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": 0.7626574145463891,
+  "eval_steps": 500,
+  "global_step": 5935,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0012850167052171678,
+      "grad_norm": 856.6189574701012,
+      "learning_rate": 2.910112359550562e-06,
+      "loss": 1.1357397079467773,
+      "step": 10
+    },
+    {
+      "epoch": 0.0025700334104343356,
+      "grad_norm": 15.552391128873273,
+      "learning_rate": 3.921348314606742e-06,
+      "loss": 0.7902592658996582,
+      "step": 20
+    },
+    {
+      "epoch": 0.0038550501156515036,
+      "grad_norm": 15.294031112683584,
+      "learning_rate": 4.932584269662922e-06,
+      "loss": 0.7169486045837402,
+      "step": 30
+    },
+    {
+      "epoch": 0.005140066820868671,
+      "grad_norm": 14.61793688455095,
+      "learning_rate": 5.943820224719102e-06,
+      "loss": 0.6879290580749512,
+      "step": 40
+    },
+    {
+      "epoch": 0.006425083526085839,
+      "grad_norm": 17.865572565928804,
+      "learning_rate": 6.955056179775282e-06,
+      "loss": 0.6837553977966309,
+      "step": 50
+    },
+    {
+      "epoch": 0.007710100231303007,
+      "grad_norm": 78.02976802685448,
+      "learning_rate": 7.966292134831462e-06,
+      "loss": 0.6830409049987793,
+      "step": 60
+    },
+    {
+      "epoch": 0.008995116936520175,
+      "grad_norm": 13.787148293237584,
+      "learning_rate": 8.977528089887641e-06,
+      "loss": 0.6678519725799561,
+      "step": 70
+    },
+    {
+      "epoch": 0.010280133641737342,
+      "grad_norm": 59.88701484053759,
+      "learning_rate": 9.988764044943822e-06,
+      "loss": 0.671458625793457,
+      "step": 80
+    },
+    {
+      "epoch": 0.01156515034695451,
+      "grad_norm": 14.98452507287525,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.6946072578430176,
+      "step": 90
+    },
+    {
+      "epoch": 0.012850167052171678,
+      "grad_norm": 9.949341410744578,
+      "learning_rate": 1.2011235955056182e-05,
+      "loss": 0.6994287014007569,
+      "step": 100
+    },
+    {
+      "epoch": 0.014135183757388847,
+      "grad_norm": 20.083005842913845,
+      "learning_rate": 1.3022471910112362e-05,
+      "loss": 0.654233169555664,
+      "step": 110
+    },
+    {
+      "epoch": 0.015420200462606014,
+      "grad_norm": 52.69058821223778,
+      "learning_rate": 1.403370786516854e-05,
+      "loss": 0.6701527595520019,
+      "step": 120
+    },
+    {
+      "epoch": 0.01670521716782318,
+      "grad_norm": 33.06654632843674,
+      "learning_rate": 1.504494382022472e-05,
+      "loss": 0.6579795837402344,
+      "step": 130
+    },
+    {
+      "epoch": 0.01799023387304035,
+      "grad_norm": 26.181216707950213,
+      "learning_rate": 1.60561797752809e-05,
+      "loss": 0.6791215419769288,
+      "step": 140
+    },
+    {
+      "epoch": 0.01927525057825752,
+      "grad_norm": 25.305185028986585,
+      "learning_rate": 1.706741573033708e-05,
+      "loss": 0.6644459724426269,
+      "step": 150
+    },
+    {
+      "epoch": 0.020560267283474685,
+      "grad_norm": 4.88393687591383,
+      "learning_rate": 1.8078651685393256e-05,
+      "loss": 0.6924647808074951,
+      "step": 160
+    },
+    {
+      "epoch": 0.021845283988691854,
+      "grad_norm": 7.644492046355705,
+      "learning_rate": 1.908988764044944e-05,
+      "loss": 0.6735517024993897,
+      "step": 170
+    },
+    {
+      "epoch": 0.02313030069390902,
+      "grad_norm": 4.0610145968849345,
+      "learning_rate": 2e-05,
+      "loss": 0.6673793315887451,
+      "step": 180
+    },
+    {
+      "epoch": 0.02441531739912619,
+      "grad_norm": 9.704924216779212,
+      "learning_rate": 2e-05,
+      "loss": 0.6797656059265137,
+      "step": 190
+    },
+    {
+      "epoch": 0.025700334104343355,
+      "grad_norm": 5.615408309945265,
+      "learning_rate": 2e-05,
+      "loss": 0.684023666381836,
+      "step": 200
+    },
+    {
+      "epoch": 0.026985350809560524,
+      "grad_norm": 5.423083132935791,
+      "learning_rate": 2e-05,
+      "loss": 0.6746089935302735,
+      "step": 210
+    },
+    {
+      "epoch": 0.028270367514777694,
+      "grad_norm": 8.79838661662345,
+      "learning_rate": 2e-05,
+      "loss": 0.6778485774993896,
+      "step": 220
+    },
+    {
+      "epoch": 0.02955538421999486,
+      "grad_norm": 3.389056327858364,
+      "learning_rate": 2e-05,
+      "loss": 0.6816946029663086,
+      "step": 230
+    },
+    {
+      "epoch": 0.03084040092521203,
+      "grad_norm": 3.2806365569635534,
+      "learning_rate": 2e-05,
+      "loss": 0.6739953517913818,
+      "step": 240
+    },
+    {
+      "epoch": 0.0321254176304292,
+      "grad_norm": 2.714387951683758,
+      "learning_rate": 2e-05,
+      "loss": 0.6525035858154297,
+      "step": 250
+    },
+    {
+      "epoch": 0.03341043433564636,
+      "grad_norm": 3.158516093565955,
+      "learning_rate": 2e-05,
+      "loss": 0.6718252182006836,
+      "step": 260
+    },
+    {
+      "epoch": 0.03469545104086353,
+      "grad_norm": 5.772344202126518,
+      "learning_rate": 2e-05,
+      "loss": 0.647340965270996,
+      "step": 270
+    },
+    {
+      "epoch": 0.0359804677460807,
+      "grad_norm": 5.502021173920203,
+      "learning_rate": 2e-05,
+      "loss": 0.6496793746948242,
+      "step": 280
+    },
+    {
+      "epoch": 0.03726548445129787,
+      "grad_norm": 2.68008821860417,
+      "learning_rate": 2e-05,
+      "loss": 0.6709864616394043,
+      "step": 290
+    },
+    {
+      "epoch": 0.03855050115651504,
+      "grad_norm": 1.7973533832528188,
+      "learning_rate": 2e-05,
+      "loss": 0.6566804885864258,
+      "step": 300
+    },
+    {
+      "epoch": 0.0398355178617322,
+      "grad_norm": 6.51867504563315,
+      "learning_rate": 2e-05,
+      "loss": 0.6657544136047363,
+      "step": 310
+    },
+    {
+      "epoch": 0.04112053456694937,
+      "grad_norm": 2.362613498811483,
+      "learning_rate": 2e-05,
+      "loss": 0.6502259731292724,
+      "step": 320
+    },
+    {
+      "epoch": 0.04240555127216654,
+      "grad_norm": 3.907152138683563,
+      "learning_rate": 2e-05,
+      "loss": 0.6525882244110107,
+      "step": 330
+    },
+    {
+      "epoch": 0.04369056797738371,
+      "grad_norm": 3.794880030440209,
+      "learning_rate": 2e-05,
+      "loss": 0.6532615661621094,
+      "step": 340
+    },
+    {
+      "epoch": 0.04497558468260087,
+      "grad_norm": 8.769127576551398,
+      "learning_rate": 2e-05,
+      "loss": 0.6508918285369873,
+      "step": 350
+    },
+    {
+      "epoch": 0.04626060138781804,
+      "grad_norm": 2.239128040127585,
+      "learning_rate": 2e-05,
+      "loss": 0.6526441097259521,
+      "step": 360
+    },
+    {
+      "epoch": 0.04754561809303521,
+      "grad_norm": 3.6055555343647714,
+      "learning_rate": 2e-05,
+      "loss": 0.681905460357666,
+      "step": 370
+    },
+    {
+      "epoch": 0.04883063479825238,
+      "grad_norm": 2.4554667315087624,
+      "learning_rate": 2e-05,
+      "loss": 0.6597890377044677,
+      "step": 380
+    },
+    {
+      "epoch": 0.05011565150346955,
+      "grad_norm": 1.8867583686163962,
+      "learning_rate": 2e-05,
+      "loss": 0.6532648086547852,
+      "step": 390
+    },
+    {
+      "epoch": 0.05140066820868671,
+      "grad_norm": 1.3820672524188946,
+      "learning_rate": 2e-05,
+      "loss": 0.653763723373413,
+      "step": 400
+    },
+    {
+      "epoch": 0.05268568491390388,
+      "grad_norm": 4.353588095279891,
+      "learning_rate": 2e-05,
+      "loss": 0.6449775695800781,
+      "step": 410
+    },
+    {
+      "epoch": 0.05397070161912105,
+      "grad_norm": 4.764374381297121,
+      "learning_rate": 2e-05,
+      "loss": 0.6555165767669677,
+      "step": 420
+    },
+    {
+      "epoch": 0.05525571832433822,
+      "grad_norm": 14.727196337345648,
+      "learning_rate": 2e-05,
+      "loss": 0.6529878139495849,
+      "step": 430
+    },
+    {
+      "epoch": 0.05654073502955539,
+      "grad_norm": 3.744804024990321,
+      "learning_rate": 2e-05,
+      "loss": 0.6880834579467774,
+      "step": 440
+    },
+    {
+      "epoch": 0.05782575173477255,
+      "grad_norm": 9.9949909991041,
+      "learning_rate": 2e-05,
+      "loss": 0.6793588161468506,
+      "step": 450
+    },
+    {
+      "epoch": 0.05911076843998972,
+      "grad_norm": 4.6856351506601595,
+      "learning_rate": 2e-05,
+      "loss": 0.6606217384338379,
+      "step": 460
+    },
+    {
+      "epoch": 0.06039578514520689,
+      "grad_norm": 2.5156498724259575,
+      "learning_rate": 2e-05,
+      "loss": 0.6711392879486084,
+      "step": 470
+    },
+    {
+      "epoch": 0.06168080185042406,
+      "grad_norm": 3.205814816900315,
+      "learning_rate": 2e-05,
+      "loss": 0.6852891921997071,
+      "step": 480
+    },
+    {
+      "epoch": 0.06296581855564122,
+      "grad_norm": 1.839545887284569,
+      "learning_rate": 2e-05,
+      "loss": 0.6825028896331787,
+      "step": 490
+    },
+    {
+      "epoch": 0.0642508352608584,
+      "grad_norm": 2.4792665803649783,
+      "learning_rate": 2e-05,
+      "loss": 0.6519015789031982,
+      "step": 500
+    },
+    {
+      "epoch": 0.06553585196607556,
+      "grad_norm": 1.9240510741369987,
+      "learning_rate": 2e-05,
+      "loss": 0.648968505859375,
+      "step": 510
+    },
+    {
+      "epoch": 0.06682086867129272,
+      "grad_norm": 2.5695541275375082,
+      "learning_rate": 2e-05,
+      "loss": 0.6466852188110351,
+      "step": 520
+    },
+    {
+      "epoch": 0.0681058853765099,
+      "grad_norm": 11.336408841436898,
+      "learning_rate": 2e-05,
+      "loss": 0.6370265483856201,
+      "step": 530
+    },
+    {
+      "epoch": 0.06939090208172706,
+      "grad_norm": 2.7714988957574405,
+      "learning_rate": 2e-05,
+      "loss": 0.6360719680786133,
+      "step": 540
+    },
+    {
+      "epoch": 0.07067591878694424,
+      "grad_norm": 2.5399185097295414,
+      "learning_rate": 2e-05,
+      "loss": 0.6560811996459961,
+      "step": 550
+    },
+    {
+      "epoch": 0.0719609354921614,
+      "grad_norm": 4.439208424776135,
+      "learning_rate": 2e-05,
+      "loss": 0.644292163848877,
+      "step": 560
+    },
+    {
+      "epoch": 0.07324595219737856,
+      "grad_norm": 5.696317025602095,
+      "learning_rate": 2e-05,
+      "loss": 0.6401437282562256,
+      "step": 570
+    },
+    {
+      "epoch": 0.07453096890259574,
+      "grad_norm": 2.664013724679843,
+      "learning_rate": 2e-05,
+      "loss": 0.6491299629211426,
+      "step": 580
+    },
+    {
+      "epoch": 0.0758159856078129,
+      "grad_norm": 2.1745058771672237,
+      "learning_rate": 2e-05,
+      "loss": 0.63185133934021,
+      "step": 590
+    },
+    {
+      "epoch": 0.07710100231303008,
+      "grad_norm": 3.0153668076131943,
+      "learning_rate": 2e-05,
+      "loss": 0.6398284912109375,
+      "step": 600
+    },
+    {
+      "epoch": 0.07838601901824724,
+      "grad_norm": 4.211123263285935,
+      "learning_rate": 2e-05,
+      "loss": 0.645693016052246,
+      "step": 610
+    },
+    {
+      "epoch": 0.0796710357234644,
+      "grad_norm": 9.078438471416925,
+      "learning_rate": 2e-05,
+      "loss": 0.6547775268554688,
+      "step": 620
+    },
+    {
+      "epoch": 0.08095605242868158,
+      "grad_norm": 1.5458422123622622,
+      "learning_rate": 2e-05,
+      "loss": 0.6409534454345703,
+      "step": 630
+    },
+    {
+      "epoch": 0.08224106913389874,
+      "grad_norm": 6.94798620371141,
+      "learning_rate": 2e-05,
+      "loss": 0.6655917644500733,
+      "step": 640
+    },
+    {
+      "epoch": 0.0835260858391159,
+      "grad_norm": 1.271505509202168,
+      "learning_rate": 2e-05,
+      "loss": 0.6486823558807373,
+      "step": 650
+    },
+    {
+      "epoch": 0.08481110254433308,
+      "grad_norm": 1.0317616781253849,
+      "learning_rate": 2e-05,
+      "loss": 0.6482730388641358,
+      "step": 660
+    },
+    {
+      "epoch": 0.08609611924955024,
+      "grad_norm": 1.1601265249889448,
+      "learning_rate": 2e-05,
+      "loss": 0.6461831569671631,
+      "step": 670
+    },
+    {
+      "epoch": 0.08738113595476742,
+      "grad_norm": 1.5955812719491171,
+      "learning_rate": 2e-05,
+      "loss": 0.6477439403533936,
+      "step": 680
+    },
+    {
+      "epoch": 0.08866615265998458,
+      "grad_norm": 1.2974816270859992,
+      "learning_rate": 2e-05,
+      "loss": 0.630802059173584,
+      "step": 690
+    },
+    {
+      "epoch": 0.08995116936520174,
+      "grad_norm": 1.802080446959691,
+      "learning_rate": 2e-05,
+      "loss": 0.6312804698944092,
+      "step": 700
+    },
+    {
+      "epoch": 0.09123618607041892,
+      "grad_norm": 1.2734807542979136,
+      "learning_rate": 2e-05,
+      "loss": 0.6359667778015137,
+      "step": 710
+    },
+    {
+      "epoch": 0.09252120277563608,
+      "grad_norm": 1.089715475560824,
+      "learning_rate": 2e-05,
+      "loss": 0.6190359115600585,
+      "step": 720
+    },
+    {
+      "epoch": 0.09380621948085326,
+      "grad_norm": 1.4111528747217608,
+      "learning_rate": 2e-05,
+      "loss": 0.6462649822235107,
+      "step": 730
+    },
+    {
