning_rate": 2e-05,
+      "loss": 0.7095576286315918,
+      "step": 4000
+    },
+    {
+      "epoch": 0.12847211097939962,
+      "grad_norm": 4.0175728810676805,
+      "learning_rate": 2e-05,
+      "loss": 0.6841886043548584,
+      "step": 4010
+    },
+    {
+      "epoch": 0.1287924903085253,
+      "grad_norm": 19.73418939091534,
+      "learning_rate": 2e-05,
+      "loss": 0.6358445644378662,
+      "step": 4020
+    },
+    {
+      "epoch": 0.12911286963765098,
+      "grad_norm": 1.739996971425279,
+      "learning_rate": 2e-05,
+      "loss": 0.6840127944946289,
+      "step": 4030
+    },
+    {
+      "epoch": 0.12943324896677666,
+      "grad_norm": 1.450557707510894,
+      "learning_rate": 2e-05,
+      "loss": 0.6738371849060059,
+      "step": 4040
+    },
+    {
+      "epoch": 0.12975362829590234,
+      "grad_norm": 2.0404367072041,
+      "learning_rate": 2e-05,
+      "loss": 0.6749392509460449,
+      "step": 4050
+    },
+    {
+      "epoch": 0.13007400762502802,
+      "grad_norm": 1.6413128097895142,
+      "learning_rate": 2e-05,
+      "loss": 0.6790598869323731,
+      "step": 4060
+    },
+    {
+      "epoch": 0.13039438695415373,
+      "grad_norm": 1.2445737299020967,
+      "learning_rate": 2e-05,
+      "loss": 0.7365617752075195,
+      "step": 4070
+    },
+    {
+      "epoch": 0.1307147662832794,
+      "grad_norm": 3.445733180267043,
+      "learning_rate": 2e-05,
+      "loss": 0.6372848510742187,
+      "step": 4080
+    },
+    {
+      "epoch": 0.1310351456124051,
+      "grad_norm": 1.2285630455823986,
+      "learning_rate": 2e-05,
+      "loss": 0.6606671810150146,
+      "step": 4090
+    },
+    {
+      "epoch": 0.13135552494153077,
+      "grad_norm": 2.349488425660991,
+      "learning_rate": 2e-05,
+      "loss": 0.6286896705627442,
+      "step": 4100
+    },
+    {
+      "epoch": 0.13167590427065645,
+      "grad_norm": 1.4532007484386635,
+      "learning_rate": 2e-05,
+      "loss": 0.6365851402282715,
+      "step": 4110
+    },
+    {
+      "epoch": 0.13199628359978213,
+      "grad_norm": 0.9308693386958972,
+      "learning_rate": 2e-05,
+      "loss": 0.6765342712402344,
+      "step": 4120
+    },
+    {
+      "epoch": 0.1323166629289078,
+      "grad_norm": 1.012417770518259,
+      "learning_rate": 2e-05,
+      "loss": 0.6740757942199707,
+      "step": 4130
+    },
+    {
+      "epoch": 0.13263704225803352,
+      "grad_norm": 1.0125547138996793,
+      "learning_rate": 2e-05,
+      "loss": 0.6393099784851074,
+      "step": 4140
+    },
+    {
+      "epoch": 0.1329574215871592,
+      "grad_norm": 1.2177266588739921,
+      "learning_rate": 2e-05,
+      "loss": 0.6418939113616944,
+      "step": 4150
+    },
+    {
+      "epoch": 0.13327780091628488,
+      "grad_norm": 1.9715086658340903,
+      "learning_rate": 2e-05,
+      "loss": 0.659096622467041,
+      "step": 4160
+    },
+    {
+      "epoch": 0.13359818024541056,
+      "grad_norm": 1.5354076340036875,
+      "learning_rate": 2e-05,
+      "loss": 0.6633724212646485,
+      "step": 4170
+    },
+    {
+      "epoch": 0.13391855957453624,
+      "grad_norm": 1.379476140664975,
+      "learning_rate": 2e-05,
+      "loss": 0.653409481048584,
+      "step": 4180
+    },
+    {
+      "epoch": 0.13423893890366193,
+      "grad_norm": 1.3861045841146946,
+      "learning_rate": 2e-05,
+      "loss": 0.6600584030151367,
+      "step": 4190
+    },
+    {
+      "epoch": 0.13455931823278763,
+      "grad_norm": 1.3471659231508513,
+      "learning_rate": 2e-05,
+      "loss": 0.6790562629699707,
+      "step": 4200
+    },
+    {
+      "epoch": 0.13487969756191331,
+      "grad_norm": 1.7716447481227624,
+      "learning_rate": 2e-05,
+      "loss": 0.6599895000457764,
+      "step": 4210
+    },
+    {
+      "epoch": 0.135200076891039,
+      "grad_norm": 1.8371495895343661,
+      "learning_rate": 2e-05,
+      "loss": 0.6585530281066895,
+      "step": 4220
+    },
+    {
+      "epoch": 0.13552045622016468,
+      "grad_norm": 1.4824802128279082,
+      "learning_rate": 2e-05,
+      "loss": 0.6495269775390625,
+      "step": 4230
+    },
+    {
+      "epoch": 0.13584083554929036,
+      "grad_norm": 9.508448966165497,
+      "learning_rate": 2e-05,
+      "loss": 0.6385743618011475,
+      "step": 4240
+    },
+    {
+      "epoch": 0.13616121487841604,
+      "grad_norm": 3.7227417071686406,
