diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,32642 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 46554,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 2.1480431327061047e-05,
+      "grad_norm": 26.65201832627396,
+      "learning_rate": 4.291845493562232e-08,
+      "loss": 2.0281,
+      "step": 1
+    },
+    {
+      "epoch": 0.0002148043132706105,
+      "grad_norm": 24.545838847916603,
+      "learning_rate": 4.291845493562232e-07,
+      "loss": 1.9707,
+      "step": 10
+    },
+    {
+      "epoch": 0.000429608626541221,
+      "grad_norm": 6.925669336747074,
+      "learning_rate": 8.583690987124464e-07,
+      "loss": 1.7361,
+      "step": 20
+    },
+    {
+      "epoch": 0.0006444129398118315,
+      "grad_norm": 3.563926800405336,
+      "learning_rate": 1.2875536480686696e-06,
+      "loss": 1.5248,
+      "step": 30
+    },
+    {
+      "epoch": 0.000859217253082442,
+      "grad_norm": 2.7917255910937255,
+      "learning_rate": 1.7167381974248929e-06,
+      "loss": 1.3065,
+      "step": 40
+    },
+    {
+      "epoch": 0.0010740215663530524,
+      "grad_norm": 1.3127546793240747,
+      "learning_rate": 2.145922746781116e-06,
+      "loss": 1.2055,
+      "step": 50
+    },
+    {
+      "epoch": 0.001288825879623663,
+      "grad_norm": 1.1166152222590124,
+      "learning_rate": 2.575107296137339e-06,
+      "loss": 1.1276,
+      "step": 60
+    },
+    {
+      "epoch": 0.0015036301928942734,
+      "grad_norm": 1.1067088125847406,
+      "learning_rate": 3.004291845493563e-06,
+      "loss": 1.111,
+      "step": 70
+    },
+    {
+      "epoch": 0.001718434506164884,
+      "grad_norm": 1.0213495251064038,
+      "learning_rate": 3.4334763948497858e-06,
+      "loss": 1.0633,
+      "step": 80
+    },
+    {
+      "epoch": 0.0019332388194354942,
+      "grad_norm": 1.4698974676460892,
+      "learning_rate": 3.862660944206009e-06,
+      "loss": 1.0363,
+      "step": 90
+    },
+    {
+      "epoch": 0.002148043132706105,
+      "grad_norm": 1.1069539300208826,
+      "learning_rate": 4.291845493562232e-06,
+      "loss": 1.0595,
+      "step": 100
+    },
+    {
+      "epoch": 0.002362847445976715,
+      "grad_norm": 1.0428861367575017,
+      "learning_rate": 4.721030042918455e-06,
+      "loss": 1.0544,
+      "step": 110
+    },
+    {
+      "epoch": 0.002577651759247326,
+      "grad_norm": 1.0155126146726958,
+      "learning_rate": 5.150214592274678e-06,
+      "loss": 1.0247,
+      "step": 120
+    },
+    {
+      "epoch": 0.002792456072517936,
+      "grad_norm": 1.0457307269517822,
+      "learning_rate": 5.579399141630901e-06,
+      "loss": 1.0231,
+      "step": 130
+    },
+    {
+      "epoch": 0.003007260385788547,
+      "grad_norm": 0.9747222232804185,
+      "learning_rate": 6.008583690987126e-06,
+      "loss": 1.0112,
+      "step": 140
+    },
+    {
+      "epoch": 0.003222064699059157,
+      "grad_norm": 1.0485056497719716,
+      "learning_rate": 6.437768240343349e-06,
+      "loss": 0.9848,
+      "step": 150
+    },
+    {
+      "epoch": 0.003436869012329768,
+      "grad_norm": 1.015067491069871,
+      "learning_rate": 6.8669527896995715e-06,
+      "loss": 1.0101,
+      "step": 160
+    },
+    {
+      "epoch": 0.003651673325600378,
+      "grad_norm": 0.9810404079330955,
+      "learning_rate": 7.296137339055794e-06,
+      "loss": 0.981,
+      "step": 170
+    },
+    {
+      "epoch": 0.0038664776388709883,
+      "grad_norm": 1.018895365287239,
+      "learning_rate": 7.725321888412017e-06,
+      "loss": 0.9754,
+      "step": 180
+    },
+    {
+      "epoch": 0.004081281952141599,
+      "grad_norm": 1.079857560560485,
+      "learning_rate": 8.154506437768241e-06,
+      "loss": 0.9996,
+      "step": 190
+    },
+    {
+      "epoch": 0.00429608626541221,
+      "grad_norm": 0.9967859001475669,
+      "learning_rate": 8.583690987124465e-06,
+      "loss": 0.9952,
+      "step": 200
+    },
+    {
+      "epoch": 0.00451089057868282,
+      "grad_norm": 1.115201831741858,
+      "learning_rate": 9.012875536480687e-06,
+      "loss": 0.9742,
+      "step": 210
+    },
+    {
+      "epoch": 0.00472569489195343,
+      "grad_norm": 0.9523509165576094,
+      "learning_rate": 9.44206008583691e-06,
+      "loss": 0.9831,
+      "step": 220
+    },
+    {
+      "epoch": 0.004940499205224041,
+      "grad_norm": 1.005436826590897,
+      "learning_rate": 9.871244635193133e-06,
+      "loss": 0.9773,
+      "step": 230
+    },
+    {
+      "epoch": 0.005155303518494652,
+      "grad_norm": 0.9901616620650135,
+      "learning_rate": 1.0300429184549356e-05,
+      "loss": 0.9781,
+      "step": 240
+    },
+    {
+      "epoch": 0.0053701078317652615,
+      "grad_norm": 1.0638485393442876,
+      "learning_rate": 1.072961373390558e-05,
+      "loss": 0.96,
+      "step": 250
+    },
+    {
+      "epoch": 0.005584912145035872,
+      "grad_norm": 0.9964178960246284,
+      "learning_rate": 1.1158798283261802e-05,
+      "loss": 0.9742,
+      "step": 260
+    },
+    {
+      "epoch": 0.005799716458306483,
+      "grad_norm": 1.0370493768439413,
+      "learning_rate": 1.1587982832618028e-05,
+      "loss": 0.9602,
+      "step": 270
+    },
+    {
+      "epoch": 0.006014520771577094,
+      "grad_norm": 0.9796931302576042,
+      "learning_rate": 1.2017167381974251e-05,
+      "loss": 0.9632,
+      "step": 280
+    },
+    {
+      "epoch": 0.0062293250848477035,
+      "grad_norm": 1.031038089029062,
+      "learning_rate": 1.2446351931330473e-05,
+      "loss": 0.9593,
+      "step": 290
+    },
+    {
+      "epoch": 0.006444129398118314,
+      "grad_norm": 0.9515929962341028,
+      "learning_rate": 1.2875536480686697e-05,
+      "loss": 0.949,
+      "step": 300
+    },
+    {
+      "epoch": 0.006658933711388925,
+      "grad_norm": 0.9520559803996252,
+      "learning_rate": 1.330472103004292e-05,
+      "loss": 0.9457,
+      "step": 310
+    },
+    {
+      "epoch": 0.006873738024659536,
+      "grad_norm": 1.3890037359682255,
+      "learning_rate": 1.3733905579399143e-05,
+      "loss": 0.9584,
+      "step": 320
+    },
+    {
+      "epoch": 0.007088542337930145,
+      "grad_norm": 0.9169319166076477,
+      "learning_rate": 1.4163090128755365e-05,
+      "loss": 0.9464,
+      "step": 330
+    },
+    {
+      "epoch": 0.007303346651200756,
+      "grad_norm": 0.9772519097125956,
+      "learning_rate": 1.4592274678111589e-05,
+      "loss": 0.9614,
+      "step": 340
+    },
+    {
+      "epoch": 0.007518150964471367,
+      "grad_norm": 1.1706529511376937,
+      "learning_rate": 1.5021459227467811e-05,
+      "loss": 0.9553,
+      "step": 350
+    },
+    {
+      "epoch": 0.007732955277741977,
+      "grad_norm": 0.9436703190905599,
+      "learning_rate": 1.5450643776824035e-05,
+      "loss": 0.9417,
+      "step": 360
+    },
+    {
+      "epoch": 0.007947759591012588,
+      "grad_norm": 0.8941917268650862,
+      "learning_rate": 1.587982832618026e-05,
+      "loss": 0.9382,
+      "step": 370
+    },
+    {
+      "epoch": 0.008162563904283198,
+      "grad_norm": 0.9682494688789302,
+      "learning_rate": 1.6309012875536482e-05,
+      "loss": 0.9636,
+      "step": 380
+    },
+    {
+      "epoch": 0.008377368217553808,
+      "grad_norm": 0.968118858319082,
+      "learning_rate": 1.6738197424892706e-05,
+      "loss": 0.9406,
+      "step": 390
+    },
+    {
+      "epoch": 0.00859217253082442,
+      "grad_norm": 0.9304992691801565,
+      "learning_rate": 1.716738197424893e-05,
+      "loss": 0.9467,
+      "step": 400
+    },
+    {
+      "epoch": 0.00880697684409503,
+      "grad_norm": 0.8846995721706429,
+      "learning_rate": 1.7596566523605153e-05,
+      "loss": 0.9552,
+      "step": 410
+    },
+    {
+      "epoch": 0.00902178115736564,
+      "grad_norm": 1.0342299734011187,
+      "learning_rate": 1.8025751072961374e-05,
+      "loss": 0.9485,
+      "step": 420
+    },
+    {
+      "epoch": 0.00923658547063625,
+      "grad_norm": 1.051909712027358,
+      "learning_rate": 1.8454935622317597e-05,
+      "loss": 0.9635,
+      "step": 430
+    },
+    {
+      "epoch": 0.00945138978390686,
+      "grad_norm": 1.0942808627738883,
+      "learning_rate": 1.888412017167382e-05,
+      "loss": 0.9421,
+      "step": 440
+    },
+    {
+      "epoch": 0.009666194097177472,
+      "grad_norm": 0.8646650135642147,
+      "learning_rate": 1.9313304721030045e-05,
+      "loss": 0.977,
+      "step": 450
+    },
+    {
+      "epoch": 0.009880998410448082,
+      "grad_norm": 0.9215653231258697,
+      "learning_rate": 1.9742489270386265e-05,
+      "loss": 0.9571,
+      "step": 460
+    },
+    {
+      "epoch": 0.010095802723718692,
+      "grad_norm": 0.9199665564587306,
+      "learning_rate": 1.999999962828165e-05,
+      "loss": 0.9356,
+      "step": 470
+    },
+    {
+      "epoch": 0.010310607036989303,
+      "grad_norm": 0.9425113636574227,
+      "learning_rate": 1.9999995446450485e-05,
+      "loss": 0.9624,
+      "step": 480
+    },
+    {
+      "epoch": 0.010525411350259913,
+      "grad_norm": 0.9337828020549753,
+      "learning_rate": 1.999998661814216e-05,
+      "loss": 0.9468,
+      "step": 490
+    },
+    {
+      "epoch": 0.010740215663530523,
+      "grad_norm": 0.927989701189411,
+      "learning_rate": 1.9999973143360788e-05,
+      "loss": 0.9396,
+      "step": 500
+    },
+    {
+      "epoch": 0.010955019976801135,
+      "grad_norm": 0.9324290313999429,
+      "learning_rate": 1.999995502211262e-05,
+      "loss": 0.9563,
+      "step": 510
+    },
+    {
+      "epoch": 0.011169824290071744,
+      "grad_norm": 0.9434102759361815,
+      "learning_rate": 1.9999932254406077e-05,
+      "loss": 0.9609,
+      "step": 520
+    },
+    {
+      "epoch": 0.011384628603342354,
+      "grad_norm": 0.8724473565087025,
+      "learning_rate": 1.9999904840251743e-05,
+      "loss": 0.9427,
+      "step": 530
+    },
+    {
+      "epoch": 0.011599432916612966,
+      "grad_norm": 0.8942612386723021,
+      "learning_rate": 1.9999872779662347e-05,
+      "loss": 0.9245,
+      "step": 540
+    },
+    {
+      "epoch": 0.011814237229883576,
+      "grad_norm": 0.9928300421934061,
+      "learning_rate": 1.9999836072652794e-05,
+      "loss": 0.932,
+      "step": 550
+    },
+    {
+      "epoch": 0.012029041543154187,
+      "grad_norm": 0.8947389195896635,
+      "learning_rate": 1.999979471924014e-05,
+      "loss": 0.945,
+      "step": 560
+    },
+    {
+      "epoch": 0.012243845856424797,
+      "grad_norm": 0.8728388436746601,
+      "learning_rate": 1.9999748719443594e-05,
+      "loss": 0.9508,
+      "step": 570
+    },
+    {
+      "epoch": 0.012458650169695407,
+      "grad_norm": 1.0931390263840937,
+      "learning_rate": 1.9999698073284534e-05,
+      "loss": 0.9334,
+      "step": 580
+    },
+    {
+      "epoch": 0.012673454482966019,
+      "grad_norm": 0.8867248563676206,
+      "learning_rate": 1.9999642780786486e-05,
+      "loss": 0.9352,
+      "step": 590
+    },
+    {
+      "epoch": 0.012888258796236628,
+      "grad_norm": 0.8378993692657936,
+      "learning_rate": 1.9999582841975152e-05,
+      "loss": 0.9211,
+      "step": 600
+    },
+    {
+      "epoch": 0.013103063109507238,
+      "grad_norm": 0.944765913779485,
+      "learning_rate": 1.9999518256878377e-05,
+      "loss": 0.9455,
+      "step": 610
+    },
+    {
+      "epoch": 0.01331786742277785,
+      "grad_norm": 1.110589470050176,
+      "learning_rate": 1.999944902552617e-05,
+      "loss": 0.9312,
+      "step": 620
+    },
+    {
+      "epoch": 0.01353267173604846,
+      "grad_norm": 0.93740820263563,
+      "learning_rate": 1.9999375147950698e-05,
+      "loss": 0.9476,
+      "step": 630
+    },
+    {
+      "epoch": 0.013747476049319071,
+      "grad_norm": 0.8693919640628258,
+      "learning_rate": 1.999929662418629e-05,
+      "loss": 0.9397,
+      "step": 640
+    },
+    {
+      "epoch": 0.013962280362589681,
+      "grad_norm": 0.821330072078906,
+      "learning_rate": 1.9999213454269433e-05,
+      "loss": 0.9387,
+      "step": 650
+    },
+    {
+      "epoch": 0.01417708467586029,
+      "grad_norm": 0.8706987017404653,
+      "learning_rate": 1.999912563823877e-05,
+      "loss": 0.9293,
+      "step": 660
+    },
+    {
+      "epoch": 0.014391888989130902,
+      "grad_norm": 0.8517170283338585,
+      "learning_rate": 1.9999033176135104e-05,
+      "loss": 0.9359,
+      "step": 670
+    },
+    {
+      "epoch": 0.014606693302401512,
+      "grad_norm": 0.8579178853579097,
+      "learning_rate": 1.99989360680014e-05,
+      "loss": 0.9063,
+      "step": 680
+    },
+    {
+      "epoch": 0.014821497615672122,
+      "grad_norm": 0.8866957779531295,
+      "learning_rate": 1.999883431388278e-05,
+      "loss": 0.9354,
+      "step": 690
+    },
+    {
+      "epoch": 0.015036301928942734,
+      "grad_norm": 0.8870565536110462,
+      "learning_rate": 1.9998727913826516e-05,
+      "loss": 0.9366,
+      "step": 700
+    },
+    {
+      "epoch": 0.015251106242213344,
+      "grad_norm": 0.8987989639943288,
+      "learning_rate": 1.9998616867882052e-05,
+      "loss": 0.9369,
+      "step": 710
+    },
+    {
+      "epoch": 0.015465910555483953,
+      "grad_norm": 0.8110762813161089,
+      "learning_rate": 1.9998501176100987e-05,
+      "loss": 0.9267,
+      "step": 720
+    },
+    {
+      "epoch": 0.015680714868754563,
+      "grad_norm": 0.8393357992971544,
+      "learning_rate": 1.9998380838537075e-05,
+      "loss": 0.9275,
+      "step": 730
+    },
+    {
+      "epoch": 0.015895519182025176,
+      "grad_norm": 0.9524547482458253,
+      "learning_rate": 1.999825585524623e-05,
+      "loss": 0.9252,
+      "step": 740
+    },
+    {
+      "epoch": 0.016110323495295786,
+      "grad_norm": 0.8155389716366475,
+      "learning_rate": 1.9998126226286528e-05,
+      "loss": 0.9147,
+      "step": 750
+    },
+    {
+      "epoch": 0.016325127808566396,
+      "grad_norm": 0.8231072792576088,
+      "learning_rate": 1.9997991951718196e-05,
+      "loss": 0.9276,
+      "step": 760
+    },
+    {
+      "epoch": 0.016539932121837006,
+      "grad_norm": 0.8125297140263947,
+      "learning_rate": 1.9997853031603628e-05,
+      "loss": 0.9345,
+      "step": 770
+    },
+    {
+      "epoch": 0.016754736435107616,
+      "grad_norm": 0.9347609633412168,
+      "learning_rate": 1.9997709466007374e-05,
+      "loss": 0.9139,
+      "step": 780
+    },
+    {
+      "epoch": 0.01696954074837823,
+      "grad_norm": 0.8167131053353975,
+      "learning_rate": 1.9997561254996138e-05,
+      "loss": 0.9258,
+      "step": 790
+    },
+    {
+      "epoch": 0.01718434506164884,
+      "grad_norm": 0.8141997003853896,
+      "learning_rate": 1.9997408398638785e-05,
+      "loss": 0.9169,
+      "step": 800
+    },
+    {
+      "epoch": 0.01739914937491945,
+      "grad_norm": 0.8020228216788807,
+      "learning_rate": 1.9997250897006345e-05,
+      "loss": 0.9114,
+      "step": 810
+    },
+    {
+      "epoch": 0.01761395368819006,
+      "grad_norm": 0.7977580905585953,
+      "learning_rate": 1.9997088750171995e-05,
+      "loss": 0.9243,
+      "step": 820
+    },
+    {
+      "epoch": 0.01782875800146067,
+      "grad_norm": 0.8518531375273793,
+      "learning_rate": 1.999692195821108e-05,
+      "loss": 0.9191,
+      "step": 830
+    },
+    {
+      "epoch": 0.01804356231473128,
+      "grad_norm": 0.8006041023630638,
+      "learning_rate": 1.99967505212011e-05,
+      "loss": 0.9152,
+      "step": 840
+    },
+    {
+      "epoch": 0.01825836662800189,
+      "grad_norm": 0.793159791876333,
+      "learning_rate": 1.9996574439221708e-05,
+      "loss": 0.9235,
+      "step": 850
+    },
+    {
+      "epoch": 0.0184731709412725,
+      "grad_norm": 0.798911293614745,
+      "learning_rate": 1.9996393712354724e-05,
+      "loss": 0.907,
+      "step": 860
+    },
+    {
+      "epoch": 0.01868797525454311,
+      "grad_norm": 0.7876487421020302,
+      "learning_rate": 1.9996208340684124e-05,
+      "loss": 0.9073,
+      "step": 870
+    },
+    {
+      "epoch": 0.01890277956781372,
+      "grad_norm": 0.7656899260637665,
+      "learning_rate": 1.9996018324296036e-05,
+      "loss": 0.9191,
+      "step": 880
+    },
+    {
+      "epoch": 0.01911758388108433,
+      "grad_norm": 0.8200150022936012,
+      "learning_rate": 1.9995823663278753e-05,
+      "loss": 0.9144,
+      "step": 890
+    },
+    {
+      "epoch": 0.019332388194354944,
+      "grad_norm": 0.8730487914144835,
+      "learning_rate": 1.9995624357722726e-05,
+      "loss": 0.9239,
+      "step": 900
+    },
+    {
+      "epoch": 0.019547192507625554,
+      "grad_norm": 0.8281237932100032,
+      "learning_rate": 1.9995420407720556e-05,
+      "loss": 0.9205,
+      "step": 910
+    },
+    {
+      "epoch": 0.019761996820896164,
+      "grad_norm": 0.8232205349274369,
+      "learning_rate": 1.9995211813367014e-05,
+      "loss": 0.919,
+      "step": 920
+    },
+    {
+      "epoch": 0.019976801134166774,
+      "grad_norm": 0.8147366319849692,
+      "learning_rate": 1.9994998574759023e-05,
+      "loss": 0.907,
+      "step": 930
+    },
+    {
+      "epoch": 0.020191605447437384,
+      "grad_norm": 0.7976853731174323,
+      "learning_rate": 1.9994780691995658e-05,
+      "loss": 0.9112,
+      "step": 940
+    },
+    {
+      "epoch": 0.020406409760707993,
+      "grad_norm": 0.8274499776650804,
+      "learning_rate": 1.9994558165178165e-05,
+      "loss": 0.9228,
+      "step": 950
+    },
+    {
+      "epoch": 0.020621214073978607,
+      "grad_norm": 0.824749615058281,
+      "learning_rate": 1.9994330994409932e-05,
+      "loss": 0.9331,
+      "step": 960
+    },
+    {
+      "epoch": 0.020836018387249217,
+      "grad_norm": 0.7904631449154701,
+      "learning_rate": 1.999409917979652e-05,
+      "loss": 0.91,
+      "step": 970
+    },
+    {
+      "epoch": 0.021050822700519826,
+      "grad_norm": 0.8284138971940564,
+      "learning_rate": 1.999386272144564e-05,
+      "loss": 0.9163,
