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": 0.16273203367058758,
+  "eval_steps": 500,
+  "global_step": 5935,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0002741904526884374,
+      "grad_norm": 533.0549778145878,
+      "learning_rate": 2.910112359550562e-06,
+      "loss": 2.0392507553100585,
+      "step": 10
+    },
+    {
+      "epoch": 0.0005483809053768748,
+      "grad_norm": 129.62802977018143,
+      "learning_rate": 3.921348314606742e-06,
+      "loss": 1.1088291168212892,
+      "step": 20
+    },
+    {
+      "epoch": 0.0008225713580653121,
+      "grad_norm": 37.028700378797375,
+      "learning_rate": 4.932584269662922e-06,
+      "loss": 0.8763736724853516,
+      "step": 30
+    },
+    {
+      "epoch": 0.0010967618107537496,
+      "grad_norm": 11.415105214785655,
+      "learning_rate": 5.943820224719102e-06,
+      "loss": 0.8334451675415039,
+      "step": 40
+    },
+    {
+      "epoch": 0.0013709522634421868,
+      "grad_norm": 168.0092220161655,
+      "learning_rate": 6.955056179775282e-06,
+      "loss": 0.841560173034668,
+      "step": 50
+    },
+    {
+      "epoch": 0.0016451427161306242,
+      "grad_norm": 10.9544582922062,
+      "learning_rate": 7.966292134831462e-06,
+      "loss": 0.7757131099700928,
+      "step": 60
+    },
+    {
+      "epoch": 0.0019193331688190617,
+      "grad_norm": 52.12017293008163,
+      "learning_rate": 8.977528089887641e-06,
+      "loss": 0.7379593849182129,
+      "step": 70
+    },
+    {
+      "epoch": 0.0021935236215074993,
+      "grad_norm": 24.23368291019739,
+      "learning_rate": 9.988764044943822e-06,
+      "loss": 0.871122169494629,
+      "step": 80
+    },
+    {
+      "epoch": 0.0024677140741959365,
+      "grad_norm": 44.606225719656116,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.8103988647460938,
+      "step": 90
+    },
+    {
+      "epoch": 0.0027419045268843737,
+      "grad_norm": 19.394878252919106,
+      "learning_rate": 1.2011235955056182e-05,
+      "loss": 0.8088878631591797,
+      "step": 100
+    },
+    {
+      "epoch": 0.0030160949795728113,
+      "grad_norm": 154.65284754287825,
+      "learning_rate": 1.3022471910112362e-05,
+      "loss": 0.7940959930419922,
+      "step": 110
+    },
+    {
+      "epoch": 0.0032902854322612485,
+      "grad_norm": 58.453168158560985,
+      "learning_rate": 1.403370786516854e-05,
+      "loss": 0.8113164901733398,
+      "step": 120
+    },
+    {
+      "epoch": 0.003564475884949686,
+      "grad_norm": 121.71906195134056,
+      "learning_rate": 1.504494382022472e-05,
+      "loss": 0.802008056640625,
+      "step": 130
+    },
+    {
+      "epoch": 0.0038386663376381233,
+      "grad_norm": 7.063904217729592,
+      "learning_rate": 1.60561797752809e-05,
+      "loss": 0.7886086940765381,
+      "step": 140
+    },
+    {
+      "epoch": 0.004112856790326561,
+      "grad_norm": 5.886231107848548,
+      "learning_rate": 1.706741573033708e-05,
+      "loss": 0.8553682327270508,
+      "step": 150
+    },
+    {
+      "epoch": 0.004387047243014999,
+      "grad_norm": 17.479053516912124,
+      "learning_rate": 1.8078651685393256e-05,
+      "loss": 0.878786849975586,
+      "step": 160
+    },
+    {
+      "epoch": 0.004661237695703435,
+      "grad_norm": 15.809884339669948,
+      "learning_rate": 1.908988764044944e-05,
+      "loss": 0.8892171859741211,
+      "step": 170
+    },
+    {
+      "epoch": 0.004935428148391873,
+      "grad_norm": 15.552804520715554,
+      "learning_rate": 2e-05,
+      "loss": 0.907989501953125,
+      "step": 180
+    },
+    {
+      "epoch": 0.005209618601080311,
+      "grad_norm": 8.234516558585694,
+      "learning_rate": 2e-05,
+      "loss": 0.8959874153137207,
+      "step": 190
+    },
+    {
+      "epoch": 0.005483809053768747,
+      "grad_norm": 47.07479845477993,
+      "learning_rate": 2e-05,
+      "loss": 0.7972547054290772,
+      "step": 200
+    },
+    {
+      "epoch": 0.005757999506457185,
+      "grad_norm": 14.678891090428264,
+      "learning_rate": 2e-05,
+      "loss": 0.7945639610290527,
+      "step": 210
+    },
+    {
+      "epoch": 0.006032189959145623,
+      "grad_norm": 17.99623194706809,
+      "learning_rate": 2e-05,
+      "loss": 0.8467901229858399,
+      "step": 220
+    },
+    {
+      "epoch": 0.00630638041183406,
+      "grad_norm": 51.95652459415379,
+      "learning_rate": 2e-05,
+      "loss": 0.7964937210083007,
+      "step": 230
+    },
+    {
+      "epoch": 0.006580570864522497,
+      "grad_norm": 10.738865490640423,
+      "learning_rate": 2e-05,
+      "loss": 0.7535751342773438,
+      "step": 240
+    },
+    {
+      "epoch": 0.006854761317210935,
+      "grad_norm": 10.795608734038868,
+      "learning_rate": 2e-05,
+      "loss": 0.828792953491211,
+      "step": 250
+    },
+    {
+      "epoch": 0.007128951769899372,
+      "grad_norm": 14.551689425739292,
+      "learning_rate": 2e-05,
+      "loss": 0.8123558044433594,
+      "step": 260
+    },
+    {
+      "epoch": 0.00740314222258781,
+      "grad_norm": 26.532992867443514,
+      "learning_rate": 2e-05,
+      "loss": 0.8517896652221679,
+      "step": 270
+    },
+    {
+      "epoch": 0.007677332675276247,
+      "grad_norm": 34.07990246963607,
+      "learning_rate": 2e-05,
+      "loss": 0.8389625549316406,
+      "step": 280
+    },
+    {
+      "epoch": 0.007951523127964684,
+      "grad_norm": 27.19116883891022,
+      "learning_rate": 2e-05,
+      "loss": 0.8190082550048828,
+      "step": 290
+    },
+    {
+      "epoch": 0.008225713580653122,
+      "grad_norm": 16.718638113526183,
+      "learning_rate": 2e-05,
+      "loss": 0.8008914947509765,
+      "step": 300
+    },
+    {
+      "epoch": 0.00849990403334156,
+      "grad_norm": 27.009662680205103,
+      "learning_rate": 2e-05,
+      "loss": 0.8091920852661133,
+      "step": 310
+    },
+    {
+      "epoch": 0.008774094486029997,
+      "grad_norm": 9.031648650732553,
+      "learning_rate": 2e-05,
+      "loss": 0.8081918716430664,
+      "step": 320
+    },
+    {
+      "epoch": 0.009048284938718433,
+      "grad_norm": 15.927828911998665,
+      "learning_rate": 2e-05,
+      "loss": 0.7878878593444825,
+      "step": 330
+    },
+    {
+      "epoch": 0.00932247539140687,
+      "grad_norm": 6.019580282287438,
+      "learning_rate": 2e-05,
+      "loss": 0.8092373847961426,
+      "step": 340
+    },
+    {
+      "epoch": 0.009596665844095308,
+      "grad_norm": 10.097656044283081,
+      "learning_rate": 2e-05,
+      "loss": 0.8236124038696289,
+      "step": 350
+    },
+    {
+      "epoch": 0.009870856296783746,
+      "grad_norm": 3.898510274122292,
+      "learning_rate": 2e-05,
+      "loss": 0.8201662063598633,
+      "step": 360
+    },
+    {
+      "epoch": 0.010145046749472184,
+      "grad_norm": 10.433240846124043,
+      "learning_rate": 2e-05,
+      "loss": 0.7954160213470459,
+      "step": 370
+    },
+    {
+      "epoch": 0.010419237202160621,
+      "grad_norm": 3.9109175859167626,
+      "learning_rate": 2e-05,
+      "loss": 0.8451309204101562,
+      "step": 380
+    },
+    {
+      "epoch": 0.010693427654849059,
+      "grad_norm": 19.47785359214917,
+      "learning_rate": 2e-05,
+      "loss": 0.8227203369140625,
+      "step": 390
+    },
+    {
+      "epoch": 0.010967618107537495,
+      "grad_norm": 1.9979422322935754,
+      "learning_rate": 2e-05,
+      "loss": 0.7947024345397949,
+      "step": 400
+    },
+    {
+      "epoch": 0.011241808560225932,
+      "grad_norm": 2.1152537387703063,
+      "learning_rate": 2e-05,
+      "loss": 0.7340120315551758,
+      "step": 410
+    },
+    {
+      "epoch": 0.01151599901291437,
+      "grad_norm": 2.457083494187141,
+      "learning_rate": 2e-05,
+      "loss": 0.7300191879272461,
+      "step": 420
+    },
+    {
+      "epoch": 0.011790189465602808,
+      "grad_norm": 17.457321034227927,
+      "learning_rate": 2e-05,
+      "loss": 0.7896897315979003,
+      "step": 430
+    },
+    {
+      "epoch": 0.012064379918291245,
+      "grad_norm": 3.7757446450653047,
+      "learning_rate": 2e-05,
+      "loss": 0.7420886039733887,
+      "step": 440
+    },
+    {
+      "epoch": 0.012338570370979683,
+      "grad_norm": 10.718659609403895,
+      "learning_rate": 2e-05,
+      "loss": 0.7464280605316163,
+      "step": 450
+    },
+    {
+      "epoch": 0.01261276082366812,
+      "grad_norm": 13.096467752661693,
+      "learning_rate": 2e-05,
+      "loss": 0.7420991420745849,
+      "step": 460
+    },
+    {
+      "epoch": 0.012886951276356558,
+      "grad_norm": 7.554114077938612,
+      "learning_rate": 2e-05,
+      "loss": 0.7896678924560547,
+      "step": 470
+    },
+    {
+      "epoch": 0.013161141729044994,
+      "grad_norm": 1.9610155148576782,
+      "learning_rate": 2e-05,
+      "loss": 0.7641692161560059,
+      "step": 480
+    },
+    {
+      "epoch": 0.013435332181733432,
+      "grad_norm": 3.743726960430974,
+      "learning_rate": 2e-05,
+      "loss": 0.7724472999572753,
+      "step": 490
+    },
+    {
+      "epoch": 0.01370952263442187,
+      "grad_norm": 3.6205794374072897,
+      "learning_rate": 2e-05,
+      "loss": 0.7726649284362793,
+      "step": 500
+    },
+    {
+      "epoch": 0.013983713087110307,
+      "grad_norm": 2.751894400685354,
+      "learning_rate": 2e-05,
+      "loss": 0.7311810970306396,
+      "step": 510
+    },
+    {
+      "epoch": 0.014257903539798744,
+      "grad_norm": 15.8354606407861,
+      "learning_rate": 2e-05,
+      "loss": 0.7479830741882324,
+      "step": 520
+    },
+    {
+      "epoch": 0.014532093992487182,
+      "grad_norm": 9.741291005928273,
+      "learning_rate": 2e-05,
+      "loss": 0.758505916595459,
+      "step": 530
+    },
+    {
+      "epoch": 0.01480628444517562,
+      "grad_norm": 6.361558235058739,
+      "learning_rate": 2e-05,
+      "loss": 0.7497701644897461,
+      "step": 540
+    },
+    {
+      "epoch": 0.015080474897864056,
+      "grad_norm": 12.406630096904701,
+      "learning_rate": 2e-05,
+      "loss": 0.7578043937683105,
+      "step": 550
+    },
+    {
+      "epoch": 0.015354665350552493,
+      "grad_norm": 5.437533111150059,
+      "learning_rate": 2e-05,
+      "loss": 0.7494767665863037,
+      "step": 560
+    },
+    {
+      "epoch": 0.015628855803240933,
+      "grad_norm": 5.385414434765143,
+      "learning_rate": 2e-05,
+      "loss": 0.76002197265625,
+      "step": 570
+    },
+    {
+      "epoch": 0.01590304625592937,
+      "grad_norm": 2.514872003150109,
+      "learning_rate": 2e-05,
+      "loss": 0.753857946395874,
+      "step": 580
+    },
+    {
+      "epoch": 0.016177236708617804,
+      "grad_norm": 7.350990110651167,
+      "learning_rate": 2e-05,
+      "loss": 0.7598513603210449,
+      "step": 590
+    },
+    {
+      "epoch": 0.016451427161306244,
+      "grad_norm": 9.359850474100211,
+      "learning_rate": 2e-05,
+      "loss": 0.7978845596313476,
+      "step": 600
+    },
+    {
+      "epoch": 0.01672561761399468,
+      "grad_norm": 2.5927971608846243,
+      "learning_rate": 2e-05,
+      "loss": 0.7353355407714843,
+      "step": 610
+    },
+    {
+      "epoch": 0.01699980806668312,
+      "grad_norm": 13.580441280433185,
+      "learning_rate": 2e-05,
+      "loss": 0.7530646800994873,
+      "step": 620
+    },
+    {
+      "epoch": 0.017273998519371555,
+      "grad_norm": 7.032504711444772,
+      "learning_rate": 2e-05,
+      "loss": 0.7999831199645996,
+      "step": 630
+    },
+    {
+      "epoch": 0.017548188972059994,
+      "grad_norm": 3.7623518166414134,
+      "learning_rate": 2e-05,
+      "loss": 0.7583576202392578,
+      "step": 640
+    },
+    {
+      "epoch": 0.01782237942474843,
+      "grad_norm": 9.996121547764488,
+      "learning_rate": 2e-05,
+      "loss": 0.7126437187194824,
+      "step": 650
+    },
+    {
+      "epoch": 0.018096569877436866,
+      "grad_norm": 3.8005663709948423,
+      "learning_rate": 2e-05,
+      "loss": 0.7714213371276856,
+      "step": 660
+    },
+    {
+      "epoch": 0.018370760330125305,
+      "grad_norm": 2.2803953101797445,
+      "learning_rate": 2e-05,
+      "loss": 0.735899829864502,
+      "step": 670
+    },
+    {
+      "epoch": 0.01864495078281374,
+      "grad_norm": 2.2804717033812687,
+      "learning_rate": 2e-05,
+      "loss": 0.7463047981262207,
+      "step": 680
+    },
+    {
+      "epoch": 0.01891914123550218,
+      "grad_norm": 2.6630744686256174,
+      "learning_rate": 2e-05,
+      "loss": 0.7112466812133789,
+      "step": 690
+    },
+    {
+      "epoch": 0.019193331688190617,
+      "grad_norm": 8.59421748677637,
+      "learning_rate": 2e-05,
+      "loss": 0.8032501220703125,
+      "step": 700
+    },
+    {
+      "epoch": 0.019467522140879056,
+      "grad_norm": 6.839099911030037,
+      "learning_rate": 2e-05,
+      "loss": 0.7257285118103027,
+      "step": 710
+    },
+    {
+      "epoch": 0.019741712593567492,
+      "grad_norm": 6.527196880605798,
+      "learning_rate": 2e-05,
+      "loss": 0.7394065856933594,
+      "step": 720
+    },
+    {
+      "epoch": 0.020015903046255928,
+      "grad_norm": 4.91546686150854,
+      "learning_rate": 2e-05,
+      "loss": 0.7233449459075928,