+      "learning_rate": 2e-05,
+      "loss": 0.6491826057434082,
+      "step": 4250
+    },
+    {
+      "epoch": 0.13648159420754172,
+      "grad_norm": 1.6191391322570134,
+      "learning_rate": 2e-05,
+      "loss": 0.6381173610687256,
+      "step": 4260
+    },
+    {
+      "epoch": 0.13680197353666743,
+      "grad_norm": 2.295415384352932,
+      "learning_rate": 2e-05,
+      "loss": 0.6816682815551758,
+      "step": 4270
+    },
+    {
+      "epoch": 0.1371223528657931,
+      "grad_norm": 3.555025076718053,
+      "learning_rate": 2e-05,
+      "loss": 0.65838623046875,
+      "step": 4280
+    },
+    {
+      "epoch": 0.1374427321949188,
+      "grad_norm": 1.8199956373458717,
+      "learning_rate": 2e-05,
+      "loss": 0.643873929977417,
+      "step": 4290
+    },
+    {
+      "epoch": 0.13776311152404447,
+      "grad_norm": 5.014043066263043,
+      "learning_rate": 2e-05,
+      "loss": 0.6718840599060059,
+      "step": 4300
+    },
+    {
+      "epoch": 0.13808349085317015,
+      "grad_norm": 3.6932065567825383,
+      "learning_rate": 2e-05,
+      "loss": 0.6397347450256348,
+      "step": 4310
+    },
+    {
+      "epoch": 0.13840387018229583,
+      "grad_norm": 2.8239286018474568,
+      "learning_rate": 2e-05,
+      "loss": 0.6621566772460937,
+      "step": 4320
+    },
+    {
+      "epoch": 0.1387242495114215,
+      "grad_norm": 2.6802412140812124,
+      "learning_rate": 2e-05,
+      "loss": 0.6485812187194824,
+      "step": 4330
+    },
+    {
+      "epoch": 0.13904462884054722,
+      "grad_norm": 14.981569614940994,
+      "learning_rate": 2e-05,
+      "loss": 0.6633264064788819,
+      "step": 4340
+    },
+    {
+      "epoch": 0.1393650081696729,
+      "grad_norm": 2.3542641904056785,
+      "learning_rate": 2e-05,
+      "loss": 0.6532891750335693,
+      "step": 4350
+    },
+    {
+      "epoch": 0.13968538749879858,
+      "grad_norm": 2.1434978987275186,
+      "learning_rate": 2e-05,
+      "loss": 0.6710407257080078,
+      "step": 4360
+    },
+    {
+      "epoch": 0.14000576682792426,
+      "grad_norm": 2.440516483612938,
+      "learning_rate": 2e-05,
+      "loss": 0.6813251495361328,
+      "step": 4370
+    },
+    {
+      "epoch": 0.14032614615704994,
+      "grad_norm": 1.616569444263289,
+      "learning_rate": 2e-05,
+      "loss": 0.6308428764343261,
+      "step": 4380
+    },
+    {
+      "epoch": 0.14064652548617562,
+      "grad_norm": 2.504912035473103,
+      "learning_rate": 2e-05,
+      "loss": 0.671413516998291,
+      "step": 4390
+    },
+    {
+      "epoch": 0.14096690481530133,
+      "grad_norm": 1.7969208881063083,
+      "learning_rate": 2e-05,
+      "loss": 0.6420493602752686,
+      "step": 4400
+    },
+    {
+      "epoch": 0.141287284144427,
+      "grad_norm": 2.1097185890269823,
+      "learning_rate": 2e-05,
+      "loss": 0.663338041305542,
+      "step": 4410
+    },
+    {
+      "epoch": 0.1416076634735527,
+      "grad_norm": 2.754978657046392,
+      "learning_rate": 2e-05,
+      "loss": 0.6554645538330078,
+      "step": 4420
+    },
+    {
+      "epoch": 0.14192804280267837,
+      "grad_norm": 1.982488431740266,
+      "learning_rate": 2e-05,
+      "loss": 0.6716144561767579,
+      "step": 4430
+    },
+    {
+      "epoch": 0.14224842213180405,
+      "grad_norm": 1.0919658882240069,
+      "learning_rate": 2e-05,
+      "loss": 0.6052421569824219,
+      "step": 4440
+    },
+    {
+      "epoch": 0.14256880146092973,
+      "grad_norm": 2.1629461397046814,
+      "learning_rate": 2e-05,
+      "loss": 0.6777543544769287,
+      "step": 4450
+    },
+    {
+      "epoch": 0.14288918079005541,
+      "grad_norm": 1.2184049851316467,
+      "learning_rate": 2e-05,
+      "loss": 0.639255142211914,
+      "step": 4460
+    },
+    {
+      "epoch": 0.14320956011918112,
+      "grad_norm": 6.729538545420926,
+      "learning_rate": 2e-05,
+      "loss": 0.6360386848449707,
+      "step": 4470
+    },
+    {
+      "epoch": 0.1435299394483068,
+      "grad_norm": 3.626596705131836,
+      "learning_rate": 2e-05,
+      "loss": 0.6660489559173584,
+      "step": 4480
+    },
+    {
+      "epoch": 0.14385031877743248,
+      "grad_norm": 1.1534629577740403,
+      "learning_rate": 2e-05,
+      "loss": 0.7270738124847412,
+      "step": 4490
+    },
+    {
+      "epoch": 0.14417069810655816,
+      "grad_norm": 1.5273936802078765,
+      "learning_rate": 2e-05,