+      "step": 980
+    },
+    {
+      "epoch": 0.021265627013790436,
+      "grad_norm": 0.8529599065323894,
+      "learning_rate": 1.9993621619467163e-05,
+      "loss": 0.9098,
+      "step": 990
+    },
+    {
+      "epoch": 0.021480431327061046,
+      "grad_norm": 0.8034510770514283,
+      "learning_rate": 1.999337587397311e-05,
+      "loss": 0.9221,
+      "step": 1000
+    },
+    {
+      "epoch": 0.02169523564033166,
+      "grad_norm": 0.7866973908662558,
+      "learning_rate": 1.9993125485077675e-05,
+      "loss": 0.9284,
+      "step": 1010
+    },
+    {
+      "epoch": 0.02191003995360227,
+      "grad_norm": 0.8772012374367091,
+      "learning_rate": 1.9992870452897195e-05,
+      "loss": 0.9145,
+      "step": 1020
+    },
+    {
+      "epoch": 0.02212484426687288,
+      "grad_norm": 0.7704037718521081,
+      "learning_rate": 1.9992610777550174e-05,
+      "loss": 0.9022,
+      "step": 1030
+    },
+    {
+      "epoch": 0.02233964858014349,
+      "grad_norm": 0.8139905784316902,
+      "learning_rate": 1.999234645915727e-05,
+      "loss": 0.899,
+      "step": 1040
+    },
+    {
+      "epoch": 0.0225544528934141,
+      "grad_norm": 0.8035022366722004,
+      "learning_rate": 1.999207749784129e-05,
+      "loss": 0.9119,
+      "step": 1050
+    },
+    {
+      "epoch": 0.02276925720668471,
+      "grad_norm": 0.8018634228953854,
+      "learning_rate": 1.9991803893727213e-05,
+      "loss": 0.913,
+      "step": 1060
+    },
+    {
+      "epoch": 0.022984061519955322,
+      "grad_norm": 0.8762806370784909,
+      "learning_rate": 1.9991525646942167e-05,
+      "loss": 0.8927,
+      "step": 1070
+    },
+    {
+      "epoch": 0.023198865833225932,
+      "grad_norm": 0.8434215687788776,
+      "learning_rate": 1.999124275761544e-05,
+      "loss": 0.9035,
+      "step": 1080
+    },
+    {
+      "epoch": 0.02341367014649654,
+      "grad_norm": 0.8323692540509068,
+      "learning_rate": 1.9990955225878475e-05,
+      "loss": 0.9184,
+      "step": 1090
+    },
+    {
+      "epoch": 0.02362847445976715,
+      "grad_norm": 0.8062273577144896,
+      "learning_rate": 1.9990663051864875e-05,
+      "loss": 0.9158,
+      "step": 1100
+    },
+    {
+      "epoch": 0.02384327877303776,
+      "grad_norm": 0.7741118350909906,
+      "learning_rate": 1.9990366235710392e-05,
+      "loss": 0.8994,
+      "step": 1110
+    },
+    {
+      "epoch": 0.024058083086308375,
+      "grad_norm": 0.8409438928193937,
+      "learning_rate": 1.9990064777552947e-05,
+      "loss": 0.9039,
+      "step": 1120
+    },
+    {
+      "epoch": 0.024272887399578984,
+      "grad_norm": 0.8040510324600977,
+      "learning_rate": 1.9989758677532613e-05,
+      "loss": 0.9153,
+      "step": 1130
+    },
+    {
+      "epoch": 0.024487691712849594,
+      "grad_norm": 0.7860147216474672,
+      "learning_rate": 1.9989447935791616e-05,
+      "loss": 0.9111,
+      "step": 1140
+    },
+    {
+      "epoch": 0.024702496026120204,
+      "grad_norm": 0.7955991962712436,
+      "learning_rate": 1.9989132552474335e-05,
+      "loss": 0.9148,
+      "step": 1150
+    },
+    {
+      "epoch": 0.024917300339390814,
+      "grad_norm": 1.305671067059314,
+      "learning_rate": 1.9988812527727324e-05,
+      "loss": 0.9093,
+      "step": 1160
+    },
+    {
+      "epoch": 0.025132104652661424,
+      "grad_norm": 0.7217672239649101,
+      "learning_rate": 1.9988487861699277e-05,
+      "loss": 0.9105,
+      "step": 1170
+    },
+    {
+      "epoch": 0.025346908965932037,
+      "grad_norm": 6.271674767342428,
+      "learning_rate": 1.9988158554541047e-05,
+      "loss": 0.9047,
+      "step": 1180
+    },
+    {
+      "epoch": 0.025561713279202647,
+      "grad_norm": 0.8349965377102029,
+      "learning_rate": 1.9987824606405647e-05,
+      "loss": 0.914,
+      "step": 1190
+    },
+    {
+      "epoch": 0.025776517592473257,
+      "grad_norm": 0.8221862837050625,
+      "learning_rate": 1.9987486017448245e-05,
+      "loss": 0.9111,
+      "step": 1200
+    },
+    {
+      "epoch": 0.025991321905743867,
+      "grad_norm": 0.7579519216571478,
+      "learning_rate": 1.998714278782617e-05,
+      "loss": 0.9148,
+      "step": 1210
+    },
+    {
+      "epoch": 0.026206126219014476,
+      "grad_norm": 1.220350327147385,
+      "learning_rate": 1.9986794917698894e-05,
+      "loss": 0.9209,
+      "step": 1220
+    },
+    {
+      "epoch": 0.02642093053228509,
+      "grad_norm": 0.765028183579955,
+      "learning_rate": 1.9986442407228064e-05,
+      "loss": 0.908,
+      "step": 1230
+    },
+    {
+      "epoch": 0.0266357348455557,
+      "grad_norm": 0.7816681258072063,
+      "learning_rate": 1.998608525657747e-05,
+      "loss": 0.9077,
+      "step": 1240
+    },
+    {
+      "epoch": 0.02685053915882631,
+      "grad_norm": 0.8249985245317925,
+      "learning_rate": 1.9985723465913058e-05,
+      "loss": 0.9021,
+      "step": 1250
+    },
+    {
+      "epoch": 0.02706534347209692,
+      "grad_norm": 0.7570103217659379,
+      "learning_rate": 1.9985357035402937e-05,
+      "loss": 0.8967,
+      "step": 1260
+    },
+    {
+      "epoch": 0.02728014778536753,
+      "grad_norm": 0.7550746594070511,
+      "learning_rate": 1.998498596521737e-05,
+      "loss": 0.9088,
+      "step": 1270
+    },
+    {
+      "epoch": 0.027494952098638142,
+      "grad_norm": 0.74593111439136,
+      "learning_rate": 1.998461025552876e-05,
+      "loss": 0.914,
+      "step": 1280
+    },
+    {
+      "epoch": 0.027709756411908752,
+      "grad_norm": 0.7538206481587928,
+      "learning_rate": 1.99842299065117e-05,
+      "loss": 0.892,
+      "step": 1290
+    },
+    {
+      "epoch": 0.027924560725179362,
+      "grad_norm": 0.7808275060158067,
+      "learning_rate": 1.9983844918342907e-05,
+      "loss": 0.9034,
+      "step": 1300
+    },
+    {
+      "epoch": 0.028139365038449972,
+      "grad_norm": 0.7539145866225501,
+      "learning_rate": 1.9983455291201267e-05,
+      "loss": 0.8867,
+      "step": 1310
+    },
+    {
+      "epoch": 0.02835416935172058,
+      "grad_norm": 0.8633072463687512,
+      "learning_rate": 1.9983061025267815e-05,
+      "loss": 0.9009,
+      "step": 1320
+    },
+    {
+      "epoch": 0.02856897366499119,
+      "grad_norm": 0.7733521101210006,
+      "learning_rate": 1.9982662120725753e-05,
+      "loss": 0.912,
+      "step": 1330
+    },
+    {
+      "epoch": 0.028783777978261805,
+      "grad_norm": 0.7673039038449831,
+      "learning_rate": 1.998225857776043e-05,
+      "loss": 0.8992,
+      "step": 1340
+    },
+    {
+      "epoch": 0.028998582291532415,
+      "grad_norm": 0.7714006517358715,
+      "learning_rate": 1.9981850396559348e-05,
+      "loss": 0.8889,
+      "step": 1350
+    },
+    {
+      "epoch": 0.029213386604803025,
+      "grad_norm": 0.8178218009457255,
+      "learning_rate": 1.9981437577312167e-05,
+      "loss": 0.9063,
+      "step": 1360
+    },
+    {
+      "epoch": 0.029428190918073634,
+      "grad_norm": 0.7794714859641063,
+      "learning_rate": 1.998102012021071e-05,
+      "loss": 0.9023,
+      "step": 1370
+    },
+    {
+      "epoch": 0.029642995231344244,
+      "grad_norm": 0.8496853423209075,
+      "learning_rate": 1.998059802544894e-05,
+      "loss": 0.8999,
+      "step": 1380
+    },
+    {
+      "epoch": 0.029857799544614858,
+      "grad_norm": 0.7714878599883224,
+      "learning_rate": 1.9980171293222982e-05,
+      "loss": 0.8896,
+      "step": 1390
+    },
+    {
+      "epoch": 0.030072603857885467,
+      "grad_norm": 0.7795267501311487,
+      "learning_rate": 1.9979739923731125e-05,
+      "loss": 0.9123,
+      "step": 1400
+    },
+    {
+      "epoch": 0.030287408171156077,
+      "grad_norm": 0.7765636777336408,
+      "learning_rate": 1.9979303917173793e-05,
+      "loss": 0.8967,
+      "step": 1410
+    },
+    {
+      "epoch": 0.030502212484426687,
+      "grad_norm": 0.7854049103534458,
+      "learning_rate": 1.9978863273753583e-05,
+      "loss": 0.8921,
+      "step": 1420
+    },
+    {
+      "epoch": 0.030717016797697297,
+      "grad_norm": 0.7525748519317893,
+      "learning_rate": 1.9978417993675236e-05,
+      "loss": 0.9078,
+      "step": 1430
+    },
+    {
+      "epoch": 0.030931821110967907,
+      "grad_norm": 0.76321986852774,
+      "learning_rate": 1.997796807714565e-05,
+      "loss": 0.9121,
+      "step": 1440
+    },
+    {
+      "epoch": 0.03114662542423852,
+      "grad_norm": 0.7619696286583564,
+      "learning_rate": 1.9977513524373878e-05,
+      "loss": 0.8986,
+      "step": 1450
+    },
+    {
+      "epoch": 0.031361429737509126,
+      "grad_norm": 0.8025471522668453,
+      "learning_rate": 1.997705433557113e-05,
+      "loss": 0.8827,
+      "step": 1460
+    },
+    {
+      "epoch": 0.031576234050779736,
+      "grad_norm": 0.7763997567061249,
+      "learning_rate": 1.9976590510950766e-05,
+      "loss": 0.9019,
+      "step": 1470
+    },
+    {
+      "epoch": 0.03179103836405035,
+      "grad_norm": 0.7413217078674147,
+      "learning_rate": 1.99761220507283e-05,
+      "loss": 0.8864,
+      "step": 1480
+    },
+    {
+      "epoch": 0.03200584267732096,
+      "grad_norm": 0.7457795342330259,
+      "learning_rate": 1.99756489551214e-05,
+      "loss": 0.8927,
+      "step": 1490
+    },
+    {
+      "epoch": 0.03222064699059157,
+      "grad_norm": 0.7177239737436911,
+      "learning_rate": 1.997517122434989e-05,
+      "loss": 0.8664,
+      "step": 1500
+    },
+    {
+      "epoch": 0.03243545130386218,
+      "grad_norm": 0.729910325595943,
+      "learning_rate": 1.9974688858635748e-05,
+      "loss": 0.8872,
+      "step": 1510
+    },
+    {
+      "epoch": 0.03265025561713279,
+      "grad_norm": 0.734760164934118,
+      "learning_rate": 1.9974201858203098e-05,
+      "loss": 0.902,
+      "step": 1520
+    },
+    {
+      "epoch": 0.0328650599304034,
+      "grad_norm": 0.7602546032074973,
+      "learning_rate": 1.9973710223278236e-05,
+      "loss": 0.8948,
+      "step": 1530
+    },
+    {
+      "epoch": 0.03307986424367401,
+      "grad_norm": 0.7562657004832819,
+      "learning_rate": 1.9973213954089586e-05,
+      "loss": 0.8869,
+      "step": 1540
+    },
+    {
+      "epoch": 0.03329466855694462,
+      "grad_norm": 0.7774738334729859,
+      "learning_rate": 1.9972713050867745e-05,
+      "loss": 0.9045,
+      "step": 1550
+    },
+    {
+      "epoch": 0.03350947287021523,
+      "grad_norm": 0.7681125250896116,
+      "learning_rate": 1.997220751384546e-05,
+      "loss": 0.9013,
+      "step": 1560
+    },
+    {
+      "epoch": 0.03372427718348584,
+      "grad_norm": 0.7339330432465896,
+      "learning_rate": 1.997169734325762e-05,
+      "loss": 0.9007,
+      "step": 1570
+    },
+    {
+      "epoch": 0.03393908149675646,
+      "grad_norm": 0.7065402914181205,
+      "learning_rate": 1.9971182539341284e-05,
+      "loss": 0.8787,
+      "step": 1580
+    },
+    {
+      "epoch": 0.03415388581002707,
+      "grad_norm": 0.7801128948603702,
+      "learning_rate": 1.9970663102335645e-05,
+      "loss": 0.9026,
+      "step": 1590
+    },
+    {
+      "epoch": 0.03436869012329768,
+      "grad_norm": 0.788227685604599,
+      "learning_rate": 1.997013903248206e-05,
+      "loss": 0.8949,
+      "step": 1600
+    },
+    {
+      "epoch": 0.03458349443656829,
+      "grad_norm": 0.7478323289786412,
+      "learning_rate": 1.9969610330024043e-05,
+      "loss": 0.8807,
+      "step": 1610
+    },
+    {
+      "epoch": 0.0347982987498389,
+      "grad_norm": 0.7594500146074927,
+      "learning_rate": 1.996907699520725e-05,
+      "loss": 0.897,
+      "step": 1620
+    },
+    {
+      "epoch": 0.03501310306310951,
+      "grad_norm": 0.7557561178768684,
+      "learning_rate": 1.9968539028279493e-05,
+      "loss": 0.8817,
+      "step": 1630
+    },
+    {
+      "epoch": 0.03522790737638012,
+      "grad_norm": 0.7705545022363263,
+      "learning_rate": 1.9967996429490738e-05,
+      "loss": 0.8975,
+      "step": 1640
+    },
+    {
+      "epoch": 0.03544271168965073,
+      "grad_norm": 0.7586938534256935,
+      "learning_rate": 1.9967449199093104e-05,
+      "loss": 0.9046,
+      "step": 1650
+    },
+    {
+      "epoch": 0.03565751600292134,
+      "grad_norm": 0.7431761255095913,
+      "learning_rate": 1.996689733734086e-05,
+      "loss": 0.8873,
+      "step": 1660
+    },
+    {
+      "epoch": 0.03587232031619195,
+      "grad_norm": 0.7208377192401944,
+      "learning_rate": 1.9966340844490427e-05,
+      "loss": 0.8785,
+      "step": 1670
+    },
+    {
+      "epoch": 0.03608712462946256,
+      "grad_norm": 0.7506584753315972,
+      "learning_rate": 1.9965779720800383e-05,
+      "loss": 0.8684,
+      "step": 1680
+    },
+    {
+      "epoch": 0.03630192894273317,
+      "grad_norm": 0.7391526908534235,
+      "learning_rate": 1.9965213966531443e-05,
+      "loss": 0.8925,
+      "step": 1690
+    },
+    {
+      "epoch": 0.03651673325600378,
+      "grad_norm": 0.7088294334006218,
+      "learning_rate": 1.996464358194649e-05,
+      "loss": 0.8812,
+      "step": 1700
+    },
+    {
+      "epoch": 0.03673153756927439,
+      "grad_norm": 0.7199397102089583,
+      "learning_rate": 1.9964068567310552e-05,
+      "loss": 0.9021,
+      "step": 1710
+    },
+    {
+      "epoch": 0.036946341882545,
+      "grad_norm": 0.7386582371472061,
+      "learning_rate": 1.996348892289081e-05,
+      "loss": 0.8847,
+      "step": 1720
+    },
+    {
+      "epoch": 0.03716114619581561,
+      "grad_norm": 0.7508464385857595,
+      "learning_rate": 1.9962904648956585e-05,
+      "loss": 0.8753,
+      "step": 1730
+    },
+    {
+      "epoch": 0.03737595050908622,
+      "grad_norm": 0.7470256078534874,
+      "learning_rate": 1.9962315745779372e-05,
+      "loss": 0.8963,
+      "step": 1740
+    },
+    {
+      "epoch": 0.03759075482235683,
+      "grad_norm": 0.7418761066889206,
+      "learning_rate": 1.9961722213632794e-05,
+      "loss": 0.8963,
+      "step": 1750
+    },
+    {
+      "epoch": 0.03780555913562744,
+      "grad_norm": 0.7616668970329441,
+      "learning_rate": 1.9961124052792636e-05,
+      "loss": 0.8795,
+      "step": 1760
+    },
+    {
+      "epoch": 0.03802036344889805,
+      "grad_norm": 0.8233002415920407,
+      "learning_rate": 1.996052126353684e-05,
+      "loss": 0.8716,
+      "step": 1770
+    },
+    {
+      "epoch": 0.03823516776216866,
+      "grad_norm": 0.7383776636172839,
+      "learning_rate": 1.995991384614548e-05,
+      "loss": 0.8813,
+      "step": 1780
+    },
+    {
+      "epoch": 0.03844997207543927,
+      "grad_norm": 0.7101146565285336,
+      "learning_rate": 1.9959301800900795e-05,
+      "loss": 0.8744,
+      "step": 1790
+    },
+    {
+      "epoch": 0.03866477638870989,
+      "grad_norm": 0.7797901881755945,
+      "learning_rate": 1.9958685128087175e-05,
+      "loss": 0.8739,
+      "step": 1800
+    },
+    {
+      "epoch": 0.0388795807019805,
+      "grad_norm": 0.7607914365027031,
+      "learning_rate": 1.995806382799115e-05,
+      "loss": 0.8769,
+      "step": 1810
+    },
+    {
+      "epoch": 0.03909438501525111,
+      "grad_norm": 0.7186167152089645,
+      "learning_rate": 1.9957437900901408e-05,
+      "loss": 0.8728,
+      "step": 1820
+    },
+    {
+      "epoch": 0.03930918932852172,
+      "grad_norm": 0.7500460414609829,
+      "learning_rate": 1.9956807347108787e-05,
+      "loss": 0.886,
+      "step": 1830
+    },
+    {
+      "epoch": 0.03952399364179233,
+      "grad_norm": 0.8448420531394487,
+      "learning_rate": 1.9956172166906267e-05,
+      "loss": 0.876,
+      "step": 1840
+    },
+    {
+      "epoch": 0.03973879795506294,
+      "grad_norm": 0.7616239143421761,
+      "learning_rate": 1.9955532360588986e-05,
+      "loss": 0.8731,
+      "step": 1850
+    },
+    {
+      "epoch": 0.03995360226833355,
+      "grad_norm": 0.7516220625045015,
+      "learning_rate": 1.9954887928454232e-05,
+      "loss": 0.8895,
+      "step": 1860
+    },
+    {
+      "epoch": 0.04016840658160416,
+      "grad_norm": 0.6940486511611167,
+      "learning_rate": 1.9954238870801434e-05,
+      "loss": 0.8816,
+      "step": 1870
+    },
+    {
+      "epoch": 0.04038321089487477,
+      "grad_norm": 0.7207106268872732,
+      "learning_rate": 1.995358518793218e-05,
+      "loss": 0.8786,
+      "step": 1880
+    },
+    {
+      "epoch": 0.04059801520814538,
+      "grad_norm": 0.7003859763965479,
+      "learning_rate": 1.99529268801502e-05,
+      "loss": 0.8909,
+      "step": 1890
+    },
+    {
+      "epoch": 0.04081281952141599,
+      "grad_norm": 0.7143855349325927,
+      "learning_rate": 1.995226394776137e-05,
+      "loss": 0.876,
+      "step": 1900
+    },
+    {
+      "epoch": 0.041027623834686604,
+      "grad_norm": 1.027903124400268,
+      "learning_rate": 1.9951596391073732e-05,
+      "loss": 0.8845,
+      "step": 1910
+    },
+    {
+      "epoch": 0.041242428147957214,
+      "grad_norm": 0.7561863103739813,
+      "learning_rate": 1.9950924210397453e-05,
+      "loss": 0.8805,
+      "step": 1920
+    },
+    {
+      "epoch": 0.04145723246122782,
+      "grad_norm": 0.7070451516725674,
+      "learning_rate": 1.9950247406044867e-05,
+      "loss": 0.8866,
+      "step": 1930
+    },
+    {
+      "epoch": 0.04167203677449843,
+      "grad_norm": 0.7414269087308623,
+      "learning_rate": 1.9949565978330447e-05,
+      "loss": 0.8967,
+      "step": 1940
+    },
+    {
+      "epoch": 0.04188684108776904,
+      "grad_norm": 0.7282024119626752,
+      "learning_rate": 1.9948879927570823e-05,
+      "loss": 0.883,
+      "step": 1950
+    },
+    {
+      "epoch": 0.04210164540103965,
+      "grad_norm": 0.7460528091826555,
+      "learning_rate": 1.9948189254084758e-05,
+      "loss": 0.8701,
+      "step": 1960
+    },
+    {
+      "epoch": 0.04231644971431026,
+      "grad_norm": 0.6963758847916436,
+      "learning_rate": 1.9947493958193176e-05,
+      "loss": 0.8748,
+      "step": 1970
+    },
+    {
+      "epoch": 0.04253125402758087,
