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": 0.3816965721268249,
+  "eval_steps": 500,
+  "global_step": 5935,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0006431281754453662,
+      "grad_norm": 264.8507844691241,
+      "learning_rate": 2.910112359550562e-06,
+      "loss": 1.4924835205078124,
+      "step": 10
+    },
+    {
+      "epoch": 0.0012862563508907324,
+      "grad_norm": 88.15687006273669,
+      "learning_rate": 3.921348314606742e-06,
+      "loss": 0.9952472686767578,
+      "step": 20
+    },
+    {
+      "epoch": 0.0019293845263360988,
+      "grad_norm": 21.15530249818695,
+      "learning_rate": 4.932584269662922e-06,
+      "loss": 0.889136791229248,
+      "step": 30
+    },
+    {
+      "epoch": 0.002572512701781465,
+      "grad_norm": 76.3435856512596,
+      "learning_rate": 5.943820224719102e-06,
+      "loss": 0.8875094413757324,
+      "step": 40
+    },
+    {
+      "epoch": 0.003215640877226831,
+      "grad_norm": 112.76291878162131,
+      "learning_rate": 6.955056179775282e-06,
+      "loss": 0.8803050994873047,
+      "step": 50
+    },
+    {
+      "epoch": 0.0038587690526721975,
+      "grad_norm": 63.07959530117864,
+      "learning_rate": 7.966292134831462e-06,
+      "loss": 0.8109166145324707,
+      "step": 60
+    },
+    {
+      "epoch": 0.004501897228117563,
+      "grad_norm": 32.36448507808977,
+      "learning_rate": 8.977528089887641e-06,
+      "loss": 0.8291963577270508,
+      "step": 70
+    },
+    {
+      "epoch": 0.00514502540356293,
+      "grad_norm": 39.66677708538367,
+      "learning_rate": 9.988764044943822e-06,
+      "loss": 0.8319297790527344,
+      "step": 80
+    },
+    {
+      "epoch": 0.005788153579008296,
+      "grad_norm": 15.73002157903958,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.8473474502563476,
+      "step": 90
+    },
+    {
+      "epoch": 0.006431281754453662,
+      "grad_norm": 29.04799205808129,
+      "learning_rate": 1.2011235955056182e-05,
+      "loss": 0.9536348342895508,
+      "step": 100
+    },
+    {
+      "epoch": 0.007074409929899029,
+      "grad_norm": 12.933556167568856,
+      "learning_rate": 1.3022471910112362e-05,
+      "loss": 0.8515434265136719,
+      "step": 110
+    },
+    {
+      "epoch": 0.007717538105344395,
+      "grad_norm": 13.216488664960004,
+      "learning_rate": 1.403370786516854e-05,
+      "loss": 0.8182208061218261,
+      "step": 120
+    },
+    {
+      "epoch": 0.008360666280789762,
+      "grad_norm": 12.840168166981233,
+      "learning_rate": 1.504494382022472e-05,
+      "loss": 0.8101629257202149,
+      "step": 130
+    },
+    {
+      "epoch": 0.009003794456235127,
+      "grad_norm": 16.647322053470525,
+      "learning_rate": 1.60561797752809e-05,
+      "loss": 0.7964733123779297,
+      "step": 140
+    },
+    {
+      "epoch": 0.009646922631680493,
+      "grad_norm": 8.26209213346245,
+      "learning_rate": 1.706741573033708e-05,
+      "loss": 0.8133392333984375,
+      "step": 150
+    },
+    {
+      "epoch": 0.01029005080712586,
+      "grad_norm": 103.09422370567049,
+      "learning_rate": 1.8078651685393256e-05,
+      "loss": 0.8019628524780273,
+      "step": 160
+    },
+    {
+      "epoch": 0.010933178982571226,
+      "grad_norm": 135.25179279143012,
+      "learning_rate": 1.908988764044944e-05,
+      "loss": 0.8667960166931152,
+      "step": 170
+    },
+    {
+      "epoch": 0.011576307158016592,
+      "grad_norm": 9.266415446110027,
+      "learning_rate": 2e-05,
+      "loss": 0.927884292602539,
+      "step": 180
+    },
+    {
+      "epoch": 0.012219435333461958,
+      "grad_norm": 19.616452125470513,
+      "learning_rate": 2e-05,
+      "loss": 0.9137126922607421,
+      "step": 190
+    },
+    {
+      "epoch": 0.012862563508907325,
+      "grad_norm": 9.521484959449038,
+      "learning_rate": 2e-05,
+      "loss": 0.9231036186218262,
+      "step": 200
+    },
+    {
+      "epoch": 0.013505691684352691,
+      "grad_norm": 3.841058878335388,
+      "learning_rate": 2e-05,
+      "loss": 0.880716609954834,
+      "step": 210
+    },
+    {
+      "epoch": 0.014148819859798057,
+      "grad_norm": 3.716737538053141,
+      "learning_rate": 2e-05,
+      "loss": 0.8603705406188965,
+      "step": 220
+    },
+    {
+      "epoch": 0.014791948035243424,
+      "grad_norm": 2.192635493713876,
+      "learning_rate": 2e-05,
+      "loss": 0.8351604461669921,
+      "step": 230
+    },
+    {
+      "epoch": 0.01543507621068879,
+      "grad_norm": 5.677647000147936,
+      "learning_rate": 2e-05,
+      "loss": 0.8132261276245117,
+      "step": 240
+    },
+    {
+      "epoch": 0.016078204386134158,
+      "grad_norm": 1.5356879007633566,
+      "learning_rate": 2e-05,
+      "loss": 0.8112307548522949,
+      "step": 250
+    },
+    {
+      "epoch": 0.016721332561579524,
+      "grad_norm": 1.751089237236811,
+      "learning_rate": 2e-05,
+      "loss": 0.7884010314941406,
+      "step": 260
+    },
+    {
+      "epoch": 0.01736446073702489,
+      "grad_norm": 3.480172236841913,
+      "learning_rate": 2e-05,
+      "loss": 0.7914746761322021,
+      "step": 270
+    },
+    {
+      "epoch": 0.018007588912470254,
+      "grad_norm": 2.0628477216575267,
+      "learning_rate": 2e-05,
+      "loss": 0.8090205192565918,
+      "step": 280
+    },
+    {
+      "epoch": 0.01865071708791562,
+      "grad_norm": 2.780990923043418,
+      "learning_rate": 2e-05,
+      "loss": 0.8231932640075683,
+      "step": 290
+    },
+    {
+      "epoch": 0.019293845263360986,
+      "grad_norm": 5.662473120221767,
+      "learning_rate": 2e-05,
+      "loss": 0.8489857673645019,
+      "step": 300
+    },
+    {
+      "epoch": 0.019936973438806353,
+      "grad_norm": 2.4856170219885727,
+      "learning_rate": 2e-05,
+      "loss": 0.8187873840332032,
+      "step": 310
+    },
+    {
+      "epoch": 0.02058010161425172,
+      "grad_norm": 1.8272788861417755,
+      "learning_rate": 2e-05,
+      "loss": 0.8340753555297852,
+      "step": 320
+    },
+    {
+      "epoch": 0.021223229789697085,
+      "grad_norm": 1.7623855646455684,
+      "learning_rate": 2e-05,
+      "loss": 0.7915143013000489,
+      "step": 330
+    },
+    {
+      "epoch": 0.02186635796514245,
+      "grad_norm": 1.9529760399471496,
+      "learning_rate": 2e-05,
+      "loss": 0.806821632385254,
+      "step": 340
+    },
+    {
+      "epoch": 0.022509486140587818,
+      "grad_norm": 3.908266701430372,
+      "learning_rate": 2e-05,
+      "loss": 0.7811341762542725,
+      "step": 350
+    },
+    {
+      "epoch": 0.023152614316033184,
+      "grad_norm": 6.946061799441667,
+      "learning_rate": 2e-05,
+      "loss": 0.8082803726196289,
+      "step": 360
+    },
+    {
+      "epoch": 0.02379574249147855,
+      "grad_norm": 4.948502227773737,
+      "learning_rate": 2e-05,
+      "loss": 0.8174592971801757,
+      "step": 370
+    },
+    {
+      "epoch": 0.024438870666923917,
+      "grad_norm": 10.39411913582579,
+      "learning_rate": 2e-05,
+      "loss": 0.7916179180145264,
+      "step": 380
+    },
+    {
+      "epoch": 0.025081998842369283,
+      "grad_norm": 2.157269114934209,
+      "learning_rate": 2e-05,
+      "loss": 0.8338122367858887,
+      "step": 390
+    },
+    {
+      "epoch": 0.02572512701781465,
+      "grad_norm": 5.266447865073407,
+      "learning_rate": 2e-05,
+      "loss": 0.827884578704834,
+      "step": 400
+    },
+    {
+      "epoch": 0.026368255193260016,
+      "grad_norm": 2.2186077419502372,
+      "learning_rate": 2e-05,
+      "loss": 0.811268424987793,
+      "step": 410
+    },
+    {
+      "epoch": 0.027011383368705382,
+      "grad_norm": 3.8503522479510415,
+      "learning_rate": 2e-05,
+      "loss": 0.7956947326660156,
+      "step": 420
+    },
+    {
+      "epoch": 0.02765451154415075,
+      "grad_norm": 4.682678958377554,
+      "learning_rate": 2e-05,
+      "loss": 0.823284912109375,
+      "step": 430
+    },
+    {
+      "epoch": 0.028297639719596115,
+      "grad_norm": 16.1664579170008,
+      "learning_rate": 2e-05,
+      "loss": 0.790956974029541,
+      "step": 440
+    },
+    {
+      "epoch": 0.02894076789504148,
+      "grad_norm": 9.145864327657529,
+      "learning_rate": 2e-05,
+      "loss": 0.8481451988220214,
+      "step": 450
+    },
+    {
+      "epoch": 0.029583896070486847,
+      "grad_norm": 9.984067276500117,
+      "learning_rate": 2e-05,
+      "loss": 0.8152891159057617,
+      "step": 460
+    },
+    {
+      "epoch": 0.030227024245932214,
+      "grad_norm": 7.200814844093958,
+      "learning_rate": 2e-05,
+      "loss": 0.8899850845336914,
+      "step": 470
+    },
+    {
+      "epoch": 0.03087015242137758,
+      "grad_norm": 4.2706772329543154,
+      "learning_rate": 2e-05,
+      "loss": 0.874454402923584,
+      "step": 480
+    },
+    {
+      "epoch": 0.03151328059682295,
+      "grad_norm": 1.8811353914333682,
+      "learning_rate": 2e-05,
+      "loss": 0.8309457778930665,
+      "step": 490
+    },
+    {
+      "epoch": 0.032156408772268316,
+      "grad_norm": 3.6385280608953763,
+      "learning_rate": 2e-05,
+      "loss": 0.8326511383056641,
+      "step": 500
+    },
+    {
+      "epoch": 0.03279953694771368,
+      "grad_norm": 1.3990140763602645,
+      "learning_rate": 2e-05,
+      "loss": 0.8103435516357422,
+      "step": 510
+    },
+    {
+      "epoch": 0.03344266512315905,
+      "grad_norm": 2.9528430814571807,
+      "learning_rate": 2e-05,
+      "loss": 0.8075934410095215,
+      "step": 520
+    },
+    {
+      "epoch": 0.034085793298604415,
+      "grad_norm": 1.7217483375429188,
+      "learning_rate": 2e-05,
+      "loss": 0.8287566184997559,
+      "step": 530
+    },
+    {
+      "epoch": 0.03472892147404978,
+      "grad_norm": 5.998438084861422,
+      "learning_rate": 2e-05,
+      "loss": 0.8061090469360351,
+      "step": 540
+    },
+    {
+      "epoch": 0.03537204964949515,
+      "grad_norm": 3.8694678808258725,
+      "learning_rate": 2e-05,
+      "loss": 0.8062751770019532,
+      "step": 550
+    },
+    {
+      "epoch": 0.03601517782494051,
+      "grad_norm": 2.9933818726818253,
+      "learning_rate": 2e-05,
+      "loss": 0.7965433120727539,
+      "step": 560
+    },
+    {
+      "epoch": 0.036658306000385874,
+      "grad_norm": 3.6993272754271627,
+      "learning_rate": 2e-05,
+      "loss": 0.8091463088989258,
+      "step": 570
+    },
+    {
+      "epoch": 0.03730143417583124,
+      "grad_norm": 1.261375926095548,
+      "learning_rate": 2e-05,
+      "loss": 0.7848992824554444,
+      "step": 580
+    },
+    {
+      "epoch": 0.037944562351276606,
+      "grad_norm": 1.2808782089307036,
+      "learning_rate": 2e-05,
+      "loss": 0.7949764251708984,
+      "step": 590
+    },
+    {
+      "epoch": 0.03858769052672197,
+      "grad_norm": 1.5363171733181304,
+      "learning_rate": 2e-05,
+      "loss": 0.7794778347015381,
+      "step": 600
+    },
+    {
+      "epoch": 0.03923081870216734,
+      "grad_norm": 1.692974926087992,
+      "learning_rate": 2e-05,
+      "loss": 0.7702068328857422,
+      "step": 610
+    },
+    {
+      "epoch": 0.039873946877612705,
+      "grad_norm": 2.212928194205319,
+      "learning_rate": 2e-05,
+      "loss": 0.8085812568664551,
+      "step": 620
+    },
+    {
+      "epoch": 0.04051707505305807,
+      "grad_norm": 1.2629369394284296,
+      "learning_rate": 2e-05,
+      "loss": 0.7806320190429688,
+      "step": 630
+    },
+    {
+      "epoch": 0.04116020322850344,
+      "grad_norm": 1.3484483120225834,
+      "learning_rate": 2e-05,
+      "loss": 0.8086847305297852,
+      "step": 640
+    },
+    {
+      "epoch": 0.041803331403948804,
+      "grad_norm": 3.702908463061738,
+      "learning_rate": 2e-05,
+      "loss": 0.7896449565887451,
+      "step": 650
+    },
+    {
+      "epoch": 0.04244645957939417,
+      "grad_norm": 6.029659806191006,
+      "learning_rate": 2e-05,
+      "loss": 0.8226722717285156,
+      "step": 660
+    },
+    {
+      "epoch": 0.04308958775483954,
+      "grad_norm": 7.370459449597133,
+      "learning_rate": 2e-05,
+      "loss": 0.7921708106994629,
+      "step": 670
+    },
+    {
+      "epoch": 0.0437327159302849,
+      "grad_norm": 2.7983562630986714,
+      "learning_rate": 2e-05,
+      "loss": 0.7904531478881835,
+      "step": 680
+    },
+    {
+      "epoch": 0.04437584410573027,
+      "grad_norm": 2.0098922533957477,
+      "learning_rate": 2e-05,
+      "loss": 0.7867809295654297,
+      "step": 690
+    },
+    {
+      "epoch": 0.045018972281175636,
+      "grad_norm": 5.897837867051966,
+      "learning_rate": 2e-05,
+      "loss": 0.758472204208374,
+      "step": 700
+    },
+    {
+      "epoch": 0.045662100456621,
+      "grad_norm": 3.7159389044440885,
+      "learning_rate": 2e-05,
+      "loss": 0.8133069992065429,
+      "step": 710
+    },
+    {
+      "epoch": 0.04630522863206637,
+      "grad_norm": 2.906657663938742,
+      "learning_rate": 2e-05,
+      "loss": 0.7981234550476074,
+      "step": 720
+    },
+    {
+      "epoch": 0.046948356807511735,
+      "grad_norm": 2.826396533617791,
+      "learning_rate": 2e-05,
+      "loss": 0.7824822902679444,