+      "loss": 0.66806001663208,
+      "step": 4500
+    },
+    {
+      "epoch": 0.14449107743568385,
+      "grad_norm": 1.0857943782617103,
+      "learning_rate": 2e-05,
+      "loss": 0.6453025341033936,
+      "step": 4510
+    },
+    {
+      "epoch": 0.14481145676480953,
+      "grad_norm": 1.4204053183923306,
+      "learning_rate": 2e-05,
+      "loss": 0.6665804862976075,
+      "step": 4520
+    },
+    {
+      "epoch": 0.1451318360939352,
+      "grad_norm": 2.770236841939151,
+      "learning_rate": 2e-05,
+      "loss": 0.6432864189147949,
+      "step": 4530
+    },
+    {
+      "epoch": 0.14545221542306092,
+      "grad_norm": 1.3119479716888935,
+      "learning_rate": 2e-05,
+      "loss": 0.656611967086792,
+      "step": 4540
+    },
+    {
+      "epoch": 0.1457725947521866,
+      "grad_norm": 1.259813141300481,
+      "learning_rate": 2e-05,
+      "loss": 0.6752814292907715,
+      "step": 4550
+    },
+    {
+      "epoch": 0.14609297408131228,
+      "grad_norm": 1.4511032127265469,
+      "learning_rate": 2e-05,
+      "loss": 0.6449288368225098,
+      "step": 4560
+    },
+    {
+      "epoch": 0.14641335341043796,
+      "grad_norm": 1.061010382277649,
+      "learning_rate": 2e-05,
+      "loss": 0.6769460201263428,
+      "step": 4570
+    },
+    {
+      "epoch": 0.14673373273956364,
+      "grad_norm": 1.252923043946276,
+      "learning_rate": 2e-05,
+      "loss": 0.6905078411102294,
+      "step": 4580
+    },
+    {
+      "epoch": 0.14705411206868932,
+      "grad_norm": 2.1232260147742252,
+      "learning_rate": 2e-05,
+      "loss": 0.6253070831298828,
+      "step": 4590
+    },
+    {
+      "epoch": 0.147374491397815,
+      "grad_norm": 1.1626353370713662,
+      "learning_rate": 2e-05,
+      "loss": 0.6894829750061036,
+      "step": 4600
+    },
+    {
+      "epoch": 0.1476948707269407,
+      "grad_norm": 1.5893300869943574,
+      "learning_rate": 2e-05,
+      "loss": 0.6677133560180664,
+      "step": 4610
+    },
+    {
+      "epoch": 0.1480152500560664,
+      "grad_norm": 1.5294628078622055,
+      "learning_rate": 2e-05,
+      "loss": 0.6319296836853028,
+      "step": 4620
+    },
+    {
+      "epoch": 0.14833562938519207,
+      "grad_norm": 1.97668656943159,
+      "learning_rate": 2e-05,
+      "loss": 0.6283845424652099,
+      "step": 4630
+    },
+    {
+      "epoch": 0.14865600871431775,
+      "grad_norm": 1.5344286541814958,
+      "learning_rate": 2e-05,
+      "loss": 0.6669802665710449,
+      "step": 4640
+    },
+    {
+      "epoch": 0.14897638804344343,
+      "grad_norm": 2.0675263045634913,
+      "learning_rate": 2e-05,
+      "loss": 0.6275815010070801,
+      "step": 4650
+    },
+    {
+      "epoch": 0.1492967673725691,
+      "grad_norm": 1.7263557892047314,
+      "learning_rate": 2e-05,
+      "loss": 0.6830034255981445,
+      "step": 4660
+    },
+    {
+      "epoch": 0.14961714670169482,
+      "grad_norm": 6.433224792075453,
+      "learning_rate": 2e-05,
+      "loss": 0.6525765895843506,
+      "step": 4670
+    },
+    {
+      "epoch": 0.1499375260308205,
+      "grad_norm": 1.4597038948397596,
+      "learning_rate": 2e-05,
+      "loss": 0.6589030742645263,
+      "step": 4680
+    },
+    {
+      "epoch": 0.15025790535994618,
+      "grad_norm": 4.824814614092905,
+      "learning_rate": 2e-05,
+      "loss": 0.6419610500335693,
+      "step": 4690
+    },
+    {
+      "epoch": 0.15057828468907186,
+      "grad_norm": 3.768242606849537,
+      "learning_rate": 2e-05,
+      "loss": 0.6277103424072266,
+      "step": 4700
+    },
+    {
+      "epoch": 0.15089866401819754,
+      "grad_norm": 3.0284514283479402,
+      "learning_rate": 2e-05,
+      "loss": 0.6320689201354981,
+      "step": 4710
+    },
+    {
+      "epoch": 0.15121904334732322,
+      "grad_norm": 1.991976725944676,
+      "learning_rate": 2e-05,
+      "loss": 0.6563403129577636,
+      "step": 4720
+    },
+    {
+      "epoch": 0.1515394226764489,
+      "grad_norm": 1.8676620435911406,
+      "learning_rate": 2e-05,
+      "loss": 0.6397079467773438,
+      "step": 4730
+    },
+    {
+      "epoch": 0.1518598020055746,
+      "grad_norm": 2.2885703993013076,
+      "learning_rate": 2e-05,
+      "loss": 0.6857260704040528,
+      "step": 4740
+    },
+    {
+      "epoch": 0.1521801813347003,
+      "grad_norm": 9.026114825724205,
+      "learning_rate": 2e-05,