+      "grad_norm": 0.7069516648127419,
+      "learning_rate": 1.9946794040219147e-05,
+      "loss": 0.8832,
+      "step": 1980
+    },
+    {
+      "epoch": 0.04274605834085148,
+      "grad_norm": 0.746440584689567,
+      "learning_rate": 1.9946089500487888e-05,
+      "loss": 0.8733,
+      "step": 1990
+    },
+    {
+      "epoch": 0.04296086265412209,
+      "grad_norm": 0.7146001602504596,
+      "learning_rate": 1.994538033932675e-05,
+      "loss": 0.88,
+      "step": 2000
+    },
+    {
+      "epoch": 0.0431756669673927,
+      "grad_norm": 0.8314467011125141,
+      "learning_rate": 1.9944666557065257e-05,
+      "loss": 0.8876,
+      "step": 2010
+    },
+    {
+      "epoch": 0.04339047128066332,
+      "grad_norm": 0.8015245377512796,
+      "learning_rate": 1.994394815403506e-05,
+      "loss": 0.9035,
+      "step": 2020
+    },
+    {
+      "epoch": 0.04360527559393393,
+      "grad_norm": 0.7222397980014357,
+      "learning_rate": 1.9943225130569967e-05,
+      "loss": 0.8846,
+      "step": 2030
+    },
+    {
+      "epoch": 0.04382007990720454,
+      "grad_norm": 0.9345503125138953,
+      "learning_rate": 1.9942497487005922e-05,
+      "loss": 0.875,
+      "step": 2040
+    },
+    {
+      "epoch": 0.04403488422047515,
+      "grad_norm": 0.777701186759307,
+      "learning_rate": 1.9941765223681032e-05,
+      "loss": 0.8886,
+      "step": 2050
+    },
+    {
+      "epoch": 0.04424968853374576,
+      "grad_norm": 0.7503022644965333,
+      "learning_rate": 1.9941028340935536e-05,
+      "loss": 0.8579,
+      "step": 2060
+    },
+    {
+      "epoch": 0.04446449284701637,
+      "grad_norm": 0.719946551023971,
+      "learning_rate": 1.9940286839111832e-05,
+      "loss": 0.8802,
+      "step": 2070
+    },
+    {
+      "epoch": 0.04467929716028698,
+      "grad_norm": 0.749093812327845,
+      "learning_rate": 1.9939540718554445e-05,
+      "loss": 0.8834,
+      "step": 2080
+    },
+    {
+      "epoch": 0.04489410147355759,
+      "grad_norm": 0.7422208424583762,
+      "learning_rate": 1.993878997961007e-05,
+      "loss": 0.8853,
+      "step": 2090
+    },
+    {
+      "epoch": 0.0451089057868282,
+      "grad_norm": 0.7374573183269133,
+      "learning_rate": 1.993803462262753e-05,
+      "loss": 0.891,
+      "step": 2100
+    },
+    {
+      "epoch": 0.04532371010009881,
+      "grad_norm": 0.7350782974678935,
+      "learning_rate": 1.99372746479578e-05,
+      "loss": 0.9003,
+      "step": 2110
+    },
+    {
+      "epoch": 0.04553851441336942,
+      "grad_norm": 0.7281319667850538,
+      "learning_rate": 1.9936510055954002e-05,
+      "loss": 0.8683,
+      "step": 2120
+    },
+    {
+      "epoch": 0.045753318726640034,
+      "grad_norm": 0.7965856691084933,
+      "learning_rate": 1.9935740846971404e-05,
+      "loss": 0.8818,
+      "step": 2130
+    },
+    {
+      "epoch": 0.045968123039910644,
+      "grad_norm": 0.7049379609150542,
+      "learning_rate": 1.9934967021367417e-05,
+      "loss": 0.8873,
+      "step": 2140
+    },
+    {
+      "epoch": 0.046182927353181254,
+      "grad_norm": 0.6866133102632582,
+      "learning_rate": 1.9934188579501596e-05,
+      "loss": 0.8799,
+      "step": 2150
+    },
+    {
+      "epoch": 0.046397731666451864,
+      "grad_norm": 0.7086840811762911,
+      "learning_rate": 1.993340552173564e-05,
+      "loss": 0.8724,
+      "step": 2160
+    },
+    {
+      "epoch": 0.04661253597972247,
+      "grad_norm": 0.7205449804498884,
+      "learning_rate": 1.99326178484334e-05,
+      "loss": 0.8906,
+      "step": 2170
+    },
+    {
+      "epoch": 0.04682734029299308,
+      "grad_norm": 0.7271300532378566,
+      "learning_rate": 1.9931825559960867e-05,
+      "loss": 0.8739,
+      "step": 2180
+    },
+    {
+      "epoch": 0.04704214460626369,
+      "grad_norm": 0.782395233795208,
+      "learning_rate": 1.993102865668617e-05,
+      "loss": 0.8766,
+      "step": 2190
+    },
+    {
+      "epoch": 0.0472569489195343,
+      "grad_norm": 0.7344948505203559,
+      "learning_rate": 1.9930227138979595e-05,
+      "loss": 0.8825,
+      "step": 2200
+    },
+    {
+      "epoch": 0.04747175323280491,
+      "grad_norm": 0.6708701924983489,
+      "learning_rate": 1.992942100721356e-05,
+      "loss": 0.8659,
+      "step": 2210
+    },
+    {
+      "epoch": 0.04768655754607552,
+      "grad_norm": 0.6843127757880155,
+      "learning_rate": 1.992861026176264e-05,
+      "loss": 0.8689,
+      "step": 2220
+    },
+    {
+      "epoch": 0.04790136185934613,
+      "grad_norm": 0.7601120607749401,
+      "learning_rate": 1.992779490300354e-05,
+      "loss": 0.8809,
+      "step": 2230
+    },
+    {
+      "epoch": 0.04811616617261675,
+      "grad_norm": 0.7771578485534071,
+      "learning_rate": 1.9926974931315114e-05,
+      "loss": 0.874,
+      "step": 2240
+    },
+    {
+      "epoch": 0.04833097048588736,
+      "grad_norm": 0.6677106942558194,
+      "learning_rate": 1.9926150347078363e-05,
+      "loss": 0.8715,
+      "step": 2250
+    },
+    {
+      "epoch": 0.04854577479915797,
+      "grad_norm": 0.6825562101263428,
+      "learning_rate": 1.9925321150676426e-05,
+      "loss": 0.8747,
+      "step": 2260
+    },
+    {
+      "epoch": 0.04876057911242858,
+      "grad_norm": 0.7870531733277122,
+      "learning_rate": 1.992448734249459e-05,
+      "loss": 0.876,
+      "step": 2270
+    },
+    {
+      "epoch": 0.04897538342569919,
+      "grad_norm": 0.7487545797908844,
+      "learning_rate": 1.9923648922920284e-05,
+      "loss": 0.8842,
+      "step": 2280
+    },
+    {
+      "epoch": 0.0491901877389698,
+      "grad_norm": 0.7615635965125641,
+      "learning_rate": 1.9922805892343073e-05,
+      "loss": 0.8758,
+      "step": 2290
+    },
+    {
+      "epoch": 0.04940499205224041,
+      "grad_norm": 0.7376131287808475,
+      "learning_rate": 1.9921958251154673e-05,
+      "loss": 0.8767,
+      "step": 2300
+    },
+    {
+      "epoch": 0.04961979636551102,
+      "grad_norm": 0.7011021964691581,
+      "learning_rate": 1.9921105999748937e-05,
+      "loss": 0.8785,
+      "step": 2310
+    },
+    {
+      "epoch": 0.04983460067878163,
+      "grad_norm": 0.6991405501900602,
+      "learning_rate": 1.9920249138521864e-05,
+      "loss": 0.8659,
+      "step": 2320
+    },
+    {
+      "epoch": 0.05004940499205224,
+      "grad_norm": 0.8709849609023845,
+      "learning_rate": 1.9919387667871586e-05,
+      "loss": 0.8831,
+      "step": 2330
+    },
+    {
+      "epoch": 0.05026420930532285,
+      "grad_norm": 0.7150713210870077,
+      "learning_rate": 1.9918521588198395e-05,
+      "loss": 0.8758,
+      "step": 2340
+    },
+    {
+      "epoch": 0.050479013618593464,
+      "grad_norm": 0.7286077482741432,
+      "learning_rate": 1.9917650899904704e-05,
+      "loss": 0.8807,
+      "step": 2350
+    },
+    {
+      "epoch": 0.050693817931864074,
+      "grad_norm": 0.7213279597207197,
+      "learning_rate": 1.9916775603395078e-05,
+      "loss": 0.8884,
+      "step": 2360
+    },
+    {
+      "epoch": 0.050908622245134684,
+      "grad_norm": 0.7135061250388653,
+      "learning_rate": 1.9915895699076224e-05,
+      "loss": 0.8759,
+      "step": 2370
+    },
+    {
+      "epoch": 0.051123426558405294,
+      "grad_norm": 0.7313831738364597,
+      "learning_rate": 1.9915011187356988e-05,
+      "loss": 0.857,
+      "step": 2380
+    },
+    {
+      "epoch": 0.051338230871675904,
+      "grad_norm": 0.6876477115175337,
+      "learning_rate": 1.991412206864835e-05,
+      "loss": 0.8572,
+      "step": 2390
+    },
+    {
+      "epoch": 0.051553035184946513,
+      "grad_norm": 0.6945309755588273,
+      "learning_rate": 1.9913228343363448e-05,
+      "loss": 0.8652,
+      "step": 2400
+    },
+    {
+      "epoch": 0.05176783949821712,
+      "grad_norm": 0.704316576659832,
+      "learning_rate": 1.991233001191754e-05,
+      "loss": 0.8601,
+      "step": 2410
+    },
+    {
+      "epoch": 0.05198264381148773,
+      "grad_norm": 0.6985006179988886,
+      "learning_rate": 1.9911427074728043e-05,
+      "loss": 0.8859,
+      "step": 2420
+    },
+    {
+      "epoch": 0.05219744812475834,
+      "grad_norm": 0.7524066390625885,
+      "learning_rate": 1.9910519532214498e-05,
+      "loss": 0.8892,
+      "step": 2430
+    },
+    {
+      "epoch": 0.05241225243802895,
+      "grad_norm": 0.6901836096027437,
+      "learning_rate": 1.990960738479859e-05,
+      "loss": 0.8846,
+      "step": 2440
+    },
+    {
+      "epoch": 0.05262705675129956,
+      "grad_norm": 0.6849358351981636,
+      "learning_rate": 1.9908690632904157e-05,
+      "loss": 0.8634,
+      "step": 2450
+    },
+    {
+      "epoch": 0.05284186106457018,
+      "grad_norm": 0.765461828081668,
+      "learning_rate": 1.9907769276957156e-05,
+      "loss": 0.8746,
+      "step": 2460
+    },
+    {
+      "epoch": 0.05305666537784079,
+      "grad_norm": 0.7679215224483907,
+      "learning_rate": 1.9906843317385696e-05,
+      "loss": 0.8717,
+      "step": 2470
+    },
+    {
+      "epoch": 0.0532714696911114,
+      "grad_norm": 0.7102215793244879,
+      "learning_rate": 1.9905912754620028e-05,
+      "loss": 0.8768,
+      "step": 2480
+    },
+    {
+      "epoch": 0.05348627400438201,
+      "grad_norm": 0.7328716303001253,
+      "learning_rate": 1.9904977589092526e-05,
+      "loss": 0.8762,
+      "step": 2490
+    },
+    {
+      "epoch": 0.05370107831765262,
+      "grad_norm": 0.6873393080849327,
+      "learning_rate": 1.990403782123772e-05,
+      "loss": 0.8666,
+      "step": 2500
+    },
+    {
+      "epoch": 0.05391588263092323,
+      "grad_norm": 0.7147008262903141,
+      "learning_rate": 1.990309345149227e-05,
+      "loss": 0.8702,
+      "step": 2510
+    },
+    {
+      "epoch": 0.05413068694419384,
+      "grad_norm": 0.7258515311880707,
+      "learning_rate": 1.990214448029497e-05,
+      "loss": 0.861,
+      "step": 2520
+    },
+    {
+      "epoch": 0.05434549125746445,
+      "grad_norm": 0.6864484714517523,
+      "learning_rate": 1.9901190908086768e-05,
+      "loss": 0.8607,
+      "step": 2530
+    },
+    {
+      "epoch": 0.05456029557073506,
+      "grad_norm": 0.7219498116874772,
+      "learning_rate": 1.9900232735310732e-05,
+      "loss": 0.8636,
+      "step": 2540
+    },
+    {
+      "epoch": 0.05477509988400567,
+      "grad_norm": 0.7132672252604229,
+      "learning_rate": 1.9899269962412075e-05,
+      "loss": 0.8785,
+      "step": 2550
+    },
+    {
+      "epoch": 0.054989904197276285,
+      "grad_norm": 0.6983739180648069,
+      "learning_rate": 1.989830258983815e-05,
+      "loss": 0.8688,
+      "step": 2560
+    },
+    {
+      "epoch": 0.055204708510546895,
+      "grad_norm": 0.7061201979674048,
+      "learning_rate": 1.9897330618038443e-05,
+      "loss": 0.8771,
+      "step": 2570
+    },
+    {
+      "epoch": 0.055419512823817504,
+      "grad_norm": 0.7219202198100194,
+      "learning_rate": 1.9896354047464578e-05,
+      "loss": 0.8875,
+      "step": 2580
+    },
+    {
+      "epoch": 0.055634317137088114,
+      "grad_norm": 0.8868550476280311,
+      "learning_rate": 1.9895372878570314e-05,
+      "loss": 0.8623,
+      "step": 2590
+    },
+    {
+      "epoch": 0.055849121450358724,
+      "grad_norm": 0.7475858890792529,
+      "learning_rate": 1.989438711181156e-05,
+      "loss": 0.852,
+      "step": 2600
+    },
+    {
+      "epoch": 0.056063925763629334,
+      "grad_norm": 0.6861827188596874,
+      "learning_rate": 1.9893396747646337e-05,
+      "loss": 0.8919,
+      "step": 2610
+    },
+    {
+      "epoch": 0.056278730076899944,
+      "grad_norm": 0.6806996838562623,
+      "learning_rate": 1.9892401786534827e-05,
+      "loss": 0.8581,
+      "step": 2620
+    },
+    {
+      "epoch": 0.056493534390170554,
+      "grad_norm": 0.6852549300278078,
+      "learning_rate": 1.989140222893933e-05,
+      "loss": 0.8697,
+      "step": 2630
+    },
+    {
+      "epoch": 0.05670833870344116,
+      "grad_norm": 0.6657649574580968,
+      "learning_rate": 1.989039807532429e-05,
+      "loss": 0.8807,
+      "step": 2640
+    },
+    {
+      "epoch": 0.05692314301671177,
+      "grad_norm": 0.6905459875216616,
+      "learning_rate": 1.988938932615628e-05,
+      "loss": 0.8673,
+      "step": 2650
+    },
+    {
+      "epoch": 0.05713794732998238,
+      "grad_norm": 0.7020389616663805,
+      "learning_rate": 1.9888375981904024e-05,
+      "loss": 0.8641,
+      "step": 2660
+    },
+    {
+      "epoch": 0.057352751643253,
+      "grad_norm": 0.6916187469122672,
+      "learning_rate": 1.988735804303836e-05,
+      "loss": 0.8621,
+      "step": 2670
+    },
+    {
+      "epoch": 0.05756755595652361,
+      "grad_norm": 0.7314594201824384,
+      "learning_rate": 1.988633551003228e-05,
+      "loss": 0.879,
+      "step": 2680
+    },
+    {
+      "epoch": 0.05778236026979422,
+      "grad_norm": 0.746171027813674,
+      "learning_rate": 1.9885308383360894e-05,
+      "loss": 0.8707,
+      "step": 2690
+    },
+    {
+      "epoch": 0.05799716458306483,
+      "grad_norm": 0.6634180978048528,
+      "learning_rate": 1.988427666350146e-05,
+      "loss": 0.8598,
+      "step": 2700
+    },
+    {
+      "epoch": 0.05821196889633544,
+      "grad_norm": 0.6453064675883513,
+      "learning_rate": 1.988324035093336e-05,
+      "loss": 0.8683,
+      "step": 2710
+    },
+    {
+      "epoch": 0.05842677320960605,
+      "grad_norm": 0.6912271165294316,
+      "learning_rate": 1.9882199446138116e-05,
+      "loss": 0.8746,
+      "step": 2720
+    },
+    {
+      "epoch": 0.05864157752287666,
+      "grad_norm": 0.7643854365177722,
+      "learning_rate": 1.9881153949599384e-05,
+      "loss": 0.8744,
+      "step": 2730
+    },
+    {
+      "epoch": 0.05885638183614727,
+      "grad_norm": 0.6767199632505446,
+      "learning_rate": 1.988010386180295e-05,
+      "loss": 0.8516,
+      "step": 2740
+    },
+    {
+      "epoch": 0.05907118614941788,
+      "grad_norm": 0.7108698181797346,
+      "learning_rate": 1.9879049183236735e-05,
+      "loss": 0.8695,
+      "step": 2750
+    },
+    {
+      "epoch": 0.05928599046268849,
+      "grad_norm": 0.7253238005438297,
+      "learning_rate": 1.9877989914390794e-05,
+      "loss": 0.8555,
+      "step": 2760
+    },
+    {
+      "epoch": 0.0595007947759591,
+      "grad_norm": 0.6453538421152363,
+      "learning_rate": 1.9876926055757316e-05,
+      "loss": 0.8633,
+      "step": 2770
+    },
+    {
+      "epoch": 0.059715599089229715,
+      "grad_norm": 0.7205543033570071,
+      "learning_rate": 1.9875857607830616e-05,
+      "loss": 0.8684,
+      "step": 2780
+    },
+    {
+      "epoch": 0.059930403402500325,
+      "grad_norm": 0.756612819272997,
+      "learning_rate": 1.987478457110715e-05,
+      "loss": 0.8544,
+      "step": 2790
+    },
+    {
+      "epoch": 0.060145207715770935,
+      "grad_norm": 0.7043547954742904,
+      "learning_rate": 1.9873706946085504e-05,
+      "loss": 0.8622,
+      "step": 2800
+    },
+    {
+      "epoch": 0.060360012029041545,
+      "grad_norm": 0.7593440635743444,
+      "learning_rate": 1.9872624733266386e-05,
+      "loss": 0.8834,
+      "step": 2810
+    },
+    {
+      "epoch": 0.060574816342312154,
+      "grad_norm": 0.6863494468953989,
+      "learning_rate": 1.9871537933152653e-05,
+      "loss": 0.8633,
+      "step": 2820
+    },
+    {
+      "epoch": 0.060789620655582764,
+      "grad_norm": 0.7138487939483891,
+      "learning_rate": 1.987044654624928e-05,
+      "loss": 0.8594,
+      "step": 2830
+    },
+    {
+      "epoch": 0.061004424968853374,
+      "grad_norm": 0.6981415113485295,
+      "learning_rate": 1.9869350573063376e-05,
+      "loss": 0.8677,
+      "step": 2840
+    },
+    {
+      "epoch": 0.061219229282123984,
+      "grad_norm": 0.7075330565366764,
+      "learning_rate": 1.9868250014104187e-05,
+      "loss": 0.8682,
+      "step": 2850
+    },
+    {
+      "epoch": 0.061434033595394594,
+      "grad_norm": 0.714624614534178,
+      "learning_rate": 1.9867144869883083e-05,
+      "loss": 0.8628,
+      "step": 2860
+    },
+    {
+      "epoch": 0.061648837908665204,
+      "grad_norm": 0.7137186386058576,
+      "learning_rate": 1.9866035140913568e-05,
+      "loss": 0.8521,
+      "step": 2870
+    },
+    {
+      "epoch": 0.06186364222193581,
+      "grad_norm": 0.6880470582910433,
+      "learning_rate": 1.9864920827711273e-05,
+      "loss": 0.8615,
+      "step": 2880
+    },
+    {
+      "epoch": 0.06207844653520643,
+      "grad_norm": 0.6631751007947058,
+      "learning_rate": 1.9863801930793966e-05,
+      "loss": 0.8684,
+      "step": 2890
+    },
+    {
+      "epoch": 0.06229325084847704,
+      "grad_norm": 0.680905465307777,
+      "learning_rate": 1.9862678450681537e-05,
+      "loss": 0.8573,
+      "step": 2900
+    },
+    {
+      "epoch": 0.06250805516174765,
+      "grad_norm": 0.722687320949144,
+      "learning_rate": 1.9861550387896007e-05,
+      "loss": 0.8573,
+      "step": 2910
+    },
+    {
+      "epoch": 0.06272285947501825,
+      "grad_norm": 0.707058027911016,
+      "learning_rate": 1.986041774296153e-05,
+      "loss": 0.8623,
+      "step": 2920
+    },
+    {
+      "epoch": 0.06293766378828887,
+      "grad_norm": 0.6766536125382012,
+      "learning_rate": 1.9859280516404387e-05,
+      "loss": 0.862,
+      "step": 2930
+    },
+    {
+      "epoch": 0.06315246810155947,
+      "grad_norm": 0.7142428076724684,
+      "learning_rate": 1.985813870875299e-05,
+      "loss": 0.8769,
+      "step": 2940
+    },
+    {
+      "epoch": 0.06336727241483009,
+      "grad_norm": 0.6712747602182263,
+      "learning_rate": 1.9856992320537872e-05,
+      "loss": 0.8651,
+      "step": 2950
+    },
+    {
+      "epoch": 0.0635820767281007,
+      "grad_norm": 0.7333177718791057,
+      "learning_rate": 1.9855841352291705e-05,
+      "loss": 0.8496,
+      "step": 2960
+    },
+    {
+      "epoch": 0.06379688104137131,
+      "grad_norm": 0.6738403800576495,
+      "learning_rate": 1.9854685804549282e-05,