+      "step": 730
+    },
+    {
+      "epoch": 0.0475914849829571,
+      "grad_norm": 2.164462884511523,
+      "learning_rate": 2e-05,
+      "loss": 0.7825541973114014,
+      "step": 740
+    },
+    {
+      "epoch": 0.04823461315840247,
+      "grad_norm": 19.66583604658485,
+      "learning_rate": 2e-05,
+      "loss": 0.7757862091064454,
+      "step": 750
+    },
+    {
+      "epoch": 0.048877741333847834,
+      "grad_norm": 1.6390436767603138,
+      "learning_rate": 2e-05,
+      "loss": 0.7854133605957031,
+      "step": 760
+    },
+    {
+      "epoch": 0.0495208695092932,
+      "grad_norm": 4.120236130301234,
+      "learning_rate": 2e-05,
+      "loss": 0.7875580787658691,
+      "step": 770
+    },
+    {
+      "epoch": 0.050163997684738566,
+      "grad_norm": 1.6380084823946195,
+      "learning_rate": 2e-05,
+      "loss": 0.7962754726409912,
+      "step": 780
+    },
+    {
+      "epoch": 0.05080712586018393,
+      "grad_norm": 2.6234563124222348,
+      "learning_rate": 2e-05,
+      "loss": 0.787530517578125,
+      "step": 790
+    },
+    {
+      "epoch": 0.0514502540356293,
+      "grad_norm": 4.114884694757611,
+      "learning_rate": 2e-05,
+      "loss": 0.805137825012207,
+      "step": 800
+    },
+    {
+      "epoch": 0.052093382211074665,
+      "grad_norm": 2.170132077686732,
+      "learning_rate": 2e-05,
+      "loss": 0.8011256217956543,
+      "step": 810
+    },
+    {
+      "epoch": 0.05273651038652003,
+      "grad_norm": 2.9542903597482884,
+      "learning_rate": 2e-05,
+      "loss": 0.7881402492523193,
+      "step": 820
+    },
+    {
+      "epoch": 0.0533796385619654,
+      "grad_norm": 6.152172836766025,
+      "learning_rate": 2e-05,
+      "loss": 0.8165565490722656,
+      "step": 830
+    },
+    {
+      "epoch": 0.054022766737410764,
+      "grad_norm": 2.6587595790131235,
+      "learning_rate": 2e-05,
+      "loss": 0.8002906799316406,
+      "step": 840
+    },
+    {
+      "epoch": 0.05466589491285613,
+      "grad_norm": 15.727752869746313,
+      "learning_rate": 2e-05,
+      "loss": 0.7681012153625488,
+      "step": 850
+    },
+    {
+      "epoch": 0.0553090230883015,
+      "grad_norm": 3.2974201088925668,
+      "learning_rate": 2e-05,
+      "loss": 0.7994019985198975,
+      "step": 860
+    },
+    {
+      "epoch": 0.05595215126374686,
+      "grad_norm": 2.6425295630258514,
+      "learning_rate": 2e-05,
+      "loss": 0.8333459854125976,
+      "step": 870
+    },
+    {
+      "epoch": 0.05659527943919223,
+      "grad_norm": 4.544009663459626,
+      "learning_rate": 2e-05,
+      "loss": 0.817692756652832,
+      "step": 880
+    },
+    {
+      "epoch": 0.057238407614637596,
+      "grad_norm": 5.041701731508353,
+      "learning_rate": 2e-05,
+      "loss": 0.8127239227294922,
+      "step": 890
+    },
+    {
+      "epoch": 0.05788153579008296,
+      "grad_norm": 14.446895711089793,
+      "learning_rate": 2e-05,
+      "loss": 0.7857817649841309,
+      "step": 900
+    },
+    {
+      "epoch": 0.05852466396552833,
+      "grad_norm": 4.651863422153641,
+      "learning_rate": 2e-05,
+      "loss": 0.7850156307220459,
+      "step": 910
+    },
+    {
+      "epoch": 0.059167792140973695,
+      "grad_norm": 13.407365054787249,
+      "learning_rate": 2e-05,
+      "loss": 0.7831117153167725,
+      "step": 920
+    },
+    {
+      "epoch": 0.05981092031641906,
+      "grad_norm": 4.625228730773092,
+      "learning_rate": 2e-05,
+      "loss": 0.8246305465698243,
+      "step": 930
+    },
+    {
+      "epoch": 0.06045404849186443,
+      "grad_norm": 2.6404931195979318,
+      "learning_rate": 2e-05,
+      "loss": 0.8151409149169921,
+      "step": 940
+    },
+    {
+      "epoch": 0.061097176667309794,
+      "grad_norm": 2.8606521849156867,
+      "learning_rate": 2e-05,
+      "loss": 0.803957748413086,
+      "step": 950
+    },
+    {
+      "epoch": 0.06174030484275516,
+      "grad_norm": 3.8845917481687176,
+      "learning_rate": 2e-05,
+      "loss": 0.7963566780090332,
+      "step": 960
+    },
+    {
+      "epoch": 0.062383433018200526,
+      "grad_norm": 6.292344120130513,
+      "learning_rate": 2e-05,
+      "loss": 0.7799639701843262,
+      "step": 970
+    },
+    {
+      "epoch": 0.0630265611936459,
+      "grad_norm": 3.7170800116374756,
+      "learning_rate": 2e-05,
+      "loss": 0.789764404296875,
+      "step": 980
+    },
+    {
+      "epoch": 0.06366968936909126,
+      "grad_norm": 19.199030147508488,
+      "learning_rate": 2e-05,
+      "loss": 0.8285154342651367,
+      "step": 990
+    },
+    {
+      "epoch": 0.06431281754453663,
+      "grad_norm": 3.928846417575705,
+      "learning_rate": 2e-05,
+      "loss": 0.8127799034118652,
+      "step": 1000
+    },
+    {
+      "epoch": 0.06495594571998199,
+      "grad_norm": 3.755271687428537,
+      "learning_rate": 2e-05,
+      "loss": 0.8010581970214844,
+      "step": 1010
+    },
+    {
+      "epoch": 0.06559907389542736,
+      "grad_norm": 2.913559778112332,
+      "learning_rate": 2e-05,
+      "loss": 0.7750132560729981,
+      "step": 1020
+    },
+    {
+      "epoch": 0.06624220207087272,
+      "grad_norm": 10.071576934081813,
+      "learning_rate": 2e-05,
+      "loss": 0.7811709403991699,
+      "step": 1030
+    },
+    {
+      "epoch": 0.0668853302463181,
+      "grad_norm": 3.587970695691555,
+      "learning_rate": 2e-05,
+      "loss": 0.7697386741638184,
+      "step": 1040
+    },
+    {
+      "epoch": 0.06752845842176346,
+      "grad_norm": 1.862423666481207,
+      "learning_rate": 2e-05,
+      "loss": 0.7978693962097168,
+      "step": 1050
+    },
+    {
+      "epoch": 0.06817158659720883,
+      "grad_norm": 2.9439913752393987,
+      "learning_rate": 2e-05,
+      "loss": 0.7703286647796631,
+      "step": 1060
+    },
+    {
+      "epoch": 0.06881471477265419,
+      "grad_norm": 5.602785502496395,
+      "learning_rate": 2e-05,
+      "loss": 0.7832149505615235,
+      "step": 1070
+    },
+    {
+      "epoch": 0.06945784294809956,
+      "grad_norm": 6.444227507124694,
+      "learning_rate": 2e-05,
+      "loss": 0.7846955776214599,
+      "step": 1080
+    },
+    {
+      "epoch": 0.07010097112354492,
+      "grad_norm": 8.49026324059516,
+      "learning_rate": 2e-05,
+      "loss": 0.8015947341918945,
+      "step": 1090
+    },
+    {
+      "epoch": 0.0707440992989903,
+      "grad_norm": 4.090223667504171,
+      "learning_rate": 2e-05,
+      "loss": 0.8186767578125,
+      "step": 1100
+    },
+    {
+      "epoch": 0.07138722747443565,
+      "grad_norm": 2.7765862343897587,
+      "learning_rate": 2e-05,
+      "loss": 0.7696638107299805,
+      "step": 1110
+    },
+    {
+      "epoch": 0.07203035564988101,
+      "grad_norm": 13.741738631565209,
+      "learning_rate": 2e-05,
+      "loss": 0.8072344779968261,
+      "step": 1120
+    },
+    {
+      "epoch": 0.07267348382532639,
+      "grad_norm": 4.540314995023613,
+      "learning_rate": 2e-05,
+      "loss": 0.8362739562988282,
+      "step": 1130
+    },
+    {
+      "epoch": 0.07331661200077175,
+      "grad_norm": 1.9835520642143039,
+      "learning_rate": 2e-05,
+      "loss": 0.7769340515136719,
+      "step": 1140
+    },
+    {
+      "epoch": 0.07395974017621712,
+      "grad_norm": 11.174659008141319,
+      "learning_rate": 2e-05,
+      "loss": 0.7873312950134277,
+      "step": 1150
+    },
+    {
+      "epoch": 0.07460286835166248,
+      "grad_norm": 35.01627971607914,
+      "learning_rate": 2e-05,
+      "loss": 0.8206969261169433,
+      "step": 1160
+    },
+    {
+      "epoch": 0.07524599652710785,
+      "grad_norm": 2.8190948682505033,
+      "learning_rate": 2e-05,
+      "loss": 0.8351315498352051,
+      "step": 1170
+    },
+    {
+      "epoch": 0.07588912470255321,
+      "grad_norm": 2.060661918032204,
+      "learning_rate": 2e-05,
+      "loss": 0.8220281600952148,
+      "step": 1180
+    },
+    {
+      "epoch": 0.07653225287799859,
+      "grad_norm": 1.6772537547463753,
+      "learning_rate": 2e-05,
+      "loss": 0.7766014099121094,
+      "step": 1190
+    },
+    {
+      "epoch": 0.07717538105344394,
+      "grad_norm": 1.7149818496545355,
+      "learning_rate": 2e-05,
+      "loss": 0.7847399711608887,
+      "step": 1200
+    },
+    {
+      "epoch": 0.07781850922888932,
+      "grad_norm": 1.842032079580061,
+      "learning_rate": 2e-05,
+      "loss": 0.7880351543426514,
+      "step": 1210
+    },
+    {
+      "epoch": 0.07846163740433468,
+      "grad_norm": 7.628369058899049,
+      "learning_rate": 2e-05,
+      "loss": 0.8086511611938476,
+      "step": 1220
+    },
+    {
+      "epoch": 0.07910476557978005,
+      "grad_norm": 3.8447978426923006,
+      "learning_rate": 2e-05,
+      "loss": 0.7901057243347168,
+      "step": 1230
+    },
+    {
+      "epoch": 0.07974789375522541,
+      "grad_norm": 5.063451031975492,
+      "learning_rate": 2e-05,
+      "loss": 0.783748197555542,
+      "step": 1240
+    },
+    {
+      "epoch": 0.08039102193067078,
+      "grad_norm": 3.751566360170122,
+      "learning_rate": 2e-05,
+      "loss": 0.7723451614379883,
+      "step": 1250
+    },
+    {
+      "epoch": 0.08103415010611614,
+      "grad_norm": 3.2156103176700626,
+      "learning_rate": 2e-05,
+      "loss": 0.7751105785369873,
+      "step": 1260
+    },
+    {
+      "epoch": 0.08167727828156152,
+      "grad_norm": 2.6025577293652815,
+      "learning_rate": 2e-05,
+      "loss": 0.7603851318359375,
+      "step": 1270
+    },
+    {
+      "epoch": 0.08232040645700688,
+      "grad_norm": 1.3919904079149132,
+      "learning_rate": 2e-05,
+      "loss": 0.7791646957397461,
+      "step": 1280
+    },
+    {
+      "epoch": 0.08296353463245225,
+      "grad_norm": 1.672301310016778,
+      "learning_rate": 2e-05,
+      "loss": 0.7866832733154296,
+      "step": 1290
+    },
+    {
+      "epoch": 0.08360666280789761,
+      "grad_norm": 1.7978359862977575,
+      "learning_rate": 2e-05,
+      "loss": 0.7824535369873047,
+      "step": 1300
+    },
+    {
+      "epoch": 0.08424979098334298,
+      "grad_norm": 1.164265222279358,
+      "learning_rate": 2e-05,
+      "loss": 0.757862663269043,
+      "step": 1310
+    },
+    {
+      "epoch": 0.08489291915878834,
+      "grad_norm": 6.39635222930186,
+      "learning_rate": 2e-05,
+      "loss": 0.7761935234069824,
+      "step": 1320
+    },
+    {
+      "epoch": 0.08553604733423371,
+      "grad_norm": 1.5835398014458386,
+      "learning_rate": 2e-05,
+      "loss": 0.7623611450195312,
+      "step": 1330
+    },
+    {
+      "epoch": 0.08617917550967907,
+      "grad_norm": 1.9528892284475325,
+      "learning_rate": 2e-05,
+      "loss": 0.7652930736541748,
+      "step": 1340
+    },
+    {
+      "epoch": 0.08682230368512445,
+      "grad_norm": 1.170982466023762,
+      "learning_rate": 2e-05,
+      "loss": 0.7785554885864258,
+      "step": 1350
+    },
+    {
+      "epoch": 0.0874654318605698,
+      "grad_norm": 1.3315886378379231,
+      "learning_rate": 2e-05,
+      "loss": 0.7762284755706788,
+      "step": 1360
+    },
+    {
+      "epoch": 0.08810856003601518,
+      "grad_norm": 3.4644691896641033,
+      "learning_rate": 2e-05,
+      "loss": 0.7717214107513428,
+      "step": 1370
+    },
+    {
+      "epoch": 0.08875168821146054,
+      "grad_norm": 2.4638970172558263,
+      "learning_rate": 2e-05,
+      "loss": 0.7678469657897949,
+      "step": 1380
+    },
+    {
+      "epoch": 0.08939481638690591,
+      "grad_norm": 1.6678660828792848,
+      "learning_rate": 2e-05,
+      "loss": 0.7703881740570069,
+      "step": 1390
+    },
+    {
+      "epoch": 0.09003794456235127,
+      "grad_norm": 1.2703517703114975,
+      "learning_rate": 2e-05,
+      "loss": 0.7905326843261719,
+      "step": 1400
+    },
+    {
+      "epoch": 0.09068107273779664,
+      "grad_norm": 2.7492971396954964,
+      "learning_rate": 2e-05,
+      "loss": 0.7528151512145996,
+      "step": 1410
+    },
+    {
+      "epoch": 0.091324200913242,
+      "grad_norm": 1.85122344862107,
+      "learning_rate": 2e-05,
+      "loss": 0.7600863456726075,
+      "step": 1420
+    },
+    {
+      "epoch": 0.09196732908868738,
+      "grad_norm": 1.121790046409357,
+      "learning_rate": 2e-05,
+      "loss": 0.771851921081543,
+      "step": 1430
+    },
+    {
+      "epoch": 0.09261045726413274,
+      "grad_norm": 1.5662881662547643,
+      "learning_rate": 2e-05,
+      "loss": 0.783432149887085,
+      "step": 1440
+    },
+    {
+      "epoch": 0.09325358543957811,
+      "grad_norm": 2.0639305477971446,
+      "learning_rate": 2e-05,
+      "loss": 0.76649169921875,
+      "step": 1450
+    },
+    {
+      "epoch": 0.09389671361502347,
+      "grad_norm": 1.6044380184600715,
+      "learning_rate": 2e-05,
+      "loss": 0.7608757972717285,
+      "step": 1460
+    },
+    {
+      "epoch": 0.09453984179046884,
+      "grad_norm": 1.1572139692941597,
+      "learning_rate": 2e-05,
+      "loss": 0.7706125259399415,
+      "step": 1470
+    },
+    {
+      "epoch": 0.0951829699659142,
+      "grad_norm": 1.8138858808035307,
+      "learning_rate": 2e-05,
+      "loss": 0.7700668334960937,
+      "step": 1480
+    },