+      "loss": 0.6624780654907226,
+      "step": 4750
+    },
+    {
+      "epoch": 0.15250056066382597,
+      "grad_norm": 1.2093039223466322,
+      "learning_rate": 2e-05,
+      "loss": 0.6520965576171875,
+      "step": 4760
+    },
+    {
+      "epoch": 0.15282093999295165,
+      "grad_norm": 1.181818856856248,
+      "learning_rate": 2e-05,
+      "loss": 0.6580140113830566,
+      "step": 4770
+    },
+    {
+      "epoch": 0.15314131932207733,
+      "grad_norm": 2.0375419479696624,
+      "learning_rate": 2e-05,
+      "loss": 0.638451337814331,
+      "step": 4780
+    },
+    {
+      "epoch": 0.15346169865120302,
+      "grad_norm": 1.3297776867098974,
+      "learning_rate": 2e-05,
+      "loss": 0.6944093704223633,
+      "step": 4790
+    },
+    {
+      "epoch": 0.1537820779803287,
+      "grad_norm": 3.412108902076532,
+      "learning_rate": 2e-05,
+      "loss": 0.671916913986206,
+      "step": 4800
+    },
+    {
+      "epoch": 0.1541024573094544,
+      "grad_norm": 1.2402690456194414,
+      "learning_rate": 2e-05,
+      "loss": 0.7751090049743652,
+      "step": 4810
+    },
+    {
+      "epoch": 0.15442283663858009,
+      "grad_norm": 2.2262240674536837,
+      "learning_rate": 2e-05,
+      "loss": 0.6468109130859375,
+      "step": 4820
+    },
+    {
+      "epoch": 0.15474321596770577,
+      "grad_norm": 3.034745259104267,
+      "learning_rate": 2e-05,
+      "loss": 0.6362489700317383,
+      "step": 4830
+    },
+    {
+      "epoch": 0.15506359529683145,
+      "grad_norm": 2.6210321495796722,
+      "learning_rate": 2e-05,
+      "loss": 0.6448084831237793,
+      "step": 4840
+    },
+    {
+      "epoch": 0.15538397462595713,
+      "grad_norm": 9.144796074516247,
+      "learning_rate": 2e-05,
+      "loss": 0.6558864116668701,
+      "step": 4850
+    },
+    {
+      "epoch": 0.1557043539550828,
+      "grad_norm": 0.8700210823800482,
+      "learning_rate": 2e-05,
+      "loss": 0.6739778041839599,
+      "step": 4860
+    },
+    {
+      "epoch": 0.15602473328420852,
+      "grad_norm": 1.1782758621665466,
+      "learning_rate": 2e-05,
+      "loss": 0.6652435302734375,
+      "step": 4870
+    },
+    {
+      "epoch": 0.1563451126133342,
+      "grad_norm": 0.906863821537445,
+      "learning_rate": 2e-05,
+      "loss": 0.643802261352539,
+      "step": 4880
+    },
+    {
+      "epoch": 0.15666549194245988,
+      "grad_norm": 2.5302325947419506,
+      "learning_rate": 2e-05,
+      "loss": 0.6078256607055664,
+      "step": 4890
+    },
+    {
+      "epoch": 0.15698587127158556,
+      "grad_norm": 2.3861517493162965,
+      "learning_rate": 2e-05,
+      "loss": 0.6253044128417968,
+      "step": 4900
+    },
+    {
+      "epoch": 0.15730625060071124,
+      "grad_norm": 1.0165987246330042,
+      "learning_rate": 2e-05,
+      "loss": 0.6433412075042725,
+      "step": 4910
+    },
+    {
+      "epoch": 0.15762662992983692,
+      "grad_norm": 2.5930004662557566,
+      "learning_rate": 2e-05,
+      "loss": 0.64521803855896,
+      "step": 4920
+    },
+    {
+      "epoch": 0.1579470092589626,
+      "grad_norm": 1.5176891459200517,
+      "learning_rate": 2e-05,
+      "loss": 0.6507583618164062,
+      "step": 4930
+    },
+    {
+      "epoch": 0.1582673885880883,
+      "grad_norm": 1.435929478682162,
+      "learning_rate": 2e-05,
+      "loss": 0.6506617546081543,
+      "step": 4940
+    },
+    {
+      "epoch": 0.158587767917214,
+      "grad_norm": 1.0652184103927025,
+      "learning_rate": 2e-05,
+      "loss": 0.6530905723571777,
+      "step": 4950
+    },
+    {
+      "epoch": 0.15890814724633967,
+      "grad_norm": 1.225958561863853,
+      "learning_rate": 2e-05,
+      "loss": 0.635218334197998,
+      "step": 4960
+    },
+    {
+      "epoch": 0.15922852657546535,
+      "grad_norm": 2.6773291567524353,
+      "learning_rate": 2e-05,
+      "loss": 0.6535334587097168,
+      "step": 4970
+    },
+    {
+      "epoch": 0.15954890590459103,
+      "grad_norm": 1.7716063009667466,
+      "learning_rate": 2e-05,
+      "loss": 0.638061809539795,
+      "step": 4980
+    },
+    {
+      "epoch": 0.1598692852337167,
+      "grad_norm": 1.5508397490104118,
+      "learning_rate": 2e-05,
+      "loss": 0.6964780807495117,
+      "step": 4990
+    },
+    {
+      "epoch": 0.1601896645628424,
+      "grad_norm": 1.7916719766973201,
+      "learning_rate": 2e-05,