+      "epoch": 0.09509123618607042,
+      "grad_norm": 2.3575893938615566,
+      "learning_rate": 2e-05,
+      "loss": 0.6176736354827881,
+      "step": 740
+    },
+    {
+      "epoch": 0.09637625289128758,
+      "grad_norm": 4.099410091763762,
+      "learning_rate": 2e-05,
+      "loss": 0.6239109039306641,
+      "step": 750
+    },
+    {
+      "epoch": 0.09766126959650476,
+      "grad_norm": 4.853333713828358,
+      "learning_rate": 2e-05,
+      "loss": 0.6444282531738281,
+      "step": 760
+    },
+    {
+      "epoch": 0.09894628630172192,
+      "grad_norm": 2.5867597289295405,
+      "learning_rate": 2e-05,
+      "loss": 0.6411311626434326,
+      "step": 770
+    },
+    {
+      "epoch": 0.1002313030069391,
+      "grad_norm": 3.048638070554145,
+      "learning_rate": 2e-05,
+      "loss": 0.6209658622741699,
+      "step": 780
+    },
+    {
+      "epoch": 0.10151631971215626,
+      "grad_norm": 1.9668189975150188,
+      "learning_rate": 2e-05,
+      "loss": 0.6313361167907715,
+      "step": 790
+    },
+    {
+      "epoch": 0.10280133641737342,
+      "grad_norm": 2.7558941074380496,
+      "learning_rate": 2e-05,
+      "loss": 0.6456878185272217,
+      "step": 800
+    },
+    {
+      "epoch": 0.1040863531225906,
+      "grad_norm": 4.931251393074305,
+      "learning_rate": 2e-05,
+      "loss": 0.616514778137207,
+      "step": 810
+    },
+    {
+      "epoch": 0.10537136982780776,
+      "grad_norm": 1.9224203127091826,
+      "learning_rate": 2e-05,
+      "loss": 0.6402089595794678,
+      "step": 820
+    },
+    {
+      "epoch": 0.10665638653302494,
+      "grad_norm": 2.6967273056980985,
+      "learning_rate": 2e-05,
+      "loss": 0.6169169425964356,
+      "step": 830
+    },
+    {
+      "epoch": 0.1079414032382421,
+      "grad_norm": 2.935113359723491,
+      "learning_rate": 2e-05,
+      "loss": 0.6237776756286622,
+      "step": 840
+    },
+    {
+      "epoch": 0.10922641994345926,
+      "grad_norm": 5.369118446973871,
+      "learning_rate": 2e-05,
+      "loss": 0.6399550437927246,
+      "step": 850
+    },
+    {
+      "epoch": 0.11051143664867644,
+      "grad_norm": 2.44005948506668,
+      "learning_rate": 2e-05,
+      "loss": 0.6399797439575196,
+      "step": 860
+    },
+    {
+      "epoch": 0.1117964533538936,
+      "grad_norm": 9.032870562997907,
+      "learning_rate": 2e-05,
+      "loss": 0.6228564739227295,
+      "step": 870
+    },
+    {
+      "epoch": 0.11308147005911078,
+      "grad_norm": 2.2222412773073255,
+      "learning_rate": 2e-05,
+      "loss": 0.6560338497161865,
+      "step": 880
+    },
+    {
+      "epoch": 0.11436648676432794,
+      "grad_norm": 3.2511137004113686,
+      "learning_rate": 2e-05,
+      "loss": 0.637030029296875,
+      "step": 890
+    },
+    {
+      "epoch": 0.1156515034695451,
+      "grad_norm": 2.19881020755354,
+      "learning_rate": 2e-05,
+      "loss": 0.6965466499328613,
+      "step": 900
+    },
+    {
+      "epoch": 0.11693652017476228,
+      "grad_norm": 1.5817637004939435,
+      "learning_rate": 2e-05,
+      "loss": 0.6543152809143067,
+      "step": 910
+    },
+    {
+      "epoch": 0.11822153687997944,
+      "grad_norm": 1.1276990840531103,
+      "learning_rate": 2e-05,
+      "loss": 0.6320334434509277,
+      "step": 920
+    },
+    {
+      "epoch": 0.1195065535851966,
+      "grad_norm": 1.5922154872413146,
+      "learning_rate": 2e-05,
+      "loss": 0.617594051361084,
+      "step": 930
+    },
+    {
+      "epoch": 0.12079157029041378,
+      "grad_norm": 2.504533267963736,
+      "learning_rate": 2e-05,
+      "loss": 0.609941816329956,
+      "step": 940
+    },
+    {
+      "epoch": 0.12207658699563094,
+      "grad_norm": 1.0654835934253888,
+      "learning_rate": 2e-05,
+      "loss": 0.6268711090087891,
+      "step": 950
+    },
+    {
+      "epoch": 0.12336160370084812,
+      "grad_norm": 1.3754530051132428,
+      "learning_rate": 2e-05,
+      "loss": 0.6483760833740234,
+      "step": 960
+    },
+    {
+      "epoch": 0.12464662040606528,
+      "grad_norm": 2.0675680343817304,
+      "learning_rate": 2e-05,
+      "loss": 0.625995397567749,
+      "step": 970
+    },
+    {
+      "epoch": 0.12593163711128244,
+      "grad_norm": 0.9779567253189434,
+      "learning_rate": 2e-05,
+      "loss": 0.6271099090576172,
+      "step": 980
+    },
+    {
+      "epoch": 0.1272166538164996,
+      "grad_norm": 0.9982099110220216,
+      "learning_rate": 2e-05,
+      "loss": 0.61366286277771,
+      "step": 990
+    },
+    {
+      "epoch": 0.1285016705217168,
+      "grad_norm": 1.0658286710551574,
+      "learning_rate": 2e-05,
+      "loss": 0.6421555519104004,
+      "step": 1000
+    },
+    {
+      "epoch": 0.12978668722693396,
+      "grad_norm": 1.2925354348794225,
+      "learning_rate": 2e-05,
+      "loss": 0.6284059524536133,
+      "step": 1010
+    },
+    {
+      "epoch": 0.13107170393215112,
+      "grad_norm": 1.0082444749487007,
+      "learning_rate": 2e-05,
+      "loss": 0.6287686824798584,
+      "step": 1020
+    },
+    {
+      "epoch": 0.13235672063736828,
+      "grad_norm": 1.4045341222523071,
+      "learning_rate": 2e-05,
+      "loss": 0.6289882659912109,
+      "step": 1030
+    },
+    {
+      "epoch": 0.13364173734258544,
+      "grad_norm": 1.1867690266586628,
+      "learning_rate": 2e-05,
+      "loss": 0.6120185852050781,
+      "step": 1040
+    },
+    {
+      "epoch": 0.13492675404780263,
+      "grad_norm": 1.0489340200173105,
+      "learning_rate": 2e-05,
+      "loss": 0.6052242279052734,
+      "step": 1050
+    },
+    {
+      "epoch": 0.1362117707530198,
+      "grad_norm": 1.3799884020710995,
+      "learning_rate": 2e-05,
+      "loss": 0.6165138244628906,
+      "step": 1060
+    },
+    {
+      "epoch": 0.13749678745823696,
+      "grad_norm": 2.9370340336765803,
+      "learning_rate": 2e-05,
+      "loss": 0.6380349159240722,
+      "step": 1070
+    },
+    {
+      "epoch": 0.13878180416345412,
+      "grad_norm": 1.0910366276944916,
+      "learning_rate": 2e-05,
+      "loss": 0.6134048461914062,
+      "step": 1080
+    },
+    {
+      "epoch": 0.14006682086867128,
+      "grad_norm": 1.099783169446652,
+      "learning_rate": 2e-05,
+      "loss": 0.6116578578948975,
+      "step": 1090
+    },
+    {
+      "epoch": 0.14135183757388847,
+      "grad_norm": 6.3506737723576645,
+      "learning_rate": 2e-05,
+      "loss": 0.6087802410125732,
+      "step": 1100
+    },
+    {
+      "epoch": 0.14263685427910563,
+      "grad_norm": 1.830326258724304,
+      "learning_rate": 2e-05,
+      "loss": 0.6111043453216553,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1439218709843228,
+      "grad_norm": 1.4371773206636391,
+      "learning_rate": 2e-05,
+      "loss": 0.6332690238952636,
+      "step": 1120
+    },
+    {
+      "epoch": 0.14520688768953996,
+      "grad_norm": 1.1346896882724038,
+      "learning_rate": 2e-05,
+      "loss": 0.6117291450500488,
+      "step": 1130
+    },
+    {
+      "epoch": 0.14649190439475712,
+      "grad_norm": 1.229043968440211,
+      "learning_rate": 2e-05,
+      "loss": 0.634147834777832,
+      "step": 1140
+    },
+    {
+      "epoch": 0.1477769210999743,
+      "grad_norm": 1.0643282673936258,
+      "learning_rate": 2e-05,
+      "loss": 0.6050711631774902,
+      "step": 1150
+    },
+    {
+      "epoch": 0.14906193780519147,
+      "grad_norm": 0.9899183553255214,
+      "learning_rate": 2e-05,
+      "loss": 0.6159072875976562,
+      "step": 1160
+    },
+    {
+      "epoch": 0.15034695451040864,
+      "grad_norm": 0.9756944895379619,
+      "learning_rate": 2e-05,
+      "loss": 0.6437289714813232,
+      "step": 1170
+    },
+    {
+      "epoch": 0.1516319712156258,
+      "grad_norm": 1.5804530368583096,
+      "learning_rate": 2e-05,
+      "loss": 0.6059542655944824,
+      "step": 1180
+    },
+    {
+      "epoch": 0.15291698792084296,
+      "grad_norm": 2.653732074916257,
+      "learning_rate": 2e-05,
+      "loss": 0.6155671119689942,
+      "step": 1190
+    },
+    {
+      "epoch": 0.15420200462606015,
+      "grad_norm": 1.0526988484175226,
+      "learning_rate": 2e-05,
+      "loss": 0.607187032699585,
+      "step": 1200
+    },
+    {
+      "epoch": 0.15548702133127731,
+      "grad_norm": 1.4066545066883747,
+      "learning_rate": 2e-05,
+      "loss": 0.6067052841186523,
+      "step": 1210
+    },
+    {
+      "epoch": 0.15677203803649448,
+      "grad_norm": 1.14445816579696,
+      "learning_rate": 2e-05,
+      "loss": 0.6031442165374756,
+      "step": 1220
+    },
+    {
+      "epoch": 0.15805705474171164,
+      "grad_norm": 0.9688379604966122,
+      "learning_rate": 2e-05,
+      "loss": 0.6165518760681152,
+      "step": 1230
+    },
+    {
+      "epoch": 0.1593420714469288,
+      "grad_norm": 0.8892837373403698,
+      "learning_rate": 2e-05,
+      "loss": 0.6012338161468506,
+      "step": 1240
+    },
+    {
+      "epoch": 0.160627088152146,
+      "grad_norm": 1.4152027685469248,
+      "learning_rate": 2e-05,
+      "loss": 0.5886805534362793,
+      "step": 1250
+    },
+    {
+      "epoch": 0.16191210485736315,
+      "grad_norm": 2.1457539636692267,
+      "learning_rate": 2e-05,
+      "loss": 0.6081838130950927,
+      "step": 1260
+    },
+    {
+      "epoch": 0.16319712156258032,
+      "grad_norm": 1.7040104665899867,
+      "learning_rate": 2e-05,
+      "loss": 0.6133224487304687,
+      "step": 1270
+    },
+    {
+      "epoch": 0.16448213826779748,
+      "grad_norm": 1.2352798817335287,
+      "learning_rate": 2e-05,
+      "loss": 0.6118340492248535,
+      "step": 1280
+    },
+    {
+      "epoch": 0.16576715497301464,
+      "grad_norm": 1.2198439029279724,
+      "learning_rate": 2e-05,
+      "loss": 0.6088254451751709,
+      "step": 1290
+    },
+    {
+      "epoch": 0.1670521716782318,
+      "grad_norm": 3.6356160898778853,
+      "learning_rate": 2e-05,
+      "loss": 0.5922161102294922,
+      "step": 1300
+    },
+    {
+      "epoch": 0.168337188383449,
+      "grad_norm": 1.434121136195825,
+      "learning_rate": 2e-05,
+      "loss": 0.6113036155700684,
+      "step": 1310
+    },
+    {
+      "epoch": 0.16962220508866616,
+      "grad_norm": 1.8215041680182131,
+      "learning_rate": 2e-05,
+      "loss": 0.6138492584228515,
+      "step": 1320
+    },
+    {
+      "epoch": 0.17090722179388332,
+      "grad_norm": 0.9733584275000338,
+      "learning_rate": 2e-05,
+      "loss": 0.6006131649017334,
+      "step": 1330
+    },
+    {
+      "epoch": 0.17219223849910048,
+      "grad_norm": 0.8291757298058455,
+      "learning_rate": 2e-05,
+      "loss": 0.6175620555877686,
+      "step": 1340
+    },
+    {
+      "epoch": 0.17347725520431764,
+      "grad_norm": 1.7778569183139799,
+      "learning_rate": 2e-05,
+      "loss": 0.6048360347747803,
+      "step": 1350
+    },
+    {
+      "epoch": 0.17476227190953483,
+      "grad_norm": 2.3344157936720307,
+      "learning_rate": 2e-05,
+      "loss": 0.6050002098083496,
+      "step": 1360
+    },
+    {
+      "epoch": 0.176047288614752,
+      "grad_norm": 3.506671656079186,
+      "learning_rate": 2e-05,
+      "loss": 0.6108762741088867,
+      "step": 1370
+    },
+    {
+      "epoch": 0.17733230531996916,
+      "grad_norm": 1.1491594873900661,
+      "learning_rate": 2e-05,
+      "loss": 0.5949950695037842,
+      "step": 1380
+    },
+    {
+      "epoch": 0.17861732202518632,
+      "grad_norm": 0.9698616451139325,
+      "learning_rate": 2e-05,
+      "loss": 0.6035851955413818,
+      "step": 1390
+    },
+    {
+      "epoch": 0.17990233873040348,
+      "grad_norm": 2.3293601794536607,
+      "learning_rate": 2e-05,
+      "loss": 0.6041230201721192,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18118735543562067,
+      "grad_norm": 0.9819292447087378,
+      "learning_rate": 2e-05,
+      "loss": 0.6033549308776855,
+      "step": 1410
+    },
+    {
+      "epoch": 0.18247237214083784,
+      "grad_norm": 5.127718733826317,
+      "learning_rate": 2e-05,
+      "loss": 0.5804622650146485,
+      "step": 1420
+    },
+    {
+      "epoch": 0.183757388846055,
+      "grad_norm": 1.3831078732533943,
+      "learning_rate": 2e-05,
+      "loss": 0.5929896354675293,
+      "step": 1430
+    },
+    {
+      "epoch": 0.18504240555127216,
+      "grad_norm": 1.409835795434802,
+      "learning_rate": 2e-05,
+      "loss": 0.620157527923584,
+      "step": 1440
+    },
+    {
+      "epoch": 0.18632742225648932,
+      "grad_norm": 0.9787989385784489,
+      "learning_rate": 2e-05,
+      "loss": 0.6010499477386475,
+      "step": 1450
+    },
+    {
+      "epoch": 0.1876124389617065,
+      "grad_norm": 1.125282063521845,
+      "learning_rate": 2e-05,
+      "loss": 0.6063499450683594,
+      "step": 1460
+    },
+    {
+      "epoch": 0.18889745566692367,
+      "grad_norm": 1.1878340415798876,
+      "learning_rate": 2e-05,
+      "loss": 0.6011593818664551,
+      "step": 1470
+    },
+    {
+      "epoch": 0.19018247237214084,
+      "grad_norm": 1.0586030412435297,
+      "learning_rate": 2e-05,
+      "loss": 0.5826192378997803,
+      "step": 1480
+    },
+    {
+      "epoch": 0.191467489077358,