+      "step": 730
+    },
+    {
+      "epoch": 0.020290093498944367,
+      "grad_norm": 9.615722695591046,
+      "learning_rate": 2e-05,
+      "loss": 0.714713191986084,
+      "step": 740
+    },
+    {
+      "epoch": 0.020564283951632803,
+      "grad_norm": 2.6469047522209244,
+      "learning_rate": 2e-05,
+      "loss": 0.7250522613525391,
+      "step": 750
+    },
+    {
+      "epoch": 0.020838474404321242,
+      "grad_norm": 27.513230581702402,
+      "learning_rate": 2e-05,
+      "loss": 0.709619665145874,
+      "step": 760
+    },
+    {
+      "epoch": 0.021112664857009678,
+      "grad_norm": 2.3305572520941102,
+      "learning_rate": 2e-05,
+      "loss": 0.7650126457214356,
+      "step": 770
+    },
+    {
+      "epoch": 0.021386855309698118,
+      "grad_norm": 1.7645473455518579,
+      "learning_rate": 2e-05,
+      "loss": 0.7501170635223389,
+      "step": 780
+    },
+    {
+      "epoch": 0.021661045762386553,
+      "grad_norm": 2.706703322701269,
+      "learning_rate": 2e-05,
+      "loss": 0.7390304565429687,
+      "step": 790
+    },
+    {
+      "epoch": 0.02193523621507499,
+      "grad_norm": 1.343452945442269,
+      "learning_rate": 2e-05,
+      "loss": 0.7560922145843506,
+      "step": 800
+    },
+    {
+      "epoch": 0.02220942666776343,
+      "grad_norm": 1.8131674161495455,
+      "learning_rate": 2e-05,
+      "loss": 0.7617329597473145,
+      "step": 810
+    },
+    {
+      "epoch": 0.022483617120451865,
+      "grad_norm": 6.6257304148619,
+      "learning_rate": 2e-05,
+      "loss": 0.7030315399169922,
+      "step": 820
+    },
+    {
+      "epoch": 0.022757807573140304,
+      "grad_norm": 2.7477972541714237,
+      "learning_rate": 2e-05,
+      "loss": 0.7051456451416016,
+      "step": 830
+    },
+    {
+      "epoch": 0.02303199802582874,
+      "grad_norm": 1.8551689554260649,
+      "learning_rate": 2e-05,
+      "loss": 0.790980863571167,
+      "step": 840
+    },
+    {
+      "epoch": 0.02330618847851718,
+      "grad_norm": 4.091035715965749,
+      "learning_rate": 2e-05,
+      "loss": 0.7167482376098633,
+      "step": 850
+    },
+    {
+      "epoch": 0.023580378931205615,
+      "grad_norm": 3.13444342764459,
+      "learning_rate": 2e-05,
+      "loss": 0.7428287506103516,
+      "step": 860
+    },
+    {
+      "epoch": 0.023854569383894055,
+      "grad_norm": 7.428502868185567,
+      "learning_rate": 2e-05,
+      "loss": 0.7720952510833741,
+      "step": 870
+    },
+    {
+      "epoch": 0.02412875983658249,
+      "grad_norm": 2.738880412295684,
+      "learning_rate": 2e-05,
+      "loss": 0.7307533264160156,
+      "step": 880
+    },
+    {
+      "epoch": 0.024402950289270926,
+      "grad_norm": 5.442319700482665,
+      "learning_rate": 2e-05,
+      "loss": 0.7590367794036865,
+      "step": 890
+    },
+    {
+      "epoch": 0.024677140741959366,
+      "grad_norm": 4.209775631505361,
+      "learning_rate": 2e-05,
+      "loss": 0.7398377418518066,
+      "step": 900
+    },
+    {
+      "epoch": 0.0249513311946478,
+      "grad_norm": 3.317366374639838,
+      "learning_rate": 2e-05,
+      "loss": 0.7573592662811279,
+      "step": 910
+    },
+    {
+      "epoch": 0.02522552164733624,
+      "grad_norm": 2.21708523073645,
+      "learning_rate": 2e-05,
+      "loss": 0.7442788124084473,
+      "step": 920
+    },
+    {
+      "epoch": 0.025499712100024677,
+      "grad_norm": 5.003510596258695,
+      "learning_rate": 2e-05,
+      "loss": 0.7208542823791504,
+      "step": 930
+    },
+    {
+      "epoch": 0.025773902552713116,
+      "grad_norm": 1.8666696979850983,
+      "learning_rate": 2e-05,
+      "loss": 0.7219260215759278,
+      "step": 940
+    },
+    {
+      "epoch": 0.026048093005401552,
+      "grad_norm": 2.319707094464392,
+      "learning_rate": 2e-05,
+      "loss": 0.705035400390625,
+      "step": 950
+    },
+    {
+      "epoch": 0.026322283458089988,
+      "grad_norm": 1.559433923602148,
+      "learning_rate": 2e-05,
+      "loss": 0.7253781795501709,
+      "step": 960
+    },
+    {
+      "epoch": 0.026596473910778427,
+      "grad_norm": 3.1103622605289294,
+      "learning_rate": 2e-05,
+      "loss": 0.7434506416320801,
+      "step": 970
+    },
+    {
+      "epoch": 0.026870664363466863,
+      "grad_norm": 4.482952591971397,
+      "learning_rate": 2e-05,
+      "loss": 0.7423406600952148,
+      "step": 980
+    },
+    {
+      "epoch": 0.027144854816155303,
+      "grad_norm": 2.349643537640656,
+      "learning_rate": 2e-05,
+      "loss": 0.7380632400512696,
+      "step": 990
+    },
+    {
+      "epoch": 0.02741904526884374,
+      "grad_norm": 6.624084473754211,
+      "learning_rate": 2e-05,
+      "loss": 0.7448741912841796,
+      "step": 1000
+    },
+    {
+      "epoch": 0.027693235721532178,
+      "grad_norm": 1.959542744744174,
+      "learning_rate": 2e-05,
+      "loss": 0.7620717048645019,
+      "step": 1010
+    },
+    {
+      "epoch": 0.027967426174220614,
+      "grad_norm": 1.9359370487542638,
+      "learning_rate": 2e-05,
+      "loss": 0.7283543586730957,
+      "step": 1020
+    },
+    {
+      "epoch": 0.02824161662690905,
+      "grad_norm": 3.7510091510273424,
+      "learning_rate": 2e-05,
+      "loss": 0.7341510772705078,
+      "step": 1030
+    },
+    {
+      "epoch": 0.02851580707959749,
+      "grad_norm": 1.645233108914731,
+      "learning_rate": 2e-05,
+      "loss": 0.7321323394775391,
+      "step": 1040
+    },
+    {
+      "epoch": 0.028789997532285925,
+      "grad_norm": 5.060129114701537,
+      "learning_rate": 2e-05,
+      "loss": 0.7923129558563232,
+      "step": 1050
+    },
+    {
+      "epoch": 0.029064187984974364,
+      "grad_norm": 6.767343077930711,
+      "learning_rate": 2e-05,
+      "loss": 0.7364850997924804,
+      "step": 1060
+    },
+    {
+      "epoch": 0.0293383784376628,
+      "grad_norm": 7.157354259172915,
+      "learning_rate": 2e-05,
+      "loss": 0.7175948143005371,
+      "step": 1070
+    },
+    {
+      "epoch": 0.02961256889035124,
+      "grad_norm": 10.118526445754803,
+      "learning_rate": 2e-05,
+      "loss": 0.6956705093383789,
+      "step": 1080
+    },
+    {
+      "epoch": 0.029886759343039675,
+      "grad_norm": 3.2975128574851205,
+      "learning_rate": 2e-05,
+      "loss": 0.7435084342956543,
+      "step": 1090
+    },
+    {
+      "epoch": 0.03016094979572811,
+      "grad_norm": 3.79438108252297,
+      "learning_rate": 2e-05,
+      "loss": 0.7209542274475098,
+      "step": 1100
+    },
+    {
+      "epoch": 0.03043514024841655,
+      "grad_norm": 3.6413783178286896,
+      "learning_rate": 2e-05,
+      "loss": 0.7462982177734375,
+      "step": 1110
+    },
+    {
+      "epoch": 0.030709330701104987,
+      "grad_norm": 2.4359365603997323,
+      "learning_rate": 2e-05,
+      "loss": 0.7772771835327148,
+      "step": 1120
+    },
+    {
+      "epoch": 0.030983521153793426,
+      "grad_norm": 2.067253539210089,
+      "learning_rate": 2e-05,
+      "loss": 0.754563283920288,
+      "step": 1130
+    },
+    {
+      "epoch": 0.031257711606481865,
+      "grad_norm": 1.3603842511083,
+      "learning_rate": 2e-05,
+      "loss": 0.7163387298583984,
+      "step": 1140
+    },
+    {
+      "epoch": 0.0315319020591703,
+      "grad_norm": 7.634204958345298,
+      "learning_rate": 2e-05,
+      "loss": 0.7153648376464844,
+      "step": 1150
+    },
+    {
+      "epoch": 0.03180609251185874,
+      "grad_norm": 2.224287938133906,
+      "learning_rate": 2e-05,
+      "loss": 0.7743209838867188,
+      "step": 1160
+    },
+    {
+      "epoch": 0.03208028296454717,
+      "grad_norm": 8.328106347948369,
+      "learning_rate": 2e-05,
+      "loss": 0.7125131130218506,
+      "step": 1170
+    },
+    {
+      "epoch": 0.03235447341723561,
+      "grad_norm": 4.9140382117559716,
+      "learning_rate": 2e-05,
+      "loss": 0.734964656829834,
+      "step": 1180
+    },
+    {
+      "epoch": 0.03262866386992405,
+      "grad_norm": 1.6437427011396712,
+      "learning_rate": 2e-05,
+      "loss": 0.736287260055542,
+      "step": 1190
+    },
+    {
+      "epoch": 0.03290285432261249,
+      "grad_norm": 1.7984413046378422,
+      "learning_rate": 2e-05,
+      "loss": 0.6888255119323731,
+      "step": 1200
+    },
+    {
+      "epoch": 0.03317704477530092,
+      "grad_norm": 3.003273429621072,
+      "learning_rate": 2e-05,
+      "loss": 0.7354516506195068,
+      "step": 1210
+    },
+    {
+      "epoch": 0.03345123522798936,
+      "grad_norm": 1.581766516298146,
+      "learning_rate": 2e-05,
+      "loss": 0.7175889492034913,
+      "step": 1220
+    },
+    {
+      "epoch": 0.0337254256806778,
+      "grad_norm": 2.3247470513343185,
+      "learning_rate": 2e-05,
+      "loss": 0.714189863204956,
+      "step": 1230
+    },
+    {
+      "epoch": 0.03399961613336624,
+      "grad_norm": 7.9334494235192325,
+      "learning_rate": 2e-05,
+      "loss": 0.7136695861816407,
+      "step": 1240
+    },
+    {
+      "epoch": 0.034273806586054674,
+      "grad_norm": 49.62487496955337,
+      "learning_rate": 2e-05,
+      "loss": 0.7086945533752441,
+      "step": 1250
+    },
+    {
+      "epoch": 0.03454799703874311,
+      "grad_norm": 2.890774584879648,
+      "learning_rate": 2e-05,
+      "loss": 0.7269087791442871,
+      "step": 1260
+    },
+    {
+      "epoch": 0.034822187491431546,
+      "grad_norm": 4.796531005737684,
+      "learning_rate": 2e-05,
+      "loss": 0.7484929084777832,
+      "step": 1270
+    },
+    {
+      "epoch": 0.03509637794411999,
+      "grad_norm": 1.6372708148105157,
+      "learning_rate": 2e-05,
+      "loss": 0.6999114036560059,
+      "step": 1280
+    },
+    {
+      "epoch": 0.035370568396808424,
+      "grad_norm": 4.162341191850801,
+      "learning_rate": 2e-05,
+      "loss": 0.7129648208618165,
+      "step": 1290
+    },
+    {
+      "epoch": 0.03564475884949686,
+      "grad_norm": 4.017056556093628,
+      "learning_rate": 2e-05,
+      "loss": 0.7217267036437989,
+      "step": 1300
+    },
+    {
+      "epoch": 0.035918949302185296,
+      "grad_norm": 9.786655951652824,
+      "learning_rate": 2e-05,
+      "loss": 0.7361152648925782,
+      "step": 1310
+    },
+    {
+      "epoch": 0.03619313975487373,
+      "grad_norm": 17.798888728639277,
+      "learning_rate": 2e-05,
+      "loss": 0.7423253059387207,
+      "step": 1320
+    },
+    {
+      "epoch": 0.036467330207562175,
+      "grad_norm": 25.98230175781132,
+      "learning_rate": 2e-05,
+      "loss": 0.7079302787780761,
+      "step": 1330
+    },
+    {
+      "epoch": 0.03674152066025061,
+      "grad_norm": 4.107909151191186,
+      "learning_rate": 2e-05,
+      "loss": 0.7637813568115235,
+      "step": 1340
+    },
+    {
+      "epoch": 0.03701571111293905,
+      "grad_norm": 1.8177555687046802,
+      "learning_rate": 2e-05,
+      "loss": 0.7440331459045411,
+      "step": 1350
+    },
+    {
+      "epoch": 0.03728990156562748,
+      "grad_norm": 6.631474920460703,
+      "learning_rate": 2e-05,
+      "loss": 0.7420357704162598,
+      "step": 1360
+    },
+    {
+      "epoch": 0.037564092018315925,
+      "grad_norm": 1.182591249805126,
+      "learning_rate": 2e-05,
+      "loss": 0.7243555068969727,
+      "step": 1370
+    },
+    {
+      "epoch": 0.03783828247100436,
+      "grad_norm": 1.5239290367740697,
+      "learning_rate": 2e-05,
+      "loss": 0.7780046463012695,
+      "step": 1380
+    },
+    {
+      "epoch": 0.0381124729236928,
+      "grad_norm": 1.9611702789468477,
+      "learning_rate": 2e-05,
+      "loss": 0.7606742858886719,
+      "step": 1390
+    },
+    {
+      "epoch": 0.03838666337638123,
+      "grad_norm": 3.4538398959809564,
+      "learning_rate": 2e-05,
+      "loss": 0.7161018371582031,
+      "step": 1400
+    },
+    {
+      "epoch": 0.03866085382906967,
+      "grad_norm": 6.74405592418129,
+      "learning_rate": 2e-05,
+      "loss": 0.7269341945648193,
+      "step": 1410
+    },
+    {
+      "epoch": 0.03893504428175811,
+      "grad_norm": 5.0613023509206,
+      "learning_rate": 2e-05,
+      "loss": 0.7150623321533203,
+      "step": 1420
+    },
+    {
+      "epoch": 0.03920923473444655,
+      "grad_norm": 15.935601654785444,
+      "learning_rate": 2e-05,
+      "loss": 0.6991084098815918,
+      "step": 1430
+    },
+    {
+      "epoch": 0.039483425187134984,
+      "grad_norm": 189.58206876056943,
+      "learning_rate": 2e-05,
+      "loss": 0.715428352355957,
+      "step": 1440
+    },
+    {
+      "epoch": 0.03975761563982342,
+      "grad_norm": 13.224258279776146,
+      "learning_rate": 2e-05,
+      "loss": 0.7448666572570801,
+      "step": 1450
+    },
+    {
+      "epoch": 0.040031806092511855,
+      "grad_norm": 15.23890210961455,
+      "learning_rate": 2e-05,
+      "loss": 0.7545663833618164,
+      "step": 1460
+    },
+    {
+      "epoch": 0.0403059965452003,
+      "grad_norm": 2.089772579003186,
+      "learning_rate": 2e-05,
+      "loss": 0.7143968105316162,
+      "step": 1470
+    },
+    {
+      "epoch": 0.040580186997888734,
+      "grad_norm": 672.5570249806746,
+      "learning_rate": 2e-05,
+      "loss": 0.6955307960510254,
+      "step": 1480
+    },
+    {