+      "loss": 0.6676883697509766,
+      "step": 5000
+    },
+    {
+      "epoch": 0.1605100438919681,
+      "grad_norm": 1.276266669893281,
+      "learning_rate": 2e-05,
+      "loss": 0.6540284633636475,
+      "step": 5010
+    },
+    {
+      "epoch": 0.16083042322109378,
+      "grad_norm": 1.5685687478744843,
+      "learning_rate": 2e-05,
+      "loss": 0.6582478523254395,
+      "step": 5020
+    },
+    {
+      "epoch": 0.16115080255021946,
+      "grad_norm": 1.005049040974868,
+      "learning_rate": 2e-05,
+      "loss": 0.6629669189453125,
+      "step": 5030
+    },
+    {
+      "epoch": 0.16147118187934514,
+      "grad_norm": 1.4303298584505024,
+      "learning_rate": 2e-05,
+      "loss": 0.6253504753112793,
+      "step": 5040
+    },
+    {
+      "epoch": 0.16179156120847082,
+      "grad_norm": 1.0220554585155903,
+      "learning_rate": 2e-05,
+      "loss": 0.6526585578918457,
+      "step": 5050
+    },
+    {
+      "epoch": 0.1621119405375965,
+      "grad_norm": 1.0280643496979298,
+      "learning_rate": 2e-05,
+      "loss": 0.6478411674499511,
+      "step": 5060
+    },
+    {
+      "epoch": 0.16243231986672219,
+      "grad_norm": 3.666383020262474,
+      "learning_rate": 2e-05,
+      "loss": 0.6606351375579834,
+      "step": 5070
+    },
+    {
+      "epoch": 0.1627526991958479,
+      "grad_norm": 1.0558288109614848,
+      "learning_rate": 2e-05,
+      "loss": 0.6425009727478027,
+      "step": 5080
+    },
+    {
+      "epoch": 0.16307307852497357,
+      "grad_norm": 2.035565603350421,
+      "learning_rate": 2e-05,
+      "loss": 0.662998104095459,
+      "step": 5090
+    },
+    {
+      "epoch": 0.16339345785409926,
+      "grad_norm": 1.2669494694291425,
+      "learning_rate": 2e-05,
+      "loss": 0.6467283248901368,
+      "step": 5100
+    },
+    {
+      "epoch": 0.16371383718322494,
+      "grad_norm": 1.7420496228128168,
+      "learning_rate": 2e-05,
+      "loss": 0.652040958404541,
+      "step": 5110
+    },
+    {
+      "epoch": 0.16403421651235062,
+      "grad_norm": 2.0540553864377267,
+      "learning_rate": 2e-05,
+      "loss": 0.6633444786071777,
+      "step": 5120
+    },
+    {
+      "epoch": 0.1643545958414763,
+      "grad_norm": 2.241855520752789,
+      "learning_rate": 2e-05,
+      "loss": 0.6187747478485107,
+      "step": 5130
+    },
+    {
+      "epoch": 0.164674975170602,
+      "grad_norm": 1.557351657361623,
+      "learning_rate": 2e-05,
+      "loss": 0.6467947959899902,
+      "step": 5140
+    },
+    {
+      "epoch": 0.1649953544997277,
+      "grad_norm": 1.1327302568986537,
+      "learning_rate": 2e-05,
+      "loss": 0.642011022567749,
+      "step": 5150
+    },
+    {
+      "epoch": 0.16531573382885337,
+      "grad_norm": 0.9588289163303403,
+      "learning_rate": 2e-05,
+      "loss": 0.6492002487182618,
+      "step": 5160
+    },
+    {
+      "epoch": 0.16563611315797905,
+      "grad_norm": 1.5616988846044089,
+      "learning_rate": 2e-05,
+      "loss": 0.6779866695404053,
+      "step": 5170
+    },
+    {
+      "epoch": 0.16595649248710473,
+      "grad_norm": 1.1384343274906892,
+      "learning_rate": 2e-05,
+      "loss": 0.6015952110290528,
+      "step": 5180
+    },
+    {
+      "epoch": 0.1662768718162304,
+      "grad_norm": 2.6329617869591178,
+      "learning_rate": 2e-05,
+      "loss": 0.6037326335906983,
+      "step": 5190
+    },
+    {
+      "epoch": 0.1665972511453561,
+      "grad_norm": 1.900594873065128,
+      "learning_rate": 2e-05,
+      "loss": 0.6537670135498047,
+      "step": 5200
+    },
+    {
+      "epoch": 0.1669176304744818,
+      "grad_norm": 22.015648278448364,
+      "learning_rate": 2e-05,
+      "loss": 0.7154106140136719,
+      "step": 5210
+    },
+    {
+      "epoch": 0.16723800980360748,
+      "grad_norm": 2.7548713628830668,
+      "learning_rate": 2e-05,
+      "loss": 0.675102949142456,
+      "step": 5220
+    },
+    {
+      "epoch": 0.16755838913273316,
+      "grad_norm": 2.3813129329457663,
+      "learning_rate": 2e-05,
+      "loss": 0.6831347465515136,
+      "step": 5230
+    },
+    {
+      "epoch": 0.16787876846185884,
+      "grad_norm": 0.9874414487742339,
+      "learning_rate": 2e-05,
+      "loss": 0.6257731437683105,
+      "step": 5240
+    },
+    {
+      "epoch": 0.16819914779098452,
+      "grad_norm": 1.4178526178245647,
+      "learning_rate": 2e-05,