+      "grad_norm": 1.4030416349158283,
+      "learning_rate": 2e-05,
+      "loss": 0.5979415893554687,
+      "step": 1490
+    },
+    {
+      "epoch": 0.19275250578257516,
+      "grad_norm": 0.9882974034594119,
+      "learning_rate": 2e-05,
+      "loss": 0.5784321784973144,
+      "step": 1500
+    },
+    {
+      "epoch": 0.19403752248779235,
+      "grad_norm": 1.5827378342447296,
+      "learning_rate": 2e-05,
+      "loss": 0.5972190380096436,
+      "step": 1510
+    },
+    {
+      "epoch": 0.19532253919300951,
+      "grad_norm": 1.772835353002146,
+      "learning_rate": 2e-05,
+      "loss": 0.5883693218231201,
+      "step": 1520
+    },
+    {
+      "epoch": 0.19660755589822668,
+      "grad_norm": 0.9504309854461958,
+      "learning_rate": 2e-05,
+      "loss": 0.5878360748291016,
+      "step": 1530
+    },
+    {
+      "epoch": 0.19789257260344384,
+      "grad_norm": 3.4573479282773754,
+      "learning_rate": 2e-05,
+      "loss": 0.5961189270019531,
+      "step": 1540
+    },
+    {
+      "epoch": 0.199177589308661,
+      "grad_norm": 1.068378062020343,
+      "learning_rate": 2e-05,
+      "loss": 0.6108866691589355,
+      "step": 1550
+    },
+    {
+      "epoch": 0.2004626060138782,
+      "grad_norm": 0.9475492216550739,
+      "learning_rate": 2e-05,
+      "loss": 0.5879310607910156,
+      "step": 1560
+    },
+    {
+      "epoch": 0.20174762271909535,
+      "grad_norm": 2.6996846803236796,
+      "learning_rate": 2e-05,
+      "loss": 0.6067344665527343,
+      "step": 1570
+    },
+    {
+      "epoch": 0.20303263942431252,
+      "grad_norm": 1.503312484489778,
+      "learning_rate": 2e-05,
+      "loss": 0.5966094970703125,
+      "step": 1580
+    },
+    {
+      "epoch": 0.20431765612952968,
+      "grad_norm": 2.1144705955722105,
+      "learning_rate": 2e-05,
+      "loss": 0.601826810836792,
+      "step": 1590
+    },
+    {
+      "epoch": 0.20560267283474684,
+      "grad_norm": 2.0298939231075686,
+      "learning_rate": 2e-05,
+      "loss": 0.613339900970459,
+      "step": 1600
+    },
+    {
+      "epoch": 0.20688768953996403,
+      "grad_norm": 0.9838666275922071,
+      "learning_rate": 2e-05,
+      "loss": 0.5876136779785156,
+      "step": 1610
+    },
+    {
+      "epoch": 0.2081727062451812,
+      "grad_norm": 1.3776226412627304,
+      "learning_rate": 2e-05,
+      "loss": 0.5954419612884522,
+      "step": 1620
+    },
+    {
+      "epoch": 0.20945772295039836,
+      "grad_norm": 0.8085782872052595,
+      "learning_rate": 2e-05,
+      "loss": 0.595486831665039,
+      "step": 1630
+    },
+    {
+      "epoch": 0.21074273965561552,
+      "grad_norm": 0.8662216027678745,
+      "learning_rate": 2e-05,
+      "loss": 0.5841521263122559,
+      "step": 1640
+    },
+    {
+      "epoch": 0.21202775636083268,
+      "grad_norm": 1.2239187193929486,
+      "learning_rate": 2e-05,
+      "loss": 0.5863484859466552,
+      "step": 1650
+    },
+    {
+      "epoch": 0.21331277306604987,
+      "grad_norm": 1.0494852911396721,
+      "learning_rate": 2e-05,
+      "loss": 0.5841124534606934,
+      "step": 1660
+    },
+    {
+      "epoch": 0.21459778977126703,
+      "grad_norm": 2.981594663490346,
+      "learning_rate": 2e-05,
+      "loss": 0.5967587947845459,
+      "step": 1670
+    },
+    {
+      "epoch": 0.2158828064764842,
+      "grad_norm": 1.296867754625597,
+      "learning_rate": 2e-05,
+      "loss": 0.6040875434875488,
+      "step": 1680
+    },
+    {
+      "epoch": 0.21716782318170136,
+      "grad_norm": 1.6322116014400065,
+      "learning_rate": 2e-05,
+      "loss": 0.6083320617675781,
+      "step": 1690
+    },
+    {
+      "epoch": 0.21845283988691852,
+      "grad_norm": 1.1441615552670734,
+      "learning_rate": 2e-05,
+      "loss": 0.6085511684417725,
+      "step": 1700
+    },
+    {
+      "epoch": 0.2197378565921357,
+      "grad_norm": 0.9746942510805159,
+      "learning_rate": 2e-05,
+      "loss": 0.5959947586059571,
+      "step": 1710
+    },
+    {
+      "epoch": 0.22102287329735287,
+      "grad_norm": 1.847667452156604,
+      "learning_rate": 2e-05,
+      "loss": 0.5838113307952881,
+      "step": 1720
+    },
+    {
+      "epoch": 0.22230789000257004,
+      "grad_norm": 1.3014028403342264,
+      "learning_rate": 2e-05,
+      "loss": 0.5898797988891602,
+      "step": 1730
+    },
+    {
+      "epoch": 0.2235929067077872,
+      "grad_norm": 1.794520627282578,
+      "learning_rate": 2e-05,
+      "loss": 0.5770421504974366,
+      "step": 1740
+    },
+    {
+      "epoch": 0.22487792341300436,
+      "grad_norm": 1.181651271225444,
+      "learning_rate": 2e-05,
+      "loss": 0.5804369926452637,
+      "step": 1750
+    },
+    {
+      "epoch": 0.22616294011822155,
+      "grad_norm": 1.4302258126941458,
+      "learning_rate": 2e-05,
+      "loss": 0.5999673843383789,
+      "step": 1760
+    },
+    {
+      "epoch": 0.2274479568234387,
+      "grad_norm": 1.5184617174292843,
+      "learning_rate": 2e-05,
+      "loss": 0.601544713973999,
+      "step": 1770
+    },
+    {
+      "epoch": 0.22873297352865588,
+      "grad_norm": 1.3090615344632057,
+      "learning_rate": 2e-05,
+      "loss": 0.5916304588317871,
+      "step": 1780
+    },
+    {
+      "epoch": 0.23001799023387304,
+      "grad_norm": 1.084083761154048,
+      "learning_rate": 2e-05,
+      "loss": 0.599252462387085,
+      "step": 1790
+    },
+    {
+      "epoch": 0.2313030069390902,
+      "grad_norm": 1.06362134663052,
+      "learning_rate": 2e-05,
+      "loss": 0.5892449855804444,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23258802364430736,
+      "grad_norm": 1.3218270041853413,
+      "learning_rate": 2e-05,
+      "loss": 0.5810562133789062,
+      "step": 1810
+    },
+    {
+      "epoch": 0.23387304034952455,
+      "grad_norm": 1.8735021692385667,
+      "learning_rate": 2e-05,
+      "loss": 0.563934326171875,
+      "step": 1820
+    },
+    {
+      "epoch": 0.23515805705474171,
+      "grad_norm": 1.1029961980816403,
+      "learning_rate": 2e-05,
+      "loss": 0.6061427116394043,
+      "step": 1830
+    },
+    {
+      "epoch": 0.23644307375995888,
+      "grad_norm": 1.0956401424009607,
+      "learning_rate": 2e-05,
+      "loss": 0.580710506439209,
+      "step": 1840
+    },
+    {
+      "epoch": 0.23772809046517604,
+      "grad_norm": 0.9022890666227324,
+      "learning_rate": 2e-05,
+      "loss": 0.5831015586853028,
+      "step": 1850
+    },
+    {
+      "epoch": 0.2390131071703932,
+      "grad_norm": 0.9739413408546281,
+      "learning_rate": 2e-05,
+      "loss": 0.5803080558776855,
+      "step": 1860
+    },
+    {
+      "epoch": 0.2402981238756104,
+      "grad_norm": 0.8887115233132612,
+      "learning_rate": 2e-05,
+      "loss": 0.5808173656463623,
+      "step": 1870
+    },
+    {
+      "epoch": 0.24158314058082755,
+      "grad_norm": 1.2281390991341632,
+      "learning_rate": 2e-05,
+      "loss": 0.584747314453125,
+      "step": 1880
+    },
+    {
+      "epoch": 0.24286815728604472,
+      "grad_norm": 1.0293295686123922,
+      "learning_rate": 2e-05,
+      "loss": 0.5973689079284668,
+      "step": 1890
+    },
+    {
+      "epoch": 0.24415317399126188,
+      "grad_norm": 2.240993859946262,
+      "learning_rate": 2e-05,
+      "loss": 0.5756341934204101,
+      "step": 1900
+    },
+    {
+      "epoch": 0.24543819069647904,
+      "grad_norm": 2.542763529237098,
+      "learning_rate": 2e-05,
+      "loss": 0.568706464767456,
+      "step": 1910
+    },
+    {
+      "epoch": 0.24672320740169623,
+      "grad_norm": 1.8658118445309126,
+      "learning_rate": 2e-05,
+      "loss": 0.5815502166748047,
+      "step": 1920
+    },
+    {
+      "epoch": 0.2480082241069134,
+      "grad_norm": 1.3680858039002373,
+      "learning_rate": 2e-05,
+      "loss": 0.5713666915893555,
+      "step": 1930
+    },
+    {
+      "epoch": 0.24929324081213056,
+      "grad_norm": 1.4545423045311263,
+      "learning_rate": 2e-05,
+      "loss": 0.5888118267059326,
+      "step": 1940
+    },
+    {
+      "epoch": 0.25057825751734775,
+      "grad_norm": 1.0001999301868607,
+      "learning_rate": 2e-05,
+      "loss": 0.581107234954834,
+      "step": 1950
+    },
+    {
+      "epoch": 0.2518632742225649,
+      "grad_norm": 3.804939278531223,
+      "learning_rate": 2e-05,
+      "loss": 0.5979499816894531,
+      "step": 1960
+    },
+    {
+      "epoch": 0.25314829092778207,
+      "grad_norm": 1.6052651400683768,
+      "learning_rate": 2e-05,
+      "loss": 0.5852889060974121,
+      "step": 1970
+    },
+    {
+      "epoch": 0.2544333076329992,
+      "grad_norm": 1.465455399606115,
+      "learning_rate": 2e-05,
+      "loss": 0.5984594345092773,
+      "step": 1980
+    },
+    {
+      "epoch": 0.2557183243382164,
+      "grad_norm": 0.9532879108802271,
+      "learning_rate": 2e-05,
+      "loss": 0.5933744430541992,
+      "step": 1990
+    },
+    {
+      "epoch": 0.2570033410434336,
+      "grad_norm": 0.9971702008546999,
+      "learning_rate": 2e-05,
+      "loss": 0.5816676616668701,
+      "step": 2000
+    },
+    {
+      "epoch": 0.2582883577486507,
+      "grad_norm": 0.9234236488287841,
+      "learning_rate": 2e-05,
+      "loss": 0.5809596538543701,
+      "step": 2010
+    },
+    {
+      "epoch": 0.2595733744538679,
+      "grad_norm": 1.2689341107385237,
+      "learning_rate": 2e-05,
+      "loss": 0.5777791023254395,
+      "step": 2020
+    },
+    {
+      "epoch": 0.26085839115908505,
+      "grad_norm": 2.127441593187859,
+      "learning_rate": 2e-05,
+      "loss": 0.5782732963562012,
+      "step": 2030
+    },
+    {
+      "epoch": 0.26214340786430224,
+      "grad_norm": 1.1764073570223446,
+      "learning_rate": 2e-05,
+      "loss": 0.5909663200378418,
+      "step": 2040
+    },
+    {
+      "epoch": 0.2634284245695194,
+      "grad_norm": 0.9744698438423596,
+      "learning_rate": 2e-05,
+      "loss": 0.5853694915771485,
+      "step": 2050
+    },
+    {
+      "epoch": 0.26471344127473656,
+      "grad_norm": 1.580815044624061,
+      "learning_rate": 2e-05,
+      "loss": 0.5661839962005615,
+      "step": 2060
+    },
+    {
+      "epoch": 0.26599845797995375,
+      "grad_norm": 1.3156449239046828,
+      "learning_rate": 2e-05,
+      "loss": 0.5968900203704834,
+      "step": 2070
+    },
+    {
+      "epoch": 0.2672834746851709,
+      "grad_norm": 2.312847962435919,
+      "learning_rate": 2e-05,
+      "loss": 0.5945404052734375,
+      "step": 2080
+    },
+    {
+      "epoch": 0.2685684913903881,
+      "grad_norm": 1.1570344276848041,
+      "learning_rate": 2e-05,
+      "loss": 0.5699649810791015,
+      "step": 2090
+    },
+    {
+      "epoch": 0.26985350809560527,
+      "grad_norm": 1.2428865565902694,
+      "learning_rate": 2e-05,
+      "loss": 0.5657583236694336,
+      "step": 2100
+    },
+    {
+      "epoch": 0.2711385248008224,
+      "grad_norm": 1.0962856067397169,
+      "learning_rate": 2e-05,
+      "loss": 0.5972212791442871,
+      "step": 2110
+    },
+    {
+      "epoch": 0.2724235415060396,
+      "grad_norm": 5.978551138656896,
+      "learning_rate": 2e-05,
+      "loss": 0.5613668441772461,
+      "step": 2120
+    },
+    {
+      "epoch": 0.2737085582112567,
+      "grad_norm": 6.824885633049643,
+      "learning_rate": 2e-05,
+      "loss": 0.5738656044006347,
+      "step": 2130
+    },
+    {
+      "epoch": 0.2749935749164739,
+      "grad_norm": 1.131602053511405,
+      "learning_rate": 2e-05,
+      "loss": 0.5778485774993897,
+      "step": 2140
+    },
+    {
+      "epoch": 0.2762785916216911,
+      "grad_norm": 1.5378948200740872,
+      "learning_rate": 2e-05,
+      "loss": 0.5719347953796386,
+      "step": 2150
+    },
+    {
+      "epoch": 0.27756360832690824,
+      "grad_norm": 1.555187915201428,
+      "learning_rate": 2e-05,
+      "loss": 0.5783113479614258,
+      "step": 2160
+    },
+    {
+      "epoch": 0.27884862503212543,
+      "grad_norm": 1.4732553969834044,
+      "learning_rate": 2e-05,
+      "loss": 0.5797761440277099,
+      "step": 2170
+    },
+    {
+      "epoch": 0.28013364173734256,
+      "grad_norm": 3.6258350650082183,
+      "learning_rate": 2e-05,
+      "loss": 0.5716004848480225,
+      "step": 2180
+    },
+    {
+      "epoch": 0.28141865844255975,
+      "grad_norm": 2.1806230459926574,
+      "learning_rate": 2e-05,
+      "loss": 0.5770562171936036,
+      "step": 2190
+    },
+    {
+      "epoch": 0.28270367514777694,
+      "grad_norm": 1.5284840156681283,
+      "learning_rate": 2e-05,
+      "loss": 0.5768312454223633,
+      "step": 2200
+    },
+    {
+      "epoch": 0.2839886918529941,
+      "grad_norm": 3.2769955799751607,
+      "learning_rate": 2e-05,
+      "loss": 0.5817856788635254,
+      "step": 2210
+    },
+    {
+      "epoch": 0.28527370855821127,
+      "grad_norm": 10.980944870429466,
+      "learning_rate": 2e-05,
+      "loss": 0.5739808559417725,
+      "step": 2220
+    },
+    {
+      "epoch": 0.2865587252634284,
+      "grad_norm": 1.29206172098196,
+      "learning_rate": 2e-05,
+      "loss": 0.5830291271209717,
+      "step": 2230
+    },
+    {
+      "epoch": 0.2878437419686456,