+      "epoch": 0.04085437745057717,
+      "grad_norm": 2.499047720535622,
+      "learning_rate": 2e-05,
+      "loss": 0.7183216094970704,
+      "step": 1490
+    },
+    {
+      "epoch": 0.041128567903265606,
+      "grad_norm": 1.6820671266931866,
+      "learning_rate": 2e-05,
+      "loss": 0.7442133903503418,
+      "step": 1500
+    },
+    {
+      "epoch": 0.04140275835595405,
+      "grad_norm": 1.6322487359391142,
+      "learning_rate": 2e-05,
+      "loss": 0.7415005683898925,
+      "step": 1510
+    },
+    {
+      "epoch": 0.041676948808642485,
+      "grad_norm": 2.0768691766562974,
+      "learning_rate": 2e-05,
+      "loss": 0.7350949764251709,
+      "step": 1520
+    },
+    {
+      "epoch": 0.04195113926133092,
+      "grad_norm": 3.462638107939682,
+      "learning_rate": 2e-05,
+      "loss": 0.6958516597747803,
+      "step": 1530
+    },
+    {
+      "epoch": 0.042225329714019356,
+      "grad_norm": 1.9410342316482636,
+      "learning_rate": 2e-05,
+      "loss": 0.7019001960754394,
+      "step": 1540
+    },
+    {
+      "epoch": 0.04249952016670779,
+      "grad_norm": 2.2435234267896944,
+      "learning_rate": 2e-05,
+      "loss": 0.7038233757019043,
+      "step": 1550
+    },
+    {
+      "epoch": 0.042773710619396235,
+      "grad_norm": 50.31729516447788,
+      "learning_rate": 2e-05,
+      "loss": 0.6971765041351319,
+      "step": 1560
+    },
+    {
+      "epoch": 0.04304790107208467,
+      "grad_norm": 6.726153221416043,
+      "learning_rate": 2e-05,
+      "loss": 0.7901406764984131,
+      "step": 1570
+    },
+    {
+      "epoch": 0.04332209152477311,
+      "grad_norm": 25.290061958502374,
+      "learning_rate": 2e-05,
+      "loss": 0.7404637813568116,
+      "step": 1580
+    },
+    {
+      "epoch": 0.04359628197746154,
+      "grad_norm": 11.648387859593914,
+      "learning_rate": 2e-05,
+      "loss": 0.7325190544128418,
+      "step": 1590
+    },
+    {
+      "epoch": 0.04387047243014998,
+      "grad_norm": 3.998377511327454,
+      "learning_rate": 2e-05,
+      "loss": 0.6938001155853272,
+      "step": 1600
+    },
+    {
+      "epoch": 0.04414466288283842,
+      "grad_norm": 1.4609573707991412,
+      "learning_rate": 2e-05,
+      "loss": 0.6974159240722656,
+      "step": 1610
+    },
+    {
+      "epoch": 0.04441885333552686,
+      "grad_norm": 1.2908767847499276,
+      "learning_rate": 2e-05,
+      "loss": 0.735476303100586,
+      "step": 1620
+    },
+    {
+      "epoch": 0.04469304378821529,
+      "grad_norm": 1.5263587301999937,
+      "learning_rate": 2e-05,
+      "loss": 0.6849418640136719,
+      "step": 1630
+    },
+    {
+      "epoch": 0.04496723424090373,
+      "grad_norm": 2.71012026891521,
+      "learning_rate": 2e-05,
+      "loss": 0.6952274322509766,
+      "step": 1640
+    },
+    {
+      "epoch": 0.04524142469359217,
+      "grad_norm": 1.5738552850120824,
+      "learning_rate": 2e-05,
+      "loss": 0.7284816265106201,
+      "step": 1650
+    },
+    {
+      "epoch": 0.04551561514628061,
+      "grad_norm": 1.5435257666345013,
+      "learning_rate": 2e-05,
+      "loss": 0.7115949153900146,
+      "step": 1660
+    },
+    {
+      "epoch": 0.045789805598969044,
+      "grad_norm": 1.516425220632835,
+      "learning_rate": 2e-05,
+      "loss": 0.6966533660888672,
+      "step": 1670
+    },
+    {
+      "epoch": 0.04606399605165748,
+      "grad_norm": 1.4858677693562772,
+      "learning_rate": 2e-05,
+      "loss": 0.712862491607666,
+      "step": 1680
+    },
+    {
+      "epoch": 0.046338186504345916,
+      "grad_norm": 1.1928774431737874,
+      "learning_rate": 2e-05,
+      "loss": 0.715973949432373,
+      "step": 1690
+    },
+    {
+      "epoch": 0.04661237695703436,
+      "grad_norm": 1.3801750667103458,
+      "learning_rate": 2e-05,
+      "loss": 0.6785056114196777,
+      "step": 1700
+    },
+    {
+      "epoch": 0.046886567409722794,
+      "grad_norm": 1.1562227123885702,
+      "learning_rate": 2e-05,
+      "loss": 0.6806015014648438,
+      "step": 1710
+    },
+    {
+      "epoch": 0.04716075786241123,
+      "grad_norm": 1.966065169897555,
+      "learning_rate": 2e-05,
+      "loss": 0.6789681911468506,
+      "step": 1720
+    },
+    {
+      "epoch": 0.047434948315099666,
+      "grad_norm": 1.796240102087905,
+      "learning_rate": 2e-05,
+      "loss": 0.6643640041351319,
+      "step": 1730
+    },
+    {
+      "epoch": 0.04770913876778811,
+      "grad_norm": 1.4714599948408273,
+      "learning_rate": 2e-05,
+      "loss": 0.672383975982666,
+      "step": 1740
+    },
+    {
+      "epoch": 0.047983329220476545,
+      "grad_norm": 1.05613596470809,
+      "learning_rate": 2e-05,
+      "loss": 0.6921384334564209,
+      "step": 1750
+    },
+    {
+      "epoch": 0.04825751967316498,
+      "grad_norm": 1.168321764739896,
+      "learning_rate": 2e-05,
+      "loss": 0.6853407859802246,
+      "step": 1760
+    },
+    {
+      "epoch": 0.04853171012585342,
+      "grad_norm": 1.304231611987027,
+      "learning_rate": 2e-05,
+      "loss": 0.6831422328948975,
+      "step": 1770
+    },
+    {
+      "epoch": 0.04880590057854185,
+      "grad_norm": 1.271885101891083,
+      "learning_rate": 2e-05,
+      "loss": 0.7312606811523438,
+      "step": 1780
+    },
+    {
+      "epoch": 0.049080091031230295,
+      "grad_norm": 1.1274168419233672,
+      "learning_rate": 2e-05,
+      "loss": 0.7195809841156006,
+      "step": 1790
+    },
+    {
+      "epoch": 0.04935428148391873,
+      "grad_norm": 2.1060356897000196,
+      "learning_rate": 2e-05,
+      "loss": 0.6862109184265137,
+      "step": 1800
+    },
+    {
+      "epoch": 0.04962847193660717,
+      "grad_norm": 1.1668574261874323,
+      "learning_rate": 2e-05,
+      "loss": 0.7068725109100342,
+      "step": 1810
+    },
+    {
+      "epoch": 0.0499026623892956,
+      "grad_norm": 3.1583477159166673,
+      "learning_rate": 2e-05,
+      "loss": 0.7150220394134521,
+      "step": 1820
+    },
+    {
+      "epoch": 0.05017685284198404,
+      "grad_norm": 1.5843620172233683,
+      "learning_rate": 2e-05,
+      "loss": 0.66549072265625,
+      "step": 1830
+    },
+    {
+      "epoch": 0.05045104329467248,
+      "grad_norm": 1.2547928536390052,
+      "learning_rate": 2e-05,
+      "loss": 0.7375857353210449,
+      "step": 1840
+    },
+    {
+      "epoch": 0.05072523374736092,
+      "grad_norm": 4.29725259279178,
+      "learning_rate": 2e-05,
+      "loss": 0.6876885890960693,
+      "step": 1850
+    },
+    {
+      "epoch": 0.050999424200049354,
+      "grad_norm": 1.9025925412069744,
+      "learning_rate": 2e-05,
+      "loss": 0.7387830734252929,
+      "step": 1860
+    },
+    {
+      "epoch": 0.05127361465273779,
+      "grad_norm": 1.5406949311604028,
+      "learning_rate": 2e-05,
+      "loss": 0.6733179092407227,
+      "step": 1870
+    },
+    {
+      "epoch": 0.05154780510542623,
+      "grad_norm": 20.215212215144863,
+      "learning_rate": 2e-05,
+      "loss": 0.7654403686523438,
+      "step": 1880
+    },
+    {
+      "epoch": 0.05182199555811467,
+      "grad_norm": 1.5669531048405803,
+      "learning_rate": 2e-05,
+      "loss": 0.7350390911102295,
+      "step": 1890
+    },
+    {
+      "epoch": 0.052096186010803104,
+      "grad_norm": 7.2191546562371425,
+      "learning_rate": 2e-05,
+      "loss": 0.7414397716522216,
+      "step": 1900
+    },
+    {
+      "epoch": 0.05237037646349154,
+      "grad_norm": 1.2325274169727398,
+      "learning_rate": 2e-05,
+      "loss": 0.7655929088592529,
+      "step": 1910
+    },
+    {
+      "epoch": 0.052644566916179976,
+      "grad_norm": 1.1944179026527708,
+      "learning_rate": 2e-05,
+      "loss": 0.7075240135192871,
+      "step": 1920
+    },
+    {
+      "epoch": 0.05291875736886842,
+      "grad_norm": 1.302822879313398,
+      "learning_rate": 2e-05,
+      "loss": 0.6797014236450195,
+      "step": 1930
+    },
+    {
+      "epoch": 0.053192947821556855,
+      "grad_norm": 2.0168747405099747,
+      "learning_rate": 2e-05,
+      "loss": 0.7000571727752686,
+      "step": 1940
+    },
+    {
+      "epoch": 0.05346713827424529,
+      "grad_norm": 1.6344897385611805,
+      "learning_rate": 2e-05,
+      "loss": 0.7235137939453125,
+      "step": 1950
+    },
+    {
+      "epoch": 0.053741328726933726,
+      "grad_norm": 3.5766753006414125,
+      "learning_rate": 2e-05,
+      "loss": 0.7371991634368896,
+      "step": 1960
+    },
+    {
+      "epoch": 0.05401551917962216,
+      "grad_norm": 1.4953479982574733,
+      "learning_rate": 2e-05,
+      "loss": 0.7229987621307373,
+      "step": 1970
+    },
+    {
+      "epoch": 0.054289709632310605,
+      "grad_norm": 1.6598347475638233,
+      "learning_rate": 2e-05,
+      "loss": 0.6758342742919922,
+      "step": 1980
+    },
+    {
+      "epoch": 0.05456390008499904,
+      "grad_norm": 1.3876361065240026,
+      "learning_rate": 2e-05,
+      "loss": 0.6621166229248047,
+      "step": 1990
+    },
+    {
+      "epoch": 0.05483809053768748,
+      "grad_norm": 1.2953540155757288,
+      "learning_rate": 2e-05,
+      "loss": 0.701081657409668,
+      "step": 2000
+    },
+    {
+      "epoch": 0.05511228099037591,
+      "grad_norm": 1.0757376098134102,
+      "learning_rate": 2e-05,
+      "loss": 0.6949031829833985,
+      "step": 2010
+    },
+    {
+      "epoch": 0.055386471443064356,
+      "grad_norm": 1.3515745918409625,
+      "learning_rate": 2e-05,
+      "loss": 0.6849285125732422,
+      "step": 2020
+    },
+    {
+      "epoch": 0.05566066189575279,
+      "grad_norm": 3.4678908742290884,
+      "learning_rate": 2e-05,
+      "loss": 0.6919690608978272,
+      "step": 2030
+    },
+    {
+      "epoch": 0.05593485234844123,
+      "grad_norm": 1.9456705464214417,
+      "learning_rate": 2e-05,
+      "loss": 0.6663742065429688,
+      "step": 2040
+    },
+    {
+      "epoch": 0.05620904280112966,
+      "grad_norm": 1.2852941271459164,
+      "learning_rate": 2e-05,
+      "loss": 0.6898897171020508,
+      "step": 2050
+    },
+    {
+      "epoch": 0.0564832332538181,
+      "grad_norm": 7.958437531519753,
+      "learning_rate": 2e-05,
+      "loss": 0.7008088588714599,
+      "step": 2060
+    },
+    {
+      "epoch": 0.05675742370650654,
+      "grad_norm": 3.3113911508682685,
+      "learning_rate": 2e-05,
+      "loss": 0.6721383571624756,
+      "step": 2070
+    },
+    {
+      "epoch": 0.05703161415919498,
+      "grad_norm": 1.1175929950295507,
+      "learning_rate": 2e-05,
+      "loss": 0.7099536895751953,
+      "step": 2080
+    },
+    {
+      "epoch": 0.057305804611883414,
+      "grad_norm": 0.9164381223894205,
+      "learning_rate": 2e-05,
+      "loss": 0.6982614040374756,
+      "step": 2090
+    },
+    {
+      "epoch": 0.05757999506457185,
+      "grad_norm": 1.3064363870590416,
+      "learning_rate": 2e-05,
+      "loss": 0.6582142829895019,
+      "step": 2100
+    },
+    {
+      "epoch": 0.057854185517260286,
+      "grad_norm": 1.9295155170705978,
+      "learning_rate": 2e-05,
+      "loss": 0.7130876064300538,
+      "step": 2110
+    },
+    {
+      "epoch": 0.05812837596994873,
+      "grad_norm": 1.598328187450677,
+      "learning_rate": 2e-05,
+      "loss": 0.7056316375732422,
+      "step": 2120
+    },
+    {
+      "epoch": 0.058402566422637164,
+      "grad_norm": 1.3733360145402107,
+      "learning_rate": 2e-05,
+      "loss": 0.6908935546875,
+      "step": 2130
+    },
+    {
+      "epoch": 0.0586767568753256,
+      "grad_norm": 1.5835542241632976,
+      "learning_rate": 2e-05,
+      "loss": 0.6900417327880859,
+      "step": 2140
+    },
+    {
+      "epoch": 0.058950947328014036,
+      "grad_norm": 15.343250543565158,
+      "learning_rate": 2e-05,
+      "loss": 0.661806869506836,
+      "step": 2150
+    },
+    {
+      "epoch": 0.05922513778070248,
+      "grad_norm": 1.7535753341635816,
+      "learning_rate": 2e-05,
+      "loss": 0.6718528270721436,
+      "step": 2160
+    },
+    {
+      "epoch": 0.059499328233390915,
+      "grad_norm": 1.9376370630167024,
+      "learning_rate": 2e-05,
+      "loss": 0.690641212463379,
+      "step": 2170
+    },
+    {
+      "epoch": 0.05977351868607935,
+      "grad_norm": 2.5374679136605596,
+      "learning_rate": 2e-05,
+      "loss": 0.6993424415588378,
+      "step": 2180
+    },
+    {
+      "epoch": 0.06004770913876779,
+      "grad_norm": 8.884410141851701,
+      "learning_rate": 2e-05,
+      "loss": 0.6886792182922363,
+      "step": 2190
+    },
+    {
+      "epoch": 0.06032189959145622,
+      "grad_norm": 1.3716294166066423,
+      "learning_rate": 2e-05,
+      "loss": 0.7170482635498047,
+      "step": 2200
+    },
+    {
+      "epoch": 0.060596090044144665,
+      "grad_norm": 15.068560848475203,
+      "learning_rate": 2e-05,
+      "loss": 0.6805070877075196,
+      "step": 2210
+    },
+    {
+      "epoch": 0.0608702804968331,
+      "grad_norm": 1.3577298589073343,
+      "learning_rate": 2e-05,
+      "loss": 0.7325131893157959,
+      "step": 2220
+    },
+    {
+      "epoch": 0.06114447094952154,
+      "grad_norm": 1.5605906922192767,
+      "learning_rate": 2e-05,
+      "loss": 0.7512709617614746,
+      "step": 2230
+    },