+      "loss": 0.6240045070648194,
+      "step": 5250
+    },
+    {
+      "epoch": 0.1685195271201102,
+      "grad_norm": 2.450829823907873,
+      "learning_rate": 2e-05,
+      "loss": 0.6385599136352539,
+      "step": 5260
+    },
+    {
+      "epoch": 0.16883990644923588,
+      "grad_norm": 1.374226761063063,
+      "learning_rate": 2e-05,
+      "loss": 0.6375367164611816,
+      "step": 5270
+    },
+    {
+      "epoch": 0.1691602857783616,
+      "grad_norm": 5.242895247725081,
+      "learning_rate": 2e-05,
+      "loss": 0.7227782726287841,
+      "step": 5280
+    },
+    {
+      "epoch": 0.16948066510748727,
+      "grad_norm": 1.1799089686323416,
+      "learning_rate": 2e-05,
+      "loss": 0.6735242366790771,
+      "step": 5290
+    },
+    {
+      "epoch": 0.16980104443661295,
+      "grad_norm": 2.1663796442731655,
+      "learning_rate": 2e-05,
+      "loss": 0.6575541496276855,
+      "step": 5300
+    },
+    {
+      "epoch": 0.17012142376573863,
+      "grad_norm": 1.3419475057653019,
+      "learning_rate": 2e-05,
+      "loss": 0.6816549301147461,
+      "step": 5310
+    },
+    {
+      "epoch": 0.1704418030948643,
+      "grad_norm": 1.3945448266449434,
+      "learning_rate": 2e-05,
+      "loss": 0.6633468627929687,
+      "step": 5320
+    },
+    {
+      "epoch": 0.17076218242399,
+      "grad_norm": 1.402892779960778,
+      "learning_rate": 2e-05,
+      "loss": 0.6270906448364257,
+      "step": 5330
+    },
+    {
+      "epoch": 0.1710825617531157,
+      "grad_norm": 1.1948931691102345,
+      "learning_rate": 2e-05,
+      "loss": 0.6193373680114747,
+      "step": 5340
+    },
+    {
+      "epoch": 0.17140294108224138,
+      "grad_norm": 1.18675809400032,
+      "learning_rate": 1.999194520388485e-05,
+      "loss": 0.627802038192749,
+      "step": 5350
+    },
+    {
+      "epoch": 0.17172332041136706,
+      "grad_norm": 1.5713973510176555,
+      "learning_rate": 1.9959247303157763e-05,
+      "loss": 0.6522041320800781,
+      "step": 5360
+    },
+    {
+      "epoch": 0.17204369974049274,
+      "grad_norm": 41.45585361115185,
+      "learning_rate": 1.990149424459817e-05,
+      "loss": 0.6411617279052735,
+      "step": 5370
+    },
+    {
+      "epoch": 0.17236407906961843,
+      "grad_norm": 1.2202426199074823,
+      "learning_rate": 1.9818847538648688e-05,
+      "loss": 0.6249457359313965,
+      "step": 5380
+    },
+    {
+      "epoch": 0.1726844583987441,
+      "grad_norm": 1.5517715693701672,
+      "learning_rate": 1.971153831256721e-05,
+      "loss": 0.6244510650634766,
+      "step": 5390
+    },
+    {
+      "epoch": 0.1730048377278698,
+      "grad_norm": 3.6579219408868515,
+      "learning_rate": 1.9579866664063435e-05,
+      "loss": 0.6412816524505616,
+      "step": 5400
+    },
+    {
+      "epoch": 0.1733252170569955,
+      "grad_norm": 1.2756904564907214,
+      "learning_rate": 1.9424200822054782e-05,
+      "loss": 0.6284741401672364,
+      "step": 5410
+    },
+    {
+      "epoch": 0.17364559638612118,
+      "grad_norm": 29.021888813029918,
+      "learning_rate": 1.9244976116888626e-05,
+      "loss": 0.6464250564575196,
+      "step": 5420
+    },
+    {
+      "epoch": 0.17396597571524686,
+      "grad_norm": 1.2605703837916828,
+      "learning_rate": 1.904269376291071e-05,
+      "loss": 0.6683068752288819,
+      "step": 5430
+    },
+    {
+      "epoch": 0.17428635504437254,
+      "grad_norm": 2.468129242098605,
+      "learning_rate": 1.88179194567844e-05,
+      "loss": 0.6732934474945068,
+      "step": 5440
+    },
+    {
+      "epoch": 0.17460673437349822,
+      "grad_norm": 2.674033150088878,
+      "learning_rate": 1.8571281795480632e-05,
+      "loss": 0.6501801967620849,
+      "step": 5450
+    },
+    {
+      "epoch": 0.1749271137026239,
+      "grad_norm": 1.1817115126814564,
+      "learning_rate": 1.8303470518362808e-05,
+      "loss": 0.6125580787658691,
+      "step": 5460
+    },
+    {
+      "epoch": 0.17524749303174958,
+      "grad_norm": 3.915940908582353,
+      "learning_rate": 1.801523457828271e-05,
+      "loss": 0.6281828880310059,
+      "step": 5470
+    },
+    {
+      "epoch": 0.1755678723608753,
+      "grad_norm": 1.2661722408402352,
+      "learning_rate": 1.7707380047081796e-05,
+      "loss": 0.6136900901794433,
+      "step": 5480
+    },
+    {
+      "epoch": 0.17588825169000097,
+      "grad_norm": 0.946152528908013,