+      "grad_norm": 2.1523791789748206,
+      "learning_rate": 2e-05,
+      "loss": 0.5948679924011231,
+      "step": 2240
+    },
+    {
+      "epoch": 0.2891287586738628,
+      "grad_norm": 1.067819120270187,
+      "learning_rate": 2e-05,
+      "loss": 0.5834884166717529,
+      "step": 2250
+    },
+    {
+      "epoch": 0.2904137753790799,
+      "grad_norm": 1.6184858138371465,
+      "learning_rate": 2e-05,
+      "loss": 0.5768936157226563,
+      "step": 2260
+    },
+    {
+      "epoch": 0.2916987920842971,
+      "grad_norm": 1.042330169791891,
+      "learning_rate": 2e-05,
+      "loss": 0.5791688919067383,
+      "step": 2270
+    },
+    {
+      "epoch": 0.29298380878951424,
+      "grad_norm": 0.9140366391060554,
+      "learning_rate": 2e-05,
+      "loss": 0.5576070785522461,
+      "step": 2280
+    },
+    {
+      "epoch": 0.29426882549473143,
+      "grad_norm": 1.0365123467067447,
+      "learning_rate": 2e-05,
+      "loss": 0.5595355987548828,
+      "step": 2290
+    },
+    {
+      "epoch": 0.2955538421999486,
+      "grad_norm": 1.0029051830523124,
+      "learning_rate": 2e-05,
+      "loss": 0.5694355010986328,
+      "step": 2300
+    },
+    {
+      "epoch": 0.29683885890516576,
+      "grad_norm": 0.9405638439959776,
+      "learning_rate": 2e-05,
+      "loss": 0.5609602928161621,
+      "step": 2310
+    },
+    {
+      "epoch": 0.29812387561038295,
+      "grad_norm": 2.75210961018307,
+      "learning_rate": 2e-05,
+      "loss": 0.5824363708496094,
+      "step": 2320
+    },
+    {
+      "epoch": 0.2994088923156001,
+      "grad_norm": 0.8431934851341428,
+      "learning_rate": 2e-05,
+      "loss": 0.5718118190765381,
+      "step": 2330
+    },
+    {
+      "epoch": 0.3006939090208173,
+      "grad_norm": 1.0435924880072194,
+      "learning_rate": 2e-05,
+      "loss": 0.5788133621215821,
+      "step": 2340
+    },
+    {
+      "epoch": 0.30197892572603446,
+      "grad_norm": 1.368725386427797,
+      "learning_rate": 2e-05,
+      "loss": 0.5597520351409913,
+      "step": 2350
+    },
+    {
+      "epoch": 0.3032639424312516,
+      "grad_norm": 1.0499621664346996,
+      "learning_rate": 2e-05,
+      "loss": 0.5606437683105469,
+      "step": 2360
+    },
+    {
+      "epoch": 0.3045489591364688,
+      "grad_norm": 0.8371855986210658,
+      "learning_rate": 2e-05,
+      "loss": 0.573657751083374,
+      "step": 2370
+    },
+    {
+      "epoch": 0.3058339758416859,
+      "grad_norm": 0.8047707198512719,
+      "learning_rate": 2e-05,
+      "loss": 0.5701785087585449,
+      "step": 2380
+    },
+    {
+      "epoch": 0.3071189925469031,
+      "grad_norm": 0.7809569763370298,
+      "learning_rate": 2e-05,
+      "loss": 0.5710490703582763,
+      "step": 2390
+    },
+    {
+      "epoch": 0.3084040092521203,
+      "grad_norm": 1.0100171159941704,
+      "learning_rate": 2e-05,
+      "loss": 0.565343189239502,
+      "step": 2400
+    },
+    {
+      "epoch": 0.30968902595733744,
+      "grad_norm": 1.2754557016668542,
+      "learning_rate": 2e-05,
+      "loss": 0.5544345855712891,
+      "step": 2410
+    },
+    {
+      "epoch": 0.31097404266255463,
+      "grad_norm": 2.4025874445648867,
+      "learning_rate": 2e-05,
+      "loss": 0.563749885559082,
+      "step": 2420
+    },
+    {
+      "epoch": 0.31225905936777176,
+      "grad_norm": 2.163486421652014,
+      "learning_rate": 2e-05,
+      "loss": 0.574526596069336,
+      "step": 2430
+    },
+    {
+      "epoch": 0.31354407607298895,
+      "grad_norm": 1.019884012431456,
+      "learning_rate": 2e-05,
+      "loss": 0.5653591156005859,
+      "step": 2440
+    },
+    {
+      "epoch": 0.31482909277820614,
+      "grad_norm": 2.3917602266355744,
+      "learning_rate": 2e-05,
+      "loss": 0.565762710571289,
+      "step": 2450
+    },
+    {
+      "epoch": 0.3161141094834233,
+      "grad_norm": 2.3530676155114745,
+      "learning_rate": 2e-05,
+      "loss": 0.5656596183776855,
+      "step": 2460
+    },
+    {
+      "epoch": 0.31739912618864047,
+      "grad_norm": 4.344273226376291,
+      "learning_rate": 2e-05,
+      "loss": 0.5669883728027344,
+      "step": 2470
+    },
+    {
+      "epoch": 0.3186841428938576,
+      "grad_norm": 2.2164261375616365,
+      "learning_rate": 2e-05,
+      "loss": 0.5692079067230225,
+      "step": 2480
+    },
+    {
+      "epoch": 0.3199691595990748,
+      "grad_norm": 1.4550433504034848,
+      "learning_rate": 2e-05,
+      "loss": 0.5787978649139405,
+      "step": 2490
+    },
+    {
+      "epoch": 0.321254176304292,
+      "grad_norm": 1.0184777383345365,
+      "learning_rate": 2e-05,
+      "loss": 0.5705706596374511,
+      "step": 2500
+    },
+    {
+      "epoch": 0.3225391930095091,
+      "grad_norm": 2.5074871251802464,
+      "learning_rate": 2e-05,
+      "loss": 0.5802200317382813,
+      "step": 2510
+    },
+    {
+      "epoch": 0.3238242097147263,
+      "grad_norm": 38.59370599582083,
+      "learning_rate": 2e-05,
+      "loss": 0.5678404808044434,
+      "step": 2520
+    },
+    {
+      "epoch": 0.32510922641994344,
+      "grad_norm": 1.1689687035318401,
+      "learning_rate": 2e-05,
+      "loss": 0.5908804893493652,
+      "step": 2530
+    },
+    {
+      "epoch": 0.32639424312516063,
+      "grad_norm": 1.1686100577044995,
+      "learning_rate": 2e-05,
+      "loss": 0.5825255393981934,
+      "step": 2540
+    },
+    {
+      "epoch": 0.3276792598303778,
+      "grad_norm": 0.8999783470710406,
+      "learning_rate": 2e-05,
+      "loss": 0.5790523529052735,
+      "step": 2550
+    },
+    {
+      "epoch": 0.32896427653559496,
+      "grad_norm": 2.6260019617386976,
+      "learning_rate": 2e-05,
+      "loss": 0.5537464141845703,
+      "step": 2560
+    },
+    {
+      "epoch": 0.33024929324081215,
+      "grad_norm": 0.9405365586026401,
+      "learning_rate": 2e-05,
+      "loss": 0.5652168273925782,
+      "step": 2570
+    },
+    {
+      "epoch": 0.3315343099460293,
+      "grad_norm": 1.0505311930700645,
+      "learning_rate": 2e-05,
+      "loss": 0.5567763328552247,
+      "step": 2580
+    },
+    {
+      "epoch": 0.33281932665124647,
+      "grad_norm": 1.1404154848897663,
+      "learning_rate": 2e-05,
+      "loss": 0.5682837486267089,
+      "step": 2590
+    },
+    {
+      "epoch": 0.3341043433564636,
+      "grad_norm": 1.3569305091906358,
+      "learning_rate": 2e-05,
+      "loss": 0.5649970054626465,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3353893600616808,
+      "grad_norm": 1.9212653049618156,
+      "learning_rate": 2e-05,
+      "loss": 0.5584495544433594,
+      "step": 2610
+    },
+    {
+      "epoch": 0.336674376766898,
+      "grad_norm": 3.127810864963417,
+      "learning_rate": 2e-05,
+      "loss": 0.5491080284118652,
+      "step": 2620
+    },
+    {
+      "epoch": 0.3379593934721151,
+      "grad_norm": 4.142753390488306,
+      "learning_rate": 2e-05,
+      "loss": 0.5558376312255859,
+      "step": 2630
+    },
+    {
+      "epoch": 0.3392444101773323,
+      "grad_norm": 1.1129102204834083,
+      "learning_rate": 2e-05,
+      "loss": 0.5642284393310547,
+      "step": 2640
+    },
+    {
+      "epoch": 0.34052942688254945,
+      "grad_norm": 1.4548367058662652,
+      "learning_rate": 2e-05,
+      "loss": 0.5562495231628418,
+      "step": 2650
+    },
+    {
+      "epoch": 0.34181444358776664,
+      "grad_norm": 7.431157537048908,
+      "learning_rate": 2e-05,
+      "loss": 0.5657413482666016,
+      "step": 2660
+    },
+    {
+      "epoch": 0.3430994602929838,
+      "grad_norm": 11.771451481316966,
+      "learning_rate": 2e-05,
+      "loss": 0.5741857528686524,
+      "step": 2670
+    },
+    {
+      "epoch": 0.34438447699820096,
+      "grad_norm": 3.425817718053954,
+      "learning_rate": 2e-05,
+      "loss": 0.5649560451507568,
+      "step": 2680
+    },
+    {
+      "epoch": 0.34566949370341815,
+      "grad_norm": 1.403026259801426,
+      "learning_rate": 2e-05,
+      "loss": 0.5578023433685303,
+      "step": 2690
+    },
+    {
+      "epoch": 0.3469545104086353,
+      "grad_norm": 1.08176073460068,
+      "learning_rate": 2e-05,
+      "loss": 0.563477611541748,
+      "step": 2700
+    },
+    {
+      "epoch": 0.3482395271138525,
+      "grad_norm": 1.8379819591522175,
+      "learning_rate": 2e-05,
+      "loss": 0.5510289669036865,
+      "step": 2710
+    },
+    {
+      "epoch": 0.34952454381906967,
+      "grad_norm": 1.7677142442232125,
+      "learning_rate": 2e-05,
+      "loss": 0.5592892646789551,
+      "step": 2720
+    },
+    {
+      "epoch": 0.3508095605242868,
+      "grad_norm": 1.0682813633397992,
+      "learning_rate": 2e-05,
+      "loss": 0.5711023330688476,
+      "step": 2730
+    },
+    {
+      "epoch": 0.352094577229504,
+      "grad_norm": 1.1939874705331395,
+      "learning_rate": 2e-05,
+      "loss": 0.5497922420501709,
+      "step": 2740
+    },
+    {
+      "epoch": 0.3533795939347211,
+      "grad_norm": 1.9476474132994257,
+      "learning_rate": 2e-05,
+      "loss": 0.5518235206604004,
+      "step": 2750
+    },
+    {
+      "epoch": 0.3546646106399383,
+      "grad_norm": 1.1334509819944443,
+      "learning_rate": 2e-05,
+      "loss": 0.559119701385498,
+      "step": 2760
+    },
+    {
+      "epoch": 0.3559496273451555,
+      "grad_norm": 1.0364196016851452,
+      "learning_rate": 2e-05,
+      "loss": 0.5543029308319092,
+      "step": 2770
+    },
+    {
+      "epoch": 0.35723464405037264,
+      "grad_norm": 3.655128224331249,
+      "learning_rate": 2e-05,
+      "loss": 0.5799161434173584,
+      "step": 2780
+    },
+    {
+      "epoch": 0.35851966075558983,
+      "grad_norm": 0.8981643760931524,
+      "learning_rate": 2e-05,
+      "loss": 0.5641114234924316,
+      "step": 2790
+    },
+    {
+      "epoch": 0.35980467746080697,
+      "grad_norm": 1.698039690384192,
+      "learning_rate": 2e-05,
+      "loss": 0.5700871467590332,
+      "step": 2800
+    },
+    {
+      "epoch": 0.36108969416602416,
+      "grad_norm": 6.655614868583629,
+      "learning_rate": 2e-05,
+      "loss": 0.5688861846923828,
+      "step": 2810
+    },
+    {
+      "epoch": 0.36237471087124135,
+      "grad_norm": 1.997332388272014,
+      "learning_rate": 2e-05,
+      "loss": 0.5722577095031738,
+      "step": 2820
+    },
+    {
+      "epoch": 0.3636597275764585,
+      "grad_norm": 0.9594467060561305,
+      "learning_rate": 2e-05,
+      "loss": 0.5443659782409668,
+      "step": 2830
+    },
+    {
+      "epoch": 0.36494474428167567,
+      "grad_norm": 0.91245764004022,
+      "learning_rate": 2e-05,
+      "loss": 0.5773438453674317,
+      "step": 2840
+    },
+    {
+      "epoch": 0.3662297609868928,
+      "grad_norm": 0.8210577044882291,
+      "learning_rate": 2e-05,
+      "loss": 0.5669143676757813,
+      "step": 2850
+    },
+    {
+      "epoch": 0.36751477769211,
+      "grad_norm": 1.3195762714483341,
+      "learning_rate": 2e-05,
+      "loss": 0.5513071060180664,
+      "step": 2860
+    },
+    {
+      "epoch": 0.3687997943973272,
+      "grad_norm": 2.1431826414983166,
+      "learning_rate": 2e-05,
+      "loss": 0.5855628967285156,
+      "step": 2870
+    },
+    {
+      "epoch": 0.3700848111025443,
+      "grad_norm": 1.3999530612896527,
+      "learning_rate": 2e-05,
+      "loss": 0.5531571388244629,
+      "step": 2880
+    },
+    {
+      "epoch": 0.3713698278077615,
+      "grad_norm": 1.4235244487003083,
+      "learning_rate": 2e-05,
+      "loss": 0.5789942741394043,
+      "step": 2890
+    },
+    {
+      "epoch": 0.37265484451297864,
+      "grad_norm": 1.2793305759174294,
+      "learning_rate": 2e-05,
+      "loss": 0.5550153255462646,
+      "step": 2900
+    },
+    {
+      "epoch": 0.37393986121819583,
+      "grad_norm": 1.3352031079457276,
+      "learning_rate": 2e-05,
+      "loss": 0.5498593330383301,
+      "step": 2910
+    },
+    {
+      "epoch": 0.375224877923413,
+      "grad_norm": 0.8906129439892354,
+      "learning_rate": 2e-05,
+      "loss": 0.5600808143615723,
+      "step": 2920
+    },
+    {
+      "epoch": 0.37650989462863016,
+      "grad_norm": 0.8331902183001065,
+      "learning_rate": 2e-05,
+      "loss": 0.5451189041137695,
+      "step": 2930
+    },
+    {
+      "epoch": 0.37779491133384735,
+      "grad_norm": 0.9247178296029375,
+      "learning_rate": 2e-05,
+      "loss": 0.5684500694274902,
+      "step": 2940
+    },
+    {
+      "epoch": 0.3790799280390645,
+      "grad_norm": 0.8622451356315481,
+      "learning_rate": 2e-05,
+      "loss": 0.5382819175720215,
+      "step": 2950
+    },
+    {
+      "epoch": 0.3803649447442817,
+      "grad_norm": 1.1172742630258659,
+      "learning_rate": 2e-05,
+      "loss": 0.5757975578308105,
+      "step": 2960
+    },
+    {
+      "epoch": 0.38164996144949886,
+      "grad_norm": 1.0283177274045316,
+      "learning_rate": 2e-05,
+      "loss": 0.5533834457397461,
+      "step": 2970
+    },
+    {
+      "epoch": 0.382934978154716,
+      "grad_norm": 1.0175401634201733,
+      "learning_rate": 2e-05,
+      "loss": 0.5515987396240234,
+      "step": 2980
+    },
+    {
+      "epoch": 0.3842199948599332,