+    {
+      "epoch": 0.06141866140220997,
+      "grad_norm": 2.7582484830984075,
+      "learning_rate": 2e-05,
+      "loss": 0.6785373687744141,
+      "step": 2240
+    },
+    {
+      "epoch": 0.061692851854898416,
+      "grad_norm": 1.5143827466792619,
+      "learning_rate": 2e-05,
+      "loss": 0.6751925945281982,
+      "step": 2250
+    },
+    {
+      "epoch": 0.06196704230758685,
+      "grad_norm": 2.0161013083730874,
+      "learning_rate": 2e-05,
+      "loss": 0.7139543533325196,
+      "step": 2260
+    },
+    {
+      "epoch": 0.06224123276027529,
+      "grad_norm": 1.5476890937684233,
+      "learning_rate": 2e-05,
+      "loss": 0.6949531555175781,
+      "step": 2270
+    },
+    {
+      "epoch": 0.06251542321296373,
+      "grad_norm": 2.854149288642599,
+      "learning_rate": 2e-05,
+      "loss": 0.7019545555114746,
+      "step": 2280
+    },
+    {
+      "epoch": 0.06278961366565217,
+      "grad_norm": 2.8910444598110145,
+      "learning_rate": 2e-05,
+      "loss": 0.679999828338623,
+      "step": 2290
+    },
+    {
+      "epoch": 0.0630638041183406,
+      "grad_norm": 4.5131786865492165,
+      "learning_rate": 2e-05,
+      "loss": 0.743858814239502,
+      "step": 2300
+    },
+    {
+      "epoch": 0.06333799457102904,
+      "grad_norm": 1.6666983651920946,
+      "learning_rate": 2e-05,
+      "loss": 0.6903061866760254,
+      "step": 2310
+    },
+    {
+      "epoch": 0.06361218502371747,
+      "grad_norm": 3.9236409182044927,
+      "learning_rate": 2e-05,
+      "loss": 0.6770228385925293,
+      "step": 2320
+    },
+    {
+      "epoch": 0.06388637547640591,
+      "grad_norm": 3.5304751421823375,
+      "learning_rate": 2e-05,
+      "loss": 0.683766508102417,
+      "step": 2330
+    },
+    {
+      "epoch": 0.06416056592909435,
+      "grad_norm": 2.3440415121115508,
+      "learning_rate": 2e-05,
+      "loss": 0.685692024230957,
+      "step": 2340
+    },
+    {
+      "epoch": 0.06443475638178278,
+      "grad_norm": 2.023495973731089,
+      "learning_rate": 2e-05,
+      "loss": 0.6623273372650147,
+      "step": 2350
+    },
+    {
+      "epoch": 0.06470894683447122,
+      "grad_norm": 1.3644101729258715,
+      "learning_rate": 2e-05,
+      "loss": 0.6689095497131348,
+      "step": 2360
+    },
+    {
+      "epoch": 0.06498313728715967,
+      "grad_norm": 1.9596672876784174,
+      "learning_rate": 2e-05,
+      "loss": 0.691974401473999,
+      "step": 2370
+    },
+    {
+      "epoch": 0.0652573277398481,
+      "grad_norm": 1.3195698986606514,
+      "learning_rate": 2e-05,
+      "loss": 0.6821478843688965,
+      "step": 2380
+    },
+    {
+      "epoch": 0.06553151819253654,
+      "grad_norm": 1.7573022373302587,
+      "learning_rate": 2e-05,
+      "loss": 0.6735329627990723,
+      "step": 2390
+    },
+    {
+      "epoch": 0.06580570864522498,
+      "grad_norm": 1.53838950058506,
+      "learning_rate": 2e-05,
+      "loss": 0.748023509979248,
+      "step": 2400
+    },
+    {
+      "epoch": 0.06607989909791341,
+      "grad_norm": 1.2532586297274881,
+      "learning_rate": 2e-05,
+      "loss": 0.7004266738891601,
+      "step": 2410
+    },
+    {
+      "epoch": 0.06635408955060185,
+      "grad_norm": 1.0700352542953793,
+      "learning_rate": 2e-05,
+      "loss": 0.6959473609924316,
+      "step": 2420
+    },
+    {
+      "epoch": 0.06662828000329028,
+      "grad_norm": 4.556276241049152,
+      "learning_rate": 2e-05,
+      "loss": 0.8061809539794922,
+      "step": 2430
+    },
+    {
+      "epoch": 0.06690247045597872,
+      "grad_norm": 2.152509533567621,
+      "learning_rate": 2e-05,
+      "loss": 0.6806859016418457,
+      "step": 2440
+    },
+    {
+      "epoch": 0.06717666090866715,
+      "grad_norm": 2.844321748025424,
+      "learning_rate": 2e-05,
+      "loss": 0.6640607833862304,
+      "step": 2450
+    },
+    {
+      "epoch": 0.0674508513613556,
+      "grad_norm": 1.1844711198885922,
+      "learning_rate": 2e-05,
+      "loss": 0.6736173629760742,
+      "step": 2460
+    },
+    {
+      "epoch": 0.06772504181404404,
+      "grad_norm": 1.7441815113714467,
+      "learning_rate": 2e-05,
+      "loss": 0.6737575531005859,
+      "step": 2470
+    },
+    {
+      "epoch": 0.06799923226673248,
+      "grad_norm": 15.110436266375306,
+      "learning_rate": 2e-05,
+      "loss": 0.6922641277313233,
+      "step": 2480
+    },
+    {
+      "epoch": 0.06827342271942091,
+      "grad_norm": 2.2896995422971513,
+      "learning_rate": 2e-05,
+      "loss": 0.6922240257263184,
+      "step": 2490
+    },
+    {
+      "epoch": 0.06854761317210935,
+      "grad_norm": 3.3946567144349524,
+      "learning_rate": 2e-05,
+      "loss": 0.7507569313049316,
+      "step": 2500
+    },
+    {
+      "epoch": 0.06882180362479778,
+      "grad_norm": 2.4027551835720606,
+      "learning_rate": 2e-05,
+      "loss": 0.6868194580078125,
+      "step": 2510
+    },
+    {
+      "epoch": 0.06909599407748622,
+      "grad_norm": 2.642622943517862,
+      "learning_rate": 2e-05,
+      "loss": 0.671518611907959,
+      "step": 2520
+    },
+    {
+      "epoch": 0.06937018453017466,
+      "grad_norm": 1.5309404635845658,
+      "learning_rate": 2e-05,
+      "loss": 0.7356051921844482,
+      "step": 2530
+    },
+    {
+      "epoch": 0.06964437498286309,
+      "grad_norm": 3.0548677174511956,
+      "learning_rate": 2e-05,
+      "loss": 0.6758328437805176,
+      "step": 2540
+    },
+    {
+      "epoch": 0.06991856543555153,
+      "grad_norm": 1.193438740119827,
+      "learning_rate": 2e-05,
+      "loss": 0.6692746639251709,
+      "step": 2550
+    },
+    {
+      "epoch": 0.07019275588823998,
+      "grad_norm": 1.2975925737053815,
+      "learning_rate": 2e-05,
+      "loss": 0.6996981620788574,
+      "step": 2560
+    },
+    {
+      "epoch": 0.07046694634092841,
+      "grad_norm": 2.604623408290158,
+      "learning_rate": 2e-05,
+      "loss": 0.6900700569152832,
+      "step": 2570
+    },
+    {
+      "epoch": 0.07074113679361685,
+      "grad_norm": 1.3002551593133926,
+      "learning_rate": 2e-05,
+      "loss": 0.6943139553070068,
+      "step": 2580
+    },
+    {
+      "epoch": 0.07101532724630528,
+      "grad_norm": 1.2531714833158747,
+      "learning_rate": 2e-05,
+      "loss": 0.6756250381469726,
+      "step": 2590
+    },
+    {
+      "epoch": 0.07128951769899372,
+      "grad_norm": 1.2937652650463045,
+      "learning_rate": 2e-05,
+      "loss": 0.663780403137207,
+      "step": 2600
+    },
+    {
+      "epoch": 0.07156370815168216,
+      "grad_norm": 2.6360485085938437,
+      "learning_rate": 2e-05,
+      "loss": 0.7122066020965576,
+      "step": 2610
+    },
+    {
+      "epoch": 0.07183789860437059,
+      "grad_norm": 1.4934934194790765,
+      "learning_rate": 2e-05,
+      "loss": 0.7055806159973145,
+      "step": 2620
+    },
+    {
+      "epoch": 0.07211208905705903,
+      "grad_norm": 1.265554029972572,
+      "learning_rate": 2e-05,
+      "loss": 0.7235059261322021,
+      "step": 2630
+    },
+    {
+      "epoch": 0.07238627950974746,
+      "grad_norm": 1.9958554767672445,
+      "learning_rate": 2e-05,
+      "loss": 0.7027205467224121,
+      "step": 2640
+    },
+    {
+      "epoch": 0.07266046996243591,
+      "grad_norm": 1.2413434845129339,
+      "learning_rate": 2e-05,
+      "loss": 0.6815381526947022,
+      "step": 2650
+    },
+    {
+      "epoch": 0.07293466041512435,
+      "grad_norm": 1.1371354821804263,
+      "learning_rate": 2e-05,
+      "loss": 0.6793872356414795,
+      "step": 2660
+    },
+    {
+      "epoch": 0.07320885086781279,
+      "grad_norm": 1.509310628738819,
+      "learning_rate": 2e-05,
+      "loss": 0.6791254997253418,
+      "step": 2670
+    },
+    {
+      "epoch": 0.07348304132050122,
+      "grad_norm": 1.1060317096479009,
+      "learning_rate": 2e-05,
+      "loss": 0.6734872341156006,
+      "step": 2680
+    },
+    {
+      "epoch": 0.07375723177318966,
+      "grad_norm": 2.300558523814945,
+      "learning_rate": 2e-05,
+      "loss": 0.7129344940185547,
+      "step": 2690
+    },
+    {
+      "epoch": 0.0740314222258781,
+      "grad_norm": 1.4427965933652498,
+      "learning_rate": 2e-05,
+      "loss": 0.6588455200195312,
+      "step": 2700
+    },
+    {
+      "epoch": 0.07430561267856653,
+      "grad_norm": 1.294782697054497,
+      "learning_rate": 2e-05,
+      "loss": 0.6733306884765625,
+      "step": 2710
+    },
+    {
+      "epoch": 0.07457980313125497,
+      "grad_norm": 1.5340099682623456,
+      "learning_rate": 2e-05,
+      "loss": 0.67325758934021,
+      "step": 2720
+    },
+    {
+      "epoch": 0.0748539935839434,
+      "grad_norm": 1.206231978423083,
+      "learning_rate": 2e-05,
+      "loss": 0.6762287139892578,
+      "step": 2730
+    },
+    {
+      "epoch": 0.07512818403663185,
+      "grad_norm": 1.9615078302959699,
+      "learning_rate": 2e-05,
+      "loss": 0.6494486808776856,
+      "step": 2740
+    },
+    {
+      "epoch": 0.07540237448932029,
+      "grad_norm": 2.7069935077316156,
+      "learning_rate": 2e-05,
+      "loss": 0.6673858642578125,
+      "step": 2750
+    },
+    {
+      "epoch": 0.07567656494200872,
+      "grad_norm": 1.244130150553988,
+      "learning_rate": 2e-05,
+      "loss": 0.6522541046142578,
+      "step": 2760
+    },
+    {
+      "epoch": 0.07595075539469716,
+      "grad_norm": 1.2098821554953647,
+      "learning_rate": 2e-05,
+      "loss": 0.683398962020874,
+      "step": 2770
+    },
+    {
+      "epoch": 0.0762249458473856,
+      "grad_norm": 1.288135083537603,
+      "learning_rate": 2e-05,
+      "loss": 0.6805778503417969,
+      "step": 2780
+    },
+    {
+      "epoch": 0.07649913630007403,
+      "grad_norm": 1.6304988366424138,
+      "learning_rate": 2e-05,
+      "loss": 0.7026628017425537,
+      "step": 2790
+    },
+    {
+      "epoch": 0.07677332675276247,
+      "grad_norm": 2.7936841529325167,
+      "learning_rate": 2e-05,
+      "loss": 0.6661033630371094,
+      "step": 2800
+    },
+    {
+      "epoch": 0.0770475172054509,
+      "grad_norm": 1.0298921513328017,
+      "learning_rate": 2e-05,
+      "loss": 0.6871449470520019,
+      "step": 2810
+    },
+    {
+      "epoch": 0.07732170765813934,
+      "grad_norm": 1.1609549565762298,
+      "learning_rate": 2e-05,
+      "loss": 0.7018230438232422,
+      "step": 2820
+    },
+    {
+      "epoch": 0.07759589811082779,
+      "grad_norm": 1.4937627603643764,
+      "learning_rate": 2e-05,
+      "loss": 0.6921014785766602,
+      "step": 2830
+    },
+    {
+      "epoch": 0.07787008856351622,
+      "grad_norm": 1.06750496448494,
+      "learning_rate": 2e-05,
+      "loss": 0.6660943508148194,
+      "step": 2840
+    },
+    {
+      "epoch": 0.07814427901620466,
+      "grad_norm": 1.3483148967347722,
+      "learning_rate": 2e-05,
+      "loss": 0.6800715923309326,
+      "step": 2850
+    },
+    {
+      "epoch": 0.0784184694688931,
+      "grad_norm": 0.9156475317981986,
+      "learning_rate": 2e-05,
+      "loss": 0.6963984489440918,
+      "step": 2860
+    },
+    {
+      "epoch": 0.07869265992158153,
+      "grad_norm": 2.541415851603761,
+      "learning_rate": 2e-05,
+      "loss": 0.6630027294158936,
+      "step": 2870
+    },
+    {
+      "epoch": 0.07896685037426997,
+      "grad_norm": 16.943114843325624,
+      "learning_rate": 2e-05,
+      "loss": 0.6456388473510742,
+      "step": 2880
+    },
+    {
+      "epoch": 0.0792410408269584,
+      "grad_norm": 1.5105418227638865,
+      "learning_rate": 2e-05,
+      "loss": 0.6730733394622803,
+      "step": 2890
+    },
+    {
+      "epoch": 0.07951523127964684,
+      "grad_norm": 1.0845416608133904,
+      "learning_rate": 2e-05,
+      "loss": 0.7171577453613281,
+      "step": 2900
+    },
+    {
+      "epoch": 0.07978942173233528,
+      "grad_norm": 1.1519480568569667,
+      "learning_rate": 2e-05,
+      "loss": 0.6709290504455566,
+      "step": 2910
+    },
+    {
+      "epoch": 0.08006361218502371,
+      "grad_norm": 1.2117290732120638,
+      "learning_rate": 2e-05,
+      "loss": 0.6620571136474609,
+      "step": 2920
+    },
+    {
+      "epoch": 0.08033780263771216,
+      "grad_norm": 1.2315110130143458,
+      "learning_rate": 2e-05,
+      "loss": 0.6733248710632325,
+      "step": 2930
+    },
+    {
+      "epoch": 0.0806119930904006,
+      "grad_norm": 2.440287172046704,
+      "learning_rate": 2e-05,
+      "loss": 0.6643840789794921,
+      "step": 2940
+    },
+    {
+      "epoch": 0.08088618354308903,
+      "grad_norm": 1.679547493244505,
+      "learning_rate": 2e-05,
+      "loss": 0.6582764625549317,
+      "step": 2950
+    },
+    {
+      "epoch": 0.08116037399577747,
+      "grad_norm": 1.4899450244629049,
+      "learning_rate": 2e-05,
+      "loss": 0.6351748466491699,
+      "step": 2960
+    },
+    {
+      "epoch": 0.0814345644484659,
+      "grad_norm": 8.727623946787714,
+      "learning_rate": 2e-05,
+      "loss": 0.6684012413024902,
+      "step": 2970
+    },
+    {
+      "epoch": 0.08170875490115434,
+      "grad_norm": 1.579015056993268,
+      "learning_rate": 2e-05,
+      "loss": 0.7062510490417481,
+      "step": 2980
+    },
+    {