+    {
+      "epoch": 0.09582609814135958,
+      "grad_norm": 1.454983551543918,
+      "learning_rate": 2e-05,
+      "loss": 0.7904407501220703,
+      "step": 1490
+    },
+    {
+      "epoch": 0.09646922631680493,
+      "grad_norm": 2.0901094734966796,
+      "learning_rate": 2e-05,
+      "loss": 0.7803089141845703,
+      "step": 1500
+    },
+    {
+      "epoch": 0.09711235449225031,
+      "grad_norm": 2.567975717980828,
+      "learning_rate": 2e-05,
+      "loss": 0.7728228569030762,
+      "step": 1510
+    },
+    {
+      "epoch": 0.09775548266769567,
+      "grad_norm": 1.780447312937554,
+      "learning_rate": 2e-05,
+      "loss": 0.8009782791137695,
+      "step": 1520
+    },
+    {
+      "epoch": 0.09839861084314104,
+      "grad_norm": 6.2870046446647,
+      "learning_rate": 2e-05,
+      "loss": 0.7996140480041504,
+      "step": 1530
+    },
+    {
+      "epoch": 0.0990417390185864,
+      "grad_norm": 2.156771416116291,
+      "learning_rate": 2e-05,
+      "loss": 0.7783120632171631,
+      "step": 1540
+    },
+    {
+      "epoch": 0.09968486719403177,
+      "grad_norm": 1.2275704688595095,
+      "learning_rate": 2e-05,
+      "loss": 0.7634446620941162,
+      "step": 1550
+    },
+    {
+      "epoch": 0.10032799536947713,
+      "grad_norm": 1.793131326578086,
+      "learning_rate": 2e-05,
+      "loss": 0.7803597927093506,
+      "step": 1560
+    },
+    {
+      "epoch": 0.1009711235449225,
+      "grad_norm": 1.2370786940553424,
+      "learning_rate": 2e-05,
+      "loss": 0.7557382106781005,
+      "step": 1570
+    },
+    {
+      "epoch": 0.10161425172036787,
+      "grad_norm": 2.2314399766443676,
+      "learning_rate": 2e-05,
+      "loss": 0.7821496486663818,
+      "step": 1580
+    },
+    {
+      "epoch": 0.10225737989581324,
+      "grad_norm": 2.3186973460423723,
+      "learning_rate": 2e-05,
+      "loss": 0.7797545433044434,
+      "step": 1590
+    },
+    {
+      "epoch": 0.1029005080712586,
+      "grad_norm": 1.462537448976435,
+      "learning_rate": 2e-05,
+      "loss": 0.7844183921813965,
+      "step": 1600
+    },
+    {
+      "epoch": 0.10354363624670397,
+      "grad_norm": 1.5205367367748963,
+      "learning_rate": 2e-05,
+      "loss": 0.7902626037597656,
+      "step": 1610
+    },
+    {
+      "epoch": 0.10418676442214933,
+      "grad_norm": 3.5432682320386655,
+      "learning_rate": 2e-05,
+      "loss": 0.7723273277282715,
+      "step": 1620
+    },
+    {
+      "epoch": 0.1048298925975947,
+      "grad_norm": 1.609507484917306,
+      "learning_rate": 2e-05,
+      "loss": 0.7709271430969238,
+      "step": 1630
+    },
+    {
+      "epoch": 0.10547302077304006,
+      "grad_norm": 1.1002224278546247,
+      "learning_rate": 2e-05,
+      "loss": 0.7905303955078125,
+      "step": 1640
+    },
+    {
+      "epoch": 0.10611614894848544,
+      "grad_norm": 2.884940398494792,
+      "learning_rate": 2e-05,
+      "loss": 0.7871637344360352,
+      "step": 1650
+    },
+    {
+      "epoch": 0.1067592771239308,
+      "grad_norm": 1.2020747081839562,
+      "learning_rate": 2e-05,
+      "loss": 0.7405866146087646,
+      "step": 1660
+    },
+    {
+      "epoch": 0.10740240529937617,
+      "grad_norm": 6.606063972824954,
+      "learning_rate": 2e-05,
+      "loss": 0.7357051849365235,
+      "step": 1670
+    },
+    {
+      "epoch": 0.10804553347482153,
+      "grad_norm": 3.5578842158542865,
+      "learning_rate": 2e-05,
+      "loss": 0.7682039260864257,
+      "step": 1680
+    },
+    {
+      "epoch": 0.1086886616502669,
+      "grad_norm": 1.3072327175477532,
+      "learning_rate": 2e-05,
+      "loss": 0.7714582443237304,
+      "step": 1690
+    },
+    {
+      "epoch": 0.10933178982571226,
+      "grad_norm": 1.3765783821951507,
+      "learning_rate": 2e-05,
+      "loss": 0.7503826141357421,
+      "step": 1700
+    },
+    {
+      "epoch": 0.10997491800115763,
+      "grad_norm": 4.459887753306068,
+      "learning_rate": 2e-05,
+      "loss": 0.7501479148864746,
+      "step": 1710
+    },
+    {
+      "epoch": 0.110618046176603,
+      "grad_norm": 2.70029059251659,
+      "learning_rate": 2e-05,
+      "loss": 0.8093817710876465,
+      "step": 1720
+    },
+    {
+      "epoch": 0.11126117435204837,
+      "grad_norm": 1.7062222278246155,
+      "learning_rate": 2e-05,
+      "loss": 0.7766101360321045,
+      "step": 1730
+    },
+    {
+      "epoch": 0.11190430252749373,
+      "grad_norm": 1.328579668130039,
+      "learning_rate": 2e-05,
+      "loss": 0.7427976608276368,
+      "step": 1740
+    },
+    {
+      "epoch": 0.1125474307029391,
+      "grad_norm": 4.778582308508733,
+      "learning_rate": 2e-05,
+      "loss": 0.781195592880249,
+      "step": 1750
+    },
+    {
+      "epoch": 0.11319055887838446,
+      "grad_norm": 1.9035347037479093,
+      "learning_rate": 2e-05,
+      "loss": 0.7611964702606201,
+      "step": 1760
+    },
+    {
+      "epoch": 0.11383368705382983,
+      "grad_norm": 1.869590411200461,
+      "learning_rate": 2e-05,
+      "loss": 0.7906878471374512,
+      "step": 1770
+    },
+    {
+      "epoch": 0.11447681522927519,
+      "grad_norm": 1.551688435499546,
+      "learning_rate": 2e-05,
+      "loss": 0.7423989295959472,
+      "step": 1780
+    },
+    {
+      "epoch": 0.11511994340472056,
+      "grad_norm": 5.098013550894126,
+      "learning_rate": 2e-05,
+      "loss": 0.7568299293518066,
+      "step": 1790
+    },
+    {
+      "epoch": 0.11576307158016592,
+      "grad_norm": 2.5007704043329495,
+      "learning_rate": 2e-05,
+      "loss": 0.7778056621551513,
+      "step": 1800
+    },
+    {
+      "epoch": 0.1164061997556113,
+      "grad_norm": 1.52922937116661,
+      "learning_rate": 2e-05,
+      "loss": 0.799095630645752,
+      "step": 1810
+    },
+    {
+      "epoch": 0.11704932793105666,
+      "grad_norm": 18.126018555580437,
+      "learning_rate": 2e-05,
+      "loss": 0.7876038551330566,
+      "step": 1820
+    },
+    {
+      "epoch": 0.11769245610650203,
+      "grad_norm": 1.282142333991493,
+      "learning_rate": 2e-05,
+      "loss": 0.7696383476257325,
+      "step": 1830
+    },
+    {
+      "epoch": 0.11833558428194739,
+      "grad_norm": 1.7962688003308664,
+      "learning_rate": 2e-05,
+      "loss": 0.7794070720672608,
+      "step": 1840
+    },
+    {
+      "epoch": 0.11897871245739276,
+      "grad_norm": 1.0874142410284557,
+      "learning_rate": 2e-05,
+      "loss": 0.768028450012207,
+      "step": 1850
+    },
+    {
+      "epoch": 0.11962184063283812,
+      "grad_norm": 1.2554969962658076,
+      "learning_rate": 2e-05,
+      "loss": 0.7807290077209472,
+      "step": 1860
+    },
+    {
+      "epoch": 0.1202649688082835,
+      "grad_norm": 1.503617789481177,
+      "learning_rate": 2e-05,
+      "loss": 0.7617989540100097,
+      "step": 1870
+    },
+    {
+      "epoch": 0.12090809698372885,
+      "grad_norm": 1.3236598847407397,
+      "learning_rate": 2e-05,
+      "loss": 0.7619621276855468,
+      "step": 1880
+    },
+    {
+      "epoch": 0.12155122515917423,
+      "grad_norm": 1.3278805791234929,
+      "learning_rate": 2e-05,
+      "loss": 0.7731639862060546,
+      "step": 1890
+    },
+    {
+      "epoch": 0.12219435333461959,
+      "grad_norm": 1.3336901969204897,
+      "learning_rate": 2e-05,
+      "loss": 0.7973304748535156,
+      "step": 1900
+    },
+    {
+      "epoch": 0.12283748151006496,
+      "grad_norm": 1.2081432983403761,
+      "learning_rate": 2e-05,
+      "loss": 0.732518196105957,
+      "step": 1910
+    },
+    {
+      "epoch": 0.12348060968551032,
+      "grad_norm": 1.0660544450914864,
+      "learning_rate": 2e-05,
+      "loss": 0.7558794975280761,
+      "step": 1920
+    },
+    {
+      "epoch": 0.1241237378609557,
+      "grad_norm": 1.073493051686456,
+      "learning_rate": 2e-05,
+      "loss": 0.770344877243042,
+      "step": 1930
+    },
+    {
+      "epoch": 0.12476686603640105,
+      "grad_norm": 1.0266721503860259,
+      "learning_rate": 2e-05,
+      "loss": 0.7605429649353027,
+      "step": 1940
+    },
+    {
+      "epoch": 0.12540999421184643,
+      "grad_norm": 1.6219397326505027,
+      "learning_rate": 2e-05,
+      "loss": 0.7637803077697753,
+      "step": 1950
+    },
+    {
+      "epoch": 0.1260531223872918,
+      "grad_norm": 1.319161229050353,
+      "learning_rate": 2e-05,
+      "loss": 0.7486878871917725,
+      "step": 1960
+    },
+    {
+      "epoch": 0.12669625056273714,
+      "grad_norm": 1.4547728573258885,
+      "learning_rate": 2e-05,
+      "loss": 0.7947936534881592,
+      "step": 1970
+    },
+    {
+      "epoch": 0.12733937873818252,
+      "grad_norm": 3.0621373300129897,
+      "learning_rate": 2e-05,
+      "loss": 0.7795517921447754,
+      "step": 1980
+    },
+    {
+      "epoch": 0.1279825069136279,
+      "grad_norm": 1.288805293411811,
+      "learning_rate": 2e-05,
+      "loss": 0.7543374538421631,
+      "step": 1990
+    },
+    {
+      "epoch": 0.12862563508907326,
+      "grad_norm": 1.8078110077721685,
+      "learning_rate": 2e-05,
+      "loss": 0.7700879096984863,
+      "step": 2000
+    },
+    {
+      "epoch": 0.1292687632645186,
+      "grad_norm": 1.5167799524227923,
+      "learning_rate": 2e-05,
+      "loss": 0.756030797958374,
+      "step": 2010
+    },
+    {
+      "epoch": 0.12991189143996398,
+      "grad_norm": 5.4727720008288365,
+      "learning_rate": 2e-05,
+      "loss": 0.7306814193725586,
+      "step": 2020
+    },
+    {
+      "epoch": 0.13055501961540936,
+      "grad_norm": 1.8434468283827343,
+      "learning_rate": 2e-05,
+      "loss": 0.7443379878997802,
+      "step": 2030
+    },
+    {
+      "epoch": 0.13119814779085473,
+      "grad_norm": 5.758844487891117,
+      "learning_rate": 2e-05,
+      "loss": 0.7579936981201172,
+      "step": 2040
+    },
+    {
+      "epoch": 0.13184127596630008,
+      "grad_norm": 2.036795752708532,
+      "learning_rate": 2e-05,
+      "loss": 0.7929269790649414,
+      "step": 2050
+    },
+    {
+      "epoch": 0.13248440414174545,
+      "grad_norm": 11.047365760715172,
+      "learning_rate": 2e-05,
+      "loss": 0.775825309753418,
+      "step": 2060
+    },
+    {
+      "epoch": 0.13312753231719082,
+      "grad_norm": 3.0068995357073787,
+      "learning_rate": 2e-05,
+      "loss": 0.7827195644378662,
+      "step": 2070
+    },
+    {
+      "epoch": 0.1337706604926362,
+      "grad_norm": 2.580647228014161,
+      "learning_rate": 2e-05,
+      "loss": 0.7702327251434327,
+      "step": 2080
+    },
+    {
+      "epoch": 0.13441378866808154,
+      "grad_norm": 2.838906857307913,
+      "learning_rate": 2e-05,
+      "loss": 0.737979793548584,
+      "step": 2090
+    },
+    {
+      "epoch": 0.13505691684352691,
+      "grad_norm": 1.8141311666853606,
+      "learning_rate": 2e-05,
+      "loss": 0.7899340629577637,
+      "step": 2100
+    },
+    {
+      "epoch": 0.1357000450189723,
+      "grad_norm": 1.3333709207185118,
+      "learning_rate": 2e-05,
+      "loss": 0.7572295188903808,
+      "step": 2110
+    },
+    {
+      "epoch": 0.13634317319441766,
+      "grad_norm": 1.439515840865014,
+      "learning_rate": 2e-05,
+      "loss": 0.7655848503112793,
+      "step": 2120
+    },
+    {
+      "epoch": 0.136986301369863,
+      "grad_norm": 2.488028952759784,
+      "learning_rate": 2e-05,
+      "loss": 0.7476516723632812,
+      "step": 2130
+    },
+    {
+      "epoch": 0.13762942954530838,
+      "grad_norm": 2.35457054677663,
+      "learning_rate": 2e-05,
+      "loss": 0.7362603664398193,
+      "step": 2140
+    },
+    {
+      "epoch": 0.13827255772075375,
+      "grad_norm": 6.028666338812484,
+      "learning_rate": 2e-05,
+      "loss": 0.754909086227417,
+      "step": 2150
+    },
+    {
+      "epoch": 0.13891568589619913,
+      "grad_norm": 3.272425021648325,
+      "learning_rate": 2e-05,
+      "loss": 0.8009754180908203,
+      "step": 2160
+    },
+    {
+      "epoch": 0.13955881407164447,
+      "grad_norm": 19.02083054528344,
+      "learning_rate": 2e-05,
+      "loss": 0.8124250411987305,
+      "step": 2170
+    },
+    {
+      "epoch": 0.14020194224708984,
+      "grad_norm": 1.927417304595474,
+      "learning_rate": 2e-05,
+      "loss": 0.8851755142211915,
+      "step": 2180
+    },
+    {
+      "epoch": 0.14084507042253522,
+      "grad_norm": 3.8177334557670815,
+      "learning_rate": 2e-05,
+      "loss": 0.8132132530212403,
+      "step": 2190
+    },
+    {
+      "epoch": 0.1414881985979806,
+      "grad_norm": 1.8934981816592293,
+      "learning_rate": 2e-05,
+      "loss": 0.7725329399108887,
+      "step": 2200
+    },
+    {
+      "epoch": 0.14213132677342594,
+      "grad_norm": 1.6465630540315317,
+      "learning_rate": 2e-05,
+      "loss": 0.7803558349609375,
+      "step": 2210
+    },
+    {
+      "epoch": 0.1427744549488713,
+      "grad_norm": 3.0986858859551707,
+      "learning_rate": 2e-05,
+      "loss": 0.7484603881835937,
+      "step": 2220
+    },
+    {
+      "epoch": 0.14341758312431668,
+      "grad_norm": 8.382066865860887,
+      "learning_rate": 2e-05,
+      "loss": 0.7658943176269531,
+      "step": 2230
+    },
+    {