+      "learning_rate": 1.7380767861355252e-05,
+      "loss": 0.6320496559143066,
+      "step": 5490
+    },
+    {
+      "epoch": 0.17620863101912665,
+      "grad_norm": 17.60650673283214,
+      "learning_rate": 1.703631141478299e-05,
+      "loss": 0.6006783962249755,
+      "step": 5500
+    },
+    {
+      "epoch": 0.17652901034825233,
+      "grad_norm": 1.5147997239275703,
+      "learning_rate": 1.6674974003760705e-05,
+      "loss": 0.592758321762085,
+      "step": 5510
+    },
+    {
+      "epoch": 0.176849389677378,
+      "grad_norm": 2.47872432204946,
+      "learning_rate": 1.6297766133474556e-05,
+      "loss": 0.6384755611419678,
+      "step": 5520
+    },
+    {
+      "epoch": 0.1771697690065037,
+      "grad_norm": 0.9616006681699473,
+      "learning_rate": 1.5905742691953186e-05,
+      "loss": 0.6026595115661622,
+      "step": 5530
+    },
+    {
+      "epoch": 0.17749014833562937,
+      "grad_norm": 1.2612678281304546,
+      "learning_rate": 1.55e-05,
+      "loss": 0.6789834499359131,
+      "step": 5540
+    },
+    {
+      "epoch": 0.17781052766475508,
+      "grad_norm": 2.086017223568668,
+      "learning_rate": 1.5081672745255862e-05,
+      "loss": 0.6710739135742188,
+      "step": 5550
+    },
+    {
+      "epoch": 0.17813090699388076,
+      "grad_norm": 0.7800578252711731,
+      "learning_rate": 1.4651930808966206e-05,
+      "loss": 0.6025380611419677,
+      "step": 5560
+    },
+    {
+      "epoch": 0.17845128632300644,
+      "grad_norm": 0.9130160794118375,
+      "learning_rate": 1.4211975994326848e-05,
+      "loss": 0.680452299118042,
+      "step": 5570
+    },
+    {
+      "epoch": 0.17877166565213212,
+      "grad_norm": 10.367346484294696,
+      "learning_rate": 1.376303866555779e-05,
+      "loss": 0.6145600318908692,
+      "step": 5580
+    },
+    {
+      "epoch": 0.1790920449812578,
+      "grad_norm": 1.1591288637856112,
+      "learning_rate": 1.3306374307104164e-05,
+      "loss": 0.6312036514282227,
+      "step": 5590
+    },
+    {
+      "epoch": 0.17941242431038348,
+      "grad_norm": 1.294448822831085,
+      "learning_rate": 1.2843260012586718e-05,
+      "loss": 0.6485425472259522,
+      "step": 5600
+    },
+    {
+      "epoch": 0.1797328036395092,
+      "grad_norm": 1.0795543529658977,
+      "learning_rate": 1.2374990913320726e-05,
+      "loss": 0.6224565505981445,
+      "step": 5610
+    },
+    {
+      "epoch": 0.18005318296863487,
+      "grad_norm": 3.0766424272626947,
+      "learning_rate": 1.1902876556391247e-05,
+      "loss": 0.6365223884582519,
+      "step": 5620
+    },
+    {
+      "epoch": 0.18037356229776055,
+      "grad_norm": 0.8624865530291689,
+      "learning_rate": 1.1428237242413685e-05,
+      "loss": 0.5929572105407714,
+      "step": 5630
+    },
+    {
+      "epoch": 0.18069394162688623,
+      "grad_norm": 2.339708458164363,
+      "learning_rate": 1.095240033322132e-05,
+      "loss": 0.6204842567443848,
+      "step": 5640
+    },
+    {
+      "epoch": 0.18101432095601191,
+      "grad_norm": 1.025671467991908,
+      "learning_rate": 1.047669653980572e-05,
+      "loss": 0.5925289154052734,
+      "step": 5650
+    },
+    {
+      "epoch": 0.1813347002851376,
+      "grad_norm": 0.8516589988532776,
+      "learning_rate": 1.0002456200890902e-05,
+      "loss": 0.6281356334686279,
+      "step": 5660
+    },
+    {
+      "epoch": 0.18165507961426328,
+      "grad_norm": 1.402779311659343,
+      "learning_rate": 9.531005562548668e-06,
+      "loss": 0.6140302181243896,
+      "step": 5670
+    },
+    {
+      "epoch": 0.18197545894338898,
+      "grad_norm": 1.3160923613286275,
+      "learning_rate": 9.06366306925925e-06,
+      "loss": 0.6756707668304444,
+      "step": 5680
+    },
+    {
+      "epoch": 0.18229583827251467,
+      "grad_norm": 0.9742564917272837,
+      "learning_rate": 8.601735676789686e-06,
+      "loss": 0.6296013832092285,
+      "step": 5690
+    },
+    {
+      "epoch": 0.18261621760164035,
+      "grad_norm": 0.9153874280665963,
+      "learning_rate": 8.14651519720118e-06,
+      "loss": 0.6270411491394043,
+      "step": 5700
+    },
+    {
+      "epoch": 0.18293659693076603,
+      "grad_norm": 1.0535461876560501,
+      "learning_rate": 7.699274686206866e-06,
+      "loss": 0.6323529720306397,
+      "step": 5710
+    },
+    {
+      "epoch": 0.1832569762598917,
+      "grad_norm": 1.023457011630653,