+      "epoch": 0.08198294535384278,
+      "grad_norm": 1.1464807229444751,
+      "learning_rate": 2e-05,
+      "loss": 0.7049516201019287,
+      "step": 2990
+    },
+    {
+      "epoch": 0.08225713580653121,
+      "grad_norm": 5.869960507362621,
+      "learning_rate": 2e-05,
+      "loss": 0.6782597541809082,
+      "step": 3000
+    },
+    {
+      "epoch": 0.08253132625921965,
+      "grad_norm": 1.4182529764185894,
+      "learning_rate": 2e-05,
+      "loss": 0.6684169769287109,
+      "step": 3010
+    },
+    {
+      "epoch": 0.0828055167119081,
+      "grad_norm": 1.6893769027297443,
+      "learning_rate": 2e-05,
+      "loss": 0.6985646247863769,
+      "step": 3020
+    },
+    {
+      "epoch": 0.08307970716459653,
+      "grad_norm": 2.5342955627283548,
+      "learning_rate": 2e-05,
+      "loss": 0.6566522121429443,
+      "step": 3030
+    },
+    {
+      "epoch": 0.08335389761728497,
+      "grad_norm": 2.2039912383675504,
+      "learning_rate": 2e-05,
+      "loss": 0.6650538444519043,
+      "step": 3040
+    },
+    {
+      "epoch": 0.0836280880699734,
+      "grad_norm": 1.1365614280802196,
+      "learning_rate": 2e-05,
+      "loss": 0.7018685340881348,
+      "step": 3050
+    },
+    {
+      "epoch": 0.08390227852266184,
+      "grad_norm": 9.019421975743805,
+      "learning_rate": 2e-05,
+      "loss": 0.6758551597595215,
+      "step": 3060
+    },
+    {
+      "epoch": 0.08417646897535028,
+      "grad_norm": 3.6659007510462036,
+      "learning_rate": 2e-05,
+      "loss": 0.6577141761779786,
+      "step": 3070
+    },
+    {
+      "epoch": 0.08445065942803871,
+      "grad_norm": 1.5678832350052356,
+      "learning_rate": 2e-05,
+      "loss": 0.6269432067871094,
+      "step": 3080
+    },
+    {
+      "epoch": 0.08472484988072715,
+      "grad_norm": 1.727670242813246,
+      "learning_rate": 2e-05,
+      "loss": 0.6586098670959473,
+      "step": 3090
+    },
+    {
+      "epoch": 0.08499904033341558,
+      "grad_norm": 3.755002398572796,
+      "learning_rate": 2e-05,
+      "loss": 0.7226817607879639,
+      "step": 3100
+    },
+    {
+      "epoch": 0.08527323078610403,
+      "grad_norm": 2.1330764933548174,
+      "learning_rate": 2e-05,
+      "loss": 0.7848432064056396,
+      "step": 3110
+    },
+    {
+      "epoch": 0.08554742123879247,
+      "grad_norm": 7.344569339937305,
+      "learning_rate": 2e-05,
+      "loss": 0.6657351970672607,
+      "step": 3120
+    },
+    {
+      "epoch": 0.0858216116914809,
+      "grad_norm": 10.468008663230526,
+      "learning_rate": 2e-05,
+      "loss": 0.7568781852722168,
+      "step": 3130
+    },
+    {
+      "epoch": 0.08609580214416934,
+      "grad_norm": 2.1138000067535527,
+      "learning_rate": 2e-05,
+      "loss": 0.6460443496704101,
+      "step": 3140
+    },
+    {
+      "epoch": 0.08636999259685778,
+      "grad_norm": 4.547865280742984,
+      "learning_rate": 2e-05,
+      "loss": 0.6882412910461426,
+      "step": 3150
+    },
+    {
+      "epoch": 0.08664418304954621,
+      "grad_norm": 3.5154327639954515,
+      "learning_rate": 2e-05,
+      "loss": 0.6506322860717774,
+      "step": 3160
+    },
+    {
+      "epoch": 0.08691837350223465,
+      "grad_norm": 2.7275478121939303,
+      "learning_rate": 2e-05,
+      "loss": 0.6696713447570801,
+      "step": 3170
+    },
+    {
+      "epoch": 0.08719256395492309,
+      "grad_norm": 1.3787630646286215,
+      "learning_rate": 2e-05,
+      "loss": 0.6836231231689454,
+      "step": 3180
+    },
+    {
+      "epoch": 0.08746675440761152,
+      "grad_norm": 2.047844653500189,
+      "learning_rate": 2e-05,
+      "loss": 0.6477073669433594,
+      "step": 3190
+    },
+    {
+      "epoch": 0.08774094486029996,
+      "grad_norm": 1.3852125913708124,
+      "learning_rate": 2e-05,
+      "loss": 0.6763352870941162,
+      "step": 3200
+    },
+    {
+      "epoch": 0.08801513531298841,
+      "grad_norm": 2.3040797052670854,
+      "learning_rate": 2e-05,
+      "loss": 0.6652051448822022,
+      "step": 3210
+    },
+    {
+      "epoch": 0.08828932576567684,
+      "grad_norm": 14.697316208995012,
+      "learning_rate": 2e-05,
+      "loss": 0.7217847347259522,
+      "step": 3220
+    },
+    {
+      "epoch": 0.08856351621836528,
+      "grad_norm": 2.51269929770817,
+      "learning_rate": 2e-05,
+      "loss": 0.6437290191650391,
+      "step": 3230
+    },
+    {
+      "epoch": 0.08883770667105371,
+      "grad_norm": 1.0839928596474344,
+      "learning_rate": 2e-05,
+      "loss": 0.6739638328552247,
+      "step": 3240
+    },
+    {
+      "epoch": 0.08911189712374215,
+      "grad_norm": 1.4232969395596793,
+      "learning_rate": 2e-05,
+      "loss": 0.6545791625976562,
+      "step": 3250
+    },
+    {
+      "epoch": 0.08938608757643059,
+      "grad_norm": 2.7705768107632407,
+      "learning_rate": 2e-05,
+      "loss": 0.6883786678314209,
+      "step": 3260
+    },
+    {
+      "epoch": 0.08966027802911902,
+      "grad_norm": 1.6294188609939928,
+      "learning_rate": 2e-05,
+      "loss": 0.6384977340698242,
+      "step": 3270
+    },
+    {
+      "epoch": 0.08993446848180746,
+      "grad_norm": 1.594582926676938,
+      "learning_rate": 2e-05,
+      "loss": 0.6840876579284668,
+      "step": 3280
+    },
+    {
+      "epoch": 0.0902086589344959,
+      "grad_norm": 3.419979140068578,
+      "learning_rate": 2e-05,
+      "loss": 0.6966186046600342,
+      "step": 3290
+    },
+    {
+      "epoch": 0.09048284938718434,
+      "grad_norm": 1.1720984784354762,
+      "learning_rate": 2e-05,
+      "loss": 0.7145553588867187,
+      "step": 3300
+    },
+    {
+      "epoch": 0.09075703983987278,
+      "grad_norm": 2.5865997974005905,
+      "learning_rate": 2e-05,
+      "loss": 0.670125675201416,
+      "step": 3310
+    },
+    {
+      "epoch": 0.09103123029256122,
+      "grad_norm": 1.7653476493006024,
+      "learning_rate": 2e-05,
+      "loss": 0.6601348876953125,
+      "step": 3320
+    },
+    {
+      "epoch": 0.09130542074524965,
+      "grad_norm": 1.078475603753133,
+      "learning_rate": 2e-05,
+      "loss": 0.6758632659912109,
+      "step": 3330
+    },
+    {
+      "epoch": 0.09157961119793809,
+      "grad_norm": 1.9813753893710975,
+      "learning_rate": 2e-05,
+      "loss": 0.6324165344238282,
+      "step": 3340
+    },
+    {
+      "epoch": 0.09185380165062652,
+      "grad_norm": 5.357267664807992,
+      "learning_rate": 2e-05,
+      "loss": 0.7044012069702148,
+      "step": 3350
+    },
+    {
+      "epoch": 0.09212799210331496,
+      "grad_norm": 1.9661122621370217,
+      "learning_rate": 2e-05,
+      "loss": 0.6732365608215332,
+      "step": 3360
+    },
+    {
+      "epoch": 0.0924021825560034,
+      "grad_norm": 1.463741721870027,
+      "learning_rate": 2e-05,
+      "loss": 0.6740346431732178,
+      "step": 3370
+    },
+    {
+      "epoch": 0.09267637300869183,
+      "grad_norm": 2.1100577721208684,
+      "learning_rate": 2e-05,
+      "loss": 0.6793565273284912,
+      "step": 3380
+    },
+    {
+      "epoch": 0.09295056346138028,
+      "grad_norm": 1.647800752209185,
+      "learning_rate": 2e-05,
+      "loss": 0.6734954357147217,
+      "step": 3390
+    },
+    {
+      "epoch": 0.09322475391406872,
+      "grad_norm": 5.2930839705776425,
+      "learning_rate": 2e-05,
+      "loss": 0.6649512767791748,
+      "step": 3400
+    },
+    {
+      "epoch": 0.09349894436675715,
+      "grad_norm": 1.2957879552995335,
+      "learning_rate": 2e-05,
+      "loss": 0.6656761169433594,
+      "step": 3410
+    },
+    {
+      "epoch": 0.09377313481944559,
+      "grad_norm": 2.2505114790067173,
+      "learning_rate": 2e-05,
+      "loss": 0.6644682884216309,
+      "step": 3420
+    },
+    {
+      "epoch": 0.09404732527213402,
+      "grad_norm": 3.242876354488533,
+      "learning_rate": 2e-05,
+      "loss": 0.7099063873291016,
+      "step": 3430
+    },
+    {
+      "epoch": 0.09432151572482246,
+      "grad_norm": 1.2326965640901657,
+      "learning_rate": 2e-05,
+      "loss": 0.6550569057464599,
+      "step": 3440
+    },
+    {
+      "epoch": 0.0945957061775109,
+      "grad_norm": 1.1460797675833456,
+      "learning_rate": 2e-05,
+      "loss": 0.7175025463104248,
+      "step": 3450
+    },
+    {
+      "epoch": 0.09486989663019933,
+      "grad_norm": 1.2339953246226074,
+      "learning_rate": 2e-05,
+      "loss": 0.6755492687225342,
+      "step": 3460
+    },
+    {
+      "epoch": 0.09514408708288777,
+      "grad_norm": 1.2239178623958433,
+      "learning_rate": 2e-05,
+      "loss": 0.7038548469543457,
+      "step": 3470
+    },
+    {
+      "epoch": 0.09541827753557622,
+      "grad_norm": 2.054453414781126,
+      "learning_rate": 2e-05,
+      "loss": 0.7407256126403808,
+      "step": 3480
+    },
+    {
+      "epoch": 0.09569246798826465,
+      "grad_norm": 0.962580766048047,
+      "learning_rate": 2e-05,
+      "loss": 0.6635346412658691,
+      "step": 3490
+    },
+    {
+      "epoch": 0.09596665844095309,
+      "grad_norm": 1.2912100361591508,
+      "learning_rate": 2e-05,
+      "loss": 0.7218295574188233,
+      "step": 3500
+    },
+    {
+      "epoch": 0.09624084889364153,
+      "grad_norm": 1.1758629619276566,
+      "learning_rate": 2e-05,
+      "loss": 0.6495042324066163,
+      "step": 3510
+    },
+    {
+      "epoch": 0.09651503934632996,
+      "grad_norm": 3.5366805734644338,
+      "learning_rate": 2e-05,
+      "loss": 0.6897785186767578,
+      "step": 3520
+    },
+    {
+      "epoch": 0.0967892297990184,
+      "grad_norm": 2.272234739463371,
+      "learning_rate": 2e-05,
+      "loss": 0.6400076866149902,
+      "step": 3530
+    },
+    {
+      "epoch": 0.09706342025170683,
+      "grad_norm": 1.0410457309022894,
+      "learning_rate": 2e-05,
+      "loss": 0.6666034698486328,
+      "step": 3540
+    },
+    {
+      "epoch": 0.09733761070439527,
+      "grad_norm": 1.47293937630036,
+      "learning_rate": 2e-05,
+      "loss": 0.6687037467956543,
+      "step": 3550
+    },
+    {
+      "epoch": 0.0976118011570837,
+      "grad_norm": 1.9748563321883545,
+      "learning_rate": 2e-05,
+      "loss": 0.6717099666595459,
+      "step": 3560
+    },
+    {
+      "epoch": 0.09788599160977214,
+      "grad_norm": 1.5933750546326055,
+      "learning_rate": 2e-05,
+      "loss": 0.6562514305114746,
+      "step": 3570
+    },
+    {
+      "epoch": 0.09816018206246059,
+      "grad_norm": 3.128593619633481,
+      "learning_rate": 2e-05,
+      "loss": 0.7533950805664062,
+      "step": 3580
+    },
+    {
+      "epoch": 0.09843437251514903,
+      "grad_norm": 2.891793023072984,
+      "learning_rate": 2e-05,
+      "loss": 0.6750978469848633,
+      "step": 3590
+    },
+    {
+      "epoch": 0.09870856296783746,
+      "grad_norm": 2.1897121019394965,
+      "learning_rate": 2e-05,
+      "loss": 0.6912845134735107,
+      "step": 3600
+    },
+    {
+      "epoch": 0.0989827534205259,
+      "grad_norm": 2.004346028083513,
+      "learning_rate": 2e-05,
+      "loss": 0.6787797927856445,
+      "step": 3610
+    },
+    {
+      "epoch": 0.09925694387321433,
+      "grad_norm": 2.915237046123075,
+      "learning_rate": 2e-05,
+      "loss": 0.6627199172973632,
+      "step": 3620
+    },
+    {
+      "epoch": 0.09953113432590277,
+      "grad_norm": 1.135046615758623,
+      "learning_rate": 2e-05,
+      "loss": 0.6445761680603027,
+      "step": 3630
+    },
+    {
+      "epoch": 0.0998053247785912,
+      "grad_norm": 1.3959275315420312,
+      "learning_rate": 2e-05,
+      "loss": 0.667326545715332,
+      "step": 3640
+    },
+    {
+      "epoch": 0.10007951523127964,
+      "grad_norm": 0.9909291231965143,
+      "learning_rate": 2e-05,
+      "loss": 0.6794458389282226,
+      "step": 3650
+    },
+    {
+      "epoch": 0.10035370568396808,
+      "grad_norm": 1.8845470713145256,
+      "learning_rate": 2e-05,
+      "loss": 0.6390461921691895,
+      "step": 3660
+    },
+    {
+      "epoch": 0.10062789613665653,
+      "grad_norm": 1.5618410538502554,
+      "learning_rate": 2e-05,
+      "loss": 0.6542781829833985,
+      "step": 3670
+    },
+    {
+      "epoch": 0.10090208658934496,
+      "grad_norm": 1.1053530735550423,
+      "learning_rate": 2e-05,
+      "loss": 0.6584223747253418,
+      "step": 3680
+    },
+    {
+      "epoch": 0.1011762770420334,
+      "grad_norm": 1.8050013341368214,
+      "learning_rate": 2e-05,
+      "loss": 0.6776100158691406,
+      "step": 3690
+    },
+    {
+      "epoch": 0.10145046749472184,
+      "grad_norm": 1.3509209662153294,
+      "learning_rate": 2e-05,
+      "loss": 0.7505715847015381,
+      "step": 3700
+    },
+    {
+      "epoch": 0.10172465794741027,
+      "grad_norm": 1.3171737413682911,
+      "learning_rate": 2e-05,
+      "loss": 0.6496476650238037,
+      "step": 3710
+    },
+    {
+      "epoch": 0.10199884840009871,
+      "grad_norm": 2.037728719646489,
+      "learning_rate": 2e-05,
+      "loss": 0.6601945400238037,
+      "step": 3720
+    },
+    {
+      "epoch": 0.10227303885278714,
+      "grad_norm": 0.9419204630678888,
+      "learning_rate": 2e-05,
+      "loss": 0.6456241607666016,
+      "step": 3730
+    },
+    {
+      "epoch": 0.10254722930547558,