+      "grad_norm": 1.4354977554155102,
+      "learning_rate": 2e-05,
+      "loss": 0.5658344268798828,
+      "step": 2990
+    },
+    {
+      "epoch": 0.3855050115651503,
+      "grad_norm": 1.6054939179280252,
+      "learning_rate": 2e-05,
+      "loss": 0.552898359298706,
+      "step": 3000
+    },
+    {
+      "epoch": 0.3867900282703675,
+      "grad_norm": 4.768958745403176,
+      "learning_rate": 2e-05,
+      "loss": 0.5477462768554687,
+      "step": 3010
+    },
+    {
+      "epoch": 0.3880750449755847,
+      "grad_norm": 2.256581182679907,
+      "learning_rate": 2e-05,
+      "loss": 0.564853572845459,
+      "step": 3020
+    },
+    {
+      "epoch": 0.38936006168080184,
+      "grad_norm": 1.489426132688879,
+      "learning_rate": 2e-05,
+      "loss": 0.5578964233398438,
+      "step": 3030
+    },
+    {
+      "epoch": 0.39064507838601903,
+      "grad_norm": 2.419098232303151,
+      "learning_rate": 2e-05,
+      "loss": 0.5723655700683594,
+      "step": 3040
+    },
+    {
+      "epoch": 0.39193009509123616,
+      "grad_norm": 2.123799563231888,
+      "learning_rate": 2e-05,
+      "loss": 0.5831342220306397,
+      "step": 3050
+    },
+    {
+      "epoch": 0.39321511179645335,
+      "grad_norm": 1.4674304124573507,
+      "learning_rate": 2e-05,
+      "loss": 0.5662974834442138,
+      "step": 3060
+    },
+    {
+      "epoch": 0.39450012850167054,
+      "grad_norm": 1.0844772079659155,
+      "learning_rate": 2e-05,
+      "loss": 0.5623149394989013,
+      "step": 3070
+    },
+    {
+      "epoch": 0.3957851452068877,
+      "grad_norm": 1.011440120331522,
+      "learning_rate": 2e-05,
+      "loss": 0.5465374946594238,
+      "step": 3080
+    },
+    {
+      "epoch": 0.39707016191210487,
+      "grad_norm": 1.3825804620924182,
+      "learning_rate": 2e-05,
+      "loss": 0.5732436656951905,
+      "step": 3090
+    },
+    {
+      "epoch": 0.398355178617322,
+      "grad_norm": 1.3979379788272477,
+      "learning_rate": 2e-05,
+      "loss": 0.5535111904144288,
+      "step": 3100
+    },
+    {
+      "epoch": 0.3996401953225392,
+      "grad_norm": 1.7785540477783877,
+      "learning_rate": 2e-05,
+      "loss": 0.5451873779296875,
+      "step": 3110
+    },
+    {
+      "epoch": 0.4009252120277564,
+      "grad_norm": 0.8924263896101219,
+      "learning_rate": 2e-05,
+      "loss": 0.5406699180603027,
+      "step": 3120
+    },
+    {
+      "epoch": 0.4022102287329735,
+      "grad_norm": 1.1373819643342473,
+      "learning_rate": 2e-05,
+      "loss": 0.5513473510742187,
+      "step": 3130
+    },
+    {
+      "epoch": 0.4034952454381907,
+      "grad_norm": 1.0994386799297136,
+      "learning_rate": 2e-05,
+      "loss": 0.5355706691741944,
+      "step": 3140
+    },
+    {
+      "epoch": 0.40478026214340784,
+      "grad_norm": 1.0162640376389322,
+      "learning_rate": 2e-05,
+      "loss": 0.5479872703552247,
+      "step": 3150
+    },
+    {
+      "epoch": 0.40606527884862503,
+      "grad_norm": 3.1408471214997737,
+      "learning_rate": 2e-05,
+      "loss": 0.5464457988739013,
+      "step": 3160
+    },
+    {
+      "epoch": 0.4073502955538422,
+      "grad_norm": 0.7889751095078529,
+      "learning_rate": 2e-05,
+      "loss": 0.5393443584442139,
+      "step": 3170
+    },
+    {
+      "epoch": 0.40863531225905936,
+      "grad_norm": 1.310208063993292,
+      "learning_rate": 2e-05,
+      "loss": 0.5311233520507812,
+      "step": 3180
+    },
+    {
+      "epoch": 0.40992032896427655,
+      "grad_norm": 1.0487335227673358,
+      "learning_rate": 2e-05,
+      "loss": 0.5296288967132569,
+      "step": 3190
+    },
+    {
+      "epoch": 0.4112053456694937,
+      "grad_norm": 2.992668998010161,
+      "learning_rate": 2e-05,
+      "loss": 0.5394338130950928,
+      "step": 3200
+    },
+    {
+      "epoch": 0.4124903623747109,
+      "grad_norm": 1.4618475187035649,
+      "learning_rate": 2e-05,
+      "loss": 0.5431808471679688,
+      "step": 3210
+    },
+    {
+      "epoch": 0.41377537907992806,
+      "grad_norm": 0.9966604131325582,
+      "learning_rate": 2e-05,
+      "loss": 0.5678199291229248,
+      "step": 3220
+    },
+    {
+      "epoch": 0.4150603957851452,
+      "grad_norm": 1.141167921833559,
+      "learning_rate": 2e-05,
+      "loss": 0.5430806159973145,
+      "step": 3230
+    },
+    {
+      "epoch": 0.4163454124903624,
+      "grad_norm": 1.3296564258720673,
+      "learning_rate": 2e-05,
+      "loss": 0.5400397300720214,
+      "step": 3240
+    },
+    {
+      "epoch": 0.4176304291955795,
+      "grad_norm": 0.8728817029194036,
+      "learning_rate": 2e-05,
+      "loss": 0.5367677688598633,
+      "step": 3250
+    },
+    {
+      "epoch": 0.4189154459007967,
+      "grad_norm": 1.9655576984853433,
+      "learning_rate": 2e-05,
+      "loss": 0.538156795501709,
+      "step": 3260
+    },
+    {
+      "epoch": 0.4202004626060139,
+      "grad_norm": 2.8498337489131664,
+      "learning_rate": 2e-05,
+      "loss": 0.5624152183532715,
+      "step": 3270
+    },
+    {
+      "epoch": 0.42148547931123104,
+      "grad_norm": 1.3040895005506978,
+      "learning_rate": 2e-05,
+      "loss": 0.536140251159668,
+      "step": 3280
+    },
+    {
+      "epoch": 0.4227704960164482,
+      "grad_norm": 1.0826496626720115,
+      "learning_rate": 2e-05,
+      "loss": 0.5399922847747802,
+      "step": 3290
+    },
+    {
+      "epoch": 0.42405551272166536,
+      "grad_norm": 0.9303254763735685,
+      "learning_rate": 2e-05,
+      "loss": 0.5406898498535156,
+      "step": 3300
+    },
+    {
+      "epoch": 0.42534052942688255,
+      "grad_norm": 1.0334771258157418,
+      "learning_rate": 2e-05,
+      "loss": 0.5344079971313477,
+      "step": 3310
+    },
+    {
+      "epoch": 0.42662554613209974,
+      "grad_norm": 0.8892420113375932,
+      "learning_rate": 2e-05,
+      "loss": 0.5460707664489746,
+      "step": 3320
+    },
+    {
+      "epoch": 0.4279105628373169,
+      "grad_norm": 0.8565975212568558,
+      "learning_rate": 2e-05,
+      "loss": 0.5327538967132568,
+      "step": 3330
+    },
+    {
+      "epoch": 0.42919557954253407,
+      "grad_norm": 0.8562616233729462,
+      "learning_rate": 2e-05,
+      "loss": 0.537261962890625,
+      "step": 3340
+    },
+    {
+      "epoch": 0.4304805962477512,
+      "grad_norm": 1.5081862652030171,
+      "learning_rate": 2e-05,
+      "loss": 0.5425304412841797,
+      "step": 3350
+    },
+    {
+      "epoch": 0.4317656129529684,
+      "grad_norm": 1.2953143366832827,
+      "learning_rate": 2e-05,
+      "loss": 0.5529687404632568,
+      "step": 3360
+    },
+    {
+      "epoch": 0.4330506296581856,
+      "grad_norm": 0.8777856526113486,
+      "learning_rate": 2e-05,
+      "loss": 0.5515317440032959,
+      "step": 3370
+    },
+    {
+      "epoch": 0.4343356463634027,
+      "grad_norm": 0.8249396892026306,
+      "learning_rate": 2e-05,
+      "loss": 0.5435605525970459,
+      "step": 3380
+    },
+    {
+      "epoch": 0.4356206630686199,
+      "grad_norm": 0.9357579121046564,
+      "learning_rate": 2e-05,
+      "loss": 0.5655780792236328,
+      "step": 3390
+    },
+    {
+      "epoch": 0.43690567977383704,
+      "grad_norm": 1.2392930618814428,
+      "learning_rate": 2e-05,
+      "loss": 0.5421281337738038,
+      "step": 3400
+    },
+    {
+      "epoch": 0.43819069647905423,
+      "grad_norm": 3.356297884879854,
+      "learning_rate": 2e-05,
+      "loss": 0.5325273036956787,
+      "step": 3410
+    },
+    {
+      "epoch": 0.4394757131842714,
+      "grad_norm": 1.147421878334075,
+      "learning_rate": 2e-05,
+      "loss": 0.5385617733001709,
+      "step": 3420
+    },
+    {
+      "epoch": 0.44076072988948856,
+      "grad_norm": 2.129558147778632,
+      "learning_rate": 2e-05,
+      "loss": 0.5379881858825684,
+      "step": 3430
+    },
+    {
+      "epoch": 0.44204574659470575,
+      "grad_norm": 1.3015321628704448,
+      "learning_rate": 2e-05,
+      "loss": 0.5286228656768799,
+      "step": 3440
+    },
+    {
+      "epoch": 0.4433307632999229,
+      "grad_norm": 1.153693349901879,
+      "learning_rate": 2e-05,
+      "loss": 0.5413069725036621,
+      "step": 3450
+    },
+    {
+      "epoch": 0.44461578000514007,
+      "grad_norm": 2.0019159098326504,
+      "learning_rate": 2e-05,
+      "loss": 0.5547749519348144,
+      "step": 3460
+    },
+    {
+      "epoch": 0.44590079671035726,
+      "grad_norm": 1.9246055674315772,
+      "learning_rate": 2e-05,
+      "loss": 0.5637221813201905,
+      "step": 3470
+    },
+    {
+      "epoch": 0.4471858134155744,
+      "grad_norm": 0.8537519197236709,
+      "learning_rate": 2e-05,
+      "loss": 0.5237927913665772,
+      "step": 3480
+    },
+    {
+      "epoch": 0.4484708301207916,
+      "grad_norm": 3.6558936411348864,
+      "learning_rate": 2e-05,
+      "loss": 0.5587848663330078,
+      "step": 3490
+    },
+    {
+      "epoch": 0.4497558468260087,
+      "grad_norm": 1.3201621568548494,
+      "learning_rate": 2e-05,
+      "loss": 0.5530472755432129,
+      "step": 3500
+    },
+    {
+      "epoch": 0.4510408635312259,
+      "grad_norm": 1.8301021636187471,
+      "learning_rate": 2e-05,
+      "loss": 0.5419929504394532,
+      "step": 3510
+    },
+    {
+      "epoch": 0.4523258802364431,
+      "grad_norm": 1.197796150606289,
+      "learning_rate": 2e-05,
+      "loss": 0.5336377620697021,
+      "step": 3520
+    },
+    {
+      "epoch": 0.45361089694166024,
+      "grad_norm": 0.9921230033510117,
+      "learning_rate": 2e-05,
+      "loss": 0.5336089134216309,
+      "step": 3530
+    },
+    {
+      "epoch": 0.4548959136468774,
+      "grad_norm": 1.025255749559333,
+      "learning_rate": 2e-05,
+      "loss": 0.5249067306518554,
+      "step": 3540
+    },
+    {
+      "epoch": 0.45618093035209456,
+      "grad_norm": 2.203141895189741,
+      "learning_rate": 2e-05,
+      "loss": 0.5060970783233643,
+      "step": 3550
+    },
+    {
+      "epoch": 0.45746594705731175,
+      "grad_norm": 0.8653968117714671,
+      "learning_rate": 2e-05,
+      "loss": 0.5262903213500977,
+      "step": 3560
+    },
+    {
+      "epoch": 0.45875096376252894,
+      "grad_norm": 0.8315769887373267,
+      "learning_rate": 2e-05,
+      "loss": 0.5510161876678467,
+      "step": 3570
+    },
+    {
+      "epoch": 0.4600359804677461,
+      "grad_norm": 0.9242326080787269,
+      "learning_rate": 2e-05,
+      "loss": 0.5491601467132569,
+      "step": 3580
+    },
+    {
+      "epoch": 0.46132099717296327,
+      "grad_norm": 1.0335836367856064,
+      "learning_rate": 2e-05,
+      "loss": 0.5123754501342773,
+      "step": 3590
+    },
+    {
+      "epoch": 0.4626060138781804,
+      "grad_norm": 0.8444669871736762,
+      "learning_rate": 2e-05,
+      "loss": 0.5372439384460449,
+      "step": 3600
+    },
+    {
+      "epoch": 0.4638910305833976,
+      "grad_norm": 0.8317913031895797,
+      "learning_rate": 2e-05,
+      "loss": 0.5363702297210693,
+      "step": 3610
+    },
+    {
+      "epoch": 0.4651760472886147,
+      "grad_norm": 0.9531691499259727,
+      "learning_rate": 2e-05,
+      "loss": 0.5279970169067383,
+      "step": 3620
+    },
+    {
+      "epoch": 0.4664610639938319,
+      "grad_norm": 0.7581195638180102,
+      "learning_rate": 2e-05,
+      "loss": 0.5252976894378663,
+      "step": 3630
+    },
+    {
+      "epoch": 0.4677460806990491,
+      "grad_norm": 0.9039517198307234,
+      "learning_rate": 2e-05,
+      "loss": 0.5378138542175293,
+      "step": 3640
+    },
+    {
+      "epoch": 0.46903109740426624,
+      "grad_norm": 1.185405763226095,
+      "learning_rate": 2e-05,
+      "loss": 0.5387648582458496,
+      "step": 3650
+    },
+    {
+      "epoch": 0.47031611410948343,
+      "grad_norm": 1.2023594830243058,
+      "learning_rate": 2e-05,
+      "loss": 0.5225371837615966,
+      "step": 3660
+    },
+    {
+      "epoch": 0.47160113081470056,
+      "grad_norm": 0.9828603398547323,
+      "learning_rate": 2e-05,
+      "loss": 0.5241292953491211,
+      "step": 3670
+    },
+    {
+      "epoch": 0.47288614751991775,
+      "grad_norm": 0.8642992181465071,
+      "learning_rate": 2e-05,
+      "loss": 0.5223854064941407,
+      "step": 3680
+    },
+    {
+      "epoch": 0.47417116422513494,
+      "grad_norm": 0.8676448316025713,
+      "learning_rate": 2e-05,
+      "loss": 0.5226163387298584,
+      "step": 3690
+    },
+    {
+      "epoch": 0.4754561809303521,
+      "grad_norm": 0.8185991382730614,
+      "learning_rate": 2e-05,
+      "loss": 0.5234535217285157,
+      "step": 3700
+    },
+    {
+      "epoch": 0.47674119763556927,
+      "grad_norm": 0.7962298986101759,
+      "learning_rate": 2e-05,
+      "loss": 0.5217377185821533,
+      "step": 3710
+    },
+    {
+      "epoch": 0.4780262143407864,
+      "grad_norm": 1.1902737506527137,
+      "learning_rate": 2e-05,
+      "loss": 0.5335277557373047,
+      "step": 3720
+    },
+    {
+      "epoch": 0.4793112310460036,
+      "grad_norm": 1.041408138289439,
+      "learning_rate": 2e-05,
+      "loss": 0.5273212909698486,
+      "step": 3730
+    },
+    {
+      "epoch": 0.4805962477512208,