+      "learning_rate": 7.261264882983024e-06,
+      "loss": 0.5944320678710937,
+      "step": 5720
+    },
+    {
+      "epoch": 0.1835773555890174,
+      "grad_norm": 0.9284587193795387,
+      "learning_rate": 6.8337107123900386e-06,
+      "loss": 0.6186958312988281,
+      "step": 5730
+    },
+    {
+      "epoch": 0.18389773491814307,
+      "grad_norm": 0.9641647776994864,
+      "learning_rate": 6.417807859384881e-06,
+      "loss": 0.6085702419281006,
+      "step": 5740
+    },
+    {
+      "epoch": 0.18421811424726878,
+      "grad_norm": 0.9922021501417239,
+      "learning_rate": 6.014719425205009e-06,
+      "loss": 0.625992727279663,
+      "step": 5750
+    },
+    {
+      "epoch": 0.18453849357639446,
+      "grad_norm": 0.8470281513133205,
+      "learning_rate": 5.6255726746749275e-06,
+      "loss": 0.6102766036987305,
+      "step": 5760
+    },
+    {
+      "epoch": 0.18485887290552014,
+      "grad_norm": 0.9620553369133289,
+      "learning_rate": 5.251455883731728e-06,
+      "loss": 0.6041369438171387,
+      "step": 5770
+    },
+    {
+      "epoch": 0.18517925223464582,
+      "grad_norm": 1.2069129310710403,
+      "learning_rate": 4.893415295985812e-06,
+      "loss": 0.5952677726745605,
+      "step": 5780
+    },
+    {
+      "epoch": 0.1854996315637715,
+      "grad_norm": 0.9295965424074922,
+      "learning_rate": 4.552452196827885e-06,
+      "loss": 0.6252847194671631,
+      "step": 5790
+    },
+    {
+      "epoch": 0.18582001089289718,
+      "grad_norm": 0.8672854001505957,
+      "learning_rate": 4.229520113264785e-06,
+      "loss": 0.6149491786956787,
+      "step": 5800
+    },
+    {
+      "epoch": 0.18614039022202286,
+      "grad_norm": 1.3649768723219773,
+      "learning_rate": 3.925522147314915e-06,
+      "loss": 0.6161471843719483,
+      "step": 5810
+    },
+    {
+      "epoch": 0.18646076955114857,
+      "grad_norm": 1.0317537906891645,
+      "learning_rate": 3.641308450420675e-06,
+      "loss": 0.5850464820861816,
+      "step": 5820
+    },
+    {
+      "epoch": 0.18678114888027425,
+      "grad_norm": 1.4397997700402234,
+      "learning_rate": 3.37767384594087e-06,
+      "loss": 0.6269999980926514,
+      "step": 5830
+    },
+    {
+      "epoch": 0.18710152820939993,
+      "grad_norm": 1.2664803684641663,
+      "learning_rate": 3.1353556063719357e-06,
+      "loss": 0.6332378387451172,
+      "step": 5840
+    },
+    {
+      "epoch": 0.1874219075385256,
+      "grad_norm": 1.0424305170614336,
+      "learning_rate": 2.9150313915141474e-06,
+      "loss": 0.5994098663330079,
+      "step": 5850
+    },
+    {
+      "epoch": 0.1877422868676513,
+      "grad_norm": 1.0068635714990892,
+      "learning_rate": 2.717317353348891e-06,
+      "loss": 0.593483304977417,
+      "step": 5860
+    },
+    {
+      "epoch": 0.18806266619677697,
+      "grad_norm": 1.0676291280310168,
+      "learning_rate": 2.5427664129268253e-06,
+      "loss": 0.5811244964599609,
+      "step": 5870
+    },
+    {
+      "epoch": 0.18838304552590268,
+      "grad_norm": 1.148575018287044,
+      "learning_rate": 2.391866714085717e-06,
+      "loss": 0.6025257110595703,
+      "step": 5880
+    },
+    {
+      "epoch": 0.18870342485502836,
+      "grad_norm": 0.873459017201487,
+      "learning_rate": 2.2650402583222753e-06,
+      "loss": 0.6060872554779053,
+      "step": 5890
+    },
+    {
+      "epoch": 0.18902380418415404,
+      "grad_norm": 5.56774114078358,
+      "learning_rate": 2.1626417246356398e-06,
+      "loss": 0.6008509635925293,
+      "step": 5900
+    },
+    {
+      "epoch": 0.18934418351327972,
+      "grad_norm": 0.8726206236052376,
+      "learning_rate": 2.084957477642894e-06,
+      "loss": 0.6007570266723633,
+      "step": 5910
+    },
+    {
+      "epoch": 0.1896645628424054,
+      "grad_norm": 1.2211667917785838,
+      "learning_rate": 2.0322047667405147e-06,
+      "loss": 0.6200607299804688,
+      "step": 5920
+    },
+    {
+      "epoch": 0.18998494217153108,
+      "grad_norm": 1.5703110754021823,
+      "learning_rate": 2.0045311185513342e-06,
+      "loss": 0.6589871406555176,
+      "step": 5930
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 5935,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 2035,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2810342056919040.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}