+      "epoch": 0.14406071129976203,
+      "grad_norm": 6.593608367142845,
+      "learning_rate": 2e-05,
+      "loss": 0.776557445526123,
+      "step": 2240
+    },
+    {
+      "epoch": 0.1447038394752074,
+      "grad_norm": 1.213232468498106,
+      "learning_rate": 2e-05,
+      "loss": 0.7852232933044434,
+      "step": 2250
+    },
+    {
+      "epoch": 0.14534696765065278,
+      "grad_norm": 1.4140321832714806,
+      "learning_rate": 2e-05,
+      "loss": 0.730495834350586,
+      "step": 2260
+    },
+    {
+      "epoch": 0.14599009582609815,
+      "grad_norm": 2.7318043792980076,
+      "learning_rate": 2e-05,
+      "loss": 0.7522841453552246,
+      "step": 2270
+    },
+    {
+      "epoch": 0.1466332240015435,
+      "grad_norm": 1.8036361425819734,
+      "learning_rate": 2e-05,
+      "loss": 0.7685168266296387,
+      "step": 2280
+    },
+    {
+      "epoch": 0.14727635217698887,
+      "grad_norm": 1.402413177619317,
+      "learning_rate": 2e-05,
+      "loss": 0.7466054916381836,
+      "step": 2290
+    },
+    {
+      "epoch": 0.14791948035243424,
+      "grad_norm": 1.1772138434845212,
+      "learning_rate": 2e-05,
+      "loss": 0.7492839813232421,
+      "step": 2300
+    },
+    {
+      "epoch": 0.1485626085278796,
+      "grad_norm": 1.6698537238489968,
+      "learning_rate": 2e-05,
+      "loss": 0.7467974662780762,
+      "step": 2310
+    },
+    {
+      "epoch": 0.14920573670332496,
+      "grad_norm": 2.8739935368632534,
+      "learning_rate": 2e-05,
+      "loss": 0.7602452754974365,
+      "step": 2320
+    },
+    {
+      "epoch": 0.14984886487877033,
+      "grad_norm": 1.7893375931088018,
+      "learning_rate": 2e-05,
+      "loss": 0.7592387199401855,
+      "step": 2330
+    },
+    {
+      "epoch": 0.1504919930542157,
+      "grad_norm": 1.6844605472151313,
+      "learning_rate": 2e-05,
+      "loss": 0.7549863815307617,
+      "step": 2340
+    },
+    {
+      "epoch": 0.15113512122966108,
+      "grad_norm": 2.0647222812625343,
+      "learning_rate": 2e-05,
+      "loss": 0.7837786674499512,
+      "step": 2350
+    },
+    {
+      "epoch": 0.15177824940510642,
+      "grad_norm": 1.300194120632371,
+      "learning_rate": 2e-05,
+      "loss": 0.7406606197357177,
+      "step": 2360
+    },
+    {
+      "epoch": 0.1524213775805518,
+      "grad_norm": 3.0794355708028873,
+      "learning_rate": 2e-05,
+      "loss": 0.7331182479858398,
+      "step": 2370
+    },
+    {
+      "epoch": 0.15306450575599717,
+      "grad_norm": 3.0317399406455205,
+      "learning_rate": 2e-05,
+      "loss": 0.7706002712249755,
+      "step": 2380
+    },
+    {
+      "epoch": 0.15370763393144254,
+      "grad_norm": 1.7372752802710871,
+      "learning_rate": 2e-05,
+      "loss": 0.760322904586792,
+      "step": 2390
+    },
+    {
+      "epoch": 0.1543507621068879,
+      "grad_norm": 1.29200077856553,
+      "learning_rate": 2e-05,
+      "loss": 0.7545608043670654,
+      "step": 2400
+    },
+    {
+      "epoch": 0.15499389028233326,
+      "grad_norm": 3.62457038105731,
+      "learning_rate": 2e-05,
+      "loss": 0.7407674312591552,
+      "step": 2410
+    },
+    {
+      "epoch": 0.15563701845777864,
+      "grad_norm": 1.5509317891309322,
+      "learning_rate": 2e-05,
+      "loss": 0.7521041393280029,
+      "step": 2420
+    },
+    {
+      "epoch": 0.156280146633224,
+      "grad_norm": 2.765862287338793,
+      "learning_rate": 2e-05,
+      "loss": 0.7846248626708985,
+      "step": 2430
+    },
+    {
+      "epoch": 0.15692327480866936,
+      "grad_norm": 2.260496474412049,
+      "learning_rate": 2e-05,
+      "loss": 0.719977331161499,
+      "step": 2440
+    },
+    {
+      "epoch": 0.15756640298411473,
+      "grad_norm": 1.3617342907067571,
+      "learning_rate": 2e-05,
+      "loss": 0.7524391174316406,
+      "step": 2450
+    },
+    {
+      "epoch": 0.1582095311595601,
+      "grad_norm": 1.2079742642832751,
+      "learning_rate": 2e-05,
+      "loss": 0.7564692974090577,
+      "step": 2460
+    },
+    {
+      "epoch": 0.15885265933500547,
+      "grad_norm": 1.9822312494182506,
+      "learning_rate": 2e-05,
+      "loss": 0.7288062095642089,
+      "step": 2470
+    },
+    {
+      "epoch": 0.15949578751045082,
+      "grad_norm": 1.3826404260340441,
+      "learning_rate": 2e-05,
+      "loss": 0.7597402572631836,
+      "step": 2480
+    },
+    {
+      "epoch": 0.1601389156858962,
+      "grad_norm": 2.4988397348734845,
+      "learning_rate": 2e-05,
+      "loss": 0.7602372646331788,
+      "step": 2490
+    },
+    {
+      "epoch": 0.16078204386134157,
+      "grad_norm": 1.4496907385102675,
+      "learning_rate": 2e-05,
+      "loss": 0.7745979309082032,
+      "step": 2500
+    },
+    {
+      "epoch": 0.16142517203678694,
+      "grad_norm": 3.4506115545259637,
+      "learning_rate": 2e-05,
+      "loss": 0.7605818748474121,
+      "step": 2510
+    },
+    {
+      "epoch": 0.16206830021223229,
+      "grad_norm": 1.5778314606797577,
+      "learning_rate": 2e-05,
+      "loss": 0.7648172378540039,
+      "step": 2520
+    },
+    {
+      "epoch": 0.16271142838767766,
+      "grad_norm": 3.0574821209635523,
+      "learning_rate": 2e-05,
+      "loss": 0.7540977478027344,
+      "step": 2530
+    },
+    {
+      "epoch": 0.16335455656312303,
+      "grad_norm": 2.791657390522,
+      "learning_rate": 2e-05,
+      "loss": 0.7353564262390136,
+      "step": 2540
+    },
+    {
+      "epoch": 0.1639976847385684,
+      "grad_norm": 8.621296600823664,
+      "learning_rate": 2e-05,
+      "loss": 0.7619668006896972,
+      "step": 2550
+    },
+    {
+      "epoch": 0.16464081291401375,
+      "grad_norm": 2.309710716291024,
+      "learning_rate": 2e-05,
+      "loss": 0.7794631958007813,
+      "step": 2560
+    },
+    {
+      "epoch": 0.16528394108945912,
+      "grad_norm": 1.3087109567906041,
+      "learning_rate": 2e-05,
+      "loss": 0.7914319515228272,
+      "step": 2570
+    },
+    {
+      "epoch": 0.1659270692649045,
+      "grad_norm": 4.404388201427118,
+      "learning_rate": 2e-05,
+      "loss": 0.7710864543914795,
+      "step": 2580
+    },
+    {
+      "epoch": 0.16657019744034987,
+      "grad_norm": 4.576157160942443,
+      "learning_rate": 2e-05,
+      "loss": 0.7495160102844238,
+      "step": 2590
+    },
+    {
+      "epoch": 0.16721332561579522,
+      "grad_norm": 1.5270451940500906,
+      "learning_rate": 2e-05,
+      "loss": 0.7589169979095459,
+      "step": 2600
+    },
+    {
+      "epoch": 0.1678564537912406,
+      "grad_norm": 1.9472599552382834,
+      "learning_rate": 2e-05,
+      "loss": 0.7595278739929199,
+      "step": 2610
+    },
+    {
+      "epoch": 0.16849958196668596,
+      "grad_norm": 1.3117407867752597,
+      "learning_rate": 2e-05,
+      "loss": 0.7396194458007812,
+      "step": 2620
+    },
+    {
+      "epoch": 0.16914271014213134,
+      "grad_norm": 2.457990989497276,
+      "learning_rate": 2e-05,
+      "loss": 0.7509166717529296,
+      "step": 2630
+    },
+    {
+      "epoch": 0.16978583831757668,
+      "grad_norm": 1.4102844574406168,
+      "learning_rate": 2e-05,
+      "loss": 0.7455813407897949,
+      "step": 2640
+    },
+    {
+      "epoch": 0.17042896649302205,
+      "grad_norm": 1.992704744838049,
+      "learning_rate": 2e-05,
+      "loss": 0.7828609466552734,
+      "step": 2650
+    },
+    {
+      "epoch": 0.17107209466846743,
+      "grad_norm": 1.249460645209259,
+      "learning_rate": 2e-05,
+      "loss": 0.7517303466796875,
+      "step": 2660
+    },
+    {
+      "epoch": 0.1717152228439128,
+      "grad_norm": 1.24728336899326,
+      "learning_rate": 2e-05,
+      "loss": 0.7468644618988037,
+      "step": 2670
+    },
+    {
+      "epoch": 0.17235835101935815,
+      "grad_norm": 1.1435876844139452,
+      "learning_rate": 2e-05,
+      "loss": 0.7520998954772949,
+      "step": 2680
+    },
+    {
+      "epoch": 0.17300147919480352,
+      "grad_norm": 2.9158357081479327,
+      "learning_rate": 2e-05,
+      "loss": 0.7450613975524902,
+      "step": 2690
+    },
+    {
+      "epoch": 0.1736446073702489,
+      "grad_norm": 7.595576550094812,
+      "learning_rate": 2e-05,
+      "loss": 0.7426441192626954,
+      "step": 2700
+    },
+    {
+      "epoch": 0.17428773554569427,
+      "grad_norm": 5.865319885985607,
+      "learning_rate": 2e-05,
+      "loss": 0.7606967449188232,
+      "step": 2710
+    },
+    {
+      "epoch": 0.1749308637211396,
+      "grad_norm": 1.2832040369148703,
+      "learning_rate": 2e-05,
+      "loss": 0.7491473197937012,
+      "step": 2720
+    },
+    {
+      "epoch": 0.17557399189658499,
+      "grad_norm": 3.10581277586597,
+      "learning_rate": 2e-05,
+      "loss": 0.7583821773529053,
+      "step": 2730
+    },
+    {
+      "epoch": 0.17621712007203036,
+      "grad_norm": 2.1289051550911227,
+      "learning_rate": 2e-05,
+      "loss": 0.7230392456054687,
+      "step": 2740
+    },
+    {
+      "epoch": 0.17686024824747573,
+      "grad_norm": 1.7640715564966836,
+      "learning_rate": 2e-05,
+      "loss": 0.7431951045989991,
+      "step": 2750
+    },
+    {
+      "epoch": 0.17750337642292108,
+      "grad_norm": 12.418363911566479,
+      "learning_rate": 2e-05,
+      "loss": 0.7820055961608887,
+      "step": 2760
+    },
+    {
+      "epoch": 0.17814650459836645,
+      "grad_norm": 3.4018065508546167,
+      "learning_rate": 2e-05,
+      "loss": 0.758150291442871,
+      "step": 2770
+    },
+    {
+      "epoch": 0.17878963277381182,
+      "grad_norm": 3.079905954505876,
+      "learning_rate": 2e-05,
+      "loss": 0.7261004447937012,
+      "step": 2780
+    },
+    {
+      "epoch": 0.1794327609492572,
+      "grad_norm": 1.4672366276490973,
+      "learning_rate": 2e-05,
+      "loss": 0.7363761901855469,
+      "step": 2790
+    },
+    {
+      "epoch": 0.18007588912470254,
+      "grad_norm": 1.9025508233475796,
+      "learning_rate": 2e-05,
+      "loss": 0.7601851463317871,
+      "step": 2800
+    },
+    {
+      "epoch": 0.18071901730014792,
+      "grad_norm": 1.5196213468498716,
+      "learning_rate": 2e-05,
+      "loss": 0.7611308574676514,
+      "step": 2810
+    },
+    {
+      "epoch": 0.1813621454755933,
+      "grad_norm": 1.909778442981796,
+      "learning_rate": 2e-05,
+      "loss": 0.7633378505706787,
+      "step": 2820
+    },
+    {
+      "epoch": 0.18200527365103866,
+      "grad_norm": 1.393412036904703,
+      "learning_rate": 2e-05,
+      "loss": 0.7511731624603272,
+      "step": 2830
+    },
+    {
+      "epoch": 0.182648401826484,
+      "grad_norm": 1.3714633012275805,
+      "learning_rate": 2e-05,
+      "loss": 0.7424302577972413,
+      "step": 2840
+    },
+    {
+      "epoch": 0.18329153000192938,
+      "grad_norm": 11.20197126426714,
+      "learning_rate": 2e-05,
+      "loss": 0.7554800987243653,
+      "step": 2850
+    },
+    {
+      "epoch": 0.18393465817737475,
+      "grad_norm": 1.9673930111498328,
+      "learning_rate": 2e-05,
+      "loss": 0.7782904624938964,
+      "step": 2860
+    },
+    {
+      "epoch": 0.18457778635282013,
+      "grad_norm": 10.537034997455336,
+      "learning_rate": 2e-05,
+      "loss": 0.7415932655334473,
+      "step": 2870
+    },
+    {
+      "epoch": 0.18522091452826547,
+      "grad_norm": 1.4241190817394258,
+      "learning_rate": 2e-05,
+      "loss": 0.7521487236022949,
+      "step": 2880
+    },
+    {
+      "epoch": 0.18586404270371085,
+      "grad_norm": 1.934465648951276,
+      "learning_rate": 2e-05,
+      "loss": 0.7793794631958008,
+      "step": 2890
+    },
+    {
+      "epoch": 0.18650717087915622,
+      "grad_norm": 4.141312287372932,
+      "learning_rate": 2e-05,
+      "loss": 0.7429276466369629,
+      "step": 2900
+    },
+    {
+      "epoch": 0.1871502990546016,
+      "grad_norm": 2.121755026383068,
+      "learning_rate": 2e-05,
+      "loss": 0.746807861328125,
+      "step": 2910
+    },
+    {
+      "epoch": 0.18779342723004694,
+      "grad_norm": 3.535800894214126,
+      "learning_rate": 2e-05,
+      "loss": 0.7623048782348633,
+      "step": 2920
+    },
+    {
+      "epoch": 0.1884365554054923,
+      "grad_norm": 1.1694107870159447,
+      "learning_rate": 2e-05,
+      "loss": 0.7261816978454589,
+      "step": 2930
+    },
+    {
+      "epoch": 0.18907968358093769,
+      "grad_norm": 1.6031608257386858,
+      "learning_rate": 2e-05,
+      "loss": 0.7551745891571044,
+      "step": 2940
+    },
+    {
+      "epoch": 0.18972281175638306,
+      "grad_norm": 1.7408090254778856,
+      "learning_rate": 2e-05,
+      "loss": 0.7328170299530029,
+      "step": 2950
+    },
+    {
+      "epoch": 0.1903659399318284,
+      "grad_norm": 0.9414858711308967,
+      "learning_rate": 2e-05,
+      "loss": 0.7597317695617676,
+      "step": 2960
+    },
+    {
+      "epoch": 0.19100906810727378,
+      "grad_norm": 5.715632347721177,
+      "learning_rate": 2e-05,
+      "loss": 0.7593686580657959,
+      "step": 2970
+    },
+    {
+      "epoch": 0.19165219628271915,
+      "grad_norm": 1.2683558980244642,
+      "learning_rate": 2e-05,
+      "loss": 0.7523900032043457,
+      "step": 2980
+    },
+    {