+      "grad_norm": 0.8241135998641578,
+      "learning_rate": 2e-05,
+      "loss": 0.5259427547454834,
+      "step": 3740
+    },
+    {
+      "epoch": 0.4818812644564379,
+      "grad_norm": 0.8935562023307105,
+      "learning_rate": 2e-05,
+      "loss": 0.5223063468933106,
+      "step": 3750
+    },
+    {
+      "epoch": 0.4831662811616551,
+      "grad_norm": 1.0126414591167698,
+      "learning_rate": 2e-05,
+      "loss": 0.5359176158905029,
+      "step": 3760
+    },
+    {
+      "epoch": 0.48445129786687224,
+      "grad_norm": 0.8057994880232076,
+      "learning_rate": 2e-05,
+      "loss": 0.5072232246398926,
+      "step": 3770
+    },
+    {
+      "epoch": 0.48573631457208943,
+      "grad_norm": 0.7836719142318709,
+      "learning_rate": 2e-05,
+      "loss": 0.5273278236389161,
+      "step": 3780
+    },
+    {
+      "epoch": 0.4870213312773066,
+      "grad_norm": 0.8494310056339948,
+      "learning_rate": 2e-05,
+      "loss": 0.5251004219055175,
+      "step": 3790
+    },
+    {
+      "epoch": 0.48830634798252376,
+      "grad_norm": 0.884991217925323,
+      "learning_rate": 2e-05,
+      "loss": 0.5339468002319336,
+      "step": 3800
+    },
+    {
+      "epoch": 0.48959136468774095,
+      "grad_norm": 4.557041648842915,
+      "learning_rate": 2e-05,
+      "loss": 0.5082455635070801,
+      "step": 3810
+    },
+    {
+      "epoch": 0.4908763813929581,
+      "grad_norm": 0.9422127831279262,
+      "learning_rate": 2e-05,
+      "loss": 0.5068249702453613,
+      "step": 3820
+    },
+    {
+      "epoch": 0.4921613980981753,
+      "grad_norm": 0.780966844061176,
+      "learning_rate": 2e-05,
+      "loss": 0.5323922634124756,
+      "step": 3830
+    },
+    {
+      "epoch": 0.49344641480339246,
+      "grad_norm": 0.8552499567648932,
+      "learning_rate": 2e-05,
+      "loss": 0.5119367599487304,
+      "step": 3840
+    },
+    {
+      "epoch": 0.4947314315086096,
+      "grad_norm": 0.8346173478373825,
+      "learning_rate": 2e-05,
+      "loss": 0.5233124256134033,
+      "step": 3850
+    },
+    {
+      "epoch": 0.4960164482138268,
+      "grad_norm": 0.7992881937056178,
+      "learning_rate": 2e-05,
+      "loss": 0.524324607849121,
+      "step": 3860
+    },
+    {
+      "epoch": 0.4973014649190439,
+      "grad_norm": 0.7329983541917202,
+      "learning_rate": 2e-05,
+      "loss": 0.5258885860443115,
+      "step": 3870
+    },
+    {
+      "epoch": 0.4985864816242611,
+      "grad_norm": 0.892204411513666,
+      "learning_rate": 2e-05,
+      "loss": 0.514957857131958,
+      "step": 3880
+    },
+    {
+      "epoch": 0.4998714983294783,
+      "grad_norm": 1.024376041601587,
+      "learning_rate": 2e-05,
+      "loss": 0.5414502143859863,
+      "step": 3890
+    },
+    {
+      "epoch": 0.5011565150346955,
+      "grad_norm": 0.8450390230296335,
+      "learning_rate": 2e-05,
+      "loss": 0.510601806640625,
+      "step": 3900
+    },
+    {
+      "epoch": 0.5024415317399126,
+      "grad_norm": 0.9244268730074975,
+      "learning_rate": 2e-05,
+      "loss": 0.5121501922607422,
+      "step": 3910
+    },
+    {
+      "epoch": 0.5037265484451298,
+      "grad_norm": 1.221985867465146,
+      "learning_rate": 2e-05,
+      "loss": 0.507999324798584,
+      "step": 3920
+    },
+    {
+      "epoch": 0.505011565150347,
+      "grad_norm": 0.8608351991696312,
+      "learning_rate": 2e-05,
+      "loss": 0.5270411014556885,
+      "step": 3930
+    },
+    {
+      "epoch": 0.5062965818555641,
+      "grad_norm": 0.9155568954584324,
+      "learning_rate": 2e-05,
+      "loss": 0.5433553218841553,
+      "step": 3940
+    },
+    {
+      "epoch": 0.5075815985607813,
+      "grad_norm": 0.8105404883003526,
+      "learning_rate": 2e-05,
+      "loss": 0.5276482582092286,
+      "step": 3950
+    },
+    {
+      "epoch": 0.5088666152659984,
+      "grad_norm": 1.0889391407418472,
+      "learning_rate": 2e-05,
+      "loss": 0.5249395370483398,
+      "step": 3960
+    },
+    {
+      "epoch": 0.5101516319712156,
+      "grad_norm": 0.9712484557078159,
+      "learning_rate": 2e-05,
+      "loss": 0.5297153472900391,
+      "step": 3970
+    },
+    {
+      "epoch": 0.5114366486764328,
+      "grad_norm": 0.9295866427194486,
+      "learning_rate": 2e-05,
+      "loss": 0.5375542640686035,
+      "step": 3980
+    },
+    {
+      "epoch": 0.51272166538165,
+      "grad_norm": 0.8525370982692666,
+      "learning_rate": 2e-05,
+      "loss": 0.5041474342346192,
+      "step": 3990
+    },
+    {
+      "epoch": 0.5140066820868672,
+      "grad_norm": 1.393898540314037,
+      "learning_rate": 2e-05,
+      "loss": 0.5157971382141113,
+      "step": 4000
+    },
+    {
+      "epoch": 0.5152916987920843,
+      "grad_norm": 1.412820986987432,
+      "learning_rate": 2e-05,
+      "loss": 0.5109616279602051,
+      "step": 4010
+    },
+    {
+      "epoch": 0.5165767154973014,
+      "grad_norm": 0.8490929123201654,
+      "learning_rate": 2e-05,
+      "loss": 0.5206557750701905,
+      "step": 4020
+    },
+    {
+      "epoch": 0.5178617322025186,
+      "grad_norm": 0.9585138143377944,
+      "learning_rate": 2e-05,
+      "loss": 0.5254447937011719,
+      "step": 4030
+    },
+    {
+      "epoch": 0.5191467489077358,
+      "grad_norm": 0.8938142529415174,
+      "learning_rate": 2e-05,
+      "loss": 0.5061586380004883,
+      "step": 4040
+    },
+    {
+      "epoch": 0.520431765612953,
+      "grad_norm": 1.0023344211133087,
+      "learning_rate": 2e-05,
+      "loss": 0.5149017810821533,
+      "step": 4050
+    },
+    {
+      "epoch": 0.5217167823181701,
+      "grad_norm": 1.0744495919967836,
+      "learning_rate": 2e-05,
+      "loss": 0.5346870422363281,
+      "step": 4060
+    },
+    {
+      "epoch": 0.5230017990233873,
+      "grad_norm": 0.8650535871523433,
+      "learning_rate": 2e-05,
+      "loss": 0.5328707218170166,
+      "step": 4070
+    },
+    {
+      "epoch": 0.5242868157286045,
+      "grad_norm": 0.8293734854135136,
+      "learning_rate": 2e-05,
+      "loss": 0.4980588436126709,
+      "step": 4080
+    },
+    {
+      "epoch": 0.5255718324338217,
+      "grad_norm": 0.927186213972399,
+      "learning_rate": 2e-05,
+      "loss": 0.5030704498291015,
+      "step": 4090
+    },
+    {
+      "epoch": 0.5268568491390389,
+      "grad_norm": 0.825662840581205,
+      "learning_rate": 2e-05,
+      "loss": 0.5091442584991455,
+      "step": 4100
+    },
+    {
+      "epoch": 0.5281418658442559,
+      "grad_norm": 1.1835132968483324,
+      "learning_rate": 2e-05,
+      "loss": 0.503142261505127,
+      "step": 4110
+    },
+    {
+      "epoch": 0.5294268825494731,
+      "grad_norm": 1.4680190855929387,
+      "learning_rate": 2e-05,
+      "loss": 0.5053106307983398,
+      "step": 4120
+    },
+    {
+      "epoch": 0.5307118992546903,
+      "grad_norm": 0.8408820065438916,
+      "learning_rate": 2e-05,
+      "loss": 0.5332332134246827,
+      "step": 4130
+    },
+    {
+      "epoch": 0.5319969159599075,
+      "grad_norm": 0.8779447282939558,
+      "learning_rate": 2e-05,
+      "loss": 0.519904088973999,
+      "step": 4140
+    },
+    {
+      "epoch": 0.5332819326651247,
+      "grad_norm": 0.7815891708805722,
+      "learning_rate": 2e-05,
+      "loss": 0.5154225349426269,
+      "step": 4150
+    },
+    {
+      "epoch": 0.5345669493703418,
+      "grad_norm": 0.741539107550217,
+      "learning_rate": 2e-05,
+      "loss": 0.5153237342834472,
+      "step": 4160
+    },
+    {
+      "epoch": 0.535851966075559,
+      "grad_norm": 0.8642257491550034,
+      "learning_rate": 2e-05,
+      "loss": 0.5246007442474365,
+      "step": 4170
+    },
+    {
+      "epoch": 0.5371369827807762,
+      "grad_norm": 0.8918801618456383,
+      "learning_rate": 2e-05,
+      "loss": 0.5201076984405517,
+      "step": 4180
+    },
+    {
+      "epoch": 0.5384219994859933,
+      "grad_norm": 0.811431447302449,
+      "learning_rate": 2e-05,
+      "loss": 0.5242047786712647,
+      "step": 4190
+    },
+    {
+      "epoch": 0.5397070161912105,
+      "grad_norm": 0.9247061430385739,
+      "learning_rate": 2e-05,
+      "loss": 0.5076504707336426,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5409920328964276,
+      "grad_norm": 0.804265935549892,
+      "learning_rate": 2e-05,
+      "loss": 0.4955127716064453,
+      "step": 4210
+    },
+    {
+      "epoch": 0.5422770496016448,
+      "grad_norm": 0.765336423614461,
+      "learning_rate": 2e-05,
+      "loss": 0.48396754264831543,
+      "step": 4220
+    },
+    {
+      "epoch": 0.543562066306862,
+      "grad_norm": 1.0023170827234076,
+      "learning_rate": 2e-05,
+      "loss": 0.5146978378295899,
+      "step": 4230
+    },
+    {
+      "epoch": 0.5448470830120792,
+      "grad_norm": 0.7906754242019115,
+      "learning_rate": 2e-05,
+      "loss": 0.5109517097473144,
+      "step": 4240
+    },
+    {
+      "epoch": 0.5461320997172964,
+      "grad_norm": 0.8566457409851771,
+      "learning_rate": 2e-05,
+      "loss": 0.5233433723449707,
+      "step": 4250
+    },
+    {
+      "epoch": 0.5474171164225135,
+      "grad_norm": 0.8071996796527903,
+      "learning_rate": 2e-05,
+      "loss": 0.4995774269104004,
+      "step": 4260
+    },
+    {
+      "epoch": 0.5487021331277306,
+      "grad_norm": 1.9066356773434627,
+      "learning_rate": 2e-05,
+      "loss": 0.5224987030029297,
+      "step": 4270
+    },
+    {
+      "epoch": 0.5499871498329478,
+      "grad_norm": 1.402488213668813,
+      "learning_rate": 2e-05,
+      "loss": 0.48797311782836916,
+      "step": 4280
+    },
+    {
+      "epoch": 0.551272166538165,
+      "grad_norm": 0.837017402855147,
+      "learning_rate": 2e-05,
+      "loss": 0.5141610145568848,
+      "step": 4290
+    },
+    {
+      "epoch": 0.5525571832433822,
+      "grad_norm": 0.9355414644146571,
+      "learning_rate": 2e-05,
+      "loss": 0.5177589416503906,
+      "step": 4300
+    },
+    {
+      "epoch": 0.5538421999485993,
+      "grad_norm": 0.9003555442550051,
+      "learning_rate": 2e-05,
+      "loss": 0.5202107429504395,
+      "step": 4310
+    },
+    {
+      "epoch": 0.5551272166538165,
+      "grad_norm": 0.8442387631592798,
+      "learning_rate": 2e-05,
+      "loss": 0.49821925163269043,
+      "step": 4320
+    },
+    {
+      "epoch": 0.5564122333590337,
+      "grad_norm": 0.8262309912870951,
+      "learning_rate": 2e-05,
+      "loss": 0.5158453941345215,
+      "step": 4330
+    },
+    {
+      "epoch": 0.5576972500642509,
+      "grad_norm": 0.8654096014165091,
+      "learning_rate": 2e-05,
+      "loss": 0.5056120872497558,
+      "step": 4340
+    },
+    {
+      "epoch": 0.558982266769468,
+      "grad_norm": 0.7968198572611914,
+      "learning_rate": 2e-05,
+      "loss": 0.4912233352661133,
+      "step": 4350
+    },
+    {
+      "epoch": 0.5602672834746851,
+      "grad_norm": 0.7675806823780124,
+      "learning_rate": 2e-05,
+      "loss": 0.5169336318969726,
+      "step": 4360
+    },
+    {
+      "epoch": 0.5615523001799023,
+      "grad_norm": 0.7711075382547916,
+      "learning_rate": 2e-05,
+      "loss": 0.5147769927978516,
+      "step": 4370
+    },
+    {
+      "epoch": 0.5628373168851195,
+      "grad_norm": 1.0580256598185973,
+      "learning_rate": 2e-05,
+      "loss": 0.49887537956237793,
+      "step": 4380
+    },
+    {
+      "epoch": 0.5641223335903367,
+      "grad_norm": 0.7572090059994065,
+      "learning_rate": 2e-05,
+      "loss": 0.5261769771575928,
+      "step": 4390
+    },
+    {
+      "epoch": 0.5654073502955539,
+      "grad_norm": 0.8322679297779081,
+      "learning_rate": 2e-05,
+      "loss": 0.488920783996582,
+      "step": 4400
+    },
+    {
+      "epoch": 0.566692367000771,
+      "grad_norm": 0.7253499845948285,
+      "learning_rate": 2e-05,
+      "loss": 0.5159592628479004,
+      "step": 4410
+    },
+    {
+      "epoch": 0.5679773837059882,
+      "grad_norm": 0.9813852336126379,
+      "learning_rate": 2e-05,
+      "loss": 0.5132109642028808,
+      "step": 4420
+    },
+    {
+      "epoch": 0.5692624004112053,
+      "grad_norm": 0.7355238005112201,
+      "learning_rate": 2e-05,
+      "loss": 0.4945272445678711,
+      "step": 4430
+    },
+    {
+      "epoch": 0.5705474171164225,
+      "grad_norm": 0.8316185891033692,
+      "learning_rate": 2e-05,
+      "loss": 0.503313398361206,
+      "step": 4440
+    },
+    {
+      "epoch": 0.5718324338216397,
+      "grad_norm": 1.2978466287777024,
+      "learning_rate": 2e-05,
+      "loss": 0.5190382480621338,
+      "step": 4450
+    },
+    {
+      "epoch": 0.5731174505268568,
+      "grad_norm": 0.8591006854492692,
+      "learning_rate": 2e-05,
+      "loss": 0.5209619522094726,
+      "step": 4460
+    },
+    {
+      "epoch": 0.574402467232074,
+      "grad_norm": 1.210997326381989,
+      "learning_rate": 2e-05,
+      "loss": 0.5047403335571289,
+      "step": 4470
+    },
+    {
+      "epoch": 0.5756874839372912,
+      "grad_norm": 0.8130614827063218,
+      "learning_rate": 2e-05,
+      "loss": 0.51310133934021,
+      "step": 4480
+    },
+    {
+      "epoch": 0.5769725006425084,
+      "grad_norm": 1.0236502525664122,