+      "grad_norm": 0.9978934523669455,
+      "learning_rate": 2e-05,
+      "loss": 0.6376785278320313,
+      "step": 3740
+    },
+    {
+      "epoch": 0.10282141975816401,
+      "grad_norm": 1.2463996045138312,
+      "learning_rate": 2e-05,
+      "loss": 0.6607570648193359,
+      "step": 3750
+    },
+    {
+      "epoch": 0.10309561021085246,
+      "grad_norm": 2.5633755688709,
+      "learning_rate": 2e-05,
+      "loss": 0.6273582458496094,
+      "step": 3760
+    },
+    {
+      "epoch": 0.1033698006635409,
+      "grad_norm": 1.3319182781528256,
+      "learning_rate": 2e-05,
+      "loss": 0.653649091720581,
+      "step": 3770
+    },
+    {
+      "epoch": 0.10364399111622934,
+      "grad_norm": 1.5739184819076273,
+      "learning_rate": 2e-05,
+      "loss": 0.6797492027282714,
+      "step": 3780
+    },
+    {
+      "epoch": 0.10391818156891777,
+      "grad_norm": 1.0307904527745226,
+      "learning_rate": 2e-05,
+      "loss": 0.6382014274597168,
+      "step": 3790
+    },
+    {
+      "epoch": 0.10419237202160621,
+      "grad_norm": 1.367080564181282,
+      "learning_rate": 2e-05,
+      "loss": 0.656506872177124,
+      "step": 3800
+    },
+    {
+      "epoch": 0.10446656247429464,
+      "grad_norm": 1.1306043565346662,
+      "learning_rate": 2e-05,
+      "loss": 0.6709275722503663,
+      "step": 3810
+    },
+    {
+      "epoch": 0.10474075292698308,
+      "grad_norm": 1.1484780247773891,
+      "learning_rate": 2e-05,
+      "loss": 0.6394906520843506,
+      "step": 3820
+    },
+    {
+      "epoch": 0.10501494337967152,
+      "grad_norm": 1.1364725089670198,
+      "learning_rate": 2e-05,
+      "loss": 0.6385705947875977,
+      "step": 3830
+    },
+    {
+      "epoch": 0.10528913383235995,
+      "grad_norm": 0.8720469361835742,
+      "learning_rate": 2e-05,
+      "loss": 0.648165512084961,
+      "step": 3840
+    },
+    {
+      "epoch": 0.10556332428504839,
+      "grad_norm": 1.0753251518024716,
+      "learning_rate": 2e-05,
+      "loss": 0.7535962581634521,
+      "step": 3850
+    },
+    {
+      "epoch": 0.10583751473773684,
+      "grad_norm": 2.0842578780523646,
+      "learning_rate": 2e-05,
+      "loss": 0.6594803810119629,
+      "step": 3860
+    },
+    {
+      "epoch": 0.10611170519042527,
+      "grad_norm": 1.2009333278099745,
+      "learning_rate": 2e-05,
+      "loss": 0.6642222881317139,
+      "step": 3870
+    },
+    {
+      "epoch": 0.10638589564311371,
+      "grad_norm": 1.336651409297708,
+      "learning_rate": 2e-05,
+      "loss": 0.6247560024261475,
+      "step": 3880
+    },
+    {
+      "epoch": 0.10666008609580215,
+      "grad_norm": 1.2693616052452827,
+      "learning_rate": 2e-05,
+      "loss": 0.6595481395721435,
+      "step": 3890
+    },
+    {
+      "epoch": 0.10693427654849058,
+      "grad_norm": 1.0968413832844812,
+      "learning_rate": 2e-05,
+      "loss": 0.6259689331054688,
+      "step": 3900
+    },
+    {
+      "epoch": 0.10720846700117902,
+      "grad_norm": 1.4406885841286199,
+      "learning_rate": 2e-05,
+      "loss": 0.635016393661499,
+      "step": 3910
+    },
+    {
+      "epoch": 0.10748265745386745,
+      "grad_norm": 1.1149430537042344,
+      "learning_rate": 2e-05,
+      "loss": 0.652572774887085,
+      "step": 3920
+    },
+    {
+      "epoch": 0.10775684790655589,
+      "grad_norm": 1.4284277973722026,
+      "learning_rate": 2e-05,
+      "loss": 0.6603116512298584,
+      "step": 3930
+    },
+    {
+      "epoch": 0.10803103835924432,
+      "grad_norm": 2.191046426715608,
+      "learning_rate": 2e-05,
+      "loss": 0.6569153785705566,
+      "step": 3940
+    },
+    {
+      "epoch": 0.10830522881193277,
+      "grad_norm": 0.9755637485958417,
+      "learning_rate": 2e-05,
+      "loss": 0.6817933559417725,
+      "step": 3950
+    },
+    {
+      "epoch": 0.10857941926462121,
+      "grad_norm": 0.9207426742604556,
+      "learning_rate": 2e-05,
+      "loss": 0.632993507385254,
+      "step": 3960
+    },
+    {
+      "epoch": 0.10885360971730965,
+      "grad_norm": 1.1075412445990842,
+      "learning_rate": 2e-05,
+      "loss": 0.6479109764099121,
+      "step": 3970
+    },
+    {
+      "epoch": 0.10912780016999808,
+      "grad_norm": 1.02734487686768,
+      "learning_rate": 2e-05,
+      "loss": 0.6399119853973388,
+      "step": 3980
+    },
+    {
+      "epoch": 0.10940199062268652,
+      "grad_norm": 1.0062247863415956,
+      "learning_rate": 2e-05,
+      "loss": 0.6592807292938232,
+      "step": 3990
+    },
+    {
+      "epoch": 0.10967618107537495,
+      "grad_norm": 1.7207229565430802,
+      "learning_rate": 2e-05,
+      "loss": 0.6665008544921875,
+      "step": 4000
+    },
+    {
+      "epoch": 0.10995037152806339,
+      "grad_norm": 8.645042736815913,
+      "learning_rate": 2e-05,
+      "loss": 0.6616263866424561,
+      "step": 4010
+    },
+    {
+      "epoch": 0.11022456198075183,
+      "grad_norm": 1.2238886957987656,
+      "learning_rate": 2e-05,
+      "loss": 0.6552308559417724,
+      "step": 4020
+    },
+    {
+      "epoch": 0.11049875243344026,
+      "grad_norm": 1.1553638761775333,
+      "learning_rate": 2e-05,
+      "loss": 0.6816983699798584,
+      "step": 4030
+    },
+    {
+      "epoch": 0.11077294288612871,
+      "grad_norm": 1.054398000948271,
+      "learning_rate": 2e-05,
+      "loss": 0.6408511161804199,
+      "step": 4040
+    },
+    {
+      "epoch": 0.11104713333881715,
+      "grad_norm": 0.9428030212455841,
+      "learning_rate": 2e-05,
+      "loss": 0.6279231071472168,
+      "step": 4050
+    },
+    {
+      "epoch": 0.11132132379150558,
+      "grad_norm": 1.388635722554096,
+      "learning_rate": 2e-05,
+      "loss": 0.6294130325317383,
+      "step": 4060
+    },
+    {
+      "epoch": 0.11159551424419402,
+      "grad_norm": 1.7370583970603126,
+      "learning_rate": 2e-05,
+      "loss": 0.7167839050292969,
+      "step": 4070
+    },
+    {
+      "epoch": 0.11186970469688245,
+      "grad_norm": 0.9284089201374223,
+      "learning_rate": 2e-05,
+      "loss": 0.6723546028137207,
+      "step": 4080
+    },
+    {
+      "epoch": 0.11214389514957089,
+      "grad_norm": 1.0672289545982019,
+      "learning_rate": 2e-05,
+      "loss": 0.6188468933105469,
+      "step": 4090
+    },
+    {
+      "epoch": 0.11241808560225933,
+      "grad_norm": 1.2224883678356142,
+      "learning_rate": 2e-05,
+      "loss": 0.6085614681243896,
+      "step": 4100
+    },
+    {
+      "epoch": 0.11269227605494776,
+      "grad_norm": 1.0181735306150577,
+      "learning_rate": 2e-05,
+      "loss": 0.6458304405212403,
+      "step": 4110
+    },
+    {
+      "epoch": 0.1129664665076362,
+      "grad_norm": 0.9522441326878012,
+      "learning_rate": 2e-05,
+      "loss": 0.6111600875854493,
+      "step": 4120
+    },
+    {
+      "epoch": 0.11324065696032465,
+      "grad_norm": 1.369700590577185,
+      "learning_rate": 2e-05,
+      "loss": 0.6220932960510254,
+      "step": 4130
+    },
+    {
+      "epoch": 0.11351484741301308,
+      "grad_norm": 1.02959520927332,
+      "learning_rate": 2e-05,
+      "loss": 0.6411037445068359,
+      "step": 4140
+    },
+    {
+      "epoch": 0.11378903786570152,
+      "grad_norm": 1.0188167622915398,
+      "learning_rate": 2e-05,
+      "loss": 0.6618610382080078,
+      "step": 4150
+    },
+    {
+      "epoch": 0.11406322831838996,
+      "grad_norm": 1.1604217926496347,
+      "learning_rate": 2e-05,
+      "loss": 0.6259605407714843,
+      "step": 4160
+    },
+    {
+      "epoch": 0.11433741877107839,
+      "grad_norm": 1.3052055907048221,
+      "learning_rate": 2e-05,
+      "loss": 0.6735570430755615,
+      "step": 4170
+    },
+    {
+      "epoch": 0.11461160922376683,
+      "grad_norm": 3.9972621386948806,
+      "learning_rate": 2e-05,
+      "loss": 0.6412978172302246,
+      "step": 4180
+    },
+    {
+      "epoch": 0.11488579967645526,
+      "grad_norm": 0.9250935510724153,
+      "learning_rate": 2e-05,
+      "loss": 0.6349028587341309,
+      "step": 4190
+    },
+    {
+      "epoch": 0.1151599901291437,
+      "grad_norm": 0.8483293757756017,
+      "learning_rate": 2e-05,
+      "loss": 0.6539294242858886,
+      "step": 4200
+    },
+    {
+      "epoch": 0.11543418058183214,
+      "grad_norm": 0.9013909884649278,
+      "learning_rate": 2e-05,
+      "loss": 0.6296156883239746,
+      "step": 4210
+    },
+    {
+      "epoch": 0.11570837103452057,
+      "grad_norm": 0.9380979327027728,
+      "learning_rate": 2e-05,
+      "loss": 0.6285907745361328,
+      "step": 4220
+    },
+    {
+      "epoch": 0.11598256148720902,
+      "grad_norm": 2.2630772850038503,
+      "learning_rate": 2e-05,
+      "loss": 0.6578060150146484,
+      "step": 4230
+    },
+    {
+      "epoch": 0.11625675193989746,
+      "grad_norm": 6.909631861817211,
+      "learning_rate": 2e-05,
+      "loss": 0.6506007671356201,
+      "step": 4240
+    },
+    {
+      "epoch": 0.11653094239258589,
+      "grad_norm": 0.8808129720301275,
+      "learning_rate": 2e-05,
+      "loss": 0.6617231369018555,
+      "step": 4250
+    },
+    {
+      "epoch": 0.11680513284527433,
+      "grad_norm": 0.9456206590178073,
+      "learning_rate": 2e-05,
+      "loss": 0.6531835079193116,
+      "step": 4260
+    },
+    {
+      "epoch": 0.11707932329796276,
+      "grad_norm": 1.1627074594962703,
+      "learning_rate": 2e-05,
+      "loss": 0.6359607219696045,
+      "step": 4270
+    },
+    {
+      "epoch": 0.1173535137506512,
+      "grad_norm": 1.1919426603748915,
+      "learning_rate": 2e-05,
+      "loss": 0.6592881202697753,
+      "step": 4280
+    },
+    {
+      "epoch": 0.11762770420333964,
+      "grad_norm": 1.5858348663210315,
+      "learning_rate": 2e-05,
+      "loss": 0.6205570220947265,
+      "step": 4290
+    },
+    {
+      "epoch": 0.11790189465602807,
+      "grad_norm": 1.5199041276821281,
+      "learning_rate": 2e-05,
+      "loss": 0.631896686553955,
+      "step": 4300
+    },
+    {
+      "epoch": 0.11817608510871651,
+      "grad_norm": 9.32521636509595,
+      "learning_rate": 2e-05,
+      "loss": 0.6201819896697998,
+      "step": 4310
+    },
+    {
+      "epoch": 0.11845027556140496,
+      "grad_norm": 1.3787500933578838,
+      "learning_rate": 2e-05,
+      "loss": 0.6414646148681641,
+      "step": 4320
+    },
+    {
+      "epoch": 0.1187244660140934,
+      "grad_norm": 5.193321713943328,
+      "learning_rate": 2e-05,
+      "loss": 0.6398839473724365,
+      "step": 4330
+    },
+    {
+      "epoch": 0.11899865646678183,
+      "grad_norm": 1.0054479720570946,
+      "learning_rate": 2e-05,
+      "loss": 0.5866321563720703,
+      "step": 4340
+    },
+    {
+      "epoch": 0.11927284691947027,
+      "grad_norm": 1.7834494646230221,
+      "learning_rate": 2e-05,
+      "loss": 0.6555882930755615,
+      "step": 4350
+    },
+    {
+      "epoch": 0.1195470373721587,
+      "grad_norm": 3.2269796303236906,
+      "learning_rate": 2e-05,
+      "loss": 0.6395574092864991,
+      "step": 4360
+    },
+    {
+      "epoch": 0.11982122782484714,
+      "grad_norm": 1.033396076294825,
+      "learning_rate": 2e-05,
+      "loss": 0.6677106857299805,
+      "step": 4370
+    },
+    {
+      "epoch": 0.12009541827753557,
+      "grad_norm": 0.8846762414717687,
+      "learning_rate": 2e-05,
+      "loss": 0.6354843139648437,
+      "step": 4380
+    },
+    {
+      "epoch": 0.12036960873022401,
+      "grad_norm": 4.909904236871553,
+      "learning_rate": 2e-05,
+      "loss": 0.7135543823242188,
+      "step": 4390
+    },
+    {
+      "epoch": 0.12064379918291245,
+      "grad_norm": 1.113177327609433,
+      "learning_rate": 2e-05,
+      "loss": 0.6340306758880615,
+      "step": 4400
+    },
+    {
+      "epoch": 0.1209179896356009,
+      "grad_norm": 1.7841199253325009,
+      "learning_rate": 2e-05,
+      "loss": 0.6683245658874511,
+      "step": 4410
+    },
+    {
+      "epoch": 0.12119218008828933,
+      "grad_norm": 8.356137648680004,
+      "learning_rate": 2e-05,
+      "loss": 0.6603833675384522,
+      "step": 4420
+    },
+    {
+      "epoch": 0.12146637054097777,
+      "grad_norm": 1.4361812174766755,
+      "learning_rate": 2e-05,
+      "loss": 0.6408306121826172,
+      "step": 4430
+    },
+    {
+      "epoch": 0.1217405609936662,
+      "grad_norm": 1.1874981344620894,
+      "learning_rate": 2e-05,
+      "loss": 0.6650359153747558,
+      "step": 4440
+    },
+    {
+      "epoch": 0.12201475144635464,
+      "grad_norm": 4.563992000468378,
+      "learning_rate": 2e-05,
+      "loss": 0.6764625072479248,
+      "step": 4450
+    },
+    {
+      "epoch": 0.12228894189904307,
+      "grad_norm": 1.7866638362790574,
+      "learning_rate": 2e-05,
+      "loss": 0.685894775390625,
+      "step": 4460
+    },
+    {
+      "epoch": 0.12256313235173151,
+      "grad_norm": 1.8495475399267947,
+      "learning_rate": 2e-05,
+      "loss": 0.6657548427581788,
+      "step": 4470
+    },
+    {
+      "epoch": 0.12283732280441995,
+      "grad_norm": 1.9288699113594114,
+      "learning_rate": 2e-05,
+      "loss": 0.6878697872161865,
+      "step": 4480
+    },
+    {
+      "epoch": 0.12311151325710838,