+      "epoch": 0.19229532445816452,
+      "grad_norm": 2.489750094364173,
+      "learning_rate": 2e-05,
+      "loss": 0.732703685760498,
+      "step": 2990
+    },
+    {
+      "epoch": 0.19293845263360987,
+      "grad_norm": 1.49133813750742,
+      "learning_rate": 2e-05,
+      "loss": 0.7828021049499512,
+      "step": 3000
+    },
+    {
+      "epoch": 0.19358158080905524,
+      "grad_norm": 1.4214984175732421,
+      "learning_rate": 2e-05,
+      "loss": 0.7481946468353271,
+      "step": 3010
+    },
+    {
+      "epoch": 0.19422470898450062,
+      "grad_norm": 1.0844799546145014,
+      "learning_rate": 2e-05,
+      "loss": 0.7648484230041503,
+      "step": 3020
+    },
+    {
+      "epoch": 0.194867837159946,
+      "grad_norm": 1.127788371561288,
+      "learning_rate": 2e-05,
+      "loss": 0.7447958946228027,
+      "step": 3030
+    },
+    {
+      "epoch": 0.19551096533539133,
+      "grad_norm": 1.3078419072376322,
+      "learning_rate": 2e-05,
+      "loss": 0.7542384624481201,
+      "step": 3040
+    },
+    {
+      "epoch": 0.1961540935108367,
+      "grad_norm": 2.95742255475418,
+      "learning_rate": 2e-05,
+      "loss": 0.7495403289794922,
+      "step": 3050
+    },
+    {
+      "epoch": 0.19679722168628208,
+      "grad_norm": 1.279951429720371,
+      "learning_rate": 2e-05,
+      "loss": 0.7891281127929688,
+      "step": 3060
+    },
+    {
+      "epoch": 0.19744034986172745,
+      "grad_norm": 1.0857892842689383,
+      "learning_rate": 2e-05,
+      "loss": 0.7577287197113037,
+      "step": 3070
+    },
+    {
+      "epoch": 0.1980834780371728,
+      "grad_norm": 2.061292399437917,
+      "learning_rate": 2e-05,
+      "loss": 0.7466474056243897,
+      "step": 3080
+    },
+    {
+      "epoch": 0.19872660621261817,
+      "grad_norm": 1.2472705804465951,
+      "learning_rate": 2e-05,
+      "loss": 0.7725540161132812,
+      "step": 3090
+    },
+    {
+      "epoch": 0.19936973438806355,
+      "grad_norm": 2.602858720884864,
+      "learning_rate": 2e-05,
+      "loss": 0.757532787322998,
+      "step": 3100
+    },
+    {
+      "epoch": 0.20001286256350892,
+      "grad_norm": 1.2739220157051647,
+      "learning_rate": 2e-05,
+      "loss": 0.7565276622772217,
+      "step": 3110
+    },
+    {
+      "epoch": 0.20065599073895427,
+      "grad_norm": 1.9033801807830197,
+      "learning_rate": 2e-05,
+      "loss": 0.7335265159606934,
+      "step": 3120
+    },
+    {
+      "epoch": 0.20129911891439964,
+      "grad_norm": 1.3554367927107596,
+      "learning_rate": 2e-05,
+      "loss": 0.7226738929748535,
+      "step": 3130
+    },
+    {
+      "epoch": 0.201942247089845,
+      "grad_norm": 1.7133014744131059,
+      "learning_rate": 2e-05,
+      "loss": 0.7206092357635498,
+      "step": 3140
+    },
+    {
+      "epoch": 0.20258537526529038,
+      "grad_norm": 1.8996628629000465,
+      "learning_rate": 2e-05,
+      "loss": 0.7250766754150391,
+      "step": 3150
+    },
+    {
+      "epoch": 0.20322850344073573,
+      "grad_norm": 1.9309128279722985,
+      "learning_rate": 2e-05,
+      "loss": 0.7418807983398438,
+      "step": 3160
+    },
+    {
+      "epoch": 0.2038716316161811,
+      "grad_norm": 2.393159552837916,
+      "learning_rate": 2e-05,
+      "loss": 0.7575026512145996,
+      "step": 3170
+    },
+    {
+      "epoch": 0.20451475979162648,
+      "grad_norm": 2.2546034439817726,
+      "learning_rate": 2e-05,
+      "loss": 0.7346158504486084,
+      "step": 3180
+    },
+    {
+      "epoch": 0.20515788796707185,
+      "grad_norm": 1.2090270418321072,
+      "learning_rate": 2e-05,
+      "loss": 0.7440875053405762,
+      "step": 3190
+    },
+    {
+      "epoch": 0.2058010161425172,
+      "grad_norm": 4.615898281722801,
+      "learning_rate": 2e-05,
+      "loss": 0.7345485687255859,
+      "step": 3200
+    },
+    {
+      "epoch": 0.20644414431796257,
+      "grad_norm": 0.9575602149167448,
+      "learning_rate": 2e-05,
+      "loss": 0.7389266967773438,
+      "step": 3210
+    },
+    {
+      "epoch": 0.20708727249340794,
+      "grad_norm": 1.3497946921262038,
+      "learning_rate": 2e-05,
+      "loss": 0.7349342346191406,
+      "step": 3220
+    },
+    {
+      "epoch": 0.20773040066885332,
+      "grad_norm": 2.266999812584626,
+      "learning_rate": 2e-05,
+      "loss": 0.7468548774719238,
+      "step": 3230
+    },
+    {
+      "epoch": 0.20837352884429866,
+      "grad_norm": 1.5673277162176498,
+      "learning_rate": 2e-05,
+      "loss": 0.7460700988769531,
+      "step": 3240
+    },
+    {
+      "epoch": 0.20901665701974403,
+      "grad_norm": 2.1281590647751867,
+      "learning_rate": 2e-05,
+      "loss": 0.7338404655456543,
+      "step": 3250
+    },
+    {
+      "epoch": 0.2096597851951894,
+      "grad_norm": 1.2264672144337803,
+      "learning_rate": 2e-05,
+      "loss": 0.7515013694763184,
+      "step": 3260
+    },
+    {
+      "epoch": 0.21030291337063478,
+      "grad_norm": 1.3589675543517392,
+      "learning_rate": 2e-05,
+      "loss": 0.7714920043945312,
+      "step": 3270
+    },
+    {
+      "epoch": 0.21094604154608013,
+      "grad_norm": 3.5719666218385147,
+      "learning_rate": 2e-05,
+      "loss": 0.7446717739105224,
+      "step": 3280
+    },
+    {
+      "epoch": 0.2115891697215255,
+      "grad_norm": 1.0813049253615596,
+      "learning_rate": 2e-05,
+      "loss": 0.729767894744873,
+      "step": 3290
+    },
+    {
+      "epoch": 0.21223229789697087,
+      "grad_norm": 0.9847719905484558,
+      "learning_rate": 2e-05,
+      "loss": 0.7469213008880615,
+      "step": 3300
+    },
+    {
+      "epoch": 0.21287542607241625,
+      "grad_norm": 1.2117464433321323,
+      "learning_rate": 2e-05,
+      "loss": 0.7576287746429443,
+      "step": 3310
+    },
+    {
+      "epoch": 0.2135185542478616,
+      "grad_norm": 1.3070127777611826,
+      "learning_rate": 2e-05,
+      "loss": 0.7229054927825928,
+      "step": 3320
+    },
+    {
+      "epoch": 0.21416168242330696,
+      "grad_norm": 1.1586711440620237,
+      "learning_rate": 2e-05,
+      "loss": 0.7159502983093262,
+      "step": 3330
+    },
+    {
+      "epoch": 0.21480481059875234,
+      "grad_norm": 1.8859067634706321,
+      "learning_rate": 2e-05,
+      "loss": 0.7311949729919434,
+      "step": 3340
+    },
+    {
+      "epoch": 0.2154479387741977,
+      "grad_norm": 2.393176793568911,
+      "learning_rate": 2e-05,
+      "loss": 0.7426108837127685,
+      "step": 3350
+    },
+    {
+      "epoch": 0.21609106694964306,
+      "grad_norm": 1.0872827405788912,
+      "learning_rate": 2e-05,
+      "loss": 0.7625868797302247,
+      "step": 3360
+    },
+    {
+      "epoch": 0.21673419512508843,
+      "grad_norm": 1.6050660583020135,
+      "learning_rate": 2e-05,
+      "loss": 0.7236311912536622,
+      "step": 3370
+    },
+    {
+      "epoch": 0.2173773233005338,
+      "grad_norm": 1.3933442745801043,
+      "learning_rate": 2e-05,
+      "loss": 0.743280553817749,
+      "step": 3380
+    },
+    {
+      "epoch": 0.21802045147597915,
+      "grad_norm": 1.894926633803295,
+      "learning_rate": 2e-05,
+      "loss": 0.7516873359680176,
+      "step": 3390
+    },
+    {
+      "epoch": 0.21866357965142452,
+      "grad_norm": 2.2498284801850525,
+      "learning_rate": 2e-05,
+      "loss": 0.7211009979248046,
+      "step": 3400
+    },
+    {
+      "epoch": 0.2193067078268699,
+      "grad_norm": 1.264197693407813,
+      "learning_rate": 2e-05,
+      "loss": 0.7462459564208984,
+      "step": 3410
+    },
+    {
+      "epoch": 0.21994983600231527,
+      "grad_norm": 21.98895784739288,
+      "learning_rate": 2e-05,
+      "loss": 0.7361465454101562,
+      "step": 3420
+    },
+    {
+      "epoch": 0.22059296417776061,
+      "grad_norm": 1.1214579111301475,
+      "learning_rate": 2e-05,
+      "loss": 0.7557395458221435,
+      "step": 3430
+    },
+    {
+      "epoch": 0.221236092353206,
+      "grad_norm": 1.1777608301516223,
+      "learning_rate": 2e-05,
+      "loss": 0.712726354598999,
+      "step": 3440
+    },
+    {
+      "epoch": 0.22187922052865136,
+      "grad_norm": 1.0290202252238054,
+      "learning_rate": 2e-05,
+      "loss": 0.7395824432373047,
+      "step": 3450
+    },
+    {
+      "epoch": 0.22252234870409673,
+      "grad_norm": 1.9911323760501045,
+      "learning_rate": 2e-05,
+      "loss": 0.7352674007415771,
+      "step": 3460
+    },
+    {
+      "epoch": 0.22316547687954208,
+      "grad_norm": 1.2107791721990242,
+      "learning_rate": 2e-05,
+      "loss": 0.7151026725769043,
+      "step": 3470
+    },
+    {
+      "epoch": 0.22380860505498745,
+      "grad_norm": 1.7236006720593955,
+      "learning_rate": 2e-05,
+      "loss": 0.7477761268615722,
+      "step": 3480
+    },
+    {
+      "epoch": 0.22445173323043283,
+      "grad_norm": 2.278121680661386,
+      "learning_rate": 2e-05,
+      "loss": 0.7548860073089599,
+      "step": 3490
+    },
+    {
+      "epoch": 0.2250948614058782,
+      "grad_norm": 1.3643047206413652,
+      "learning_rate": 2e-05,
+      "loss": 0.7573224544525147,
+      "step": 3500
+    },
+    {
+      "epoch": 0.22573798958132354,
+      "grad_norm": 2.0643631313256714,
+      "learning_rate": 2e-05,
+      "loss": 0.7669732570648193,
+      "step": 3510
+    },
+    {
+      "epoch": 0.22638111775676892,
+      "grad_norm": 0.9853596965229803,
+      "learning_rate": 2e-05,
+      "loss": 0.7209647178649903,
+      "step": 3520
+    },
+    {
+      "epoch": 0.2270242459322143,
+      "grad_norm": 1.0519923519832508,
+      "learning_rate": 2e-05,
+      "loss": 0.7520937919616699,
+      "step": 3530
+    },
+    {
+      "epoch": 0.22766737410765966,
+      "grad_norm": 1.0761752319926652,
+      "learning_rate": 2e-05,
+      "loss": 0.7350026130676269,
+      "step": 3540
+    },
+    {
+      "epoch": 0.228310502283105,
+      "grad_norm": 2.962457855780589,
+      "learning_rate": 2e-05,
+      "loss": 0.7190482139587402,
+      "step": 3550
+    },
+    {
+      "epoch": 0.22895363045855038,
+      "grad_norm": 1.0780608984541828,
+      "learning_rate": 2e-05,
+      "loss": 0.7538261413574219,
+      "step": 3560
+    },
+    {
+      "epoch": 0.22959675863399576,
+      "grad_norm": 0.961122991249304,
+      "learning_rate": 2e-05,
+      "loss": 0.7133353233337403,
+      "step": 3570
+    },
+    {
+      "epoch": 0.23023988680944113,
+      "grad_norm": 0.9791332825550016,
+      "learning_rate": 2e-05,
+      "loss": 0.7245227813720703,
+      "step": 3580
+    },
+    {
+      "epoch": 0.23088301498488648,
+      "grad_norm": 2.169927580161253,
+      "learning_rate": 2e-05,
+      "loss": 0.7545035362243653,
+      "step": 3590
+    },
+    {
+      "epoch": 0.23152614316033185,
+      "grad_norm": 4.985635526209392,
+      "learning_rate": 2e-05,
+      "loss": 0.7207778930664063,
+      "step": 3600
+    },
+    {
+      "epoch": 0.23216927133577722,
+      "grad_norm": 1.0689943863751645,
+      "learning_rate": 2e-05,
+      "loss": 0.730648136138916,
+      "step": 3610
+    },
+    {
+      "epoch": 0.2328123995112226,
+      "grad_norm": 1.1799195803451448,
+      "learning_rate": 2e-05,
+      "loss": 0.7492318630218506,
+      "step": 3620
+    },
+    {
+      "epoch": 0.23345552768666794,
+      "grad_norm": 2.1485492458802575,
+      "learning_rate": 2e-05,
+      "loss": 0.7677487373352051,
+      "step": 3630
+    },
+    {
+      "epoch": 0.23409865586211331,
+      "grad_norm": 1.1057215523859423,
+      "learning_rate": 2e-05,
+      "loss": 0.7381268978118897,
+      "step": 3640
+    },
+    {
+      "epoch": 0.2347417840375587,
+      "grad_norm": 0.897648870431795,
+      "learning_rate": 2e-05,
+      "loss": 0.7404172897338868,
+      "step": 3650
+    },
+    {
+      "epoch": 0.23538491221300406,
+      "grad_norm": 3.531578092679105,
+      "learning_rate": 2e-05,
+      "loss": 0.7522658348083496,
+      "step": 3660
+    },
+    {
+      "epoch": 0.2360280403884494,
+      "grad_norm": 1.0228218486975473,
+      "learning_rate": 2e-05,
+      "loss": 0.7196244716644287,
+      "step": 3670
+    },
+    {
+      "epoch": 0.23667116856389478,
+      "grad_norm": 2.0551736719061897,
+      "learning_rate": 2e-05,
+      "loss": 0.7346316337585449,
+      "step": 3680
+    },
+    {
+      "epoch": 0.23731429673934015,
+      "grad_norm": 1.0974059128784066,
+      "learning_rate": 2e-05,
+      "loss": 0.7254920482635498,
+      "step": 3690
+    },
+    {
+      "epoch": 0.23795742491478553,
+      "grad_norm": 1.3935304437470286,
+      "learning_rate": 2e-05,
+      "loss": 0.7414390087127686,
+      "step": 3700
+    },
+    {
+      "epoch": 0.23860055309023087,
+      "grad_norm": 1.0013311067122783,
+      "learning_rate": 2e-05,
+      "loss": 0.7595787048339844,
+      "step": 3710
+    },
+    {
+      "epoch": 0.23924368126567624,
+      "grad_norm": 1.954358944346554,
+      "learning_rate": 2e-05,
+      "loss": 0.7554808616638183,
+      "step": 3720
+    },
+    {
+      "epoch": 0.23988680944112162,
+      "grad_norm": 2.9163383985698634,
+      "learning_rate": 2e-05,
+      "loss": 0.7291338920593262,
+      "step": 3730
+    },
+    {