+      "learning_rate": 2e-05,
+      "loss": 0.5052859783172607,
+      "step": 4490
+    },
+    {
+      "epoch": 0.5782575173477256,
+      "grad_norm": 0.7905721438801276,
+      "learning_rate": 2e-05,
+      "loss": 0.5106428146362305,
+      "step": 4500
+    },
+    {
+      "epoch": 0.5795425340529426,
+      "grad_norm": 0.8152154981528194,
+      "learning_rate": 2e-05,
+      "loss": 0.49611482620239256,
+      "step": 4510
+    },
+    {
+      "epoch": 0.5808275507581598,
+      "grad_norm": 0.7824319580994991,
+      "learning_rate": 2e-05,
+      "loss": 0.4803518772125244,
+      "step": 4520
+    },
+    {
+      "epoch": 0.582112567463377,
+      "grad_norm": 0.7399753164356996,
+      "learning_rate": 2e-05,
+      "loss": 0.5259390354156495,
+      "step": 4530
+    },
+    {
+      "epoch": 0.5833975841685942,
+      "grad_norm": 0.7680298465582936,
+      "learning_rate": 2e-05,
+      "loss": 0.4961595058441162,
+      "step": 4540
+    },
+    {
+      "epoch": 0.5846826008738114,
+      "grad_norm": 0.9465049047332362,
+      "learning_rate": 2e-05,
+      "loss": 0.4950114250183105,
+      "step": 4550
+    },
+    {
+      "epoch": 0.5859676175790285,
+      "grad_norm": 2.502840865583999,
+      "learning_rate": 2e-05,
+      "loss": 0.5089702606201172,
+      "step": 4560
+    },
+    {
+      "epoch": 0.5872526342842457,
+      "grad_norm": 0.8141311487817591,
+      "learning_rate": 2e-05,
+      "loss": 0.4885561943054199,
+      "step": 4570
+    },
+    {
+      "epoch": 0.5885376509894629,
+      "grad_norm": 0.7869425519272899,
+      "learning_rate": 2e-05,
+      "loss": 0.47963666915893555,
+      "step": 4580
+    },
+    {
+      "epoch": 0.5898226676946801,
+      "grad_norm": 0.8773411300442306,
+      "learning_rate": 2e-05,
+      "loss": 0.5025991439819336,
+      "step": 4590
+    },
+    {
+      "epoch": 0.5911076843998972,
+      "grad_norm": 0.8364171111573889,
+      "learning_rate": 2e-05,
+      "loss": 0.5023444652557373,
+      "step": 4600
+    },
+    {
+      "epoch": 0.5923927011051143,
+      "grad_norm": 0.8284238552838473,
+      "learning_rate": 2e-05,
+      "loss": 0.49907331466674804,
+      "step": 4610
+    },
+    {
+      "epoch": 0.5936777178103315,
+      "grad_norm": 0.783480414368962,
+      "learning_rate": 2e-05,
+      "loss": 0.5003586769104004,
+      "step": 4620
+    },
+    {
+      "epoch": 0.5949627345155487,
+      "grad_norm": 0.917927401369239,
+      "learning_rate": 2e-05,
+      "loss": 0.49894137382507325,
+      "step": 4630
+    },
+    {
+      "epoch": 0.5962477512207659,
+      "grad_norm": 0.7752427636525809,
+      "learning_rate": 2e-05,
+      "loss": 0.4990223407745361,
+      "step": 4640
+    },
+    {
+      "epoch": 0.5975327679259831,
+      "grad_norm": 0.7367735234905566,
+      "learning_rate": 2e-05,
+      "loss": 0.48964557647705076,
+      "step": 4650
+    },
+    {
+      "epoch": 0.5988177846312002,
+      "grad_norm": 0.7509096085536084,
+      "learning_rate": 2e-05,
+      "loss": 0.5013049125671387,
+      "step": 4660
+    },
+    {
+      "epoch": 0.6001028013364174,
+      "grad_norm": 0.8017122679405846,
+      "learning_rate": 2e-05,
+      "loss": 0.4996955871582031,
+      "step": 4670
+    },
+    {
+      "epoch": 0.6013878180416345,
+      "grad_norm": 1.2923423781205177,
+      "learning_rate": 2e-05,
+      "loss": 0.49192109107971194,
+      "step": 4680
+    },
+    {
+      "epoch": 0.6026728347468517,
+      "grad_norm": 0.9573498162819248,
+      "learning_rate": 2e-05,
+      "loss": 0.5013810157775879,
+      "step": 4690
+    },
+    {
+      "epoch": 0.6039578514520689,
+      "grad_norm": 0.8151866127899103,
+      "learning_rate": 2e-05,
+      "loss": 0.5075257778167724,
+      "step": 4700
+    },
+    {
+      "epoch": 0.605242868157286,
+      "grad_norm": 0.8663750065300752,
+      "learning_rate": 2e-05,
+      "loss": 0.5030995845794678,
+      "step": 4710
+    },
+    {
+      "epoch": 0.6065278848625032,
+      "grad_norm": 1.7751782466164017,
+      "learning_rate": 2e-05,
+      "loss": 0.4892571449279785,
+      "step": 4720
+    },
+    {
+      "epoch": 0.6078129015677204,
+      "grad_norm": 1.0113198598075126,
+      "learning_rate": 2e-05,
+      "loss": 0.509736442565918,
+      "step": 4730
+    },
+    {
+      "epoch": 0.6090979182729376,
+      "grad_norm": 0.8712620566883285,
+      "learning_rate": 2e-05,
+      "loss": 0.5105513572692871,
+      "step": 4740
+    },
+    {
+      "epoch": 0.6103829349781548,
+      "grad_norm": 0.7822826623078739,
+      "learning_rate": 2e-05,
+      "loss": 0.5033591747283935,
+      "step": 4750
+    },
+    {
+      "epoch": 0.6116679516833718,
+      "grad_norm": 0.8092735884511209,
+      "learning_rate": 2e-05,
+      "loss": 0.5063093662261963,
+      "step": 4760
+    },
+    {
+      "epoch": 0.612952968388589,
+      "grad_norm": 0.9728916735115027,
+      "learning_rate": 2e-05,
+      "loss": 0.4826228618621826,
+      "step": 4770
+    },
+    {
+      "epoch": 0.6142379850938062,
+      "grad_norm": 0.9104141074900824,
+      "learning_rate": 2e-05,
+      "loss": 0.48541889190673826,
+      "step": 4780
+    },
+    {
+      "epoch": 0.6155230017990234,
+      "grad_norm": 1.1940811066832335,
+      "learning_rate": 2e-05,
+      "loss": 0.5228432655334473,
+      "step": 4790
+    },
+    {
+      "epoch": 0.6168080185042406,
+      "grad_norm": 0.9398566457091766,
+      "learning_rate": 2e-05,
+      "loss": 0.48981599807739257,
+      "step": 4800
+    },
+    {
+      "epoch": 0.6180930352094577,
+      "grad_norm": 0.7730517710383639,
+      "learning_rate": 2e-05,
+      "loss": 0.5024430274963378,
+      "step": 4810
+    },
+    {
+      "epoch": 0.6193780519146749,
+      "grad_norm": 0.8235572252753366,
+      "learning_rate": 2e-05,
+      "loss": 0.4938220500946045,
+      "step": 4820
+    },
+    {
+      "epoch": 0.6206630686198921,
+      "grad_norm": 0.7970760403588683,
+      "learning_rate": 2e-05,
+      "loss": 0.489678955078125,
+      "step": 4830
+    },
+    {
+      "epoch": 0.6219480853251093,
+      "grad_norm": 0.7850106082602308,
+      "learning_rate": 2e-05,
+      "loss": 0.4984504222869873,
+      "step": 4840
+    },
+    {
+      "epoch": 0.6232331020303264,
+      "grad_norm": 0.8210859780133003,
+      "learning_rate": 2e-05,
+      "loss": 0.5064220428466797,
+      "step": 4850
+    },
+    {
+      "epoch": 0.6245181187355435,
+      "grad_norm": 0.8171403850112406,
+      "learning_rate": 2e-05,
+      "loss": 0.47418646812438964,
+      "step": 4860
+    },
+    {
+      "epoch": 0.6258031354407607,
+      "grad_norm": 0.7773259617951196,
+      "learning_rate": 2e-05,
+      "loss": 0.5049054622650146,
+      "step": 4870
+    },
+    {
+      "epoch": 0.6270881521459779,
+      "grad_norm": 1.038368730755559,
+      "learning_rate": 2e-05,
+      "loss": 0.4879733085632324,
+      "step": 4880
+    },
+    {
+      "epoch": 0.6283731688511951,
+      "grad_norm": 0.8269477391316369,
+      "learning_rate": 2e-05,
+      "loss": 0.49253435134887696,
+      "step": 4890
+    },
+    {
+      "epoch": 0.6296581855564123,
+      "grad_norm": 0.9228960081079675,
+      "learning_rate": 2e-05,
+      "loss": 0.5066840171813964,
+      "step": 4900
+    },
+    {
+      "epoch": 0.6309432022616294,
+      "grad_norm": 2.0877713907036473,
+      "learning_rate": 2e-05,
+      "loss": 0.501807975769043,
+      "step": 4910
+    },
+    {
+      "epoch": 0.6322282189668466,
+      "grad_norm": 0.8625511489390753,
+      "learning_rate": 2e-05,
+      "loss": 0.5112593650817872,
+      "step": 4920
+    },
+    {
+      "epoch": 0.6335132356720637,
+      "grad_norm": 0.8308733958285156,
+      "learning_rate": 2e-05,
+      "loss": 0.5196004390716553,
+      "step": 4930
+    },
+    {
+      "epoch": 0.6347982523772809,
+      "grad_norm": 0.849836168759295,
+      "learning_rate": 2e-05,
+      "loss": 0.49697341918945315,
+      "step": 4940
+    },
+    {
+      "epoch": 0.6360832690824981,
+      "grad_norm": 0.741029812405731,
+      "learning_rate": 2e-05,
+      "loss": 0.4893211841583252,
+      "step": 4950
+    },
+    {
+      "epoch": 0.6373682857877152,
+      "grad_norm": 0.8081975039156659,
+      "learning_rate": 2e-05,
+      "loss": 0.47731471061706543,
+      "step": 4960
+    },
+    {
+      "epoch": 0.6386533024929324,
+      "grad_norm": 0.8415971689412449,
+      "learning_rate": 2e-05,
+      "loss": 0.48677778244018555,
+      "step": 4970
+    },
+    {
+      "epoch": 0.6399383191981496,
+      "grad_norm": 0.7927528702083361,
+      "learning_rate": 2e-05,
+      "loss": 0.4838107109069824,
+      "step": 4980
+    },
+    {
+      "epoch": 0.6412233359033668,
+      "grad_norm": 0.9639699643542403,
+      "learning_rate": 2e-05,
+      "loss": 0.5059066772460937,
+      "step": 4990
+    },
+    {
+      "epoch": 0.642508352608584,
+      "grad_norm": 0.8310517623053585,
+      "learning_rate": 2e-05,
+      "loss": 0.525218391418457,
+      "step": 5000
+    },
+    {
+      "epoch": 0.643793369313801,
+      "grad_norm": 0.8113766460858126,
+      "learning_rate": 2e-05,
+      "loss": 0.5149060726165772,
+      "step": 5010
+    },
+    {
+      "epoch": 0.6450783860190182,
+      "grad_norm": 0.7627832593146716,
+      "learning_rate": 2e-05,
+      "loss": 0.48557653427124026,
+      "step": 5020
+    },
+    {
+      "epoch": 0.6463634027242354,
+      "grad_norm": 0.7988163499646616,
+      "learning_rate": 2e-05,
+      "loss": 0.4841338634490967,
+      "step": 5030
+    },
+    {
+      "epoch": 0.6476484194294526,
+      "grad_norm": 0.7674769002852545,
+      "learning_rate": 2e-05,
+      "loss": 0.49269657135009765,
+      "step": 5040
+    },
+    {
+      "epoch": 0.6489334361346698,
+      "grad_norm": 0.952929885959604,
+      "learning_rate": 2e-05,
+      "loss": 0.4772337913513184,
+      "step": 5050
+    },
+    {
+      "epoch": 0.6502184528398869,
+      "grad_norm": 0.810770670832558,
+      "learning_rate": 2e-05,
+      "loss": 0.4704909324645996,
+      "step": 5060
+    },
+    {
+      "epoch": 0.6515034695451041,
+      "grad_norm": 0.9145425400350574,
+      "learning_rate": 2e-05,
+      "loss": 0.4715601444244385,
+      "step": 5070
+    },
+    {
+      "epoch": 0.6527884862503213,
+      "grad_norm": 0.8872950851424755,
+      "learning_rate": 2e-05,
+      "loss": 0.48098044395446776,
+      "step": 5080
+    },
+    {
+      "epoch": 0.6540735029555385,
+      "grad_norm": 0.8448359474479072,
+      "learning_rate": 2e-05,
+      "loss": 0.47185382843017576,
+      "step": 5090
+    },
+    {
+      "epoch": 0.6553585196607556,
+      "grad_norm": 0.9549301529412771,
+      "learning_rate": 2e-05,
+      "loss": 0.4880897045135498,
+      "step": 5100
+    },
+    {
+      "epoch": 0.6566435363659727,
+      "grad_norm": 0.8363704008974566,
+      "learning_rate": 2e-05,
+      "loss": 0.47212977409362794,
+      "step": 5110
+    },
+    {
+      "epoch": 0.6579285530711899,
+      "grad_norm": 0.8382330327043032,
+      "learning_rate": 2e-05,
+      "loss": 0.49033145904541015,
+      "step": 5120
+    },
+    {
+      "epoch": 0.6592135697764071,
+      "grad_norm": 0.8825068282377296,
+      "learning_rate": 2e-05,
+      "loss": 0.48597164154052735,
+      "step": 5130
+    },
+    {
+      "epoch": 0.6604985864816243,
+      "grad_norm": 0.8055402962501728,
+      "learning_rate": 2e-05,
+      "loss": 0.5118926525115967,
+      "step": 5140
+    },
+    {
+      "epoch": 0.6617836031868414,
+      "grad_norm": 0.8486492111132548,
+      "learning_rate": 2e-05,
+      "loss": 0.482420539855957,
+      "step": 5150
+    },
+    {
+      "epoch": 0.6630686198920586,
+      "grad_norm": 0.7662109697663915,
+      "learning_rate": 2e-05,
+      "loss": 0.4894374370574951,
+      "step": 5160
+    },
+    {
+      "epoch": 0.6643536365972758,
+      "grad_norm": 0.8412582206599033,
+      "learning_rate": 2e-05,
+      "loss": 0.4936180114746094,
+      "step": 5170
+    },
+    {
+      "epoch": 0.6656386533024929,
+      "grad_norm": 1.1695944242248755,
+      "learning_rate": 2e-05,
+      "loss": 0.49751806259155273,
+      "step": 5180
+    },
+    {
+      "epoch": 0.6669236700077101,
+      "grad_norm": 0.8098035889134314,
+      "learning_rate": 2e-05,
+      "loss": 0.5000545501708984,
+      "step": 5190
+    },
+    {
+      "epoch": 0.6682086867129272,
+      "grad_norm": 0.8105336483254787,
+      "learning_rate": 2e-05,
+      "loss": 0.4880019187927246,
+      "step": 5200
+    },
+    {
+      "epoch": 0.6694937034181444,
+      "grad_norm": 1.5053887578784888,
+      "learning_rate": 2e-05,
+      "loss": 0.4805875778198242,
+      "step": 5210
+    },
+    {
+      "epoch": 0.6707787201233616,
+      "grad_norm": 0.9784403092870857,
+      "learning_rate": 2e-05,
+      "loss": 0.4997675895690918,
+      "step": 5220
+    },
+    {
+      "epoch": 0.6720637368285788,
+      "grad_norm": 0.8010190595399024,
+      "learning_rate": 2e-05,
+      "loss": 0.4884494781494141,
+      "step": 5230
+    },
+    {
+      "epoch": 0.673348753533796,
+      "grad_norm": 0.8548760519066766,
+      "learning_rate": 2e-05,