+      "grad_norm": 0.832606879162686,
+      "learning_rate": 2e-05,
+      "loss": 0.6772202491760254,
+      "step": 4490
+    },
+    {
+      "epoch": 0.12338570370979683,
+      "grad_norm": 1.2684959078910074,
+      "learning_rate": 2e-05,
+      "loss": 0.6572877883911132,
+      "step": 4500
+    },
+    {
+      "epoch": 0.12365989416248527,
+      "grad_norm": 0.9100620504734364,
+      "learning_rate": 2e-05,
+      "loss": 0.6964794158935547,
+      "step": 4510
+    },
+    {
+      "epoch": 0.1239340846151737,
+      "grad_norm": 0.9311468668119511,
+      "learning_rate": 2e-05,
+      "loss": 0.6210803031921387,
+      "step": 4520
+    },
+    {
+      "epoch": 0.12420827506786214,
+      "grad_norm": 1.2767331744997732,
+      "learning_rate": 2e-05,
+      "loss": 0.6693114757537841,
+      "step": 4530
+    },
+    {
+      "epoch": 0.12448246552055058,
+      "grad_norm": 0.8708955193038721,
+      "learning_rate": 2e-05,
+      "loss": 0.6495052337646484,
+      "step": 4540
+    },
+    {
+      "epoch": 0.12475665597323901,
+      "grad_norm": 3.2672106558058576,
+      "learning_rate": 2e-05,
+      "loss": 0.6230100154876709,
+      "step": 4550
+    },
+    {
+      "epoch": 0.12503084642592746,
+      "grad_norm": 0.9512752059448129,
+      "learning_rate": 2e-05,
+      "loss": 0.6514958381652832,
+      "step": 4560
+    },
+    {
+      "epoch": 0.12530503687861588,
+      "grad_norm": 2.3426602209776743,
+      "learning_rate": 2e-05,
+      "loss": 0.6401823043823243,
+      "step": 4570
+    },
+    {
+      "epoch": 0.12557922733130433,
+      "grad_norm": 8.431270370614746,
+      "learning_rate": 2e-05,
+      "loss": 0.6417835235595704,
+      "step": 4580
+    },
+    {
+      "epoch": 0.12585341778399275,
+      "grad_norm": 0.8964417949699013,
+      "learning_rate": 2e-05,
+      "loss": 0.7339300155639649,
+      "step": 4590
+    },
+    {
+      "epoch": 0.1261276082366812,
+      "grad_norm": 1.08320970438144,
+      "learning_rate": 2e-05,
+      "loss": 0.6680604934692382,
+      "step": 4600
+    },
+    {
+      "epoch": 0.12640179868936963,
+      "grad_norm": 1.41576208241425,
+      "learning_rate": 2e-05,
+      "loss": 0.6463046550750733,
+      "step": 4610
+    },
+    {
+      "epoch": 0.12667598914205808,
+      "grad_norm": 1.842999503817917,
+      "learning_rate": 2e-05,
+      "loss": 0.6181574821472168,
+      "step": 4620
+    },
+    {
+      "epoch": 0.1269501795947465,
+      "grad_norm": 2.307169600378519,
+      "learning_rate": 2e-05,
+      "loss": 0.6464315414428711,
+      "step": 4630
+    },
+    {
+      "epoch": 0.12722437004743495,
+      "grad_norm": 1.2959627920676864,
+      "learning_rate": 2e-05,
+      "loss": 0.6126112937927246,
+      "step": 4640
+    },
+    {
+      "epoch": 0.1274985605001234,
+      "grad_norm": 1.1670020586654428,
+      "learning_rate": 2e-05,
+      "loss": 0.6580602645874023,
+      "step": 4650
+    },
+    {
+      "epoch": 0.12777275095281182,
+      "grad_norm": 1.042513876455037,
+      "learning_rate": 2e-05,
+      "loss": 0.6094191551208497,
+      "step": 4660
+    },
+    {
+      "epoch": 0.12804694140550027,
+      "grad_norm": 0.9328002107147049,
+      "learning_rate": 2e-05,
+      "loss": 0.637706708908081,
+      "step": 4670
+    },
+    {
+      "epoch": 0.1283211318581887,
+      "grad_norm": 0.9377593093916341,
+      "learning_rate": 2e-05,
+      "loss": 0.6559667587280273,
+      "step": 4680
+    },
+    {
+      "epoch": 0.12859532231087714,
+      "grad_norm": 0.8951429697430744,
+      "learning_rate": 2e-05,
+      "loss": 0.6032649517059326,
+      "step": 4690
+    },
+    {
+      "epoch": 0.12886951276356556,
+      "grad_norm": 1.3137988611620086,
+      "learning_rate": 2e-05,
+      "loss": 0.6768720149993896,
+      "step": 4700
+    },
+    {
+      "epoch": 0.129143703216254,
+      "grad_norm": 4.840643792501735,
+      "learning_rate": 2e-05,
+      "loss": 0.6318180084228515,
+      "step": 4710
+    },
+    {
+      "epoch": 0.12941789366894244,
+      "grad_norm": 2.259116097019283,
+      "learning_rate": 2e-05,
+      "loss": 0.6226761817932129,
+      "step": 4720
+    },
+    {
+      "epoch": 0.12969208412163089,
+      "grad_norm": 2.109693160122205,
+      "learning_rate": 2e-05,
+      "loss": 0.6313670635223388,
+      "step": 4730
+    },
+    {
+      "epoch": 0.12996627457431933,
+      "grad_norm": 1.0467624418179453,
+      "learning_rate": 2e-05,
+      "loss": 0.6680407524108887,
+      "step": 4740
+    },
+    {
+      "epoch": 0.13024046502700776,
+      "grad_norm": 2.05380153347514,
+      "learning_rate": 2e-05,
+      "loss": 0.6619737148284912,
+      "step": 4750
+    },
+    {
+      "epoch": 0.1305146554796962,
+      "grad_norm": 1.0783546178155812,
+      "learning_rate": 2e-05,
+      "loss": 0.6868871212005615,
+      "step": 4760
+    },
+    {
+      "epoch": 0.13078884593238463,
+      "grad_norm": 1.239694133832941,
+      "learning_rate": 2e-05,
+      "loss": 0.6313930511474609,
+      "step": 4770
+    },
+    {
+      "epoch": 0.13106303638507308,
+      "grad_norm": 1.124019271740094,
+      "learning_rate": 2e-05,
+      "loss": 0.6283907890319824,
+      "step": 4780
+    },
+    {
+      "epoch": 0.1313372268377615,
+      "grad_norm": 1.1658747506104035,
+      "learning_rate": 2e-05,
+      "loss": 0.6595157623291016,
+      "step": 4790
+    },
+    {
+      "epoch": 0.13161141729044995,
+      "grad_norm": 1.20932907292486,
+      "learning_rate": 2e-05,
+      "loss": 0.6121905326843262,
+      "step": 4800
+    },
+    {
+      "epoch": 0.13188560774313837,
+      "grad_norm": 3.091050723156674,
+      "learning_rate": 2e-05,
+      "loss": 0.6246075630187988,
+      "step": 4810
+    },
+    {
+      "epoch": 0.13215979819582682,
+      "grad_norm": 1.1598041846915066,
+      "learning_rate": 2e-05,
+      "loss": 0.6399677276611329,
+      "step": 4820
+    },
+    {
+      "epoch": 0.13243398864851527,
+      "grad_norm": 0.8967898024225303,
+      "learning_rate": 2e-05,
+      "loss": 0.6439048767089843,
+      "step": 4830
+    },
+    {
+      "epoch": 0.1327081791012037,
+      "grad_norm": 0.928461443349827,
+      "learning_rate": 2e-05,
+      "loss": 0.6454927444458007,
+      "step": 4840
+    },
+    {
+      "epoch": 0.13298236955389214,
+      "grad_norm": 1.5141200567038786,
+      "learning_rate": 2e-05,
+      "loss": 0.6538416862487793,
+      "step": 4850
+    },
+    {
+      "epoch": 0.13325656000658057,
+      "grad_norm": 1.0146485308874171,
+      "learning_rate": 2e-05,
+      "loss": 0.6176496028900147,
+      "step": 4860
+    },
+    {
+      "epoch": 0.13353075045926902,
+      "grad_norm": 1.8029297397830968,
+      "learning_rate": 2e-05,
+      "loss": 0.6379146575927734,
+      "step": 4870
+    },
+    {
+      "epoch": 0.13380494091195744,
+      "grad_norm": 5.591138588683559,
+      "learning_rate": 2e-05,
+      "loss": 0.6584857940673828,
+      "step": 4880
+    },
+    {
+      "epoch": 0.1340791313646459,
+      "grad_norm": 1.8727832016116324,
+      "learning_rate": 2e-05,
+      "loss": 0.6653517723083496,
+      "step": 4890
+    },
+    {
+      "epoch": 0.1343533218173343,
+      "grad_norm": 1.4521817959245416,
+      "learning_rate": 2e-05,
+      "loss": 0.6213447093963623,
+      "step": 4900
+    },
+    {
+      "epoch": 0.13462751227002276,
+      "grad_norm": 2.242378684708907,
+      "learning_rate": 2e-05,
+      "loss": 0.616880226135254,
+      "step": 4910
+    },
+    {
+      "epoch": 0.1349017027227112,
+      "grad_norm": 2.9534188779283657,
+      "learning_rate": 2e-05,
+      "loss": 0.6441243648529053,
+      "step": 4920
+    },
+    {
+      "epoch": 0.13517589317539963,
+      "grad_norm": 0.9368928158210166,
+      "learning_rate": 2e-05,
+      "loss": 0.6245801448822021,
+      "step": 4930
+    },
+    {
+      "epoch": 0.13545008362808808,
+      "grad_norm": 1.5278079773286917,
+      "learning_rate": 2e-05,
+      "loss": 0.6409152984619141,
+      "step": 4940
+    },
+    {
+      "epoch": 0.1357242740807765,
+      "grad_norm": 2.5049941294308598,
+      "learning_rate": 2e-05,
+      "loss": 0.6499926567077636,
+      "step": 4950
+    },
+    {
+      "epoch": 0.13599846453346495,
+      "grad_norm": 0.9282641726461336,
+      "learning_rate": 2e-05,
+      "loss": 0.6331137657165528,
+      "step": 4960
+    },
+    {
+      "epoch": 0.13627265498615337,
+      "grad_norm": 20.87828189339407,
+      "learning_rate": 2e-05,
+      "loss": 0.6314829826354981,
+      "step": 4970
+    },
+    {
+      "epoch": 0.13654684543884182,
+      "grad_norm": 1.1398365383866425,
+      "learning_rate": 2e-05,
+      "loss": 0.6701987743377685,
+      "step": 4980
+    },
+    {
+      "epoch": 0.13682103589153025,
+      "grad_norm": 0.9365437123476379,
+      "learning_rate": 2e-05,
+      "loss": 0.5784749031066895,
+      "step": 4990
+    },
+    {
+      "epoch": 0.1370952263442187,
+      "grad_norm": 1.1978311964314798,
+      "learning_rate": 2e-05,
+      "loss": 0.6229817390441894,
+      "step": 5000
+    },
+    {
+      "epoch": 0.13736941679690712,
+      "grad_norm": 0.9909792796870246,
+      "learning_rate": 2e-05,
+      "loss": 0.6193679809570313,
+      "step": 5010
+    },
+    {
+      "epoch": 0.13764360724959557,
+      "grad_norm": 4.5033032073277175,
+      "learning_rate": 2e-05,
+      "loss": 0.6506769180297851,
+      "step": 5020
+    },
+    {
+      "epoch": 0.13791779770228402,
+      "grad_norm": 1.9314425643193032,
+      "learning_rate": 2e-05,
+      "loss": 0.6618499755859375,
+      "step": 5030
+    },
+    {
+      "epoch": 0.13819198815497244,
+      "grad_norm": 0.9230739369671146,
+      "learning_rate": 2e-05,
+      "loss": 0.649410629272461,
+      "step": 5040
+    },
+    {
+      "epoch": 0.1384661786076609,
+      "grad_norm": 1.1163497806501543,
+      "learning_rate": 2e-05,
+      "loss": 0.6926450252532959,
+      "step": 5050
+    },
+    {
+      "epoch": 0.1387403690603493,
+      "grad_norm": 0.912538720924845,
+      "learning_rate": 2e-05,
+      "loss": 0.7174716472625733,
+      "step": 5060
+    },
+    {
+      "epoch": 0.13901455951303776,
+      "grad_norm": 1.135336646937726,
+      "learning_rate": 2e-05,
+      "loss": 0.6577661037445068,
+      "step": 5070
+    },
+    {
+      "epoch": 0.13928874996572618,
+      "grad_norm": 1.9836851331927898,
+      "learning_rate": 2e-05,
+      "loss": 0.6392465114593506,
+      "step": 5080
+    },
+    {
+      "epoch": 0.13956294041841463,
+      "grad_norm": 1.0207453193662328,
+      "learning_rate": 2e-05,
+      "loss": 0.6006183624267578,
+      "step": 5090
+    },
+    {
+      "epoch": 0.13983713087110305,
+      "grad_norm": 0.9820768061649969,
+      "learning_rate": 2e-05,
+      "loss": 0.6379579067230224,
+      "step": 5100
+    },
+    {
+      "epoch": 0.1401113213237915,
+      "grad_norm": 2.087676684541065,
+      "learning_rate": 2e-05,
+      "loss": 0.641821813583374,
+      "step": 5110
+    },
+    {
+      "epoch": 0.14038551177647995,
+      "grad_norm": 2.2230292841349626,
+      "learning_rate": 2e-05,
+      "loss": 0.657183027267456,
+      "step": 5120
+    },
+    {
+      "epoch": 0.14065970222916838,
+      "grad_norm": 1.4414983538460966,
+      "learning_rate": 2e-05,
+      "loss": 0.645259428024292,
+      "step": 5130
+    },
+    {
+      "epoch": 0.14093389268185683,
+      "grad_norm": 22.391701185064424,
+      "learning_rate": 2e-05,
+      "loss": 0.6407365322113037,
+      "step": 5140
+    },
+    {
+      "epoch": 0.14120808313454525,
+      "grad_norm": 0.8745204950076926,
+      "learning_rate": 2e-05,
+      "loss": 0.6331599235534668,
+      "step": 5150
+    },
+    {
+      "epoch": 0.1414822735872337,
+      "grad_norm": 1.1087062763938034,
+      "learning_rate": 2e-05,
+      "loss": 0.6326388359069824,
+      "step": 5160
+    },
+    {
+      "epoch": 0.14175646403992212,
+      "grad_norm": 0.9198258048241423,
+      "learning_rate": 2e-05,
+      "loss": 0.6268315315246582,
+      "step": 5170
+    },
+    {
+      "epoch": 0.14203065449261057,
+      "grad_norm": 1.0172477219241125,
+      "learning_rate": 2e-05,
+      "loss": 0.6445353507995606,
+      "step": 5180
+    },
+    {
+      "epoch": 0.142304844945299,
+      "grad_norm": 0.9724763771902007,
+      "learning_rate": 2e-05,
+      "loss": 0.6503581523895263,
+      "step": 5190
+    },
+    {
+      "epoch": 0.14257903539798744,
+      "grad_norm": 1.058679219657049,
+      "learning_rate": 2e-05,
+      "loss": 0.6169459819793701,
+      "step": 5200
+    },
+    {
+      "epoch": 0.1428532258506759,
+      "grad_norm": 6.104119348366525,
+      "learning_rate": 2e-05,
+      "loss": 0.651916217803955,
+      "step": 5210
+    },
+    {
+      "epoch": 0.1431274163033643,
+      "grad_norm": 1.1392961260844134,
+      "learning_rate": 2e-05,
+      "loss": 0.6104033946990967,
+      "step": 5220
+    },
+    {
+      "epoch": 0.14340160675605276,
+      "grad_norm": 1.1577411550628673,
+      "learning_rate": 2e-05,
+      "loss": 0.675320053100586,
+      "step": 5230
+    },
+    {
+      "epoch": 0.14367579720874118,
+      "grad_norm": 0.990333225582636,