+      "epoch": 0.240529937616567,
+      "grad_norm": 3.4997209306931922,
+      "learning_rate": 2e-05,
+      "loss": 0.786865234375,
+      "step": 3740
+    },
+    {
+      "epoch": 0.24117306579201234,
+      "grad_norm": 2.672649925634819,
+      "learning_rate": 2e-05,
+      "loss": 0.7527791023254394,
+      "step": 3750
+    },
+    {
+      "epoch": 0.2418161939674577,
+      "grad_norm": 1.3383122055071315,
+      "learning_rate": 2e-05,
+      "loss": 0.7263989925384522,
+      "step": 3760
+    },
+    {
+      "epoch": 0.24245932214290308,
+      "grad_norm": 1.294572198712939,
+      "learning_rate": 2e-05,
+      "loss": 0.7176456451416016,
+      "step": 3770
+    },
+    {
+      "epoch": 0.24310245031834846,
+      "grad_norm": 1.0321977403689269,
+      "learning_rate": 2e-05,
+      "loss": 0.7343600749969482,
+      "step": 3780
+    },
+    {
+      "epoch": 0.2437455784937938,
+      "grad_norm": 1.819613359316589,
+      "learning_rate": 2e-05,
+      "loss": 0.7157769203186035,
+      "step": 3790
+    },
+    {
+      "epoch": 0.24438870666923918,
+      "grad_norm": 1.5746899224086783,
+      "learning_rate": 2e-05,
+      "loss": 0.7444544792175293,
+      "step": 3800
+    },
+    {
+      "epoch": 0.24503183484468455,
+      "grad_norm": 1.2747884813210626,
+      "learning_rate": 2e-05,
+      "loss": 0.7165373802185059,
+      "step": 3810
+    },
+    {
+      "epoch": 0.24567496302012992,
+      "grad_norm": 1.2806982813441588,
+      "learning_rate": 2e-05,
+      "loss": 0.7362030029296875,
+      "step": 3820
+    },
+    {
+      "epoch": 0.24631809119557527,
+      "grad_norm": 2.2525458966228853,
+      "learning_rate": 2e-05,
+      "loss": 0.7462127208709717,
+      "step": 3830
+    },
+    {
+      "epoch": 0.24696121937102064,
+      "grad_norm": 5.330631085753853,
+      "learning_rate": 2e-05,
+      "loss": 0.7210140228271484,
+      "step": 3840
+    },
+    {
+      "epoch": 0.247604347546466,
+      "grad_norm": 1.0329616086380298,
+      "learning_rate": 2e-05,
+      "loss": 0.7626016139984131,
+      "step": 3850
+    },
+    {
+      "epoch": 0.2482474757219114,
+      "grad_norm": 1.3852333003997563,
+      "learning_rate": 2e-05,
+      "loss": 0.7498868465423584,
+      "step": 3860
+    },
+    {
+      "epoch": 0.24889060389735673,
+      "grad_norm": 1.5215017048315698,
+      "learning_rate": 2e-05,
+      "loss": 0.734537410736084,
+      "step": 3870
+    },
+    {
+      "epoch": 0.2495337320728021,
+      "grad_norm": 1.1273331846647028,
+      "learning_rate": 2e-05,
+      "loss": 0.7316556453704834,
+      "step": 3880
+    },
+    {
+      "epoch": 0.2501768602482475,
+      "grad_norm": 3.0633504241163783,
+      "learning_rate": 2e-05,
+      "loss": 0.7480705261230469,
+      "step": 3890
+    },
+    {
+      "epoch": 0.25081998842369285,
+      "grad_norm": 1.1896818789764612,
+      "learning_rate": 2e-05,
+      "loss": 0.7362353324890136,
+      "step": 3900
+    },
+    {
+      "epoch": 0.2514631165991382,
+      "grad_norm": 2.708363743291426,
+      "learning_rate": 2e-05,
+      "loss": 0.7096200942993164,
+      "step": 3910
+    },
+    {
+      "epoch": 0.2521062447745836,
+      "grad_norm": 1.021031072475808,
+      "learning_rate": 2e-05,
+      "loss": 0.7346982479095459,
+      "step": 3920
+    },
+    {
+      "epoch": 0.2527493729500289,
+      "grad_norm": 1.0461717317156485,
+      "learning_rate": 2e-05,
+      "loss": 0.7280598640441894,
+      "step": 3930
+    },
+    {
+      "epoch": 0.2533925011254743,
+      "grad_norm": 1.9375312333606454,
+      "learning_rate": 2e-05,
+      "loss": 0.7707145690917969,
+      "step": 3940
+    },
+    {
+      "epoch": 0.25403562930091966,
+      "grad_norm": 1.23640529871102,
+      "learning_rate": 2e-05,
+      "loss": 0.7371127128601074,
+      "step": 3950
+    },
+    {
+      "epoch": 0.25467875747636504,
+      "grad_norm": 1.5916935604411764,
+      "learning_rate": 2e-05,
+      "loss": 0.7383847236633301,
+      "step": 3960
+    },
+    {
+      "epoch": 0.2553218856518104,
+      "grad_norm": 1.2465889420348975,
+      "learning_rate": 2e-05,
+      "loss": 0.7365485191345215,
+      "step": 3970
+    },
+    {
+      "epoch": 0.2559650138272558,
+      "grad_norm": 2.683991266677957,
+      "learning_rate": 2e-05,
+      "loss": 0.745120620727539,
+      "step": 3980
+    },
+    {
+      "epoch": 0.25660814200270116,
+      "grad_norm": 1.5255465541840554,
+      "learning_rate": 2e-05,
+      "loss": 0.7617275238037109,
+      "step": 3990
+    },
+    {
+      "epoch": 0.25725127017814653,
+      "grad_norm": 2.7656928725606464,
+      "learning_rate": 2e-05,
+      "loss": 0.7161584377288819,
+      "step": 4000
+    },
+    {
+      "epoch": 0.25789439835359185,
+      "grad_norm": 1.3381173205142576,
+      "learning_rate": 2e-05,
+      "loss": 0.722429609298706,
+      "step": 4010
+    },
+    {
+      "epoch": 0.2585375265290372,
+      "grad_norm": 2.143378240963822,
+      "learning_rate": 2e-05,
+      "loss": 0.7329329490661621,
+      "step": 4020
+    },
+    {
+      "epoch": 0.2591806547044826,
+      "grad_norm": 1.5781672687213713,
+      "learning_rate": 2e-05,
+      "loss": 0.7398040771484375,
+      "step": 4030
+    },
+    {
+      "epoch": 0.25982378287992797,
+      "grad_norm": 1.3937231266231127,
+      "learning_rate": 2e-05,
+      "loss": 0.7229412078857422,
+      "step": 4040
+    },
+    {
+      "epoch": 0.26046691105537334,
+      "grad_norm": 1.5766131870956492,
+      "learning_rate": 2e-05,
+      "loss": 0.743466854095459,
+      "step": 4050
+    },
+    {
+      "epoch": 0.2611100392308187,
+      "grad_norm": 0.9968445874917599,
+      "learning_rate": 2e-05,
+      "loss": 0.7092678070068359,
+      "step": 4060
+    },
+    {
+      "epoch": 0.2617531674062641,
+      "grad_norm": 0.9336600000992251,
+      "learning_rate": 2e-05,
+      "loss": 0.7497157573699951,
+      "step": 4070
+    },
+    {
+      "epoch": 0.26239629558170946,
+      "grad_norm": 0.9497495497866886,
+      "learning_rate": 2e-05,
+      "loss": 0.724982738494873,
+      "step": 4080
+    },
+    {
+      "epoch": 0.2630394237571548,
+      "grad_norm": 2.2961911019285015,
+      "learning_rate": 2e-05,
+      "loss": 0.7381220817565918,
+      "step": 4090
+    },
+    {
+      "epoch": 0.26368255193260015,
+      "grad_norm": 0.9547493877338601,
+      "learning_rate": 2e-05,
+      "loss": 0.6838616847991943,
+      "step": 4100
+    },
+    {
+      "epoch": 0.2643256801080455,
+      "grad_norm": 1.7412946265607543,
+      "learning_rate": 2e-05,
+      "loss": 0.7001391887664795,
+      "step": 4110
+    },
+    {
+      "epoch": 0.2649688082834909,
+      "grad_norm": 1.0821157164931752,
+      "learning_rate": 2e-05,
+      "loss": 0.7096992015838623,
+      "step": 4120
+    },
+    {
+      "epoch": 0.26561193645893627,
+      "grad_norm": 2.408846113232737,
+      "learning_rate": 2e-05,
+      "loss": 0.7188594818115235,
+      "step": 4130
+    },
+    {
+      "epoch": 0.26625506463438164,
+      "grad_norm": 1.6644434842681461,
+      "learning_rate": 2e-05,
+      "loss": 0.7558166027069092,
+      "step": 4140
+    },
+    {
+      "epoch": 0.266898192809827,
+      "grad_norm": 2.5743582876246323,
+      "learning_rate": 2e-05,
+      "loss": 0.7378355979919433,
+      "step": 4150
+    },
+    {
+      "epoch": 0.2675413209852724,
+      "grad_norm": 1.2994774266444291,
+      "learning_rate": 2e-05,
+      "loss": 0.7637517929077149,
+      "step": 4160
+    },
+    {
+      "epoch": 0.2681844491607177,
+      "grad_norm": 2.1928287931928496,
+      "learning_rate": 2e-05,
+      "loss": 0.7436198711395263,
+      "step": 4170
+    },
+    {
+      "epoch": 0.2688275773361631,
+      "grad_norm": 1.4094426875428832,
+      "learning_rate": 2e-05,
+      "loss": 0.7239650249481201,
+      "step": 4180
+    },
+    {
+      "epoch": 0.26947070551160845,
+      "grad_norm": 0.9826082837222696,
+      "learning_rate": 2e-05,
+      "loss": 0.7198858261108398,
+      "step": 4190
+    },
+    {
+      "epoch": 0.27011383368705383,
+      "grad_norm": 1.0878194503318233,
+      "learning_rate": 2e-05,
+      "loss": 0.7271060466766357,
+      "step": 4200
+    },
+    {
+      "epoch": 0.2707569618624992,
+      "grad_norm": 2.1838705637028135,
+      "learning_rate": 2e-05,
+      "loss": 0.7809905052185059,
+      "step": 4210
+    },
+    {
+      "epoch": 0.2714000900379446,
+      "grad_norm": 5.047327133404171,
+      "learning_rate": 2e-05,
+      "loss": 0.744784927368164,
+      "step": 4220
+    },
+    {
+      "epoch": 0.27204321821338995,
+      "grad_norm": 3.42462283705851,
+      "learning_rate": 2e-05,
+      "loss": 0.7031085968017579,
+      "step": 4230
+    },
+    {
+      "epoch": 0.2726863463888353,
+      "grad_norm": 1.7479327520469765,
+      "learning_rate": 2e-05,
+      "loss": 0.727137279510498,
+      "step": 4240
+    },
+    {
+      "epoch": 0.27332947456428064,
+      "grad_norm": 4.873365527896015,
+      "learning_rate": 2e-05,
+      "loss": 0.7273091793060302,
+      "step": 4250
+    },
+    {
+      "epoch": 0.273972602739726,
+      "grad_norm": 1.128395835545906,
+      "learning_rate": 2e-05,
+      "loss": 0.7530433654785156,
+      "step": 4260
+    },
+    {
+      "epoch": 0.2746157309151714,
+      "grad_norm": 1.300382182756331,
+      "learning_rate": 2e-05,
+      "loss": 0.7198864936828613,
+      "step": 4270
+    },
+    {
+      "epoch": 0.27525885909061676,
+      "grad_norm": 0.9583633050292032,
+      "learning_rate": 2e-05,
+      "loss": 0.7156751155853271,
+      "step": 4280
+    },
+    {
+      "epoch": 0.27590198726606213,
+      "grad_norm": 1.4777059722236674,
+      "learning_rate": 2e-05,
+      "loss": 0.725702953338623,
+      "step": 4290
+    },
+    {
+      "epoch": 0.2765451154415075,
+      "grad_norm": 1.930139500946224,
+      "learning_rate": 2e-05,
+      "loss": 0.7401728630065918,
+      "step": 4300
+    },
+    {
+      "epoch": 0.2771882436169529,
+      "grad_norm": 1.1379184083848206,
+      "learning_rate": 2e-05,
+      "loss": 0.731195068359375,
+      "step": 4310
+    },
+    {
+      "epoch": 0.27783137179239825,
+      "grad_norm": 1.0650634262943084,
+      "learning_rate": 2e-05,
+      "loss": 0.7313991546630859,
+      "step": 4320
+    },
+    {
+      "epoch": 0.27847449996784357,
+      "grad_norm": 1.21689992254794,
+      "learning_rate": 2e-05,
+      "loss": 0.7207317352294922,
+      "step": 4330
+    },
+    {
+      "epoch": 0.27911762814328894,
+      "grad_norm": 3.051162099709167,
+      "learning_rate": 2e-05,
+      "loss": 0.7360997676849366,
+      "step": 4340
+    },
+    {
+      "epoch": 0.2797607563187343,
+      "grad_norm": 1.2276380037115646,
+      "learning_rate": 2e-05,
+      "loss": 0.7188148498535156,
+      "step": 4350
+    },
+    {
+      "epoch": 0.2804038844941797,
+      "grad_norm": 2.2703609186797724,
+      "learning_rate": 2e-05,
+      "loss": 0.71884765625,
+      "step": 4360
+    },
+    {
+      "epoch": 0.28104701266962506,
+      "grad_norm": 5.463544690584565,
+      "learning_rate": 2e-05,
+      "loss": 0.7465624809265137,
+      "step": 4370
+    },
+    {
+      "epoch": 0.28169014084507044,
+      "grad_norm": 2.8901881534175216,
+      "learning_rate": 2e-05,
+      "loss": 0.7248989105224609,
+      "step": 4380
+    },
+    {
+      "epoch": 0.2823332690205158,
+      "grad_norm": 1.6240989205208243,
+      "learning_rate": 2e-05,
+      "loss": 0.7251723289489747,
+      "step": 4390
+    },
+    {
+      "epoch": 0.2829763971959612,
+      "grad_norm": 1.1663527362573316,
+      "learning_rate": 2e-05,
+      "loss": 0.715138053894043,
+      "step": 4400
+    },
+    {
+      "epoch": 0.2836195253714065,
+      "grad_norm": 1.1410678337832476,
+      "learning_rate": 2e-05,
+      "loss": 0.716748571395874,
+      "step": 4410
+    },
+    {
+      "epoch": 0.2842626535468519,
+      "grad_norm": 1.5952455939373982,
+      "learning_rate": 2e-05,
+      "loss": 0.7237945079803467,
+      "step": 4420
+    },
+    {
+      "epoch": 0.28490578172229725,
+      "grad_norm": 1.0555426535081012,
+      "learning_rate": 2e-05,
+      "loss": 0.7176595211029053,
+      "step": 4430
+    },
+    {
+      "epoch": 0.2855489098977426,
+      "grad_norm": 18.639246785545314,
+      "learning_rate": 2e-05,
+      "loss": 0.7433196544647217,
+      "step": 4440
+    },
+    {
+      "epoch": 0.286192038073188,
+      "grad_norm": 1.0885043541115613,
+      "learning_rate": 2e-05,
+      "loss": 0.7389396667480469,
+      "step": 4450
+    },
+    {
+      "epoch": 0.28683516624863337,
+      "grad_norm": 0.987110568747816,
+      "learning_rate": 2e-05,
+      "loss": 0.703716230392456,
+      "step": 4460
+    },
+    {
+      "epoch": 0.28747829442407874,
+      "grad_norm": 1.6676871757345224,
+      "learning_rate": 2e-05,
+      "loss": 0.6983733177185059,
+      "step": 4470
+    },
+    {
+      "epoch": 0.28812142259952406,
+      "grad_norm": 2.819537990824305,
+      "learning_rate": 2e-05,
+      "loss": 0.7210405349731446,
+      "step": 4480
+    },
+    {
+      "epoch": 0.28876455077496943,