+      "loss": 0.476505184173584,
+      "step": 5240
+    },
+    {
+      "epoch": 0.674633770239013,
+      "grad_norm": 0.8195769041495217,
+      "learning_rate": 2e-05,
+      "loss": 0.48655381202697756,
+      "step": 5250
+    },
+    {
+      "epoch": 0.6759187869442302,
+      "grad_norm": 0.9996707942902721,
+      "learning_rate": 2e-05,
+      "loss": 0.4948671817779541,
+      "step": 5260
+    },
+    {
+      "epoch": 0.6772038036494474,
+      "grad_norm": 0.8124143118764754,
+      "learning_rate": 2e-05,
+      "loss": 0.4862051963806152,
+      "step": 5270
+    },
+    {
+      "epoch": 0.6784888203546646,
+      "grad_norm": 1.7939306727499553,
+      "learning_rate": 2e-05,
+      "loss": 0.4883720397949219,
+      "step": 5280
+    },
+    {
+      "epoch": 0.6797738370598818,
+      "grad_norm": 0.7325073173759571,
+      "learning_rate": 2e-05,
+      "loss": 0.4730748176574707,
+      "step": 5290
+    },
+    {
+      "epoch": 0.6810588537650989,
+      "grad_norm": 0.7941493316817206,
+      "learning_rate": 2e-05,
+      "loss": 0.497575044631958,
+      "step": 5300
+    },
+    {
+      "epoch": 0.6823438704703161,
+      "grad_norm": 0.8832942218641286,
+      "learning_rate": 2e-05,
+      "loss": 0.5093006134033203,
+      "step": 5310
+    },
+    {
+      "epoch": 0.6836288871755333,
+      "grad_norm": 0.7788091873797266,
+      "learning_rate": 2e-05,
+      "loss": 0.484604549407959,
+      "step": 5320
+    },
+    {
+      "epoch": 0.6849139038807505,
+      "grad_norm": 0.8173187385616143,
+      "learning_rate": 2e-05,
+      "loss": 0.4690967082977295,
+      "step": 5330
+    },
+    {
+      "epoch": 0.6861989205859677,
+      "grad_norm": 0.8290448212855159,
+      "learning_rate": 2e-05,
+      "loss": 0.4745177745819092,
+      "step": 5340
+    },
+    {
+      "epoch": 0.6874839372911847,
+      "grad_norm": 0.832560759615804,
+      "learning_rate": 1.999194520388485e-05,
+      "loss": 0.47368464469909666,
+      "step": 5350
+    },
+    {
+      "epoch": 0.6887689539964019,
+      "grad_norm": 0.7876007238761321,
+      "learning_rate": 1.9959247303157763e-05,
+      "loss": 0.4730240821838379,
+      "step": 5360
+    },
+    {
+      "epoch": 0.6900539707016191,
+      "grad_norm": 0.7482307497167341,
+      "learning_rate": 1.990149424459817e-05,
+      "loss": 0.47472343444824217,
+      "step": 5370
+    },
+    {
+      "epoch": 0.6913389874068363,
+      "grad_norm": 0.7880013542271783,
+      "learning_rate": 1.9818847538648688e-05,
+      "loss": 0.48226380348205566,
+      "step": 5380
+    },
+    {
+      "epoch": 0.6926240041120535,
+      "grad_norm": 0.8178772739057695,
+      "learning_rate": 1.971153831256721e-05,
+      "loss": 0.47275633811950685,
+      "step": 5390
+    },
+    {
+      "epoch": 0.6939090208172706,
+      "grad_norm": 0.9216796465463473,
+      "learning_rate": 1.9579866664063435e-05,
+      "loss": 0.48569550514221194,
+      "step": 5400
+    },
+    {
+      "epoch": 0.6951940375224878,
+      "grad_norm": 0.9715322819571414,
+      "learning_rate": 1.9424200822054782e-05,
+      "loss": 0.46499032974243165,
+      "step": 5410
+    },
+    {
+      "epoch": 0.696479054227705,
+      "grad_norm": 0.9235680013465394,
+      "learning_rate": 1.9244976116888626e-05,
+      "loss": 0.5130784034729003,
+      "step": 5420
+    },
+    {
+      "epoch": 0.6977640709329221,
+      "grad_norm": 0.914498967105097,
+      "learning_rate": 1.904269376291071e-05,
+      "loss": 0.46350908279418945,
+      "step": 5430
+    },
+    {
+      "epoch": 0.6990490876381393,
+      "grad_norm": 0.9491731332317279,
+      "learning_rate": 1.88179194567844e-05,
+      "loss": 0.47063326835632324,
+      "step": 5440
+    },
+    {
+      "epoch": 0.7003341043433564,
+      "grad_norm": 0.8377048974893545,
+      "learning_rate": 1.8571281795480632e-05,
+      "loss": 0.49516901969909666,
+      "step": 5450
+    },
+    {
+      "epoch": 0.7016191210485736,
+      "grad_norm": 0.8325698723817421,
+      "learning_rate": 1.8303470518362808e-05,
+      "loss": 0.47369985580444335,
+      "step": 5460
+    },
+    {
+      "epoch": 0.7029041377537908,
+      "grad_norm": 0.8305400502625536,
+      "learning_rate": 1.801523457828271e-05,
+      "loss": 0.46996564865112306,
+      "step": 5470
+    },
+    {
+      "epoch": 0.704189154459008,
+      "grad_norm": 0.8022782721402015,
+      "learning_rate": 1.7707380047081796e-05,
+      "loss": 0.458741569519043,
+      "step": 5480
+    },
+    {
+      "epoch": 0.7054741711642252,
+      "grad_norm": 0.8908247522028909,
+      "learning_rate": 1.7380767861355252e-05,
+      "loss": 0.4691608428955078,
+      "step": 5490
+    },
+    {
+      "epoch": 0.7067591878694423,
+      "grad_norm": 0.7765155575812603,
+      "learning_rate": 1.703631141478299e-05,
+      "loss": 0.4600387096405029,
+      "step": 5500
+    },
+    {
+      "epoch": 0.7080442045746594,
+      "grad_norm": 0.8574289816861882,
+      "learning_rate": 1.6674974003760705e-05,
+      "loss": 0.48156914710998533,
+      "step": 5510
+    },
+    {
+      "epoch": 0.7093292212798766,
+      "grad_norm": 0.9363652227899883,
+      "learning_rate": 1.6297766133474556e-05,
+      "loss": 0.46467018127441406,
+      "step": 5520
+    },
+    {
+      "epoch": 0.7106142379850938,
+      "grad_norm": 0.8619038707019582,
+      "learning_rate": 1.5905742691953186e-05,
+      "loss": 0.488844633102417,
+      "step": 5530
+    },
+    {
+      "epoch": 0.711899254690311,
+      "grad_norm": 1.00398269827733,
+      "learning_rate": 1.55e-05,
+      "loss": 0.47431125640869143,
+      "step": 5540
+    },
+    {
+      "epoch": 0.7131842713955281,
+      "grad_norm": 1.137933286481776,
+      "learning_rate": 1.5081672745255862e-05,
+      "loss": 0.46503682136535646,
+      "step": 5550
+    },
+    {
+      "epoch": 0.7144692881007453,
+      "grad_norm": 1.067412570091964,
+      "learning_rate": 1.4651930808966206e-05,
+      "loss": 0.4685947418212891,
+      "step": 5560
+    },
+    {
+      "epoch": 0.7157543048059625,
+      "grad_norm": 0.8360574805940808,
+      "learning_rate": 1.4211975994326848e-05,
+      "loss": 0.46284799575805663,
+      "step": 5570
+    },
+    {
+      "epoch": 0.7170393215111797,
+      "grad_norm": 0.8057204815080912,
+      "learning_rate": 1.376303866555779e-05,
+      "loss": 0.47574658393859864,
+      "step": 5580
+    },
+    {
+      "epoch": 0.7183243382163969,
+      "grad_norm": 0.7416623085949049,
+      "learning_rate": 1.3306374307104164e-05,
+      "loss": 0.4582655906677246,
+      "step": 5590
+    },
+    {
+      "epoch": 0.7196093549216139,
+      "grad_norm": 0.955096028720366,
+      "learning_rate": 1.2843260012586718e-05,
+      "loss": 0.46666727066040037,
+      "step": 5600
+    },
+    {
+      "epoch": 0.7208943716268311,
+      "grad_norm": 0.9416798533760223,
+      "learning_rate": 1.2374990913320726e-05,
+      "loss": 0.4426871299743652,
+      "step": 5610
+    },
+    {
+      "epoch": 0.7221793883320483,
+      "grad_norm": 0.7899505154791333,
+      "learning_rate": 1.1902876556391247e-05,
+      "loss": 0.43970518112182616,
+      "step": 5620
+    },
+    {
+      "epoch": 0.7234644050372655,
+      "grad_norm": 0.7937297096974494,
+      "learning_rate": 1.1428237242413685e-05,
+      "loss": 0.451738166809082,
+      "step": 5630
+    },
+    {
+      "epoch": 0.7247494217424827,
+      "grad_norm": 0.7630790967410468,
+      "learning_rate": 1.095240033322132e-05,
+      "loss": 0.46645317077636717,
+      "step": 5640
+    },
+    {
+      "epoch": 0.7260344384476998,
+      "grad_norm": 0.7512008048664177,
+      "learning_rate": 1.047669653980572e-05,
+      "loss": 0.46428213119506834,
+      "step": 5650
+    },
+    {
+      "epoch": 0.727319455152917,
+      "grad_norm": 0.7324752011179224,
+      "learning_rate": 1.0002456200890902e-05,
+      "loss": 0.4806793212890625,
+      "step": 5660
+    },
+    {
+      "epoch": 0.7286044718581342,
+      "grad_norm": 0.7999691339624928,
+      "learning_rate": 9.531005562548668e-06,
+      "loss": 0.4535094738006592,
+      "step": 5670
+    },
+    {
+      "epoch": 0.7298894885633513,
+      "grad_norm": 0.7223539057040752,
+      "learning_rate": 9.06366306925925e-06,
+      "loss": 0.46710891723632814,
+      "step": 5680
+    },
+    {
+      "epoch": 0.7311745052685685,
+      "grad_norm": 0.8272048949084428,
+      "learning_rate": 8.601735676789686e-06,
+      "loss": 0.4569704055786133,
+      "step": 5690
+    },
+    {
+      "epoch": 0.7324595219737856,
+      "grad_norm": 0.7876806632700706,
+      "learning_rate": 8.14651519720118e-06,
+      "loss": 0.44039597511291506,
+      "step": 5700
+    },
+    {
+      "epoch": 0.7337445386790028,
+      "grad_norm": 0.7749183278180823,
+      "learning_rate": 7.699274686206866e-06,
+      "loss": 0.4593049049377441,
+      "step": 5710
+    },
+    {
+      "epoch": 0.73502955538422,
+      "grad_norm": 0.790088864306392,
+      "learning_rate": 7.261264882983024e-06,
+      "loss": 0.43912839889526367,
+      "step": 5720
+    },
+    {
+      "epoch": 0.7363145720894372,
+      "grad_norm": 0.7689663285525302,
+      "learning_rate": 6.8337107123900386e-06,
+      "loss": 0.45433816909790037,
+      "step": 5730
+    },
+    {
+      "epoch": 0.7375995887946544,
+      "grad_norm": 0.7759262873445865,
+      "learning_rate": 6.417807859384881e-06,
+      "loss": 0.44428653717041017,
+      "step": 5740
+    },
+    {
+      "epoch": 0.7388846054998714,
+      "grad_norm": 1.544241986650257,
+      "learning_rate": 6.014719425205009e-06,
+      "loss": 0.4489415645599365,
+      "step": 5750
+    },
+    {
+      "epoch": 0.7401696222050886,
+      "grad_norm": 0.7449617662998115,
+      "learning_rate": 5.6255726746749275e-06,
+      "loss": 0.4385959625244141,
+      "step": 5760
+    },
+    {
+      "epoch": 0.7414546389103058,
+      "grad_norm": 0.8228085870270948,
+      "learning_rate": 5.251455883731728e-06,
+      "loss": 0.4706584930419922,
+      "step": 5770
+    },
+    {
+      "epoch": 0.742739655615523,
+      "grad_norm": 0.9544380198658783,
+      "learning_rate": 4.893415295985812e-06,
+      "loss": 0.4448546409606934,
+      "step": 5780
+    },
+    {
+      "epoch": 0.7440246723207402,
+      "grad_norm": 0.735149389522285,
+      "learning_rate": 4.552452196827885e-06,
+      "loss": 0.4394658088684082,
+      "step": 5790
+    },
+    {
+      "epoch": 0.7453096890259573,
+      "grad_norm": 0.8424205588790463,
+      "learning_rate": 4.229520113264785e-06,
+      "loss": 0.45569396018981934,
+      "step": 5800
+    },
+    {
+      "epoch": 0.7465947057311745,
+      "grad_norm": 0.7200072673511734,
+      "learning_rate": 3.925522147314915e-06,
+      "loss": 0.4093926429748535,
+      "step": 5810
+    },
+    {
+      "epoch": 0.7478797224363917,
+      "grad_norm": 0.747309966780883,
+      "learning_rate": 3.641308450420675e-06,
+      "loss": 0.4588127613067627,
+      "step": 5820
+    },
+    {
+      "epoch": 0.7491647391416089,
+      "grad_norm": 0.7928790729415274,
+      "learning_rate": 3.37767384594087e-06,
+      "loss": 0.4229104518890381,
+      "step": 5830
+    },
+    {
+      "epoch": 0.750449755846826,
+      "grad_norm": 0.8047685219156716,
+      "learning_rate": 3.1353556063719357e-06,
+      "loss": 0.4265634536743164,
+      "step": 5840
+    },
+    {
+      "epoch": 0.7517347725520431,
+      "grad_norm": 3.5333167070451554,
+      "learning_rate": 2.9150313915141474e-06,
+      "loss": 0.44034571647644044,
+      "step": 5850
+    },
+    {
+      "epoch": 0.7530197892572603,
+      "grad_norm": 0.7725926921127012,
+      "learning_rate": 2.717317353348891e-06,
+      "loss": 0.4585292339324951,
+      "step": 5860
+    },
+    {
+      "epoch": 0.7543048059624775,
+      "grad_norm": 0.7702696236127015,
+      "learning_rate": 2.5427664129268253e-06,
+      "loss": 0.4487299919128418,
+      "step": 5870
+    },
+    {
+      "epoch": 0.7555898226676947,
+      "grad_norm": 0.7257460662234754,
+      "learning_rate": 2.391866714085717e-06,
+      "loss": 0.4354090690612793,
+      "step": 5880
+    },
+    {
+      "epoch": 0.7568748393729119,
+      "grad_norm": 1.3317960559695148,
+      "learning_rate": 2.2650402583222753e-06,
+      "loss": 0.4356398582458496,
+      "step": 5890
+    },
+    {
+      "epoch": 0.758159856078129,
+      "grad_norm": 0.7394571684662601,
+      "learning_rate": 2.1626417246356398e-06,
+      "loss": 0.427561092376709,
+      "step": 5900
+    },
+    {
+      "epoch": 0.7594448727833462,
+      "grad_norm": 0.7507986575998497,
+      "learning_rate": 2.084957477642894e-06,
+      "loss": 0.42830848693847656,
+      "step": 5910
+    },
+    {
+      "epoch": 0.7607298894885633,
+      "grad_norm": 0.8570892232604302,
+      "learning_rate": 2.0322047667405147e-06,
+      "loss": 0.4541787147521973,
+      "step": 5920
+    },
+    {
+      "epoch": 0.7620149061937805,
+      "grad_norm": 0.7940759625197243,
+      "learning_rate": 2.0045311185513342e-06,
+      "loss": 0.44152402877807617,
+      "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
+}