+      "loss": 0.8679,
+      "step": 2970
+    },
+    {
+      "epoch": 0.06401168535464193,
+      "grad_norm": 0.6926446949090426,
+      "learning_rate": 1.985352567784753e-05,
+      "loss": 0.8737,
+      "step": 2980
+    },
+    {
+      "epoch": 0.06422648966791253,
+      "grad_norm": 0.6819388635508017,
+      "learning_rate": 1.985236097272549e-05,
+      "loss": 0.8433,
+      "step": 2990
+    },
+    {
+      "epoch": 0.06444129398118315,
+      "grad_norm": 0.6578041236132836,
+      "learning_rate": 1.985119168972435e-05,
+      "loss": 0.8419,
+      "step": 3000
+    },
+    {
+      "epoch": 0.06465609829445375,
+      "grad_norm": 0.6561607204214681,
+      "learning_rate": 1.9850017829387406e-05,
+      "loss": 0.8539,
+      "step": 3010
+    },
+    {
+      "epoch": 0.06487090260772437,
+      "grad_norm": 0.693338421429239,
+      "learning_rate": 1.98488393922601e-05,
+      "loss": 0.8482,
+      "step": 3020
+    },
+    {
+      "epoch": 0.06508570692099497,
+      "grad_norm": 0.7241505791029577,
+      "learning_rate": 1.984765637888998e-05,
+      "loss": 0.8506,
+      "step": 3030
+    },
+    {
+      "epoch": 0.06530051123426558,
+      "grad_norm": 0.6726045381693205,
+      "learning_rate": 1.9846468789826737e-05,
+      "loss": 0.8605,
+      "step": 3040
+    },
+    {
+      "epoch": 0.0655153155475362,
+      "grad_norm": 0.7093889099822586,
+      "learning_rate": 1.984527662562218e-05,
+      "loss": 0.8651,
+      "step": 3050
+    },
+    {
+      "epoch": 0.0657301198608068,
+      "grad_norm": 0.6510030697876009,
+      "learning_rate": 1.9844079886830246e-05,
+      "loss": 0.8641,
+      "step": 3060
+    },
+    {
+      "epoch": 0.06594492417407742,
+      "grad_norm": 0.6560638118957208,
+      "learning_rate": 1.9842878574007e-05,
+      "loss": 0.874,
+      "step": 3070
+    },
+    {
+      "epoch": 0.06615972848734802,
+      "grad_norm": 0.7134647721402186,
+      "learning_rate": 1.9841672687710624e-05,
+      "loss": 0.8528,
+      "step": 3080
+    },
+    {
+      "epoch": 0.06637453280061864,
+      "grad_norm": 0.6922835771738086,
+      "learning_rate": 1.9840462228501432e-05,
+      "loss": 0.8609,
+      "step": 3090
+    },
+    {
+      "epoch": 0.06658933711388924,
+      "grad_norm": 0.6573245546694103,
+      "learning_rate": 1.9839247196941862e-05,
+      "loss": 0.8644,
+      "step": 3100
+    },
+    {
+      "epoch": 0.06680414142715986,
+      "grad_norm": 0.6864765270164866,
+      "learning_rate": 1.9838027593596477e-05,
+      "loss": 0.8633,
+      "step": 3110
+    },
+    {
+      "epoch": 0.06701894574043046,
+      "grad_norm": 0.6508340450389928,
+      "learning_rate": 1.9836803419031964e-05,
+      "loss": 0.8486,
+      "step": 3120
+    },
+    {
+      "epoch": 0.06723375005370108,
+      "grad_norm": 0.6712340814966278,
+      "learning_rate": 1.983557467381713e-05,
+      "loss": 0.8576,
+      "step": 3130
+    },
+    {
+      "epoch": 0.06744855436697168,
+      "grad_norm": 0.6946743953968081,
+      "learning_rate": 1.9834341358522914e-05,
+      "loss": 0.8615,
+      "step": 3140
+    },
+    {
+      "epoch": 0.0676633586802423,
+      "grad_norm": 0.6631456581489633,
+      "learning_rate": 1.9833103473722366e-05,
+      "loss": 0.8616,
+      "step": 3150
+    },
+    {
+      "epoch": 0.06787816299351292,
+      "grad_norm": 0.7015972542735149,
+      "learning_rate": 1.983186101999067e-05,
+      "loss": 0.8616,
+      "step": 3160
+    },
+    {
+      "epoch": 0.06809296730678352,
+      "grad_norm": 0.6897254667864308,
+      "learning_rate": 1.9830613997905136e-05,
+      "loss": 0.865,
+      "step": 3170
+    },
+    {
+      "epoch": 0.06830777162005414,
+      "grad_norm": 0.6577402763346296,
+      "learning_rate": 1.982936240804518e-05,
+      "loss": 0.8544,
+      "step": 3180
+    },
+    {
+      "epoch": 0.06852257593332474,
+      "grad_norm": 0.6890291309672536,
+      "learning_rate": 1.9828106250992353e-05,
+      "loss": 0.8645,
+      "step": 3190
+    },
+    {
+      "epoch": 0.06873738024659536,
+      "grad_norm": 0.703215862838126,
+      "learning_rate": 1.982684552733033e-05,
+      "loss": 0.8728,
+      "step": 3200
+    },
+    {
+      "epoch": 0.06895218455986596,
+      "grad_norm": 0.6827501388608269,
+      "learning_rate": 1.9825580237644903e-05,
+      "loss": 0.8602,
+      "step": 3210
+    },
+    {
+      "epoch": 0.06916698887313658,
+      "grad_norm": 0.697650000738595,
+      "learning_rate": 1.9824310382523982e-05,
+      "loss": 0.8726,
+      "step": 3220
+    },
+    {
+      "epoch": 0.06938179318640718,
+      "grad_norm": 0.6721960819494287,
+      "learning_rate": 1.9823035962557608e-05,
+      "loss": 0.8605,
+      "step": 3230
+    },
+    {
+      "epoch": 0.0695965974996778,
+      "grad_norm": 0.7154662455928354,
+      "learning_rate": 1.9821756978337935e-05,
+      "loss": 0.8643,
+      "step": 3240
+    },
+    {
+      "epoch": 0.0698114018129484,
+      "grad_norm": 0.681240118026375,
+      "learning_rate": 1.9820473430459237e-05,
+      "loss": 0.8621,
+      "step": 3250
+    },
+    {
+      "epoch": 0.07002620612621901,
+      "grad_norm": 0.6580326580613558,
+      "learning_rate": 1.9819185319517915e-05,
+      "loss": 0.8466,
+      "step": 3260
+    },
+    {
+      "epoch": 0.07024101043948963,
+      "grad_norm": 0.6710767809283256,
+      "learning_rate": 1.981789264611249e-05,
+      "loss": 0.8659,
+      "step": 3270
+    },
+    {
+      "epoch": 0.07045581475276023,
+      "grad_norm": 0.6593001195695262,
+      "learning_rate": 1.9816595410843596e-05,
+      "loss": 0.8663,
+      "step": 3280
+    },
+    {
+      "epoch": 0.07067061906603085,
+      "grad_norm": 0.6628941061395422,
+      "learning_rate": 1.981529361431399e-05,
+      "loss": 0.8766,
+      "step": 3290
+    },
+    {
+      "epoch": 0.07088542337930145,
+      "grad_norm": 0.6706542522918096,
+      "learning_rate": 1.9813987257128552e-05,
+      "loss": 0.8515,
+      "step": 3300
+    },
+    {
+      "epoch": 0.07110022769257207,
+      "grad_norm": 0.7081735675645245,
+      "learning_rate": 1.981267633989428e-05,
+      "loss": 0.8536,
+      "step": 3310
+    },
+    {
+      "epoch": 0.07131503200584267,
+      "grad_norm": 0.6631068395510924,
+      "learning_rate": 1.981136086322028e-05,
+      "loss": 0.8595,
+      "step": 3320
+    },
+    {
+      "epoch": 0.07152983631911329,
+      "grad_norm": 0.6815244729783259,
+      "learning_rate": 1.98100408277178e-05,
+      "loss": 0.8452,
+      "step": 3330
+    },
+    {
+      "epoch": 0.0717446406323839,
+      "grad_norm": 0.6708767643583214,
+      "learning_rate": 1.9808716234000176e-05,
+      "loss": 0.8646,
+      "step": 3340
+    },
+    {
+      "epoch": 0.07195944494565451,
+      "grad_norm": 0.7070823636265635,
+      "learning_rate": 1.9807387082682888e-05,
+      "loss": 0.8596,
+      "step": 3350
+    },
+    {
+      "epoch": 0.07217424925892511,
+      "grad_norm": 0.6385205053779432,
+      "learning_rate": 1.980605337438352e-05,
+      "loss": 0.8596,
+      "step": 3360
+    },
+    {
+      "epoch": 0.07238905357219573,
+      "grad_norm": 0.6900045332944192,
+      "learning_rate": 1.9804715109721773e-05,
+      "loss": 0.8455,
+      "step": 3370
+    },
+    {
+      "epoch": 0.07260385788546635,
+      "grad_norm": 0.6952726462768112,
+      "learning_rate": 1.980337228931948e-05,
+      "loss": 0.854,
+      "step": 3380
+    },
+    {
+      "epoch": 0.07281866219873695,
+      "grad_norm": 0.6316666106551074,
+      "learning_rate": 1.9802024913800567e-05,
+      "loss": 0.8607,
+      "step": 3390
+    },
+    {
+      "epoch": 0.07303346651200757,
+      "grad_norm": 0.6782909648023765,
+      "learning_rate": 1.9800672983791097e-05,
+      "loss": 0.8578,
+      "step": 3400
+    },
+    {
+      "epoch": 0.07324827082527817,
+      "grad_norm": 0.703136553111246,
+      "learning_rate": 1.979931649991924e-05,
+      "loss": 0.8537,
+      "step": 3410
+    },
+    {
+      "epoch": 0.07346307513854879,
+      "grad_norm": 0.6766152461075413,
+      "learning_rate": 1.9797955462815285e-05,
+      "loss": 0.8551,
+      "step": 3420
+    },
+    {
+      "epoch": 0.07367787945181939,
+      "grad_norm": 0.6548199817442408,
+      "learning_rate": 1.9796589873111627e-05,
+      "loss": 0.848,
+      "step": 3430
+    },
+    {
+      "epoch": 0.07389268376509,
+      "grad_norm": 0.6699380907258793,
+      "learning_rate": 1.9795219731442798e-05,
+      "loss": 0.8518,
+      "step": 3440
+    },
+    {
+      "epoch": 0.07410748807836061,
+      "grad_norm": 0.646613126593096,
+      "learning_rate": 1.979384503844542e-05,
+      "loss": 0.8503,
+      "step": 3450
+    },
+    {
+      "epoch": 0.07432229239163123,
+      "grad_norm": 0.6657451032225497,
+      "learning_rate": 1.9792465794758246e-05,
+      "loss": 0.865,
+      "step": 3460
+    },
+    {
+      "epoch": 0.07453709670490183,
+      "grad_norm": 0.6566855884425971,
+      "learning_rate": 1.9791082001022137e-05,
+      "loss": 0.843,
+      "step": 3470
+    },
+    {
+      "epoch": 0.07475190101817245,
+      "grad_norm": 0.68529511952354,
+      "learning_rate": 1.978969365788007e-05,
+      "loss": 0.8569,
+      "step": 3480
+    },
+    {
+      "epoch": 0.07496670533144306,
+      "grad_norm": 0.694727034477339,
+      "learning_rate": 1.978830076597714e-05,
+      "loss": 0.8472,
+      "step": 3490
+    },
+    {
+      "epoch": 0.07518150964471366,
+      "grad_norm": 0.6631339825070333,
+      "learning_rate": 1.9786903325960545e-05,
+      "loss": 0.855,
+      "step": 3500
+    },
+    {
+      "epoch": 0.07539631395798428,
+      "grad_norm": 0.694444241956036,
+      "learning_rate": 1.9785501338479605e-05,
+      "loss": 0.8524,
+      "step": 3510
+    },
+    {
+      "epoch": 0.07561111827125488,
+      "grad_norm": 0.6945927280010864,
+      "learning_rate": 1.9784094804185755e-05,
+      "loss": 0.85,
+      "step": 3520
+    },
+    {
+      "epoch": 0.0758259225845255,
+      "grad_norm": 0.6672445340085226,
+      "learning_rate": 1.978268372373253e-05,
+      "loss": 0.8498,
+      "step": 3530
+    },
+    {
+      "epoch": 0.0760407268977961,
+      "grad_norm": 0.646988287501803,
+      "learning_rate": 1.978126809777559e-05,
+      "loss": 0.848,
+      "step": 3540
+    },
+    {
+      "epoch": 0.07625553121106672,
+      "grad_norm": 0.6911424046720633,
+      "learning_rate": 1.9779847926972703e-05,
+      "loss": 0.8652,
+      "step": 3550
+    },
+    {
+      "epoch": 0.07647033552433732,
+      "grad_norm": 0.6571056886382604,
+      "learning_rate": 1.977842321198375e-05,
+      "loss": 0.864,
+      "step": 3560
+    },
+    {
+      "epoch": 0.07668513983760794,
+      "grad_norm": 0.6668618449865177,
+      "learning_rate": 1.977699395347072e-05,
+      "loss": 0.8486,
+      "step": 3570
+    },
+    {
+      "epoch": 0.07689994415087854,
+      "grad_norm": 0.6575595446352954,
+      "learning_rate": 1.9775560152097713e-05,
+      "loss": 0.8534,
+      "step": 3580
+    },
+    {
+      "epoch": 0.07711474846414916,
+      "grad_norm": 0.658442535770228,
+      "learning_rate": 1.9774121808530944e-05,
+      "loss": 0.8554,
+      "step": 3590
+    },
+    {
+      "epoch": 0.07732955277741978,
+      "grad_norm": 0.6725808248232724,
+      "learning_rate": 1.977267892343874e-05,
+      "loss": 0.8524,
+      "step": 3600
+    },
+    {
+      "epoch": 0.07754435709069038,
+      "grad_norm": 0.6579027842938031,
+      "learning_rate": 1.9771231497491526e-05,
+      "loss": 0.8505,
+      "step": 3610
+    },
+    {
+      "epoch": 0.077759161403961,
+      "grad_norm": 0.6960301484489192,
+      "learning_rate": 1.976977953136185e-05,
+      "loss": 0.8602,
+      "step": 3620
+    },
+    {
+      "epoch": 0.0779739657172316,
+      "grad_norm": 0.6697741714896654,
+      "learning_rate": 1.9768323025724368e-05,
+      "loss": 0.8758,
+      "step": 3630
+    },
+    {
+      "epoch": 0.07818877003050222,
+      "grad_norm": 2.859419478862952,
+      "learning_rate": 1.9766861981255837e-05,
+      "loss": 0.864,
+      "step": 3640
+    },
+    {
+      "epoch": 0.07840357434377282,
+      "grad_norm": 0.6822863380857591,
+      "learning_rate": 1.9765396398635133e-05,
+      "loss": 0.8562,
+      "step": 3650
+    },
+    {
+      "epoch": 0.07861837865704344,
+      "grad_norm": 0.6500244246977851,
+      "learning_rate": 1.976392627854323e-05,
+      "loss": 0.8458,
+      "step": 3660
+    },
+    {
+      "epoch": 0.07883318297031404,
+      "grad_norm": 0.7383310944585931,
+      "learning_rate": 1.9762451621663225e-05,
+      "loss": 0.8435,
+      "step": 3670
+    },
+    {
+      "epoch": 0.07904798728358466,
+      "grad_norm": 0.692308880241196,
+      "learning_rate": 1.9760972428680304e-05,
+      "loss": 0.8506,
+      "step": 3680
+    },
+    {
+      "epoch": 0.07926279159685526,
+      "grad_norm": 0.6856631590673132,
+      "learning_rate": 1.975948870028178e-05,
+      "loss": 0.8686,
+      "step": 3690
+    },
+    {
+      "epoch": 0.07947759591012588,
+      "grad_norm": 0.6903089329869376,
+      "learning_rate": 1.9758000437157058e-05,
+      "loss": 0.8483,
+      "step": 3700
+    },
+    {
+      "epoch": 0.07969240022339649,
+      "grad_norm": 0.7078943762114845,
+      "learning_rate": 1.9756507639997663e-05,
+      "loss": 0.855,
+      "step": 3710
+    },
+    {
+      "epoch": 0.0799072045366671,
+      "grad_norm": 0.694778958880388,
+      "learning_rate": 1.975501030949721e-05,
+      "loss": 0.8738,
+      "step": 3720
+    },
+    {
+      "epoch": 0.08012200884993771,
+      "grad_norm": 0.7025809779085943,
+      "learning_rate": 1.975350844635144e-05,
+      "loss": 0.8455,
+      "step": 3730
+    },
+    {
+      "epoch": 0.08033681316320831,
+      "grad_norm": 0.7689670952873736,
+      "learning_rate": 1.9752002051258187e-05,
+      "loss": 0.8553,
+      "step": 3740
+    },
+    {
+      "epoch": 0.08055161747647893,
+      "grad_norm": 0.6448632769568776,
+      "learning_rate": 1.9750491124917396e-05,
+      "loss": 0.847,
+      "step": 3750
+    },
+    {
+      "epoch": 0.08076642178974953,
+      "grad_norm": 0.6613202041245299,
+      "learning_rate": 1.9748975668031113e-05,
+      "loss": 0.8514,
+      "step": 3760
+    },
+    {
+      "epoch": 0.08098122610302015,
+      "grad_norm": 0.6795441486901408,
+      "learning_rate": 1.974745568130349e-05,
+      "loss": 0.8551,
+      "step": 3770
+    },
+    {
+      "epoch": 0.08119603041629075,
+      "grad_norm": 0.6665715977606939,
+      "learning_rate": 1.974593116544079e-05,
+      "loss": 0.8464,
+      "step": 3780
+    },
+    {
+      "epoch": 0.08141083472956137,
+      "grad_norm": 0.6572184682125283,
+      "learning_rate": 1.974440212115138e-05,
+      "loss": 0.8535,
+      "step": 3790
+    },
+    {
+      "epoch": 0.08162563904283197,
+      "grad_norm": 0.6552167641764182,
+      "learning_rate": 1.9742868549145716e-05,
+      "loss": 0.8432,
+      "step": 3800
+    },
+    {
+      "epoch": 0.08184044335610259,
+      "grad_norm": 0.6421489707631659,
+      "learning_rate": 1.9741330450136377e-05,
+      "loss": 0.849,
+      "step": 3810
+    },
+    {
+      "epoch": 0.08205524766937321,
+      "grad_norm": 0.6485917266184518,
+      "learning_rate": 1.9739787824838036e-05,
+      "loss": 0.838,
+      "step": 3820
+    },
+    {
+      "epoch": 0.08227005198264381,
+      "grad_norm": 0.6959055244885545,
+      "learning_rate": 1.9738240673967473e-05,
+      "loss": 0.8473,
+      "step": 3830
+    },
+    {
+      "epoch": 0.08248485629591443,
+      "grad_norm": 0.713164987686896,
+      "learning_rate": 1.9736688998243562e-05,
+      "loss": 0.8528,
+      "step": 3840
+    },
+    {
+      "epoch": 0.08269966060918503,
+      "grad_norm": 0.6580917396986089,
+      "learning_rate": 1.9735132798387294e-05,
+      "loss": 0.852,
+      "step": 3850
+    },
+    {
+      "epoch": 0.08291446492245565,
+      "grad_norm": 0.7027948387761466,
+      "learning_rate": 1.9733572075121746e-05,
+      "loss": 0.8447,
+      "step": 3860
+    },
+    {
+      "epoch": 0.08312926923572625,
+      "grad_norm": 0.6861004251959052,
+      "learning_rate": 1.973200682917211e-05,
+      "loss": 0.8526,
+      "step": 3870
+    },
+    {
+      "epoch": 0.08334407354899687,
+      "grad_norm": 0.6869412094000382,
+      "learning_rate": 1.9730437061265674e-05,
+      "loss": 0.8416,
+      "step": 3880
+    },
+    {
+      "epoch": 0.08355887786226747,
+      "grad_norm": 0.6269537456004645,
+      "learning_rate": 1.9728862772131822e-05,
+      "loss": 0.8487,
+      "step": 3890
+    },
+    {
+      "epoch": 0.08377368217553809,
+      "grad_norm": 0.6491403415344045,
+      "learning_rate": 1.9727283962502054e-05,
+      "loss": 0.8678,
+      "step": 3900
+    },
+    {
+      "epoch": 0.08398848648880869,
+      "grad_norm": 0.6995859598956509,
+      "learning_rate": 1.9725700633109955e-05,
+      "loss": 0.8433,
+      "step": 3910
+    },
+    {
+      "epoch": 0.0842032908020793,
+      "grad_norm": 0.6700202312183003,
+      "learning_rate": 1.9724112784691213e-05,
+      "loss": 0.847,
+      "step": 3920
+    },
+    {
+      "epoch": 0.08441809511534992,
+      "grad_norm": 0.6485370618480597,
+      "learning_rate": 1.9722520417983618e-05,
+      "loss": 0.8544,
+      "step": 3930
+    },
+    {
+      "epoch": 0.08463289942862053,
+      "grad_norm": 0.6592424429265452,
+      "learning_rate": 1.972092353372707e-05,
+      "loss": 0.837,
+      "step": 3940
+    },
+    {
+      "epoch": 0.08484770374189114,
+      "grad_norm": 0.6555177826724199,
+      "learning_rate": 1.9719322132663547e-05,
+      "loss": 0.8654,
+      "step": 3950
+    },
+    {
+      "epoch": 0.08506250805516175,
+      "grad_norm": 0.6836389096207424,
+      "learning_rate": 1.9717716215537145e-05,
+      "loss": 0.8626,
+      "step": 3960
+    },
+    {
+      "epoch": 0.08527731236843236,
+      "grad_norm": 0.6336266324638282,