+      "grad_norm": 1.1159119973661147,
+      "learning_rate": 2e-05,
+      "loss": 0.7208063125610351,
+      "step": 4490
+    },
+    {
+      "epoch": 0.2894076789504148,
+      "grad_norm": 2.028701731731351,
+      "learning_rate": 2e-05,
+      "loss": 0.7111621856689453,
+      "step": 4500
+    },
+    {
+      "epoch": 0.2900508071258602,
+      "grad_norm": 1.1404095121439821,
+      "learning_rate": 2e-05,
+      "loss": 0.733924674987793,
+      "step": 4510
+    },
+    {
+      "epoch": 0.29069393530130555,
+      "grad_norm": 14.956390238853409,
+      "learning_rate": 2e-05,
+      "loss": 0.7344192028045654,
+      "step": 4520
+    },
+    {
+      "epoch": 0.2913370634767509,
+      "grad_norm": 1.1081603634992983,
+      "learning_rate": 2e-05,
+      "loss": 0.7081531524658203,
+      "step": 4530
+    },
+    {
+      "epoch": 0.2919801916521963,
+      "grad_norm": 1.3597792075030952,
+      "learning_rate": 2e-05,
+      "loss": 0.7079843044281006,
+      "step": 4540
+    },
+    {
+      "epoch": 0.29262331982764167,
+      "grad_norm": 1.026830290280156,
+      "learning_rate": 2e-05,
+      "loss": 0.7209130764007569,
+      "step": 4550
+    },
+    {
+      "epoch": 0.293266448003087,
+      "grad_norm": 1.1858293095567627,
+      "learning_rate": 2e-05,
+      "loss": 0.7196866035461426,
+      "step": 4560
+    },
+    {
+      "epoch": 0.29390957617853236,
+      "grad_norm": 0.9314067425777351,
+      "learning_rate": 2e-05,
+      "loss": 0.7655232906341553,
+      "step": 4570
+    },
+    {
+      "epoch": 0.29455270435397773,
+      "grad_norm": 3.04722872581358,
+      "learning_rate": 2e-05,
+      "loss": 0.7183656215667724,
+      "step": 4580
+    },
+    {
+      "epoch": 0.2951958325294231,
+      "grad_norm": 1.509439008999805,
+      "learning_rate": 2e-05,
+      "loss": 0.7322559356689453,
+      "step": 4590
+    },
+    {
+      "epoch": 0.2958389607048685,
+      "grad_norm": 4.082820588021,
+      "learning_rate": 2e-05,
+      "loss": 0.7471843719482422,
+      "step": 4600
+    },
+    {
+      "epoch": 0.29648208888031385,
+      "grad_norm": 1.0558035610252994,
+      "learning_rate": 2e-05,
+      "loss": 0.708987808227539,
+      "step": 4610
+    },
+    {
+      "epoch": 0.2971252170557592,
+      "grad_norm": 1.3225030482307876,
+      "learning_rate": 2e-05,
+      "loss": 0.7098447799682617,
+      "step": 4620
+    },
+    {
+      "epoch": 0.2977683452312046,
+      "grad_norm": 0.9689533636594124,
+      "learning_rate": 2e-05,
+      "loss": 0.7451314926147461,
+      "step": 4630
+    },
+    {
+      "epoch": 0.2984114734066499,
+      "grad_norm": 3.8494637352218346,
+      "learning_rate": 2e-05,
+      "loss": 0.6989140510559082,
+      "step": 4640
+    },
+    {
+      "epoch": 0.2990546015820953,
+      "grad_norm": 1.3895638156079348,
+      "learning_rate": 2e-05,
+      "loss": 0.7003883361816406,
+      "step": 4650
+    },
+    {
+      "epoch": 0.29969772975754067,
+      "grad_norm": 1.2480961113003382,
+      "learning_rate": 2e-05,
+      "loss": 0.7026212215423584,
+      "step": 4660
+    },
+    {
+      "epoch": 0.30034085793298604,
+      "grad_norm": 0.968584012382354,
+      "learning_rate": 2e-05,
+      "loss": 0.777205228805542,
+      "step": 4670
+    },
+    {
+      "epoch": 0.3009839861084314,
+      "grad_norm": 0.9159192602186901,
+      "learning_rate": 2e-05,
+      "loss": 0.7216925621032715,
+      "step": 4680
+    },
+    {
+      "epoch": 0.3016271142838768,
+      "grad_norm": 2.3023062895325226,
+      "learning_rate": 2e-05,
+      "loss": 0.7152534008026123,
+      "step": 4690
+    },
+    {
+      "epoch": 0.30227024245932216,
+      "grad_norm": 0.9167429036398073,
+      "learning_rate": 2e-05,
+      "loss": 0.698724889755249,
+      "step": 4700
+    },
+    {
+      "epoch": 0.30291337063476753,
+      "grad_norm": 1.5484656327185233,
+      "learning_rate": 2e-05,
+      "loss": 0.7216808319091796,
+      "step": 4710
+    },
+    {
+      "epoch": 0.30355649881021285,
+      "grad_norm": 0.9883985475355792,
+      "learning_rate": 2e-05,
+      "loss": 0.7521573066711426,
+      "step": 4720
+    },
+    {
+      "epoch": 0.3041996269856582,
+      "grad_norm": 1.659176669096098,
+      "learning_rate": 2e-05,
+      "loss": 0.7266528129577636,
+      "step": 4730
+    },
+    {
+      "epoch": 0.3048427551611036,
+      "grad_norm": 1.087329903160801,
+      "learning_rate": 2e-05,
+      "loss": 0.7182927608489991,
+      "step": 4740
+    },
+    {
+      "epoch": 0.30548588333654897,
+      "grad_norm": 2.6119903500572343,
+      "learning_rate": 2e-05,
+      "loss": 0.7173304557800293,
+      "step": 4750
+    },
+    {
+      "epoch": 0.30612901151199434,
+      "grad_norm": 2.311443446486716,
+      "learning_rate": 2e-05,
+      "loss": 0.7191170692443848,
+      "step": 4760
+    },
+    {
+      "epoch": 0.3067721396874397,
+      "grad_norm": 0.9908036813787956,
+      "learning_rate": 2e-05,
+      "loss": 0.6904977798461914,
+      "step": 4770
+    },
+    {
+      "epoch": 0.3074152678628851,
+      "grad_norm": 0.93529914993198,
+      "learning_rate": 2e-05,
+      "loss": 0.7196500301361084,
+      "step": 4780
+    },
+    {
+      "epoch": 0.30805839603833046,
+      "grad_norm": 1.8715682001850917,
+      "learning_rate": 2e-05,
+      "loss": 0.7597176551818847,
+      "step": 4790
+    },
+    {
+      "epoch": 0.3087015242137758,
+      "grad_norm": 1.4581395700833577,
+      "learning_rate": 2e-05,
+      "loss": 0.6957194805145264,
+      "step": 4800
+    },
+    {
+      "epoch": 0.30934465238922115,
+      "grad_norm": 1.4118321000123575,
+      "learning_rate": 2e-05,
+      "loss": 0.6958081722259521,
+      "step": 4810
+    },
+    {
+      "epoch": 0.3099877805646665,
+      "grad_norm": 2.0656927345538914,
+      "learning_rate": 2e-05,
+      "loss": 0.6939652442932129,
+      "step": 4820
+    },
+    {
+      "epoch": 0.3106309087401119,
+      "grad_norm": 1.700453933805578,
+      "learning_rate": 2e-05,
+      "loss": 0.7029267311096191,
+      "step": 4830
+    },
+    {
+      "epoch": 0.3112740369155573,
+      "grad_norm": 1.2693174983430136,
+      "learning_rate": 2e-05,
+      "loss": 0.7128015041351319,
+      "step": 4840
+    },
+    {
+      "epoch": 0.31191716509100265,
+      "grad_norm": 1.1720745624716569,
+      "learning_rate": 2e-05,
+      "loss": 0.7327308654785156,
+      "step": 4850
+    },
+    {
+      "epoch": 0.312560293266448,
+      "grad_norm": 2.509731278170732,
+      "learning_rate": 2e-05,
+      "loss": 0.707606315612793,
+      "step": 4860
+    },
+    {
+      "epoch": 0.3132034214418934,
+      "grad_norm": 1.5470244895474448,
+      "learning_rate": 2e-05,
+      "loss": 0.6996453285217286,
+      "step": 4870
+    },
+    {
+      "epoch": 0.3138465496173387,
+      "grad_norm": 1.2977494092316977,
+      "learning_rate": 2e-05,
+      "loss": 0.6856669425964356,
+      "step": 4880
+    },
+    {
+      "epoch": 0.3144896777927841,
+      "grad_norm": 2.5305366102455533,
+      "learning_rate": 2e-05,
+      "loss": 0.7406852722167969,
+      "step": 4890
+    },
+    {
+      "epoch": 0.31513280596822946,
+      "grad_norm": 1.1010606611871123,
+      "learning_rate": 2e-05,
+      "loss": 0.7163901329040527,
+      "step": 4900
+    },
+    {
+      "epoch": 0.31577593414367483,
+      "grad_norm": 4.566699297464212,
+      "learning_rate": 2e-05,
+      "loss": 0.7339641094207764,
+      "step": 4910
+    },
+    {
+      "epoch": 0.3164190623191202,
+      "grad_norm": 1.1001900806073777,
+      "learning_rate": 2e-05,
+      "loss": 0.7495140075683594,
+      "step": 4920
+    },
+    {
+      "epoch": 0.3170621904945656,
+      "grad_norm": 2.074518876326924,
+      "learning_rate": 2e-05,
+      "loss": 0.7341387748718262,
+      "step": 4930
+    },
+    {
+      "epoch": 0.31770531867001095,
+      "grad_norm": 2.4384817971489285,
+      "learning_rate": 2e-05,
+      "loss": 0.7034233093261719,
+      "step": 4940
+    },
+    {
+      "epoch": 0.3183484468454563,
+      "grad_norm": 2.072456043600415,
+      "learning_rate": 2e-05,
+      "loss": 0.7102450370788574,
+      "step": 4950
+    },
+    {
+      "epoch": 0.31899157502090164,
+      "grad_norm": 1.1620228222217182,
+      "learning_rate": 2e-05,
+      "loss": 0.6992386817932129,
+      "step": 4960
+    },
+    {
+      "epoch": 0.319634703196347,
+      "grad_norm": 1.0905557154791223,
+      "learning_rate": 2e-05,
+      "loss": 0.7454713821411133,
+      "step": 4970
+    },
+    {
+      "epoch": 0.3202778313717924,
+      "grad_norm": 1.0722624110453984,
+      "learning_rate": 2e-05,
+      "loss": 0.7722179412841796,
+      "step": 4980
+    },
+    {
+      "epoch": 0.32092095954723776,
+      "grad_norm": 1.1818987144389994,
+      "learning_rate": 2e-05,
+      "loss": 0.7095654964447021,
+      "step": 4990
+    },
+    {
+      "epoch": 0.32156408772268313,
+      "grad_norm": 2.244493704606387,
+      "learning_rate": 2e-05,
+      "loss": 0.7169167995452881,
+      "step": 5000
+    },
+    {
+      "epoch": 0.3222072158981285,
+      "grad_norm": 1.0516013816410592,
+      "learning_rate": 2e-05,
+      "loss": 0.7189145565032959,
+      "step": 5010
+    },
+    {
+      "epoch": 0.3228503440735739,
+      "grad_norm": 0.807078238829302,
+      "learning_rate": 2e-05,
+      "loss": 0.6756699562072754,
+      "step": 5020
+    },
+    {
+      "epoch": 0.32349347224901925,
+      "grad_norm": 1.060465182902963,
+      "learning_rate": 2e-05,
+      "loss": 0.718842601776123,
+      "step": 5030
+    },
+    {
+      "epoch": 0.32413660042446457,
+      "grad_norm": 0.9944004353991349,
+      "learning_rate": 2e-05,
+      "loss": 0.6963611602783203,
+      "step": 5040
+    },
+    {
+      "epoch": 0.32477972859990994,
+      "grad_norm": 1.6707213565044037,
+      "learning_rate": 2e-05,
+      "loss": 0.6979356288909913,
+      "step": 5050
+    },
+    {
+      "epoch": 0.3254228567753553,
+      "grad_norm": 1.0851275106965528,
+      "learning_rate": 2e-05,
+      "loss": 0.7029341697692871,
+      "step": 5060
+    },
+    {
+      "epoch": 0.3260659849508007,
+      "grad_norm": 1.2208014018780948,
+      "learning_rate": 2e-05,
+      "loss": 0.7180676937103272,
+      "step": 5070
+    },
+    {
+      "epoch": 0.32670911312624606,
+      "grad_norm": 1.046857019378032,
+      "learning_rate": 2e-05,
+      "loss": 0.6998267650604248,
+      "step": 5080
+    },
+    {
+      "epoch": 0.32735224130169144,
+      "grad_norm": 1.061279463246683,
+      "learning_rate": 2e-05,
+      "loss": 0.6928921699523926,
+      "step": 5090
+    },
+    {
+      "epoch": 0.3279953694771368,
+      "grad_norm": 1.2035472525859028,
+      "learning_rate": 2e-05,
+      "loss": 0.729271411895752,
+      "step": 5100
+    },
+    {
+      "epoch": 0.3286384976525822,
+      "grad_norm": 1.322589231168056,
+      "learning_rate": 2e-05,
+      "loss": 0.70732741355896,
+      "step": 5110
+    },
+    {
+      "epoch": 0.3292816258280275,
+      "grad_norm": 1.5735255656754488,
+      "learning_rate": 2e-05,
+      "loss": 0.7230887889862061,
+      "step": 5120
+    },
+    {
+      "epoch": 0.3299247540034729,
+      "grad_norm": 5.351941362537579,
+      "learning_rate": 2e-05,
+      "loss": 0.7041518211364746,
+      "step": 5130
+    },
+    {
+      "epoch": 0.33056788217891825,
+      "grad_norm": 1.0941849716750445,
+      "learning_rate": 2e-05,
+      "loss": 0.709419059753418,
+      "step": 5140
+    },
+    {
+      "epoch": 0.3312110103543636,
+      "grad_norm": 1.8677567118488376,
+      "learning_rate": 2e-05,
+      "loss": 0.7113753318786621,
+      "step": 5150
+    },
+    {
+      "epoch": 0.331854138529809,
+      "grad_norm": 1.2706540303085747,
+      "learning_rate": 2e-05,
+      "loss": 0.7233946800231934,
+      "step": 5160
+    },
+    {
+      "epoch": 0.33249726670525437,
+      "grad_norm": 1.6589897106308442,
+      "learning_rate": 2e-05,
+      "loss": 0.7174046993255615,
+      "step": 5170
+    },
+    {
+      "epoch": 0.33314039488069974,
+      "grad_norm": 3.758063713951005,
+      "learning_rate": 2e-05,
+      "loss": 0.7290914058685303,
+      "step": 5180
+    },
+    {
+      "epoch": 0.3337835230561451,
+      "grad_norm": 1.0756987643452771,
+      "learning_rate": 2e-05,
+      "loss": 0.7174134731292725,
+      "step": 5190
+    },
+    {
+      "epoch": 0.33442665123159043,
+      "grad_norm": 2.7235800367049463,
+      "learning_rate": 2e-05,
+      "loss": 0.7555623531341553,
+      "step": 5200
+    },
+    {
+      "epoch": 0.3350697794070358,
+      "grad_norm": 3.1514561116640136,
+      "learning_rate": 2e-05,
+      "loss": 0.74488844871521,
+      "step": 5210
+    },
+    {
+      "epoch": 0.3357129075824812,
+      "grad_norm": 1.055101035543414,
+      "learning_rate": 2e-05,
+      "loss": 0.7059333324432373,
+      "step": 5220
+    },
+    {
+      "epoch": 0.33635603575792655,
+      "grad_norm": 1.4130343579292037,
+      "learning_rate": 2e-05,
+      "loss": 0.7198103904724121,
+      "step": 5230
+    },
+    {
+      "epoch": 0.3369991639333719,
+      "grad_norm": 1.4096070513560626,