+      "learning_rate": 2e-05,
+      "loss": 0.6172122001647949,
+      "step": 5240
+    },
+    {
+      "epoch": 0.14394998766142963,
+      "grad_norm": 1.7480284043685863,
+      "learning_rate": 2e-05,
+      "loss": 0.5739638328552246,
+      "step": 5250
+    },
+    {
+      "epoch": 0.14422417811411806,
+      "grad_norm": 1.1169472954287158,
+      "learning_rate": 2e-05,
+      "loss": 0.6423290252685547,
+      "step": 5260
+    },
+    {
+      "epoch": 0.1444983685668065,
+      "grad_norm": 23.45393392551487,
+      "learning_rate": 2e-05,
+      "loss": 0.7626509666442871,
+      "step": 5270
+    },
+    {
+      "epoch": 0.14477255901949493,
+      "grad_norm": 1.1564136989753562,
+      "learning_rate": 2e-05,
+      "loss": 0.5942586898803711,
+      "step": 5280
+    },
+    {
+      "epoch": 0.14504674947218338,
+      "grad_norm": 1.5797111989892583,
+      "learning_rate": 2e-05,
+      "loss": 0.6128039360046387,
+      "step": 5290
+    },
+    {
+      "epoch": 0.14532093992487183,
+      "grad_norm": 1.0024235249594795,
+      "learning_rate": 2e-05,
+      "loss": 0.6589185237884522,
+      "step": 5300
+    },
+    {
+      "epoch": 0.14559513037756025,
+      "grad_norm": 1.150101064758784,
+      "learning_rate": 2e-05,
+      "loss": 0.6360054016113281,
+      "step": 5310
+    },
+    {
+      "epoch": 0.1458693208302487,
+      "grad_norm": 1.7084327334102327,
+      "learning_rate": 2e-05,
+      "loss": 0.6344101905822754,
+      "step": 5320
+    },
+    {
+      "epoch": 0.14614351128293712,
+      "grad_norm": 2.209658604090048,
+      "learning_rate": 2e-05,
+      "loss": 0.6441127777099609,
+      "step": 5330
+    },
+    {
+      "epoch": 0.14641770173562557,
+      "grad_norm": 0.8721213743349046,
+      "learning_rate": 2e-05,
+      "loss": 0.633753490447998,
+      "step": 5340
+    },
+    {
+      "epoch": 0.146691892188314,
+      "grad_norm": 1.0475475377298475,
+      "learning_rate": 1.999194520388485e-05,
+      "loss": 0.5961577892303467,
+      "step": 5350
+    },
+    {
+      "epoch": 0.14696608264100244,
+      "grad_norm": 1.2445919915570411,
+      "learning_rate": 1.9959247303157763e-05,
+      "loss": 0.6229787349700928,
+      "step": 5360
+    },
+    {
+      "epoch": 0.14724027309369087,
+      "grad_norm": 0.9791820606129884,
+      "learning_rate": 1.990149424459817e-05,
+      "loss": 0.6159355163574218,
+      "step": 5370
+    },
+    {
+      "epoch": 0.14751446354637932,
+      "grad_norm": 0.9148104708703945,
+      "learning_rate": 1.9818847538648688e-05,
+      "loss": 0.5957195281982421,
+      "step": 5380
+    },
+    {
+      "epoch": 0.14778865399906777,
+      "grad_norm": 0.8994676746910212,
+      "learning_rate": 1.971153831256721e-05,
+      "loss": 0.6053611755371093,
+      "step": 5390
+    },
+    {
+      "epoch": 0.1480628444517562,
+      "grad_norm": 0.8553991007640382,
+      "learning_rate": 1.9579866664063435e-05,
+      "loss": 0.6179040908813477,
+      "step": 5400
+    },
+    {
+      "epoch": 0.14833703490444464,
+      "grad_norm": 1.2563771929255254,
+      "learning_rate": 1.9424200822054782e-05,
+      "loss": 0.6267671108245849,
+      "step": 5410
+    },
+    {
+      "epoch": 0.14861122535713306,
+      "grad_norm": 1.0620578990475011,
+      "learning_rate": 1.9244976116888626e-05,
+      "loss": 0.64296875,
+      "step": 5420
+    },
+    {
+      "epoch": 0.1488854158098215,
+      "grad_norm": 2.0607475263842323,
+      "learning_rate": 1.904269376291071e-05,
+      "loss": 0.6543359279632568,
+      "step": 5430
+    },
+    {
+      "epoch": 0.14915960626250993,
+      "grad_norm": 1.125275731336469,
+      "learning_rate": 1.88179194567844e-05,
+      "loss": 0.6579808712005615,
+      "step": 5440
+    },
+    {
+      "epoch": 0.14943379671519838,
+      "grad_norm": 0.8336058240485836,
+      "learning_rate": 1.8571281795480632e-05,
+      "loss": 0.6401700496673584,
+      "step": 5450
+    },
+    {
+      "epoch": 0.1497079871678868,
+      "grad_norm": 1.0531176713346853,
+      "learning_rate": 1.8303470518362808e-05,
+      "loss": 0.6423959732055664,
+      "step": 5460
+    },
+    {
+      "epoch": 0.14998217762057525,
+      "grad_norm": 0.9305775401527474,
+      "learning_rate": 1.801523457828271e-05,
+      "loss": 0.6211867332458496,
+      "step": 5470
+    },
+    {
+      "epoch": 0.1502563680732637,
+      "grad_norm": 1.109913291862116,
+      "learning_rate": 1.7707380047081796e-05,
+      "loss": 0.6827861309051514,
+      "step": 5480
+    },
+    {
+      "epoch": 0.15053055852595212,
+      "grad_norm": 0.9822286159167692,
+      "learning_rate": 1.7380767861355252e-05,
+      "loss": 0.6433372497558594,
+      "step": 5490
+    },
+    {
+      "epoch": 0.15080474897864057,
+      "grad_norm": 1.0217415483641756,
+      "learning_rate": 1.703631141478299e-05,
+      "loss": 0.6526628017425538,
+      "step": 5500
+    },
+    {
+      "epoch": 0.151078939431329,
+      "grad_norm": 1.0826469588561494,
+      "learning_rate": 1.6674974003760705e-05,
+      "loss": 0.6787196636199951,
+      "step": 5510
+    },
+    {
+      "epoch": 0.15135312988401745,
+      "grad_norm": 0.8845899530843044,
+      "learning_rate": 1.6297766133474556e-05,
+      "loss": 0.672211742401123,
+      "step": 5520
+    },
+    {
+      "epoch": 0.15162732033670587,
+      "grad_norm": 2.7084541029563867,
+      "learning_rate": 1.5905742691953186e-05,
+      "loss": 0.6030494689941406,
+      "step": 5530
+    },
+    {
+      "epoch": 0.15190151078939432,
+      "grad_norm": 5.064550510065152,
+      "learning_rate": 1.55e-05,
+      "loss": 0.5881963253021241,
+      "step": 5540
+    },
+    {
+      "epoch": 0.15217570124208274,
+      "grad_norm": 1.9405464827318075,
+      "learning_rate": 1.5081672745255862e-05,
+      "loss": 0.6467166900634765,
+      "step": 5550
+    },
+    {
+      "epoch": 0.1524498916947712,
+      "grad_norm": 1.1087249493089857,
+      "learning_rate": 1.4651930808966206e-05,
+      "loss": 0.6357633113861084,
+      "step": 5560
+    },
+    {
+      "epoch": 0.15272408214745964,
+      "grad_norm": 1.1362863198178426,
+      "learning_rate": 1.4211975994326848e-05,
+      "loss": 0.6300517082214355,
+      "step": 5570
+    },
+    {
+      "epoch": 0.15299827260014806,
+      "grad_norm": 0.8748153534987523,
+      "learning_rate": 1.376303866555779e-05,
+      "loss": 0.6353620529174805,
+      "step": 5580
+    },
+    {
+      "epoch": 0.1532724630528365,
+      "grad_norm": 0.9885771785096913,
+      "learning_rate": 1.3306374307104164e-05,
+      "loss": 0.6294742584228515,
+      "step": 5590
+    },
+    {
+      "epoch": 0.15354665350552493,
+      "grad_norm": 0.7221676591681324,
+      "learning_rate": 1.2843260012586718e-05,
+      "loss": 0.5856698513031006,
+      "step": 5600
+    },
+    {
+      "epoch": 0.15382084395821338,
+      "grad_norm": 0.9479133799071373,
+      "learning_rate": 1.2374990913320726e-05,
+      "loss": 0.5971508026123047,
+      "step": 5610
+    },
+    {
+      "epoch": 0.1540950344109018,
+      "grad_norm": 1.0603187805330192,
+      "learning_rate": 1.1902876556391247e-05,
+      "loss": 0.6023536682128906,
+      "step": 5620
+    },
+    {
+      "epoch": 0.15436922486359025,
+      "grad_norm": 1.4587369211314318,
+      "learning_rate": 1.1428237242413685e-05,
+      "loss": 0.5937503814697266,
+      "step": 5630
+    },
+    {
+      "epoch": 0.15464341531627868,
+      "grad_norm": 0.890155363117968,
+      "learning_rate": 1.095240033322132e-05,
+      "loss": 0.5944953918457031,
+      "step": 5640
+    },
+    {
+      "epoch": 0.15491760576896713,
+      "grad_norm": 1.0608960749784877,
+      "learning_rate": 1.047669653980572e-05,
+      "loss": 0.6116889953613281,
+      "step": 5650
+    },
+    {
+      "epoch": 0.15519179622165558,
+      "grad_norm": 1.0092019186113759,
+      "learning_rate": 1.0002456200890902e-05,
+      "loss": 0.6418744564056397,
+      "step": 5660
+    },
+    {
+      "epoch": 0.155465986674344,
+      "grad_norm": 1.2478371338268945,
+      "learning_rate": 9.531005562548668e-06,
+      "loss": 0.6281229972839355,
+      "step": 5670
+    },
+    {
+      "epoch": 0.15574017712703245,
+      "grad_norm": 0.8609291027455624,
+      "learning_rate": 9.06366306925925e-06,
+      "loss": 0.6666102409362793,
+      "step": 5680
+    },
+    {
+      "epoch": 0.15601436757972087,
+      "grad_norm": 1.022269136035383,
+      "learning_rate": 8.601735676789686e-06,
+      "loss": 0.5726197242736817,
+      "step": 5690
+    },
+    {
+      "epoch": 0.15628855803240932,
+      "grad_norm": 0.8716380690452823,
+      "learning_rate": 8.14651519720118e-06,
+      "loss": 0.6102499008178711,
+      "step": 5700
+    },
+    {
+      "epoch": 0.15656274848509774,
+      "grad_norm": 0.9304419996503162,
+      "learning_rate": 7.699274686206866e-06,
+      "loss": 0.5969880104064942,
+      "step": 5710
+    },
+    {
+      "epoch": 0.1568369389377862,
+      "grad_norm": 0.9372889294683869,
+      "learning_rate": 7.261264882983024e-06,
+      "loss": 0.5699095726013184,
+      "step": 5720
+    },
+    {
+      "epoch": 0.1571111293904746,
+      "grad_norm": 0.7688781847028398,
+      "learning_rate": 6.8337107123900386e-06,
+      "loss": 0.5727814674377442,
+      "step": 5730
+    },
+    {
+      "epoch": 0.15738531984316306,
+      "grad_norm": 0.854387144312868,
+      "learning_rate": 6.417807859384881e-06,
+      "loss": 0.5921380519866943,
+      "step": 5740
+    },
+    {
+      "epoch": 0.15765951029585148,
+      "grad_norm": 2.8099485859512794,
+      "learning_rate": 6.014719425205009e-06,
+      "loss": 0.6002985954284668,
+      "step": 5750
+    },
+    {
+      "epoch": 0.15793370074853993,
+      "grad_norm": 0.9643777613948312,
+      "learning_rate": 5.6255726746749275e-06,
+      "loss": 0.6102675437927246,
+      "step": 5760
+    },
+    {
+      "epoch": 0.15820789120122838,
+      "grad_norm": 7.952118304957494,
+      "learning_rate": 5.251455883731728e-06,
+      "loss": 0.598834228515625,
+      "step": 5770
+    },
+    {
+      "epoch": 0.1584820816539168,
+      "grad_norm": 1.0170294663170043,
+      "learning_rate": 4.893415295985812e-06,
+      "loss": 0.6264277458190918,
+      "step": 5780
+    },
+    {
+      "epoch": 0.15875627210660526,
+      "grad_norm": 0.859651999814259,
+      "learning_rate": 4.552452196827885e-06,
+      "loss": 0.6501161098480225,
+      "step": 5790
+    },
+    {
+      "epoch": 0.15903046255929368,
+      "grad_norm": 0.8149654920396862,
+      "learning_rate": 4.229520113264785e-06,
+      "loss": 0.6601780891418457,
+      "step": 5800
+    },
+    {
+      "epoch": 0.15930465301198213,
+      "grad_norm": 7.211333876916126,
+      "learning_rate": 3.925522147314915e-06,
+      "loss": 0.6759274482727051,
+      "step": 5810
+    },
+    {
+      "epoch": 0.15957884346467055,
+      "grad_norm": 1.3757522851107338,
+      "learning_rate": 3.641308450420675e-06,
+      "loss": 0.5789789199829102,
+      "step": 5820
+    },
+    {
+      "epoch": 0.159853033917359,
+      "grad_norm": 0.7942883517184334,
+      "learning_rate": 3.37767384594087e-06,
+      "loss": 0.6161562919616699,
+      "step": 5830
+    },
+    {
+      "epoch": 0.16012722437004742,
+      "grad_norm": 0.9567267976355488,
+      "learning_rate": 3.1353556063719357e-06,
+      "loss": 0.5677800178527832,
+      "step": 5840
+    },
+    {
+      "epoch": 0.16040141482273587,
+      "grad_norm": 1.1946950016950628,
+      "learning_rate": 2.9150313915141474e-06,
+      "loss": 0.586619520187378,
+      "step": 5850
+    },
+    {
+      "epoch": 0.16067560527542432,
+      "grad_norm": 0.7831679028757641,
+      "learning_rate": 2.717317353348891e-06,
+      "loss": 0.6182596206665039,
+      "step": 5860
+    },
+    {
+      "epoch": 0.16094979572811274,
+      "grad_norm": 1.4768052554318936,
+      "learning_rate": 2.5427664129268253e-06,
+      "loss": 0.6375679016113281,
+      "step": 5870
+    },
+    {
+      "epoch": 0.1612239861808012,
+      "grad_norm": 0.8112440363807435,
+      "learning_rate": 2.391866714085717e-06,
+      "loss": 0.5907173156738281,
+      "step": 5880
+    },
+    {
+      "epoch": 0.16149817663348962,
+      "grad_norm": 1.3947895297906168,
+      "learning_rate": 2.2650402583222753e-06,
+      "loss": 0.5791441917419433,
+      "step": 5890
+    },
+    {
+      "epoch": 0.16177236708617806,
+      "grad_norm": 0.9678010696159037,
+      "learning_rate": 2.1626417246356398e-06,
+      "loss": 0.6151109218597413,
+      "step": 5900
+    },
+    {
+      "epoch": 0.1620465575388665,
+      "grad_norm": 0.8421374108012089,
+      "learning_rate": 2.084957477642894e-06,
+      "loss": 0.6073257446289062,
+      "step": 5910
+    },
+    {
+      "epoch": 0.16232074799155494,
+      "grad_norm": 1.377490035979538,
+      "learning_rate": 2.0322047667405147e-06,
+      "loss": 0.6207810401916504,
+      "step": 5920
+    },
+    {
+      "epoch": 0.16259493844424336,
+      "grad_norm": 1.0820792328125313,
+      "learning_rate": 2.0045311185513342e-06,
+      "loss": 0.5762025356292725,
+      "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
+}