+      "learning_rate": 1.971610578309404e-05,
+      "loss": 0.8488,
+      "step": 3970
+    },
+    {
+      "epoch": 0.08549211668170296,
+      "grad_norm": 0.6198253547328094,
+      "learning_rate": 1.9714490836082527e-05,
+      "loss": 0.8606,
+      "step": 3980
+    },
+    {
+      "epoch": 0.08570692099497358,
+      "grad_norm": 0.6537044949678656,
+      "learning_rate": 1.9712871375252983e-05,
+      "loss": 0.8457,
+      "step": 3990
+    },
+    {
+      "epoch": 0.08592172530824418,
+      "grad_norm": 0.6678736312065985,
+      "learning_rate": 1.9711247401357886e-05,
+      "loss": 0.835,
+      "step": 4000
+    },
+    {
+      "epoch": 0.0861365296215148,
+      "grad_norm": 0.6692336566643887,
+      "learning_rate": 1.970961891515181e-05,
+      "loss": 0.8566,
+      "step": 4010
+    },
+    {
+      "epoch": 0.0863513339347854,
+      "grad_norm": 0.6585733096564904,
+      "learning_rate": 1.9707985917391435e-05,
+      "loss": 0.8339,
+      "step": 4020
+    },
+    {
+      "epoch": 0.08656613824805602,
+      "grad_norm": 0.6303363377268884,
+      "learning_rate": 1.9706348408835522e-05,
+      "loss": 0.8593,
+      "step": 4030
+    },
+    {
+      "epoch": 0.08678094256132664,
+      "grad_norm": 0.6782552039693405,
+      "learning_rate": 1.9704706390244943e-05,
+      "loss": 0.8601,
+      "step": 4040
+    },
+    {
+      "epoch": 0.08699574687459724,
+      "grad_norm": 0.6324566209058814,
+      "learning_rate": 1.970305986238265e-05,
+      "loss": 0.8513,
+      "step": 4050
+    },
+    {
+      "epoch": 0.08721055118786786,
+      "grad_norm": 0.6534333164738351,
+      "learning_rate": 1.970140882601371e-05,
+      "loss": 0.8526,
+      "step": 4060
+    },
+    {
+      "epoch": 0.08742535550113846,
+      "grad_norm": 0.6319924406902253,
+      "learning_rate": 1.9699753281905266e-05,
+      "loss": 0.8432,
+      "step": 4070
+    },
+    {
+      "epoch": 0.08764015981440908,
+      "grad_norm": 0.6523877824199336,
+      "learning_rate": 1.9698093230826566e-05,
+      "loss": 0.8472,
+      "step": 4080
+    },
+    {
+      "epoch": 0.08785496412767968,
+      "grad_norm": 0.6485572561784627,
+      "learning_rate": 1.9696428673548948e-05,
+      "loss": 0.8581,
+      "step": 4090
+    },
+    {
+      "epoch": 0.0880697684409503,
+      "grad_norm": 0.6747875245775294,
+      "learning_rate": 1.969475961084584e-05,
+      "loss": 0.8268,
+      "step": 4100
+    },
+    {
+      "epoch": 0.0882845727542209,
+      "grad_norm": 0.6183565239941164,
+      "learning_rate": 1.9693086043492778e-05,
+      "loss": 0.8336,
+      "step": 4110
+    },
+    {
+      "epoch": 0.08849937706749152,
+      "grad_norm": 0.6659244460477072,
+      "learning_rate": 1.9691407972267377e-05,
+      "loss": 0.8718,
+      "step": 4120
+    },
+    {
+      "epoch": 0.08871418138076212,
+      "grad_norm": 0.6537327783323973,
+      "learning_rate": 1.968972539794935e-05,
+      "loss": 0.8369,
+      "step": 4130
+    },
+    {
+      "epoch": 0.08892898569403274,
+      "grad_norm": 0.7320247153435835,
+      "learning_rate": 1.96880383213205e-05,
+      "loss": 0.8539,
+      "step": 4140
+    },
+    {
+      "epoch": 0.08914379000730335,
+      "grad_norm": 0.6260363637735203,
+      "learning_rate": 1.9686346743164726e-05,
+      "loss": 0.847,
+      "step": 4150
+    },
+    {
+      "epoch": 0.08935859432057396,
+      "grad_norm": 0.65678546506254,
+      "learning_rate": 1.9684650664268016e-05,
+      "loss": 0.8428,
+      "step": 4160
+    },
+    {
+      "epoch": 0.08957339863384457,
+      "grad_norm": 0.6552849199787741,
+      "learning_rate": 1.9682950085418446e-05,
+      "loss": 0.8336,
+      "step": 4170
+    },
+    {
+      "epoch": 0.08978820294711518,
+      "grad_norm": 0.6302616145397817,
+      "learning_rate": 1.9681245007406192e-05,
+      "loss": 0.8452,
+      "step": 4180
+    },
+    {
+      "epoch": 0.09000300726038579,
+      "grad_norm": 0.664218834980882,
+      "learning_rate": 1.9679535431023512e-05,
+      "loss": 0.8528,
+      "step": 4190
+    },
+    {
+      "epoch": 0.0902178115736564,
+      "grad_norm": 0.6382635022742835,
+      "learning_rate": 1.9677821357064758e-05,
+      "loss": 0.837,
+      "step": 4200
+    },
+    {
+      "epoch": 0.09043261588692701,
+      "grad_norm": 0.6287924929580544,
+      "learning_rate": 1.967610278632637e-05,
+      "loss": 0.8456,
+      "step": 4210
+    },
+    {
+      "epoch": 0.09064742020019761,
+      "grad_norm": 0.6735082120990712,
+      "learning_rate": 1.9674379719606874e-05,
+      "loss": 0.8307,
+      "step": 4220
+    },
+    {
+      "epoch": 0.09086222451346823,
+      "grad_norm": 0.6377012982256299,
+      "learning_rate": 1.9672652157706897e-05,
+      "loss": 0.8483,
+      "step": 4230
+    },
+    {
+      "epoch": 0.09107702882673883,
+      "grad_norm": 0.6593721140765066,
+      "learning_rate": 1.9670920101429142e-05,
+      "loss": 0.8441,
+      "step": 4240
+    },
+    {
+      "epoch": 0.09129183314000945,
+      "grad_norm": 0.674319812375293,
+      "learning_rate": 1.9669183551578414e-05,
+      "loss": 0.8243,
+      "step": 4250
+    },
+    {
+      "epoch": 0.09150663745328007,
+      "grad_norm": 0.627013344868889,
+      "learning_rate": 1.9667442508961585e-05,
+      "loss": 0.8425,
+      "step": 4260
+    },
+    {
+      "epoch": 0.09172144176655067,
+      "grad_norm": 0.6958437594298889,
+      "learning_rate": 1.9665696974387633e-05,
+      "loss": 0.8526,
+      "step": 4270
+    },
+    {
+      "epoch": 0.09193624607982129,
+      "grad_norm": 0.6044037023198737,
+      "learning_rate": 1.966394694866762e-05,
+      "loss": 0.8533,
+      "step": 4280
+    },
+    {
+      "epoch": 0.09215105039309189,
+      "grad_norm": 0.6542299584474508,
+      "learning_rate": 1.9662192432614683e-05,
+      "loss": 0.8565,
+      "step": 4290
+    },
+    {
+      "epoch": 0.09236585470636251,
+      "grad_norm": 0.6390677549567899,
+      "learning_rate": 1.9660433427044064e-05,
+      "loss": 0.8494,
+      "step": 4300
+    },
+    {
+      "epoch": 0.09258065901963311,
+      "grad_norm": 0.679504215836474,
+      "learning_rate": 1.9658669932773072e-05,
+      "loss": 0.8587,
+      "step": 4310
+    },
+    {
+      "epoch": 0.09279546333290373,
+      "grad_norm": 0.6627368866357531,
+      "learning_rate": 1.965690195062112e-05,
+      "loss": 0.8355,
+      "step": 4320
+    },
+    {
+      "epoch": 0.09301026764617433,
+      "grad_norm": 0.6349256488418371,
+      "learning_rate": 1.9655129481409695e-05,
+      "loss": 0.8476,
+      "step": 4330
+    },
+    {
+      "epoch": 0.09322507195944495,
+      "grad_norm": 0.6397326363467064,
+      "learning_rate": 1.9653352525962363e-05,
+      "loss": 0.8424,
+      "step": 4340
+    },
+    {
+      "epoch": 0.09343987627271555,
+      "grad_norm": 0.6437203998695848,
+      "learning_rate": 1.9651571085104796e-05,
+      "loss": 0.8575,
+      "step": 4350
+    },
+    {
+      "epoch": 0.09365468058598617,
+      "grad_norm": 0.6296219759406441,
+      "learning_rate": 1.9649785159664723e-05,
+      "loss": 0.8444,
+      "step": 4360
+    },
+    {
+      "epoch": 0.09386948489925678,
+      "grad_norm": 0.6364710421208644,
+      "learning_rate": 1.964799475047198e-05,
+      "loss": 0.8409,
+      "step": 4370
+    },
+    {
+      "epoch": 0.09408428921252739,
+      "grad_norm": 0.664099321204506,
+      "learning_rate": 1.9646199858358475e-05,
+      "loss": 0.8537,
+      "step": 4380
+    },
+    {
+      "epoch": 0.094299093525798,
+      "grad_norm": 0.6145796569330872,
+      "learning_rate": 1.96444004841582e-05,
+      "loss": 0.835,
+      "step": 4390
+    },
+    {
+      "epoch": 0.0945138978390686,
+      "grad_norm": 0.6258749417882592,
+      "learning_rate": 1.964259662870723e-05,
+      "loss": 0.8423,
+      "step": 4400
+    },
+    {
+      "epoch": 0.09472870215233922,
+      "grad_norm": 0.6768124086471206,
+      "learning_rate": 1.9640788292843722e-05,
+      "loss": 0.8526,
+      "step": 4410
+    },
+    {
+      "epoch": 0.09494350646560983,
+      "grad_norm": 0.6679825227131061,
+      "learning_rate": 1.963897547740792e-05,
+      "loss": 0.8593,
+      "step": 4420
+    },
+    {
+      "epoch": 0.09515831077888044,
+      "grad_norm": 0.6708318644795358,
+      "learning_rate": 1.9637158183242138e-05,
+      "loss": 0.8403,
+      "step": 4430
+    },
+    {
+      "epoch": 0.09537311509215105,
+      "grad_norm": 0.6182453595533811,
+      "learning_rate": 1.9635336411190786e-05,
+      "loss": 0.8464,
+      "step": 4440
+    },
+    {
+      "epoch": 0.09558791940542166,
+      "grad_norm": 0.6518263259411956,
+      "learning_rate": 1.963351016210034e-05,
+      "loss": 0.8406,
+      "step": 4450
+    },
+    {
+      "epoch": 0.09580272371869226,
+      "grad_norm": 0.6695930906879859,
+      "learning_rate": 1.9631679436819363e-05,
+      "loss": 0.8461,
+      "step": 4460
+    },
+    {
+      "epoch": 0.09601752803196288,
+      "grad_norm": 0.6609509728832942,
+      "learning_rate": 1.9629844236198502e-05,
+      "loss": 0.8533,
+      "step": 4470
+    },
+    {
+      "epoch": 0.0962323323452335,
+      "grad_norm": 0.6657143200018495,
+      "learning_rate": 1.9628004561090474e-05,
+      "loss": 0.8357,
+      "step": 4480
+    },
+    {
+      "epoch": 0.0964471366585041,
+      "grad_norm": 0.6630542341667736,
+      "learning_rate": 1.9626160412350085e-05,
+      "loss": 0.8322,
+      "step": 4490
+    },
+    {
+      "epoch": 0.09666194097177472,
+      "grad_norm": 0.6314767253481299,
+      "learning_rate": 1.9624311790834217e-05,
+      "loss": 0.8385,
+      "step": 4500
+    },
+    {
+      "epoch": 0.09687674528504532,
+      "grad_norm": 0.6903306283162439,
+      "learning_rate": 1.962245869740182e-05,
+      "loss": 0.8557,
+      "step": 4510
+    },
+    {
+      "epoch": 0.09709154959831594,
+      "grad_norm": 0.669279047897614,
+      "learning_rate": 1.9620601132913936e-05,
+      "loss": 0.8552,
+      "step": 4520
+    },
+    {
+      "epoch": 0.09730635391158654,
+      "grad_norm": 0.6814182363619116,
+      "learning_rate": 1.9618739098233676e-05,
+      "loss": 0.8416,
+      "step": 4530
+    },
+    {
+      "epoch": 0.09752115822485716,
+      "grad_norm": 0.6462039533634677,
+      "learning_rate": 1.9616872594226232e-05,
+      "loss": 0.8443,
+      "step": 4540
+    },
+    {
+      "epoch": 0.09773596253812776,
+      "grad_norm": 0.6397436767943191,
+      "learning_rate": 1.9615001621758867e-05,
+      "loss": 0.8401,
+      "step": 4550
+    },
+    {
+      "epoch": 0.09795076685139838,
+      "grad_norm": 0.661032708042101,
+      "learning_rate": 1.9613126181700932e-05,
+      "loss": 0.8313,
+      "step": 4560
+    },
+    {
+      "epoch": 0.09816557116466898,
+      "grad_norm": 0.6481039328379562,
+      "learning_rate": 1.9611246274923844e-05,
+      "loss": 0.834,
+      "step": 4570
+    },
+    {
+      "epoch": 0.0983803754779396,
+      "grad_norm": 0.6747780419159984,
+      "learning_rate": 1.960936190230109e-05,
+      "loss": 0.8326,
+      "step": 4580
+    },
+    {
+      "epoch": 0.09859517979121021,
+      "grad_norm": 0.661376547720958,
+      "learning_rate": 1.960747306470825e-05,
+      "loss": 0.8294,
+      "step": 4590
+    },
+    {
+      "epoch": 0.09880998410448082,
+      "grad_norm": 0.650547641584137,
+      "learning_rate": 1.9605579763022966e-05,
+      "loss": 0.8426,
+      "step": 4600
+    },
+    {
+      "epoch": 0.09902478841775143,
+      "grad_norm": 0.6682948635311629,
+      "learning_rate": 1.960368199812495e-05,
+      "loss": 0.8419,
+      "step": 4610
+    },
+    {
+      "epoch": 0.09923959273102204,
+      "grad_norm": 0.6467278134601648,
+      "learning_rate": 1.9601779770896007e-05,
+      "loss": 0.8293,
+      "step": 4620
+    },
+    {
+      "epoch": 0.09945439704429265,
+      "grad_norm": 0.6325684184789404,
+      "learning_rate": 1.9599873082219992e-05,
+      "loss": 0.8338,
+      "step": 4630
+    },
+    {
+      "epoch": 0.09966920135756326,
+      "grad_norm": 0.6688907540062,
+      "learning_rate": 1.9597961932982845e-05,
+      "loss": 0.8363,
+      "step": 4640
+    },
+    {
+      "epoch": 0.09988400567083387,
+      "grad_norm": 0.654845343648715,
+      "learning_rate": 1.9596046324072586e-05,
+      "loss": 0.8416,
+      "step": 4650
+    },
+    {
+      "epoch": 0.10009880998410448,
+      "grad_norm": 0.6320139560546627,
+      "learning_rate": 1.9594126256379286e-05,
+      "loss": 0.863,
+      "step": 4660
+    },
+    {
+      "epoch": 0.10031361429737509,
+      "grad_norm": 0.6691994371938742,
+      "learning_rate": 1.9592201730795112e-05,
+      "loss": 0.8456,
+      "step": 4670
+    },
+    {
+      "epoch": 0.1005284186106457,
+      "grad_norm": 0.6606839474988468,
+      "learning_rate": 1.9590272748214283e-05,
+      "loss": 0.8302,
+      "step": 4680
+    },
+    {
+      "epoch": 0.10074322292391631,
+      "grad_norm": 0.6262068989715459,
+      "learning_rate": 1.9588339309533103e-05,
+      "loss": 0.8377,
+      "step": 4690
+    },
+    {
+      "epoch": 0.10095802723718693,
+      "grad_norm": 0.6243559896905192,
+      "learning_rate": 1.9586401415649935e-05,
+      "loss": 0.8396,
+      "step": 4700
+    },
+    {
+      "epoch": 0.10117283155045753,
+      "grad_norm": 0.6687990805968839,
+      "learning_rate": 1.958445906746522e-05,
+      "loss": 0.8354,
+      "step": 4710
+    },
+    {
+      "epoch": 0.10138763586372815,
+      "grad_norm": 0.6310170587812323,
+      "learning_rate": 1.9582512265881467e-05,
+      "loss": 0.8211,
+      "step": 4720
+    },
+    {
+      "epoch": 0.10160244017699875,
+      "grad_norm": 0.6388753086935843,
+      "learning_rate": 1.958056101180325e-05,
+      "loss": 0.8441,
+      "step": 4730
+    },
+    {
+      "epoch": 0.10181724449026937,
+      "grad_norm": 0.6638753058560826,
+      "learning_rate": 1.9578605306137216e-05,
+      "loss": 0.8358,
+      "step": 4740
+    },
+    {
+      "epoch": 0.10203204880353997,
+      "grad_norm": 0.6302752971289505,
+      "learning_rate": 1.9576645149792083e-05,
+      "loss": 0.8395,
+      "step": 4750
+    },
+    {
+      "epoch": 0.10224685311681059,
+      "grad_norm": 0.6458064089587059,
+      "learning_rate": 1.957468054367863e-05,
+      "loss": 0.8401,
+      "step": 4760
+    },
+    {
+      "epoch": 0.10246165743008119,
+      "grad_norm": 0.646382816972308,
+      "learning_rate": 1.957271148870971e-05,
+      "loss": 0.8435,
+      "step": 4770
+    },
+    {
+      "epoch": 0.10267646174335181,
+      "grad_norm": 0.6155538183074262,
+      "learning_rate": 1.9570737985800237e-05,
+      "loss": 0.8306,
+      "step": 4780
+    },
+    {
+      "epoch": 0.10289126605662241,
+      "grad_norm": 0.6367578606631397,
+      "learning_rate": 1.95687600358672e-05,
+      "loss": 0.8369,
+      "step": 4790
+    },
+    {
+      "epoch": 0.10310607036989303,
+      "grad_norm": 0.6404541919457326,
+      "learning_rate": 1.956677763982964e-05,
+      "loss": 0.8254,
+      "step": 4800
+    },
+    {
+      "epoch": 0.10332087468316364,
+      "grad_norm": 0.6215108376628148,
+      "learning_rate": 1.9564790798608682e-05,
+      "loss": 0.8584,
+      "step": 4810
+    },
+    {
+      "epoch": 0.10353567899643425,
+      "grad_norm": 0.6338570435141742,
+      "learning_rate": 1.9562799513127507e-05,
+      "loss": 0.8324,
+      "step": 4820
+    },
+    {
+      "epoch": 0.10375048330970486,
+      "grad_norm": 0.6019370576161179,
+      "learning_rate": 1.956080378431136e-05,
+      "loss": 0.8559,
+      "step": 4830
+    },
+    {
+      "epoch": 0.10396528762297547,
+      "grad_norm": 0.6297162423533607,
+      "learning_rate": 1.9558803613087548e-05,
+      "loss": 0.8462,
+      "step": 4840
+    },
+    {
+      "epoch": 0.10418009193624608,
+      "grad_norm": 0.6045449219846318,
+      "learning_rate": 1.9556799000385454e-05,
+      "loss": 0.8386,
+      "step": 4850
+    },
+    {
+      "epoch": 0.10439489624951669,
+      "grad_norm": 0.6549861670806117,
+      "learning_rate": 1.9554789947136508e-05,
+      "loss": 0.8306,
+      "step": 4860
+    },
+    {
+      "epoch": 0.1046097005627873,
+      "grad_norm": 0.6630879044217777,
+      "learning_rate": 1.9552776454274223e-05,
+      "loss": 0.8314,
+      "step": 4870
+    },
+    {
+      "epoch": 0.1048245048760579,
+      "grad_norm": 0.651950768006272,
+      "learning_rate": 1.955075852273416e-05,
+      "loss": 0.8385,
+      "step": 4880
+    },
+    {
+      "epoch": 0.10503930918932852,
+      "grad_norm": 0.6626536087969357,
+      "learning_rate": 1.9548736153453943e-05,
+      "loss": 0.8363,
+      "step": 4890
+    },
+    {
+      "epoch": 0.10525411350259913,
+      "grad_norm": 0.6886579304677174,
+      "learning_rate": 1.9546709347373265e-05,
+      "loss": 0.8499,
+      "step": 4900
+    },
+    {
+      "epoch": 0.10546891781586974,
+      "grad_norm": 0.6312997525023037,
+      "learning_rate": 1.954467810543388e-05,
+      "loss": 0.8235,
+      "step": 4910
+    },
+    {
+      "epoch": 0.10568372212914036,
+      "grad_norm": 0.6365142217446823,
+      "learning_rate": 1.954264242857959e-05,
+      "loss": 0.8356,
+      "step": 4920
+    },
+    {
+      "epoch": 0.10589852644241096,
+      "grad_norm": 0.6746025653030885,
+      "learning_rate": 1.954060231775628e-05,
+      "loss": 0.8528,
+      "step": 4930
+    },
+    {
+      "epoch": 0.10611333075568158,
+      "grad_norm": 0.6281429344046033,
+      "learning_rate": 1.9538557773911878e-05,
+      "loss": 0.8473,
+      "step": 4940
+    },
+    {
+      "epoch": 0.10632813506895218,
+      "grad_norm": 0.6505897210485625,
+      "learning_rate": 1.9536508797996377e-05,
+      "loss": 0.8377,
+      "step": 4950
+    },
+    {
+      "epoch": 0.1065429393822228,
+      "grad_norm": 0.6758994881928211,
+      "learning_rate": 1.953445539096183e-05,
+      "loss": 0.8544,
+      "step": 4960