+      "learning_rate": 2e-05,
+      "loss": 0.7354572296142579,
+      "step": 5240
+    },
+    {
+      "epoch": 0.3376422921088173,
+      "grad_norm": 1.0013907169889942,
+      "learning_rate": 2e-05,
+      "loss": 0.7371484756469726,
+      "step": 5250
+    },
+    {
+      "epoch": 0.3382854202842627,
+      "grad_norm": 2.4814083432347918,
+      "learning_rate": 2e-05,
+      "loss": 0.6908042430877686,
+      "step": 5260
+    },
+    {
+      "epoch": 0.33892854845970805,
+      "grad_norm": 1.1896903623310542,
+      "learning_rate": 2e-05,
+      "loss": 0.7036655426025391,
+      "step": 5270
+    },
+    {
+      "epoch": 0.33957167663515336,
+      "grad_norm": 0.9835382071466582,
+      "learning_rate": 2e-05,
+      "loss": 0.6827800750732422,
+      "step": 5280
+    },
+    {
+      "epoch": 0.34021480481059874,
+      "grad_norm": 1.0748801522625755,
+      "learning_rate": 2e-05,
+      "loss": 0.6930392742156982,
+      "step": 5290
+    },
+    {
+      "epoch": 0.3408579329860441,
+      "grad_norm": 9.223986394588502,
+      "learning_rate": 2e-05,
+      "loss": 0.6980859279632569,
+      "step": 5300
+    },
+    {
+      "epoch": 0.3415010611614895,
+      "grad_norm": 2.160181710587223,
+      "learning_rate": 2e-05,
+      "loss": 0.7207084655761719,
+      "step": 5310
+    },
+    {
+      "epoch": 0.34214418933693486,
+      "grad_norm": 3.1123964133530375,
+      "learning_rate": 2e-05,
+      "loss": 0.7246676445007324,
+      "step": 5320
+    },
+    {
+      "epoch": 0.34278731751238023,
+      "grad_norm": 1.364387759042788,
+      "learning_rate": 2e-05,
+      "loss": 0.6927510738372803,
+      "step": 5330
+    },
+    {
+      "epoch": 0.3434304456878256,
+      "grad_norm": 0.9824019090068365,
+      "learning_rate": 2e-05,
+      "loss": 0.7154762268066406,
+      "step": 5340
+    },
+    {
+      "epoch": 0.344073573863271,
+      "grad_norm": 3.0683757100501206,
+      "learning_rate": 1.999194520388485e-05,
+      "loss": 0.6943968772888184,
+      "step": 5350
+    },
+    {
+      "epoch": 0.3447167020387163,
+      "grad_norm": 1.1191601252812802,
+      "learning_rate": 1.9959247303157763e-05,
+      "loss": 0.6733108043670655,
+      "step": 5360
+    },
+    {
+      "epoch": 0.34535983021416167,
+      "grad_norm": 1.443883124530169,
+      "learning_rate": 1.990149424459817e-05,
+      "loss": 0.7091413021087647,
+      "step": 5370
+    },
+    {
+      "epoch": 0.34600295838960704,
+      "grad_norm": 1.0514627967337997,
+      "learning_rate": 1.9818847538648688e-05,
+      "loss": 0.7165494441986084,
+      "step": 5380
+    },
+    {
+      "epoch": 0.3466460865650524,
+      "grad_norm": 1.4260436532422882,
+      "learning_rate": 1.971153831256721e-05,
+      "loss": 0.7044605255126953,
+      "step": 5390
+    },
+    {
+      "epoch": 0.3472892147404978,
+      "grad_norm": 0.9874331175686362,
+      "learning_rate": 1.9579866664063435e-05,
+      "loss": 0.7165324211120605,
+      "step": 5400
+    },
+    {
+      "epoch": 0.34793234291594316,
+      "grad_norm": 0.8873933734307349,
+      "learning_rate": 1.9424200822054782e-05,
+      "loss": 0.6822521209716796,
+      "step": 5410
+    },
+    {
+      "epoch": 0.34857547109138853,
+      "grad_norm": 1.128493485271591,
+      "learning_rate": 1.9244976116888626e-05,
+      "loss": 0.6852675914764405,
+      "step": 5420
+    },
+    {
+      "epoch": 0.3492185992668339,
+      "grad_norm": 1.4196444825257328,
+      "learning_rate": 1.904269376291071e-05,
+      "loss": 0.7181539058685302,
+      "step": 5430
+    },
+    {
+      "epoch": 0.3498617274422792,
+      "grad_norm": 1.119676407804779,
+      "learning_rate": 1.88179194567844e-05,
+      "loss": 0.7114235877990722,
+      "step": 5440
+    },
+    {
+      "epoch": 0.3505048556177246,
+      "grad_norm": 4.861953224440987,
+      "learning_rate": 1.8571281795480632e-05,
+      "loss": 0.7523653984069825,
+      "step": 5450
+    },
+    {
+      "epoch": 0.35114798379316997,
+      "grad_norm": 1.1236663024856386,
+      "learning_rate": 1.8303470518362808e-05,
+      "loss": 0.7040934562683105,
+      "step": 5460
+    },
+    {
+      "epoch": 0.35179111196861534,
+      "grad_norm": 0.9339002627996613,
+      "learning_rate": 1.801523457828271e-05,
+      "loss": 0.7564759731292725,
+      "step": 5470
+    },
+    {
+      "epoch": 0.3524342401440607,
+      "grad_norm": 1.858929134395736,
+      "learning_rate": 1.7707380047081796e-05,
+      "loss": 0.6958681106567383,
+      "step": 5480
+    },
+    {
+      "epoch": 0.3530773683195061,
+      "grad_norm": 0.816354702563157,
+      "learning_rate": 1.7380767861355252e-05,
+      "loss": 0.7180388450622559,
+      "step": 5490
+    },
+    {
+      "epoch": 0.35372049649495146,
+      "grad_norm": 1.2009381208842937,
+      "learning_rate": 1.703631141478299e-05,
+      "loss": 0.7174203872680665,
+      "step": 5500
+    },
+    {
+      "epoch": 0.35436362467039684,
+      "grad_norm": 1.0720136208216389,
+      "learning_rate": 1.6674974003760705e-05,
+      "loss": 0.7131974220275878,
+      "step": 5510
+    },
+    {
+      "epoch": 0.35500675284584216,
+      "grad_norm": 1.1995397833903698,
+      "learning_rate": 1.6297766133474556e-05,
+      "loss": 0.7016341686248779,
+      "step": 5520
+    },
+    {
+      "epoch": 0.35564988102128753,
+      "grad_norm": 1.1422601032243749,
+      "learning_rate": 1.5905742691953186e-05,
+      "loss": 0.684351110458374,
+      "step": 5530
+    },
+    {
+      "epoch": 0.3562930091967329,
+      "grad_norm": 0.9840502040512998,
+      "learning_rate": 1.55e-05,
+      "loss": 0.7468593120574951,
+      "step": 5540
+    },
+    {
+      "epoch": 0.3569361373721783,
+      "grad_norm": 1.080107796883931,
+      "learning_rate": 1.5081672745255862e-05,
+      "loss": 0.690840482711792,
+      "step": 5550
+    },
+    {
+      "epoch": 0.35757926554762365,
+      "grad_norm": 0.9688569545507764,
+      "learning_rate": 1.4651930808966206e-05,
+      "loss": 0.7117425441741944,
+      "step": 5560
+    },
+    {
+      "epoch": 0.358222393723069,
+      "grad_norm": 1.1228242843704186,
+      "learning_rate": 1.4211975994326848e-05,
+      "loss": 0.6876376628875732,
+      "step": 5570
+    },
+    {
+      "epoch": 0.3588655218985144,
+      "grad_norm": 1.0850177989049035,
+      "learning_rate": 1.376303866555779e-05,
+      "loss": 0.6947522640228272,
+      "step": 5580
+    },
+    {
+      "epoch": 0.35950865007395977,
+      "grad_norm": 1.1161677398561476,
+      "learning_rate": 1.3306374307104164e-05,
+      "loss": 0.6996616363525391,
+      "step": 5590
+    },
+    {
+      "epoch": 0.3601517782494051,
+      "grad_norm": 1.460812025366727,
+      "learning_rate": 1.2843260012586718e-05,
+      "loss": 0.6984384059906006,
+      "step": 5600
+    },
+    {
+      "epoch": 0.36079490642485046,
+      "grad_norm": 1.5006435989348337,
+      "learning_rate": 1.2374990913320726e-05,
+      "loss": 0.6947767257690429,
+      "step": 5610
+    },
+    {
+      "epoch": 0.36143803460029583,
+      "grad_norm": 1.0720089200870861,
+      "learning_rate": 1.1902876556391247e-05,
+      "loss": 0.7189748287200928,
+      "step": 5620
+    },
+    {
+      "epoch": 0.3620811627757412,
+      "grad_norm": 1.3347117101508061,
+      "learning_rate": 1.1428237242413685e-05,
+      "loss": 0.6703412055969238,
+      "step": 5630
+    },
+    {
+      "epoch": 0.3627242909511866,
+      "grad_norm": 0.9351449245361221,
+      "learning_rate": 1.095240033322132e-05,
+      "loss": 0.6984054565429687,
+      "step": 5640
+    },
+    {
+      "epoch": 0.36336741912663195,
+      "grad_norm": 1.2277731616255392,
+      "learning_rate": 1.047669653980572e-05,
+      "loss": 0.6937325954437256,
+      "step": 5650
+    },
+    {
+      "epoch": 0.3640105473020773,
+      "grad_norm": 1.2087840578416604,
+      "learning_rate": 1.0002456200890902e-05,
+      "loss": 0.6907111167907715,
+      "step": 5660
+    },
+    {
+      "epoch": 0.36465367547752264,
+      "grad_norm": 1.018215163709065,
+      "learning_rate": 9.531005562548668e-06,
+      "loss": 0.6939154624938965,
+      "step": 5670
+    },
+    {
+      "epoch": 0.365296803652968,
+      "grad_norm": 0.9802178214934197,
+      "learning_rate": 9.06366306925925e-06,
+      "loss": 0.7418316841125489,
+      "step": 5680
+    },
+    {
+      "epoch": 0.3659399318284134,
+      "grad_norm": 0.9055814143530228,
+      "learning_rate": 8.601735676789686e-06,
+      "loss": 0.6801801204681397,
+      "step": 5690
+    },
+    {
+      "epoch": 0.36658306000385876,
+      "grad_norm": 0.8293132642990987,
+      "learning_rate": 8.14651519720118e-06,
+      "loss": 0.6952208995819091,
+      "step": 5700
+    },
+    {
+      "epoch": 0.36722618817930414,
+      "grad_norm": 1.1838675009534916,
+      "learning_rate": 7.699274686206866e-06,
+      "loss": 0.6944610595703125,
+      "step": 5710
+    },
+    {
+      "epoch": 0.3678693163547495,
+      "grad_norm": 1.25373363909008,
+      "learning_rate": 7.261264882983024e-06,
+      "loss": 0.6859086513519287,
+      "step": 5720
+    },
+    {
+      "epoch": 0.3685124445301949,
+      "grad_norm": 1.5583929721128569,
+      "learning_rate": 6.8337107123900386e-06,
+      "loss": 0.6726075649261475,
+      "step": 5730
+    },
+    {
+      "epoch": 0.36915557270564026,
+      "grad_norm": 0.9152637770183092,
+      "learning_rate": 6.417807859384881e-06,
+      "loss": 0.6806456565856933,
+      "step": 5740
+    },
+    {
+      "epoch": 0.3697987008810856,
+      "grad_norm": 0.9418161595133706,
+      "learning_rate": 6.014719425205009e-06,
+      "loss": 0.6885168075561523,
+      "step": 5750
+    },
+    {
+      "epoch": 0.37044182905653095,
+      "grad_norm": 1.0084234288849065,
+      "learning_rate": 5.6255726746749275e-06,
+      "loss": 0.661384391784668,
+      "step": 5760
+    },
+    {
+      "epoch": 0.3710849572319763,
+      "grad_norm": 1.0559945386186083,
+      "learning_rate": 5.251455883731728e-06,
+      "loss": 0.7189984321594238,
+      "step": 5770
+    },
+    {
+      "epoch": 0.3717280854074217,
+      "grad_norm": 1.568218290836974,
+      "learning_rate": 4.893415295985812e-06,
+      "loss": 0.6970807552337647,
+      "step": 5780
+    },
+    {
+      "epoch": 0.37237121358286707,
+      "grad_norm": 0.9185094936372992,
+      "learning_rate": 4.552452196827885e-06,
+      "loss": 0.677029800415039,
+      "step": 5790
+    },
+    {
+      "epoch": 0.37301434175831244,
+      "grad_norm": 0.8901731281561018,
+      "learning_rate": 4.229520113264785e-06,
+      "loss": 0.718800163269043,
+      "step": 5800
+    },
+    {
+      "epoch": 0.3736574699337578,
+      "grad_norm": 1.7470373047246883,
+      "learning_rate": 3.925522147314915e-06,
+      "loss": 0.7066672325134278,
+      "step": 5810
+    },
+    {
+      "epoch": 0.3743005981092032,
+      "grad_norm": 5.415721884037847,
+      "learning_rate": 3.641308450420675e-06,
+      "loss": 0.6964803218841553,
+      "step": 5820
+    },
+    {
+      "epoch": 0.3749437262846485,
+      "grad_norm": 1.05325519241195,
+      "learning_rate": 3.37767384594087e-06,
+      "loss": 0.7134075164794922,
+      "step": 5830
+    },
+    {
+      "epoch": 0.3755868544600939,
+      "grad_norm": 3.1325973047940363,
+      "learning_rate": 3.1353556063719357e-06,
+      "loss": 0.6560983180999755,
+      "step": 5840
+    },
+    {
+      "epoch": 0.37622998263553925,
+      "grad_norm": 1.1132375722169887,
+      "learning_rate": 2.9150313915141474e-06,
+      "loss": 0.67728590965271,
+      "step": 5850
+    },
+    {
+      "epoch": 0.3768731108109846,
+      "grad_norm": 0.922494120402013,
+      "learning_rate": 2.717317353348891e-06,
+      "loss": 0.6633328437805176,
+      "step": 5860
+    },
+    {
+      "epoch": 0.37751623898643,
+      "grad_norm": 1.0692823446874786,
+      "learning_rate": 2.5427664129268253e-06,
+      "loss": 0.6658954620361328,
+      "step": 5870
+    },
+    {
+      "epoch": 0.37815936716187537,
+      "grad_norm": 1.226691802263373,
+      "learning_rate": 2.391866714085717e-06,
+      "loss": 0.6793697357177735,
+      "step": 5880
+    },
+    {
+      "epoch": 0.37880249533732074,
+      "grad_norm": 0.9490574083866483,
+      "learning_rate": 2.2650402583222753e-06,
+      "loss": 0.7224684715270996,
+      "step": 5890
+    },
+    {
+      "epoch": 0.3794456235127661,
+      "grad_norm": 1.5364744382591737,
+      "learning_rate": 2.1626417246356398e-06,
+      "loss": 0.6527574539184571,
+      "step": 5900
+    },
+    {
+      "epoch": 0.38008875168821143,
+      "grad_norm": 2.0111234169783105,
+      "learning_rate": 2.084957477642894e-06,
+      "loss": 0.7000278472900391,
+      "step": 5910
+    },
+    {
+      "epoch": 0.3807318798636568,
+      "grad_norm": 2.775992150012775,
+      "learning_rate": 2.0322047667405147e-06,
+      "loss": 0.6976972103118897,
+      "step": 5920
+    },
+    {
+      "epoch": 0.3813750080391022,
+      "grad_norm": 0.8660380276419226,
+      "learning_rate": 2.0045311185513342e-06,
+      "loss": 0.6765724182128906,
+      "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
+}