+    },
+    {
+      "epoch": 0.1067577436954934,
+      "grad_norm": 0.6285712244694984,
+      "learning_rate": 1.953239755376235e-05,
+      "loss": 0.8544,
+      "step": 4970
+    },
+    {
+      "epoch": 0.10697254800876402,
+      "grad_norm": 0.6210838565177652,
+      "learning_rate": 1.9530335287354102e-05,
+      "loss": 0.8304,
+      "step": 4980
+    },
+    {
+      "epoch": 0.10718735232203462,
+      "grad_norm": 0.6140776582373115,
+      "learning_rate": 1.952826859269532e-05,
+      "loss": 0.836,
+      "step": 4990
+    },
+    {
+      "epoch": 0.10740215663530524,
+      "grad_norm": 0.6671326709180452,
+      "learning_rate": 1.9526197470746283e-05,
+      "loss": 0.8343,
+      "step": 5000
+    },
+    {
+      "epoch": 0.10761696094857584,
+      "grad_norm": 0.6597003360765735,
+      "learning_rate": 1.9524121922469338e-05,
+      "loss": 0.8652,
+      "step": 5010
+    },
+    {
+      "epoch": 0.10783176526184646,
+      "grad_norm": 0.5956926521621552,
+      "learning_rate": 1.952204194882888e-05,
+      "loss": 0.8262,
+      "step": 5020
+    },
+    {
+      "epoch": 0.10804656957511707,
+      "grad_norm": 0.7019664111760795,
+      "learning_rate": 1.9519957550791372e-05,
+      "loss": 0.8429,
+      "step": 5030
+    },
+    {
+      "epoch": 0.10826137388838768,
+      "grad_norm": 0.6304099214813407,
+      "learning_rate": 1.951786872932532e-05,
+      "loss": 0.8393,
+      "step": 5040
+    },
+    {
+      "epoch": 0.1084761782016583,
+      "grad_norm": 0.6463172161616285,
+      "learning_rate": 1.951577548540129e-05,
+      "loss": 0.8387,
+      "step": 5050
+    },
+    {
+      "epoch": 0.1086909825149289,
+      "grad_norm": 0.6502922763327477,
+      "learning_rate": 1.9513677819991905e-05,
+      "loss": 0.835,
+      "step": 5060
+    },
+    {
+      "epoch": 0.10890578682819951,
+      "grad_norm": 0.618888528634145,
+      "learning_rate": 1.951157573407184e-05,
+      "loss": 0.8559,
+      "step": 5070
+    },
+    {
+      "epoch": 0.10912059114147012,
+      "grad_norm": 0.6374773953158367,
+      "learning_rate": 1.9509469228617827e-05,
+      "loss": 0.8317,
+      "step": 5080
+    },
+    {
+      "epoch": 0.10933539545474073,
+      "grad_norm": 0.6303019884270492,
+      "learning_rate": 1.9507358304608644e-05,
+      "loss": 0.8303,
+      "step": 5090
+    },
+    {
+      "epoch": 0.10955019976801134,
+      "grad_norm": 0.6561252982950152,
+      "learning_rate": 1.9505242963025133e-05,
+      "loss": 0.8273,
+      "step": 5100
+    },
+    {
+      "epoch": 0.10976500408128195,
+      "grad_norm": 0.6290165189842113,
+      "learning_rate": 1.9503123204850184e-05,
+      "loss": 0.8421,
+      "step": 5110
+    },
+    {
+      "epoch": 0.10997980839455257,
+      "grad_norm": 0.617329955446763,
+      "learning_rate": 1.9500999031068734e-05,
+      "loss": 0.8232,
+      "step": 5120
+    },
+    {
+      "epoch": 0.11019461270782317,
+      "grad_norm": 0.6107026109474121,
+      "learning_rate": 1.949887044266778e-05,
+      "loss": 0.835,
+      "step": 5130
+    },
+    {
+      "epoch": 0.11040941702109379,
+      "grad_norm": 0.6269967944152154,
+      "learning_rate": 1.949673744063636e-05,
+      "loss": 0.8305,
+      "step": 5140
+    },
+    {
+      "epoch": 0.11062422133436439,
+      "grad_norm": 0.6360564673651822,
+      "learning_rate": 1.949460002596557e-05,
+      "loss": 0.832,
+      "step": 5150
+    },
+    {
+      "epoch": 0.11083902564763501,
+      "grad_norm": 0.6328470911946484,
+      "learning_rate": 1.9492458199648564e-05,
+      "loss": 0.8273,
+      "step": 5160
+    },
+    {
+      "epoch": 0.11105382996090561,
+      "grad_norm": 0.880034167436241,
+      "learning_rate": 1.949031196268053e-05,
+      "loss": 0.8422,
+      "step": 5170
+    },
+    {
+      "epoch": 0.11126863427417623,
+      "grad_norm": 0.6718310412202478,
+      "learning_rate": 1.948816131605871e-05,
+      "loss": 0.8444,
+      "step": 5180
+    },
+    {
+      "epoch": 0.11148343858744683,
+      "grad_norm": 0.6443011347377299,
+      "learning_rate": 1.9486006260782406e-05,
+      "loss": 0.8365,
+      "step": 5190
+    },
+    {
+      "epoch": 0.11169824290071745,
+      "grad_norm": 0.6183046355489557,
+      "learning_rate": 1.948384679785295e-05,
+      "loss": 0.8545,
+      "step": 5200
+    },
+    {
+      "epoch": 0.11191304721398805,
+      "grad_norm": 0.6251393760951905,
+      "learning_rate": 1.9481682928273738e-05,
+      "loss": 0.8268,
+      "step": 5210
+    },
+    {
+      "epoch": 0.11212785152725867,
+      "grad_norm": 0.6344101557996115,
+      "learning_rate": 1.9479514653050212e-05,
+      "loss": 0.8404,
+      "step": 5220
+    },
+    {
+      "epoch": 0.11234265584052928,
+      "grad_norm": 0.5860944916243113,
+      "learning_rate": 1.9477341973189844e-05,
+      "loss": 0.8355,
+      "step": 5230
+    },
+    {
+      "epoch": 0.11255746015379989,
+      "grad_norm": 0.6186996808539741,
+      "learning_rate": 1.9475164889702175e-05,
+      "loss": 0.8479,
+      "step": 5240
+    },
+    {
+      "epoch": 0.1127722644670705,
+      "grad_norm": 0.6034601205701468,
+      "learning_rate": 1.9472983403598783e-05,
+      "loss": 0.8277,
+      "step": 5250
+    },
+    {
+      "epoch": 0.11298706878034111,
+      "grad_norm": 0.6440643260727591,
+      "learning_rate": 1.9470797515893286e-05,
+      "loss": 0.8376,
+      "step": 5260
+    },
+    {
+      "epoch": 0.11320187309361172,
+      "grad_norm": 0.6324744223203518,
+      "learning_rate": 1.946860722760135e-05,
+      "loss": 0.8311,
+      "step": 5270
+    },
+    {
+      "epoch": 0.11341667740688233,
+      "grad_norm": 0.618379973907413,
+      "learning_rate": 1.9466412539740697e-05,
+      "loss": 0.8287,
+      "step": 5280
+    },
+    {
+      "epoch": 0.11363148172015294,
+      "grad_norm": 0.6256736960626628,
+      "learning_rate": 1.9464213453331076e-05,
+      "loss": 0.8384,
+      "step": 5290
+    },
+    {
+      "epoch": 0.11384628603342355,
+      "grad_norm": 0.607811975191907,
+      "learning_rate": 1.9462009969394292e-05,
+      "loss": 0.8439,
+      "step": 5300
+    },
+    {
+      "epoch": 0.11406109034669416,
+      "grad_norm": 0.6214988694407934,
+      "learning_rate": 1.945980208895419e-05,
+      "loss": 0.8385,
+      "step": 5310
+    },
+    {
+      "epoch": 0.11427589465996477,
+      "grad_norm": 0.6260531019431299,
+      "learning_rate": 1.945758981303665e-05,
+      "loss": 0.8386,
+      "step": 5320
+    },
+    {
+      "epoch": 0.11449069897323538,
+      "grad_norm": 0.6510239330074841,
+      "learning_rate": 1.9455373142669615e-05,
+      "loss": 0.8155,
+      "step": 5330
+    },
+    {
+      "epoch": 0.114705503286506,
+      "grad_norm": 0.6114277545538802,
+      "learning_rate": 1.945315207888304e-05,
+      "loss": 0.8236,
+      "step": 5340
+    },
+    {
+      "epoch": 0.1149203075997766,
+      "grad_norm": 0.6290587853644128,
+      "learning_rate": 1.945092662270895e-05,
+      "loss": 0.8391,
+      "step": 5350
+    },
+    {
+      "epoch": 0.11513511191304722,
+      "grad_norm": 0.6252788540877645,
+      "learning_rate": 1.9448696775181393e-05,
+      "loss": 0.8302,
+      "step": 5360
+    },
+    {
+      "epoch": 0.11534991622631782,
+      "grad_norm": 0.6366366260996406,
+      "learning_rate": 1.9446462537336462e-05,
+      "loss": 0.832,
+      "step": 5370
+    },
+    {
+      "epoch": 0.11556472053958844,
+      "grad_norm": 0.6403035006648804,
+      "learning_rate": 1.9444223910212297e-05,
+      "loss": 0.8401,
+      "step": 5380
+    },
+    {
+      "epoch": 0.11577952485285904,
+      "grad_norm": 0.6471061463773111,
+      "learning_rate": 1.9441980894849068e-05,
+      "loss": 0.822,
+      "step": 5390
+    },
+    {
+      "epoch": 0.11599432916612966,
+      "grad_norm": 0.6485069683086401,
+      "learning_rate": 1.9439733492288986e-05,
+      "loss": 0.8327,
+      "step": 5400
+    },
+    {
+      "epoch": 0.11620913347940026,
+      "grad_norm": 0.6195424958899896,
+      "learning_rate": 1.9437481703576303e-05,
+      "loss": 0.827,
+      "step": 5410
+    },
+    {
+      "epoch": 0.11642393779267088,
+      "grad_norm": 0.6067271492911613,
+      "learning_rate": 1.9435225529757307e-05,
+      "loss": 0.8378,
+      "step": 5420
+    },
+    {
+      "epoch": 0.11663874210594148,
+      "grad_norm": 0.6333870629436046,
+      "learning_rate": 1.9432964971880325e-05,
+      "loss": 0.8274,
+      "step": 5430
+    },
+    {
+      "epoch": 0.1168535464192121,
+      "grad_norm": 0.6349621748512131,
+      "learning_rate": 1.9430700030995724e-05,
+      "loss": 0.8371,
+      "step": 5440
+    },
+    {
+      "epoch": 0.11706835073248271,
+      "grad_norm": 0.6132008334219502,
+      "learning_rate": 1.94284307081559e-05,
+      "loss": 0.8254,
+      "step": 5450
+    },
+    {
+      "epoch": 0.11728315504575332,
+      "grad_norm": 0.626720073182986,
+      "learning_rate": 1.942615700441529e-05,
+      "loss": 0.8296,
+      "step": 5460
+    },
+    {
+      "epoch": 0.11749795935902393,
+      "grad_norm": 0.6280587181806514,
+      "learning_rate": 1.9423878920830366e-05,
+      "loss": 0.8228,
+      "step": 5470
+    },
+    {
+      "epoch": 0.11771276367229454,
+      "grad_norm": 0.6514270560053967,
+      "learning_rate": 1.942159645845963e-05,
+      "loss": 0.8262,
+      "step": 5480
+    },
+    {
+      "epoch": 0.11792756798556515,
+      "grad_norm": 0.6130650794491808,
+      "learning_rate": 1.9419309618363637e-05,
+      "loss": 0.8379,
+      "step": 5490
+    },
+    {
+      "epoch": 0.11814237229883576,
+      "grad_norm": 0.637159008327422,
+      "learning_rate": 1.9417018401604947e-05,
+      "loss": 0.8282,
+      "step": 5500
+    },
+    {
+      "epoch": 0.11835717661210637,
+      "grad_norm": 0.6188423823463596,
+      "learning_rate": 1.9414722809248182e-05,
+      "loss": 0.8305,
+      "step": 5510
+    },
+    {
+      "epoch": 0.11857198092537698,
+      "grad_norm": 0.6245770269033344,
+      "learning_rate": 1.9412422842359976e-05,
+      "loss": 0.833,
+      "step": 5520
+    },
+    {
+      "epoch": 0.1187867852386476,
+      "grad_norm": 0.6365044259718952,
+      "learning_rate": 1.9410118502009003e-05,
+      "loss": 0.8388,
+      "step": 5530
+    },
+    {
+      "epoch": 0.1190015895519182,
+      "grad_norm": 0.6328160935271273,
+      "learning_rate": 1.9407809789265973e-05,
+      "loss": 0.845,
+      "step": 5540
+    },
+    {
+      "epoch": 0.11921639386518881,
+      "grad_norm": 0.630311910667844,
+      "learning_rate": 1.9405496705203628e-05,
+      "loss": 0.8345,
+      "step": 5550
+    },
+    {
+      "epoch": 0.11943119817845943,
+      "grad_norm": 0.638696910751977,
+      "learning_rate": 1.9403179250896733e-05,
+      "loss": 0.8416,
+      "step": 5560
+    },
+    {
+      "epoch": 0.11964600249173003,
+      "grad_norm": 0.6247248724186333,
+      "learning_rate": 1.940085742742209e-05,
+      "loss": 0.8331,
+      "step": 5570
+    },
+    {
+      "epoch": 0.11986080680500065,
+      "grad_norm": 0.6213341924856995,
+      "learning_rate": 1.9398531235858525e-05,
+      "loss": 0.8347,
+      "step": 5580
+    },
+    {
+      "epoch": 0.12007561111827125,
+      "grad_norm": 0.6121503915927947,
+      "learning_rate": 1.9396200677286907e-05,
+      "loss": 0.8323,
+      "step": 5590
+    },
+    {
+      "epoch": 0.12029041543154187,
+      "grad_norm": 0.6189647502053036,
+      "learning_rate": 1.939386575279012e-05,
+      "loss": 0.8457,
+      "step": 5600
+    },
+    {
+      "epoch": 0.12050521974481247,
+      "grad_norm": 0.6135493571505611,
+      "learning_rate": 1.939152646345308e-05,
+      "loss": 0.8368,
+      "step": 5610
+    },
+    {
+      "epoch": 0.12072002405808309,
+      "grad_norm": 0.6378689788750032,
+      "learning_rate": 1.9389182810362738e-05,
+      "loss": 0.8449,
+      "step": 5620
+    },
+    {
+      "epoch": 0.12093482837135369,
+      "grad_norm": 0.6294801432407113,
+      "learning_rate": 1.938683479460806e-05,
+      "loss": 0.8271,
+      "step": 5630
+    },
+    {
+      "epoch": 0.12114963268462431,
+      "grad_norm": 0.6328552710008448,
+      "learning_rate": 1.9384482417280056e-05,
+      "loss": 0.8262,
+      "step": 5640
+    },
+    {
+      "epoch": 0.12136443699789491,
+      "grad_norm": 0.6186605157387922,
+      "learning_rate": 1.9382125679471745e-05,
+      "loss": 0.8278,
+      "step": 5650
+    },
+    {
+      "epoch": 0.12157924131116553,
+      "grad_norm": 0.6334724170139352,
+      "learning_rate": 1.9379764582278185e-05,
+      "loss": 0.8284,
+      "step": 5660
+    },
+    {
+      "epoch": 0.12179404562443615,
+      "grad_norm": 0.6283387797940377,
+      "learning_rate": 1.9377399126796454e-05,
+      "loss": 0.8359,
+      "step": 5670
+    },
+    {
+      "epoch": 0.12200884993770675,
+      "grad_norm": 0.6055428009166745,
+      "learning_rate": 1.9375029314125658e-05,
+      "loss": 0.8267,
+      "step": 5680
+    },
+    {
+      "epoch": 0.12222365425097736,
+      "grad_norm": 0.6034275177420789,
+      "learning_rate": 1.9372655145366922e-05,
+      "loss": 0.8252,
+      "step": 5690
+    },
+    {
+      "epoch": 0.12243845856424797,
+      "grad_norm": 0.6100632464810285,
+      "learning_rate": 1.93702766216234e-05,
+      "loss": 0.8284,
+      "step": 5700
+    },
+    {
+      "epoch": 0.12265326287751858,
+      "grad_norm": 0.6654383228152203,
+      "learning_rate": 1.936789374400027e-05,
+      "loss": 0.8124,
+      "step": 5710
+    },
+    {
+      "epoch": 0.12286806719078919,
+      "grad_norm": 0.6202336240170405,
+      "learning_rate": 1.9365506513604725e-05,
+      "loss": 0.828,
+      "step": 5720
+    },
+    {
+      "epoch": 0.1230828715040598,
+      "grad_norm": 0.6181456688535198,
+      "learning_rate": 1.936311493154599e-05,
+      "loss": 0.8446,
+      "step": 5730
+    },
+    {
+      "epoch": 0.12329767581733041,
+      "grad_norm": 0.619325566527615,
+      "learning_rate": 1.9360718998935315e-05,
+      "loss": 0.835,
+      "step": 5740
+    },
+    {
+      "epoch": 0.12351248013060102,
+      "grad_norm": 0.6293306658393487,
+      "learning_rate": 1.9358318716885955e-05,
+      "loss": 0.8225,
+      "step": 5750
+    },
+    {
+      "epoch": 0.12372728444387163,
+      "grad_norm": 0.6231931933033569,
+      "learning_rate": 1.9355914086513205e-05,
+      "loss": 0.845,
+      "step": 5760
+    },
+    {
+      "epoch": 0.12394208875714224,
+      "grad_norm": 0.6298812123623225,
+      "learning_rate": 1.9353505108934363e-05,
+      "loss": 0.829,
+      "step": 5770
+    },
+    {
+      "epoch": 0.12415689307041286,
+      "grad_norm": 0.6200961201702063,
+      "learning_rate": 1.9351091785268762e-05,
+      "loss": 0.8269,
+      "step": 5780
+    },
+    {
+      "epoch": 0.12437169738368346,
+      "grad_norm": 0.6118430477057613,
+      "learning_rate": 1.9348674116637747e-05,
+      "loss": 0.8308,
+      "step": 5790
+    },
+    {
+      "epoch": 0.12458650169695408,
+      "grad_norm": 0.6127169022228522,
+      "learning_rate": 1.934625210416468e-05,
+      "loss": 0.8188,
+      "step": 5800
+    },
+    {
+      "epoch": 0.12480130601022468,
+      "grad_norm": 0.5858112550024495,
+      "learning_rate": 1.9343825748974946e-05,
+      "loss": 0.8225,
+      "step": 5810
+    },
+    {
+      "epoch": 0.1250161103234953,
+      "grad_norm": 0.6431362667704915,
+      "learning_rate": 1.9341395052195943e-05,
+      "loss": 0.8251,
+      "step": 5820
+    },
+    {
+      "epoch": 0.12523091463676592,
+      "grad_norm": 0.6177184812180734,
+      "learning_rate": 1.9338960014957094e-05,
+      "loss": 0.8117,
+      "step": 5830
+    },
+    {
+      "epoch": 0.1254457189500365,
+      "grad_norm": 7.261555237494985,
+      "learning_rate": 1.9336520638389828e-05,
+      "loss": 0.8424,
+      "step": 5840
+    },
+    {
+      "epoch": 0.12566052326330712,
+      "grad_norm": 0.6165137803342475,
+      "learning_rate": 1.9334076923627602e-05,
+      "loss": 0.8468,
+      "step": 5850
+    },
+    {
+      "epoch": 0.12587532757657774,
+      "grad_norm": 0.6652429602651283,
+      "learning_rate": 1.9331628871805882e-05,
+      "loss": 0.8241,
+      "step": 5860
+    },
+    {
+      "epoch": 0.12609013188984836,
+      "grad_norm": 0.5977417751851771,
+      "learning_rate": 1.9329176484062147e-05,
+      "loss": 0.815,
+      "step": 5870
+    },
+    {
+      "epoch": 0.12630493620311894,
+      "grad_norm": 0.6143551332297007,
+      "learning_rate": 1.9326719761535896e-05,
+      "loss": 0.829,
+      "step": 5880
+    },
+    {
+      "epoch": 0.12651974051638956,
+      "grad_norm": 0.5722771504606794,
+      "learning_rate": 1.932425870536864e-05,
+      "loss": 0.8276,
+      "step": 5890
+    },
+    {
+      "epoch": 0.12673454482966018,
+      "grad_norm": 0.6233204435595209,
+      "learning_rate": 1.9321793316703904e-05,
+      "loss": 0.8257,
+      "step": 5900
+    },
+    {
+      "epoch": 0.1269493491429308,
+      "grad_norm": 0.6060690392530359,
+      "learning_rate": 1.9319323596687226e-05,
+      "loss": 0.8361,
+      "step": 5910
+    },
+    {
+      "epoch": 0.1271641534562014,
+      "grad_norm": 0.6503234853479353,
+      "learning_rate": 1.9316849546466154e-05,
+      "loss": 0.8534,
+      "step": 5920
+    },
+    {
+      "epoch": 0.127378957769472,
+      "grad_norm": 0.6577873265981129,
+      "learning_rate": 1.931437116719025e-05,
+      "loss": 0.8263,
+      "step": 5930
+    },
+    {
+      "epoch": 0.12759376208274262,
+      "grad_norm": 0.6504948800648593,
+      "learning_rate": 1.9311888460011096e-05,
+      "loss": 0.8294,
+      "step": 5940
+    },
+    {
+      "epoch": 0.12780856639601323,
+      "grad_norm": 0.6317328887950812,
+      "learning_rate": 1.9309401426082263e-05,
+      "loss": 0.8313,
+      "step": 5950
+    },
+    {
+      "epoch": 0.12802337070928385,
+      "grad_norm": 0.6211031800695695,
+      "learning_rate": 1.9306910066559358e-05,
+      "loss": 0.8355,
+      "step": 5960
+    },
+    {
+      "epoch": 0.12823817502255444,
+      "grad_norm": 0.6055270936106261,
+      "learning_rate": 1.9304414382599977e-05,
+      "loss": 0.8246,
+      "step": 5970
+    },
+    {
+      "epoch": 0.12845297933582506,
+      "grad_norm": 0.5997807600765583,
+      "learning_rate": 1.9301914375363746e-05,
+      "loss": 0.8056,
+      "step": 5980
+    },
+    {
+      "epoch": 0.12866778364909567,
+      "grad_norm": 0.6211427083875907,
+      "learning_rate": 1.9299410046012277e-05,
+      "loss": 0.8315,
+      "step": 5990
+    },
+    {
+      "epoch": 0.1288825879623663,
+      "grad_norm": 0.5688864716592172,
+      "learning_rate": 1.9296901395709206e-05,
+      "loss": 0.8257,
+      "step": 6000
+    },
+    {
+      "epoch": 0.1290973922756369,
+      "grad_norm": 0.6012936384715789,
+      "learning_rate": 1.929438842562017e-05,
+      "loss": 0.8351,
+      "step": 6010
+    },
+    {
+      "epoch": 0.1293121965889075,
+      "grad_norm": 0.6005677088042237,
+      "learning_rate": 1.929187113691282e-05,
+      "loss": 0.8357,
+      "step": 6020
+    },
+    {
+      "epoch": 0.1295270009021781,
+      "grad_norm": 0.6160401614681437,
+      "learning_rate": 1.9289349530756804e-05,
+      "loss": 0.843,
+      "step": 6030
+    },
+    {
+      "epoch": 0.12974180521544873,
+      "grad_norm": 0.6124623308197319,
+      "learning_rate": 1.9286823608323785e-05,
+      "loss": 0.8233,
+      "step": 6040
+    },
+    {
+      "epoch": 0.12995660952871935,
+      "grad_norm": 0.6203159062875466,
+      "learning_rate": 1.9284293370787424e-05,
+      "loss": 0.8303,
+      "step": 6050
+    },
+    {
+      "epoch": 0.13017141384198994,
+      "grad_norm": 0.6107868080049285,
+      "learning_rate": 1.9281758819323393e-05,
+      "loss": 0.8236,
+      "step": 6060
+    },
+    {
+      "epoch": 0.13038621815526055,
+      "grad_norm": 0.6132700985840815,
+      "learning_rate": 1.9279219955109366e-05,
+      "loss": 0.8307,
+      "step": 6070
+    },
+    {
+      "epoch": 0.13060102246853117,
+      "grad_norm": 0.6055289750075414,
+      "learning_rate": 1.927667677932502e-05,
+      "loss": 0.8127,
+      "step": 6080
+    },
+    {
+      "epoch": 0.1308158267818018,
+      "grad_norm": 0.6116418582015993,
+      "learning_rate": 1.9274129293152037e-05,
+      "loss": 0.832,
+      "step": 6090
+    },
+    {
+      "epoch": 0.1310306310950724,
+      "grad_norm": 0.6179783492037558,
+      "learning_rate": 1.92715774977741e-05,
+      "loss": 0.829,
+      "step": 6100
+    },
+    {
+      "epoch": 0.131245435408343,
+      "grad_norm": 0.5940118498029715,
+      "learning_rate": 1.9269021394376896e-05,
+      "loss": 0.8176,
+      "step": 6110
+    },
+    {
+      "epoch": 0.1314602397216136,
+      "grad_norm": 0.6031829297041618,
+      "learning_rate": 1.9266460984148116e-05,
+      "loss": 0.8279,
+      "step": 6120
+    },
+    {
+      "epoch": 0.13167504403488423,
+      "grad_norm": 0.5930057021061649,
+      "learning_rate": 1.9263896268277448e-05,
+      "loss": 0.8139,
+      "step": 6130
+    },
+    {
+      "epoch": 0.13188984834815484,
+      "grad_norm": 0.6095258155514939,
+      "learning_rate": 1.9261327247956575e-05,
+      "loss": 0.8163,
+      "step": 6140
+    },
+    {
+      "epoch": 0.13210465266142543,
+      "grad_norm": 0.600045191978913,
+      "learning_rate": 1.9258753924379196e-05,
+      "loss": 0.8167,
+      "step": 6150
+    },
+    {
+      "epoch": 0.13231945697469605,
+      "grad_norm": 0.6983897516026605,
+      "learning_rate": 1.9256176298740997e-05,
+      "loss": 0.828,
+      "step": 6160
+    },
+    {
+      "epoch": 0.13253426128796666,
+      "grad_norm": 0.6132138821976615,
+      "learning_rate": 1.925359437223967e-05,
+      "loss": 0.8201,
+      "step": 6170
+    },
+    {
+      "epoch": 0.13274906560123728,
+      "grad_norm": 0.6155911518334949,
+      "learning_rate": 1.9251008146074895e-05,
+      "loss": 0.822,
+      "step": 6180
+    },
+    {
+      "epoch": 0.13296386991450787,
+      "grad_norm": 1.3805735351829016,
+      "learning_rate": 1.924841762144836e-05,
+      "loss": 0.8297,
+      "step": 6190
+    },
+    {
+      "epoch": 0.1331786742277785,
+      "grad_norm": 0.7417405322736338,
+      "learning_rate": 1.924582279956375e-05,
+      "loss": 0.8178,
+      "step": 6200
+    },
+    {
+      "epoch": 0.1333934785410491,
+      "grad_norm": 0.6590382875105805,
+      "learning_rate": 1.9243223681626734e-05,
+      "loss": 0.8177,
+      "step": 6210
+    },
+    {
+      "epoch": 0.13360828285431972,
+      "grad_norm": 0.5947961735015622,
+      "learning_rate": 1.9240620268845e-05,
+      "loss": 0.8375,
+      "step": 6220
+    },
+    {
+      "epoch": 0.13382308716759034,
+      "grad_norm": 0.6151652104414234,
+      "learning_rate": 1.9238012562428204e-05,
+      "loss": 0.8136,
+      "step": 6230
+    },
+    {
+      "epoch": 0.13403789148086093,
+      "grad_norm": 0.6081687195694883,
+      "learning_rate": 1.9235400563588027e-05,
+      "loss": 0.8092,
+      "step": 6240
+    },
+    {
+      "epoch": 0.13425269579413154,
+      "grad_norm": 0.609223276457847,
+      "learning_rate": 1.923278427353812e-05,
+      "loss": 0.8176,
+      "step": 6250
+    },
+    {
+      "epoch": 0.13446750010740216,
+      "grad_norm": 0.6057832795531277,
+      "learning_rate": 1.9230163693494134e-05,
+      "loss": 0.8245,
+      "step": 6260
+    },
+    {
+      "epoch": 0.13468230442067278,
+      "grad_norm": 0.7218938650204069,
+      "learning_rate": 1.922753882467372e-05,
+      "loss": 0.829,
+      "step": 6270
+    },
+    {
+      "epoch": 0.13489710873394337,
+      "grad_norm": 0.5956376678659784,
+      "learning_rate": 1.9224909668296518e-05,
+      "loss": 0.8323,
+      "step": 6280
+    },
+    {
+      "epoch": 0.13511191304721398,
+      "grad_norm": 0.6055774590690349,
+      "learning_rate": 1.922227622558416e-05,
+      "loss": 0.8325,
+      "step": 6290
+    },
+    {
+      "epoch": 0.1353267173604846,
+      "grad_norm": 0.63594852369417,
+      "learning_rate": 1.9219638497760272e-05,
+      "loss": 0.8236,
+      "step": 6300
+    },
+    {
+      "epoch": 0.13554152167375522,
+      "grad_norm": 0.6615251477336661,
+      "learning_rate": 1.921699648605046e-05,
+      "loss": 0.8299,
+      "step": 6310
+    },
+    {
+      "epoch": 0.13575632598702583,
+      "grad_norm": 0.5939888969308088,
+      "learning_rate": 1.921435019168234e-05,
+      "loss": 0.8314,
+      "step": 6320
+    },
+    {
+      "epoch": 0.13597113030029642,
+      "grad_norm": 0.6224118503409543,
+      "learning_rate": 1.9211699615885505e-05,
+      "loss": 0.8105,
+      "step": 6330
+    },
+    {
+      "epoch": 0.13618593461356704,
+      "grad_norm": 0.6349901674353197,
+      "learning_rate": 1.920904475989153e-05,
+      "loss": 0.8267,
+      "step": 6340
+    },
+    {
+      "epoch": 0.13640073892683766,
+      "grad_norm": 0.6184766342855774,
+      "learning_rate": 1.9206385624934002e-05,
+      "loss": 0.8235,
+      "step": 6350
+    },
+    {
+      "epoch": 0.13661554324010827,
+      "grad_norm": 0.593593688668399,
+      "learning_rate": 1.9203722212248473e-05,
+      "loss": 0.8179,
+      "step": 6360
+    },
+    {
+      "epoch": 0.13683034755337886,
+      "grad_norm": 0.5904445744420688,
+      "learning_rate": 1.9201054523072497e-05,
+      "loss": 0.8278,
+      "step": 6370
+    },
+    {
+      "epoch": 0.13704515186664948,
+      "grad_norm": 0.6197705191484348,
+      "learning_rate": 1.9198382558645608e-05,
+      "loss": 0.8315,
+      "step": 6380
+    },
+    {
+      "epoch": 0.1372599561799201,
+      "grad_norm": 0.5970982729709278,
+      "learning_rate": 1.9195706320209326e-05,
+      "loss": 0.823,
+      "step": 6390
+    },
+    {
+      "epoch": 0.1374747604931907,
+      "grad_norm": 0.5860292711215872,
+      "learning_rate": 1.9193025809007164e-05,
+      "loss": 0.8029,
+      "step": 6400
+    },
+    {
+      "epoch": 0.1376895648064613,
+      "grad_norm": 0.5880366108161124,
+      "learning_rate": 1.919034102628462e-05,
+      "loss": 0.8206,
+      "step": 6410
+    },
+    {
+      "epoch": 0.13790436911973192,
+      "grad_norm": 0.6454586755047972,
+      "learning_rate": 1.918765197328916e-05,
+      "loss": 0.8252,
+      "step": 6420
+    },
+    {
+      "epoch": 0.13811917343300253,
+      "grad_norm": 0.5855160919576355,
+      "learning_rate": 1.918495865127026e-05,
+      "loss": 0.8177,
+      "step": 6430
+    },
+    {
+      "epoch": 0.13833397774627315,
+      "grad_norm": 0.568749463356773,
+      "learning_rate": 1.9182261061479357e-05,
+      "loss": 0.8168,
+      "step": 6440
+    },
+    {
+      "epoch": 0.13854878205954377,
+      "grad_norm": 0.578722846016812,
+      "learning_rate": 1.9179559205169885e-05,
+      "loss": 0.8283,
+      "step": 6450
+    },
+    {
+      "epoch": 0.13876358637281436,
+      "grad_norm": 0.5933587203351999,
+      "learning_rate": 1.9176853083597257e-05,
+      "loss": 0.8366,
+      "step": 6460
+    },
+    {
+      "epoch": 0.13897839068608497,
+      "grad_norm": 0.679493941830241,
+      "learning_rate": 1.9174142698018864e-05,
+      "loss": 0.8439,
+      "step": 6470
+    },
+    {
+      "epoch": 0.1391931949993556,
+      "grad_norm": 0.6027286672996572,
+      "learning_rate": 1.9171428049694082e-05,
+      "loss": 0.8181,
+      "step": 6480
+    },
+    {
+      "epoch": 0.1394079993126262,
+      "grad_norm": 0.6104128555070933,
+      "learning_rate": 1.916870913988427e-05,
+      "loss": 0.8276,
+      "step": 6490
+    },
+    {
+      "epoch": 0.1396228036258968,
+      "grad_norm": 0.6013008901028725,
+      "learning_rate": 1.9165985969852757e-05,
+      "loss": 0.8274,
+      "step": 6500
+    },
+    {
+      "epoch": 0.1398376079391674,
+      "grad_norm": 0.6122352927576057,
+      "learning_rate": 1.916325854086486e-05,
+      "loss": 0.824,
+      "step": 6510
+    },
+    {
+      "epoch": 0.14005241225243803,
+      "grad_norm": 0.6300545859500599,
+      "learning_rate": 1.916052685418788e-05,
+      "loss": 0.8308,
+      "step": 6520
+    },
+    {
+      "epoch": 0.14026721656570865,
+      "grad_norm": 0.583338621183778,
+      "learning_rate": 1.9157790911091082e-05,
+      "loss": 0.8186,
+      "step": 6530
+    },
+    {
+      "epoch": 0.14048202087897926,
+      "grad_norm": 0.5947321272096432,
+      "learning_rate": 1.9155050712845722e-05,
+      "loss": 0.8141,
+      "step": 6540
+    },
+    {
+      "epoch": 0.14069682519224985,
+      "grad_norm": 0.6508513713242783,
+      "learning_rate": 1.915230626072502e-05,
+      "loss": 0.8241,
+      "step": 6550
+    },
+    {
+      "epoch": 0.14091162950552047,
+      "grad_norm": 0.5902491680834664,
+      "learning_rate": 1.914955755600419e-05,
+      "loss": 0.812,
+      "step": 6560
+    },
+    {
+      "epoch": 0.14112643381879109,
+      "grad_norm": 0.6223882649259539,
+      "learning_rate": 1.91468045999604e-05,
+      "loss": 0.8205,
+      "step": 6570
+    },
+    {
+      "epoch": 0.1413412381320617,
+      "grad_norm": 0.6039015945397973,
+      "learning_rate": 1.9144047393872818e-05,
+      "loss": 0.8254,
+      "step": 6580
+    },
+    {
+      "epoch": 0.1415560424453323,
+      "grad_norm": 4.480885455321173,
+      "learning_rate": 1.9141285939022563e-05,
+      "loss": 0.8123,
+      "step": 6590
+    },
+    {
+      "epoch": 0.1417708467586029,
+      "grad_norm": 0.6004465875420536,
+      "learning_rate": 1.9138520236692747e-05,
+      "loss": 0.8316,
+      "step": 6600
+    },
+    {
+      "epoch": 0.14198565107187353,
+      "grad_norm": 0.5877543081764919,
+      "learning_rate": 1.9135750288168446e-05,
+      "loss": 0.8194,
+      "step": 6610
+    },
+    {
+      "epoch": 0.14220045538514414,
+      "grad_norm": 0.6060063198959242,
+      "learning_rate": 1.9132976094736707e-05,
+      "loss": 0.8122,
+      "step": 6620
+    },
+    {
+      "epoch": 0.14241525969841473,
+      "grad_norm": 0.6105433238582144,
+      "learning_rate": 1.9130197657686555e-05,
+      "loss": 0.8144,
+      "step": 6630
+    },
+    {
+      "epoch": 0.14263006401168535,
+      "grad_norm": 0.5854211970675489,
+      "learning_rate": 1.9127414978308987e-05,
+      "loss": 0.826,
+      "step": 6640
+    },
+    {
+      "epoch": 0.14284486832495596,
+      "grad_norm": 0.6132307714509498,
+      "learning_rate": 1.9124628057896972e-05,
+      "loss": 0.8096,
+      "step": 6650
+    },
+    {
+      "epoch": 0.14305967263822658,
+      "grad_norm": 0.6097804703131902,
+      "learning_rate": 1.912183689774544e-05,
+      "loss": 0.8137,
+      "step": 6660
+    },
+    {
+      "epoch": 0.1432744769514972,
+      "grad_norm": 0.5755382542472285,
+      "learning_rate": 1.91190414991513e-05,
+      "loss": 0.8191,
+      "step": 6670
+    },
+    {
+      "epoch": 0.1434892812647678,
+      "grad_norm": 0.6077737497744028,
+      "learning_rate": 1.9116241863413433e-05,
+      "loss": 0.8327,
+      "step": 6680
+    },
+    {
+      "epoch": 0.1437040855780384,
+      "grad_norm": 0.6330293724055359,
+      "learning_rate": 1.9113437991832678e-05,
+      "loss": 0.8159,
+      "step": 6690
+    },
+    {
+      "epoch": 0.14391888989130902,
+      "grad_norm": 0.6019952588522437,
+      "learning_rate": 1.911062988571185e-05,
+      "loss": 0.8103,
+      "step": 6700
+    },
+    {
+      "epoch": 0.14413369420457964,
+      "grad_norm": 0.6157593320641529,
+      "learning_rate": 1.9107817546355726e-05,
+      "loss": 0.8192,
+      "step": 6710
+    },
+    {
+      "epoch": 0.14434849851785023,
+      "grad_norm": 0.6262865372100354,
+      "learning_rate": 1.9105000975071062e-05,
+      "loss": 0.8102,
+      "step": 6720
+    },
+    {
+      "epoch": 0.14456330283112084,
+      "grad_norm": 0.598604944782719,
+      "learning_rate": 1.9102180173166565e-05,
+      "loss": 0.8178,
+      "step": 6730
+    },
+    {
+      "epoch": 0.14477810714439146,
+      "grad_norm": 0.6130319012297948,
+      "learning_rate": 1.909935514195292e-05,
+      "loss": 0.8146,
+      "step": 6740
+    },
+    {
+      "epoch": 0.14499291145766208,
+      "grad_norm": 0.6051680516635761,
+      "learning_rate": 1.9096525882742766e-05,
+      "loss": 0.8243,
+      "step": 6750
+    },
+    {
+      "epoch": 0.1452077157709327,
+      "grad_norm": 0.5945658988091994,
+      "learning_rate": 1.909369239685072e-05,
+      "loss": 0.8215,
+      "step": 6760
+    },
+    {
+      "epoch": 0.14542252008420328,
+      "grad_norm": 0.6014406118512295,
+      "learning_rate": 1.9090854685593344e-05,
+      "loss": 0.8152,
+      "step": 6770
+    },
+    {
+      "epoch": 0.1456373243974739,
+      "grad_norm": 0.6069056436911239,
+      "learning_rate": 1.9088012750289185e-05,
+      "loss": 0.8089,
+      "step": 6780
+    },
+    {
+      "epoch": 0.14585212871074452,
+      "grad_norm": 0.6384017208108504,
+      "learning_rate": 1.908516659225874e-05,
+      "loss": 0.8185,
+      "step": 6790
+    },
+    {
+      "epoch": 0.14606693302401513,
+      "grad_norm": 0.6183552826731581,
+      "learning_rate": 1.9082316212824467e-05,
+      "loss": 0.829,
+      "step": 6800
+    },
+    {
+      "epoch": 0.14628173733728572,
+      "grad_norm": 0.6033726613256812,
+      "learning_rate": 1.9079461613310793e-05,
+      "loss": 0.8368,
+      "step": 6810
+    },
+    {
+      "epoch": 0.14649654165055634,
+      "grad_norm": 0.6767597702311462,
+      "learning_rate": 1.90766027950441e-05,
+      "loss": 0.8312,
+      "step": 6820
+    },
+    {
+      "epoch": 0.14671134596382696,
+      "grad_norm": 0.6047536333583156,
+      "learning_rate": 1.9073739759352728e-05,
+      "loss": 0.8334,
+      "step": 6830
+    },
+    {
+      "epoch": 0.14692615027709757,
+      "grad_norm": 0.5868050666630678,
+      "learning_rate": 1.9070872507566988e-05,
+      "loss": 0.8243,
+      "step": 6840
+    },
+    {
+      "epoch": 0.14714095459036816,
+      "grad_norm": 0.5987664541670297,
+      "learning_rate": 1.906800104101914e-05,
+      "loss": 0.8259,
+      "step": 6850
+    },
+    {
+      "epoch": 0.14735575890363878,
+      "grad_norm": 0.6171104544958446,
+      "learning_rate": 1.9065125361043403e-05,
+      "loss": 0.8265,
+      "step": 6860
+    },
+    {
+      "epoch": 0.1475705632169094,
+      "grad_norm": 0.5898285999163112,
+      "learning_rate": 1.9062245468975958e-05,
+      "loss": 0.8208,
+      "step": 6870
+    },
+    {
+      "epoch": 0.14778536753018,
+      "grad_norm": 0.6178554712481548,
+      "learning_rate": 1.9059361366154937e-05,
+      "loss": 0.8269,
+      "step": 6880
+    },
+    {
+      "epoch": 0.14800017184345063,
+      "grad_norm": 0.5673710580373017,
+      "learning_rate": 1.9056473053920436e-05,
+      "loss": 0.8027,
+      "step": 6890
+    },
+    {
+      "epoch": 0.14821497615672122,
+      "grad_norm": 0.599452012891556,
+      "learning_rate": 1.9053580533614502e-05,
+      "loss": 0.841,
+      "step": 6900
+    },
+    {
+      "epoch": 0.14842978046999183,
+      "grad_norm": 0.5874536149399996,
+      "learning_rate": 1.905068380658114e-05,
+      "loss": 0.805,
+      "step": 6910
+    },
+    {
+      "epoch": 0.14864458478326245,
+      "grad_norm": 0.5970361855515987,
+      "learning_rate": 1.9047782874166308e-05,
+      "loss": 0.8067,
+      "step": 6920
+    },
+    {
+      "epoch": 0.14885938909653307,
+      "grad_norm": 0.6177964172907193,
+      "learning_rate": 1.9044877737717916e-05,
+      "loss": 0.8372,
+      "step": 6930
+    },
+    {
+      "epoch": 0.14907419340980366,
+      "grad_norm": 0.5961958626534212,
+      "learning_rate": 1.904196839858583e-05,
+      "loss": 0.8336,
+      "step": 6940
+    },
+    {
+      "epoch": 0.14928899772307427,
+      "grad_norm": 0.6016838781672733,
+      "learning_rate": 1.9039054858121872e-05,
+      "loss": 0.8138,
+      "step": 6950
+    },
+    {
+      "epoch": 0.1495038020363449,
+      "grad_norm": 0.5772199742586428,
+      "learning_rate": 1.903613711767981e-05,
+      "loss": 0.8061,
+      "step": 6960
+    },
+    {
+      "epoch": 0.1497186063496155,
+      "grad_norm": 0.6091085341764657,
+      "learning_rate": 1.9033215178615363e-05,
+      "loss": 0.8292,
+      "step": 6970
+    },
+    {
+      "epoch": 0.14993341066288612,
+      "grad_norm": 0.6111170798001496,
+      "learning_rate": 1.903028904228621e-05,
+      "loss": 0.8123,
+      "step": 6980
+    },
+    {
+      "epoch": 0.1501482149761567,
+      "grad_norm": 0.641212096959498,
+      "learning_rate": 1.902735871005197e-05,
+      "loss": 0.8281,
+      "step": 6990
+    },
+    {
+      "epoch": 0.15036301928942733,
+      "grad_norm": 0.601677168468839,
+      "learning_rate": 1.9024424183274216e-05,
+      "loss": 0.8302,
+      "step": 7000
+    },
+    {
+      "epoch": 0.15057782360269795,
+      "grad_norm": 0.6001519478847499,
+      "learning_rate": 1.9021485463316468e-05,
+      "loss": 0.8328,
+      "step": 7010
+    },
+    {
+      "epoch": 0.15079262791596856,
+      "grad_norm": 0.590756165100162,
+      "learning_rate": 1.9018542551544205e-05,
+      "loss": 0.8283,
+      "step": 7020
+    },
+    {
+      "epoch": 0.15100743222923915,
+      "grad_norm": 0.5930250026896874,
+      "learning_rate": 1.9015595449324837e-05,
+      "loss": 0.8211,
+      "step": 7030
+    },
+    {
+      "epoch": 0.15122223654250977,
+      "grad_norm": 0.6342243168511045,
+      "learning_rate": 1.9012644158027727e-05,
+      "loss": 0.8136,
+      "step": 7040
+    },
+    {
+      "epoch": 0.15143704085578039,
+      "grad_norm": 0.6288449639482769,
+      "learning_rate": 1.900968867902419e-05,
+      "loss": 0.8161,
+      "step": 7050
+    },
+    {
+      "epoch": 0.151651845169051,
+      "grad_norm": 0.5885895763623495,
+      "learning_rate": 1.9006729013687488e-05,
+      "loss": 0.823,
+      "step": 7060
+    },
+    {
+      "epoch": 0.1518666494823216,
+      "grad_norm": 0.5945227181413594,
+      "learning_rate": 1.900376516339282e-05,
+      "loss": 0.8184,
+      "step": 7070
+    },
+    {
+      "epoch": 0.1520814537955922,
+      "grad_norm": 0.6038107671354757,
+      "learning_rate": 1.9000797129517326e-05,
+      "loss": 0.8338,
+      "step": 7080
+    },
+    {
+      "epoch": 0.15229625810886283,
+      "grad_norm": 0.6179750378509373,
+      "learning_rate": 1.8997824913440102e-05,
+      "loss": 0.819,
+      "step": 7090
+    },
+    {
+      "epoch": 0.15251106242213344,
+      "grad_norm": 0.580594427759617,
+      "learning_rate": 1.8994848516542187e-05,
+      "loss": 0.8116,
+      "step": 7100
+    },
+    {
+      "epoch": 0.15272586673540406,
+      "grad_norm": 0.6629610839791461,
+      "learning_rate": 1.899186794020655e-05,
+      "loss": 0.8164,
+      "step": 7110
+    },
+    {
+      "epoch": 0.15294067104867465,
+      "grad_norm": 0.630038582706391,
+      "learning_rate": 1.898888318581811e-05,
+      "loss": 0.8197,
+      "step": 7120
+    },
+    {
+      "epoch": 0.15315547536194526,
+      "grad_norm": 0.6060433882973041,
+      "learning_rate": 1.8985894254763734e-05,
+      "loss": 0.8345,
+      "step": 7130
+    },
+    {
+      "epoch": 0.15337027967521588,
+      "grad_norm": 0.5763806285603041,
+      "learning_rate": 1.8982901148432214e-05,
+      "loss": 0.8195,
+      "step": 7140
+    },
+    {
+      "epoch": 0.1535850839884865,
+      "grad_norm": 0.5858697966874623,
+      "learning_rate": 1.89799038682143e-05,
+      "loss": 0.8108,
+      "step": 7150
+    },
+    {
+      "epoch": 0.1537998883017571,
+      "grad_norm": 0.6192381232542434,
+      "learning_rate": 1.8976902415502664e-05,
+      "loss": 0.8096,
+      "step": 7160
+    },
+    {
+      "epoch": 0.1540146926150277,
+      "grad_norm": 0.5775749857070134,
+      "learning_rate": 1.8973896791691925e-05,
+      "loss": 0.8152,
+      "step": 7170
+    },
+    {
+      "epoch": 0.15422949692829832,
+      "grad_norm": 0.6222213814962546,
+      "learning_rate": 1.8970886998178648e-05,
+      "loss": 0.8409,
+      "step": 7180
+    },
+    {
+      "epoch": 0.15444430124156894,
+      "grad_norm": 0.6088269913390629,
+      "learning_rate": 1.8967873036361316e-05,
+      "loss": 0.8327,
+      "step": 7190
+    },
+    {
+      "epoch": 0.15465910555483955,
+      "grad_norm": 0.5930210210719912,
+      "learning_rate": 1.896485490764037e-05,
+      "loss": 0.8115,
+      "step": 7200
+    },
+    {
+      "epoch": 0.15487390986811014,
+      "grad_norm": 0.6000706146439299,
+      "learning_rate": 1.8961832613418173e-05,
+      "loss": 0.8238,
+      "step": 7210
+    },
+    {
+      "epoch": 0.15508871418138076,
+      "grad_norm": 0.6006800299840814,
+      "learning_rate": 1.895880615509903e-05,
+      "loss": 0.8145,
+      "step": 7220
+    },
+    {
+      "epoch": 0.15530351849465138,
+      "grad_norm": 0.6451495489221977,
+      "learning_rate": 1.8955775534089168e-05,
+      "loss": 0.8222,
+      "step": 7230
+    },
+    {
+      "epoch": 0.155518322807922,
+      "grad_norm": 0.5789768463565536,
+      "learning_rate": 1.8952740751796776e-05,
+      "loss": 0.8025,
+      "step": 7240
+    },
+    {
+      "epoch": 0.15573312712119258,
+      "grad_norm": 0.5853169790059398,
+      "learning_rate": 1.8949701809631945e-05,
+      "loss": 0.8204,
+      "step": 7250
+    },
+    {
+      "epoch": 0.1559479314344632,
+      "grad_norm": 0.5695304385263977,
+      "learning_rate": 1.894665870900672e-05,
+      "loss": 0.8125,
+      "step": 7260
+    },
+    {
+      "epoch": 0.15616273574773382,
+      "grad_norm": 0.6049599632857573,
+      "learning_rate": 1.8943611451335075e-05,
+      "loss": 0.8107,
+      "step": 7270
+    },
+    {
+      "epoch": 0.15637754006100443,
+      "grad_norm": 0.573259462488128,
+      "learning_rate": 1.89405600380329e-05,
+      "loss": 0.8135,
+      "step": 7280
+    },
+    {
+      "epoch": 0.15659234437427502,
+      "grad_norm": 0.5940723497985662,
+      "learning_rate": 1.893750447051804e-05,
+      "loss": 0.808,
+      "step": 7290
+    },
+    {
+      "epoch": 0.15680714868754564,
+      "grad_norm": 0.641894313123313,
+      "learning_rate": 1.893444475021025e-05,
+      "loss": 0.8177,
+      "step": 7300
+    },
+    {
+      "epoch": 0.15702195300081626,
+      "grad_norm": 0.5858730208906369,
+      "learning_rate": 1.8931380878531228e-05,
+      "loss": 0.8286,
+      "step": 7310
+    },
+    {
+      "epoch": 0.15723675731408687,
+      "grad_norm": 0.582476355104475,
+      "learning_rate": 1.892831285690459e-05,
+      "loss": 0.8257,
+      "step": 7320
+    },
+    {
+      "epoch": 0.1574515616273575,
+      "grad_norm": 0.5832307163787261,
+      "learning_rate": 1.892524068675589e-05,
+      "loss": 0.8007,
+      "step": 7330
+    },
+    {
+      "epoch": 0.15766636594062808,
+      "grad_norm": 0.5790768352438785,
+      "learning_rate": 1.892216436951261e-05,
+      "loss": 0.7953,
+      "step": 7340
+    },
+    {
+      "epoch": 0.1578811702538987,
+      "grad_norm": 0.633690146863657,
+      "learning_rate": 1.8919083906604144e-05,
+      "loss": 0.8205,
+      "step": 7350
+    },
+    {
+      "epoch": 0.1580959745671693,
+      "grad_norm": 0.617991296431158,
+      "learning_rate": 1.891599929946183e-05,
+      "loss": 0.8098,
+      "step": 7360
+    },
+    {
+      "epoch": 0.15831077888043993,
+      "grad_norm": 0.5974548816213079,
+      "learning_rate": 1.8912910549518924e-05,
+      "loss": 0.8079,
+      "step": 7370
+    },
+    {
+      "epoch": 0.15852558319371052,
+      "grad_norm": 0.6020174180448247,
+      "learning_rate": 1.89098176582106e-05,
+      "loss": 0.8177,
+      "step": 7380
+    },
+    {
+      "epoch": 0.15874038750698113,
+      "grad_norm": 0.5899662948515287,
+      "learning_rate": 1.8906720626973975e-05,
+      "loss": 0.8175,
+      "step": 7390
+    },
+    {
+      "epoch": 0.15895519182025175,
+      "grad_norm": 0.5872850170519595,
+      "learning_rate": 1.8903619457248072e-05,
+      "loss": 0.8005,
+      "step": 7400
+    },
+    {
+      "epoch": 0.15916999613352237,
+      "grad_norm": 0.5852430036799516,
+      "learning_rate": 1.890051415047384e-05,
+      "loss": 0.8219,
+      "step": 7410
+    },
+    {
+      "epoch": 0.15938480044679298,
+      "grad_norm": 0.5917524465188728,
+      "learning_rate": 1.8897404708094154e-05,
+      "loss": 0.8072,
+      "step": 7420
+    },
+    {
+      "epoch": 0.15959960476006357,
+      "grad_norm": 0.5508421705343692,
+      "learning_rate": 1.8894291131553817e-05,
+      "loss": 0.8198,
+      "step": 7430
+    },
+    {
+      "epoch": 0.1598144090733342,
+      "grad_norm": 0.5852065142165715,
+      "learning_rate": 1.889117342229954e-05,
+      "loss": 0.8194,
+      "step": 7440
+    },
+    {
+      "epoch": 0.1600292133866048,
+      "grad_norm": 0.5801695841837192,
+      "learning_rate": 1.8888051581779964e-05,
+      "loss": 0.8203,
+      "step": 7450
+    },
+    {
+      "epoch": 0.16024401769987542,
+      "grad_norm": 0.602952041149204,
+      "learning_rate": 1.888492561144564e-05,
+      "loss": 0.8059,
+      "step": 7460
+    },
+    {
+      "epoch": 0.160458822013146,
+      "grad_norm": 0.6431449708685523,
+      "learning_rate": 1.8881795512749046e-05,
+      "loss": 0.8326,
+      "step": 7470
+    },
+    {
+      "epoch": 0.16067362632641663,
+      "grad_norm": 0.6196861452682948,
+      "learning_rate": 1.887866128714458e-05,
+      "loss": 0.8457,
+      "step": 7480
+    },
+    {
+      "epoch": 0.16088843063968725,
+      "grad_norm": 0.5649025093930304,
+      "learning_rate": 1.887552293608855e-05,
+      "loss": 0.8113,
+      "step": 7490
+    },
+    {
+      "epoch": 0.16110323495295786,
+      "grad_norm": 0.63666785388803,
+      "learning_rate": 1.8872380461039184e-05,
+      "loss": 0.8264,
+      "step": 7500
+    },
+    {
+      "epoch": 0.16131803926622845,
+      "grad_norm": 0.6005193455583923,
+      "learning_rate": 1.8869233863456627e-05,
+      "loss": 0.819,
+      "step": 7510
+    },
+    {
+      "epoch": 0.16153284357949907,
+      "grad_norm": 0.6077387623232092,
+      "learning_rate": 1.8866083144802938e-05,
+      "loss": 0.8185,
+      "step": 7520
+    },
+    {
+      "epoch": 0.16174764789276969,
+      "grad_norm": 0.5938972114256622,
+      "learning_rate": 1.8862928306542093e-05,
+      "loss": 0.8225,
+      "step": 7530
+    },
+    {
+      "epoch": 0.1619624522060403,
+      "grad_norm": 0.6156548912606762,
+      "learning_rate": 1.8859769350139982e-05,
+      "loss": 0.8263,
+      "step": 7540
+    },
+    {
+      "epoch": 0.16217725651931092,
+      "grad_norm": 0.6056374183349044,
+      "learning_rate": 1.8856606277064407e-05,
+      "loss": 0.8198,
+      "step": 7550
+    },
+    {
+      "epoch": 0.1623920608325815,
+      "grad_norm": 0.5663370767459714,
+      "learning_rate": 1.8853439088785084e-05,
+      "loss": 0.8159,
+      "step": 7560
+    },
+    {
+      "epoch": 0.16260686514585213,
+      "grad_norm": 0.5775902577797735,
+      "learning_rate": 1.885026778677364e-05,
+      "loss": 0.8139,
+      "step": 7570
+    },
+    {
+      "epoch": 0.16282166945912274,
+      "grad_norm": 0.591831820384521,
+      "learning_rate": 1.884709237250361e-05,
+      "loss": 0.8176,
+      "step": 7580
+    },
+    {
+      "epoch": 0.16303647377239336,
+      "grad_norm": 0.5987459436284528,
+      "learning_rate": 1.884391284745045e-05,
+      "loss": 0.8246,
+      "step": 7590
+    },
+    {
+      "epoch": 0.16325127808566395,
+      "grad_norm": 0.5985317736090139,
+      "learning_rate": 1.8840729213091514e-05,
+      "loss": 0.8211,
+      "step": 7600
+    },
+    {
+      "epoch": 0.16346608239893456,
+      "grad_norm": 0.5967857623504043,
+      "learning_rate": 1.8837541470906076e-05,
+      "loss": 0.8068,
+      "step": 7610
+    },
+    {
+      "epoch": 0.16368088671220518,
+      "grad_norm": 0.5809065765322244,
+      "learning_rate": 1.883434962237531e-05,
+      "loss": 0.8008,
+      "step": 7620
+    },
+    {
+      "epoch": 0.1638956910254758,
+      "grad_norm": 0.5851551177907041,
+      "learning_rate": 1.8831153668982304e-05,
+      "loss": 0.8204,
+      "step": 7630
+    },
+    {
+      "epoch": 0.16411049533874641,
+      "grad_norm": 0.5917914776651239,
+      "learning_rate": 1.882795361221205e-05,
+      "loss": 0.8209,
+      "step": 7640
+    },
+    {
+      "epoch": 0.164325299652017,
+      "grad_norm": 0.6224876855421713,
+      "learning_rate": 1.882474945355145e-05,
+      "loss": 0.7925,
+      "step": 7650
+    },
+    {
+      "epoch": 0.16454010396528762,
+      "grad_norm": 0.5603554031670195,
+      "learning_rate": 1.8821541194489307e-05,
+      "loss": 0.8085,
+      "step": 7660
+    },
+    {
+      "epoch": 0.16475490827855824,
+      "grad_norm": 0.5489520906362242,
+      "learning_rate": 1.8818328836516334e-05,
+      "loss": 0.8161,
+      "step": 7670
+    },
+    {
+      "epoch": 0.16496971259182885,
+      "grad_norm": 0.6169881588767844,
+      "learning_rate": 1.8815112381125146e-05,
+      "loss": 0.8297,
+      "step": 7680
+    },
+    {
+      "epoch": 0.16518451690509944,
+      "grad_norm": 0.5901587326877425,
+      "learning_rate": 1.8811891829810257e-05,
+      "loss": 0.8172,
+      "step": 7690
+    },
+    {
+      "epoch": 0.16539932121837006,
+      "grad_norm": 0.561361319802584,
+      "learning_rate": 1.8808667184068098e-05,
+      "loss": 0.8172,
+      "step": 7700
+    },
+    {
+      "epoch": 0.16561412553164068,
+      "grad_norm": 0.5859380687978089,
+      "learning_rate": 1.880543844539699e-05,
+      "loss": 0.8145,
+      "step": 7710
+    },
+    {
+      "epoch": 0.1658289298449113,
+      "grad_norm": 0.5693560253720011,
+      "learning_rate": 1.880220561529716e-05,
+      "loss": 0.8259,
+      "step": 7720
+    },
+    {
+      "epoch": 0.16604373415818188,
+      "grad_norm": 0.6279685202690973,
+      "learning_rate": 1.8798968695270735e-05,
+      "loss": 0.8073,
+      "step": 7730
+    },
+    {
+      "epoch": 0.1662585384714525,
+      "grad_norm": 0.5868609499267986,
+      "learning_rate": 1.879572768682174e-05,
+      "loss": 0.8045,
+      "step": 7740
+    },
+    {
+      "epoch": 0.16647334278472312,
+      "grad_norm": 0.5949413009843553,
+      "learning_rate": 1.8792482591456115e-05,
+      "loss": 0.821,
+      "step": 7750
+    },
+    {
+      "epoch": 0.16668814709799373,
+      "grad_norm": 0.5734770401983357,
+      "learning_rate": 1.8789233410681675e-05,
+      "loss": 0.8037,
+      "step": 7760
+    },
+    {
+      "epoch": 0.16690295141126435,
+      "grad_norm": 0.6165809896599992,
+      "learning_rate": 1.8785980146008146e-05,
+      "loss": 0.8229,
+      "step": 7770
+    },
+    {
+      "epoch": 0.16711775572453494,
+      "grad_norm": 0.5747676549015432,
+      "learning_rate": 1.8782722798947154e-05,
+      "loss": 0.8145,
+      "step": 7780
+    },
+    {
+      "epoch": 0.16733256003780556,
+      "grad_norm": 0.587504015936523,
+      "learning_rate": 1.8779461371012224e-05,
+      "loss": 0.8096,
+      "step": 7790
+    },
+    {
+      "epoch": 0.16754736435107617,
+      "grad_norm": 0.6174090349249043,
+      "learning_rate": 1.877619586371876e-05,
+      "loss": 0.8142,
+      "step": 7800
+    },
+    {
+      "epoch": 0.1677621686643468,
+      "grad_norm": 0.5847105333987745,
+      "learning_rate": 1.8772926278584077e-05,
+      "loss": 0.8247,
+      "step": 7810
+    },
+    {
+      "epoch": 0.16797697297761738,
+      "grad_norm": 0.588578056608747,
+      "learning_rate": 1.8769652617127388e-05,
+      "loss": 0.8043,
+      "step": 7820
+    },
+    {
+      "epoch": 0.168191777290888,
+      "grad_norm": 0.577693735778559,
+      "learning_rate": 1.8766374880869785e-05,
+      "loss": 0.8224,
+      "step": 7830
+    },
+    {
+      "epoch": 0.1684065816041586,
+      "grad_norm": 0.5827121728285275,
+      "learning_rate": 1.8763093071334263e-05,
+      "loss": 0.8016,
+      "step": 7840
+    },
+    {
+      "epoch": 0.16862138591742923,
+      "grad_norm": 0.5897150773152798,
+      "learning_rate": 1.8759807190045715e-05,
+      "loss": 0.8078,
+      "step": 7850
+    },
+    {
+      "epoch": 0.16883619023069985,
+      "grad_norm": 0.5759666753347583,
+      "learning_rate": 1.8756517238530904e-05,
+      "loss": 0.8072,
+      "step": 7860
+    },
+    {
+      "epoch": 0.16905099454397043,
+      "grad_norm": 0.5818077020634499,
+      "learning_rate": 1.8753223218318515e-05,
+      "loss": 0.8117,
+      "step": 7870
+    },
+    {
+      "epoch": 0.16926579885724105,
+      "grad_norm": 0.6076398421069771,
+      "learning_rate": 1.87499251309391e-05,
+      "loss": 0.8033,
+      "step": 7880
+    },
+    {
+      "epoch": 0.16948060317051167,
+      "grad_norm": 0.5810094500721327,
+      "learning_rate": 1.874662297792511e-05,
+      "loss": 0.8123,
+      "step": 7890
+    },
+    {
+      "epoch": 0.16969540748378228,
+      "grad_norm": 0.6001843216635706,
+      "learning_rate": 1.874331676081088e-05,
+      "loss": 0.8239,
+      "step": 7900
+    },
+    {
+      "epoch": 0.16991021179705287,
+      "grad_norm": 0.6000314594312535,
+      "learning_rate": 1.874000648113264e-05,
+      "loss": 0.8276,
+      "step": 7910
+    },
+    {
+      "epoch": 0.1701250161103235,
+      "grad_norm": 0.5742460502628838,
+      "learning_rate": 1.8736692140428506e-05,
+      "loss": 0.8053,
+      "step": 7920
+    },
+    {
+      "epoch": 0.1703398204235941,
+      "grad_norm": 0.5832371696394795,
+      "learning_rate": 1.873337374023848e-05,
+      "loss": 0.8116,
+      "step": 7930
+    },
+    {
+      "epoch": 0.17055462473686472,
+      "grad_norm": 0.5845557501110522,
+      "learning_rate": 1.8730051282104446e-05,
+      "loss": 0.8221,
+      "step": 7940
+    },
+    {
+      "epoch": 0.1707694290501353,
+      "grad_norm": 0.6092794931268248,
+      "learning_rate": 1.8726724767570178e-05,
+      "loss": 0.8034,
+      "step": 7950
+    },
+    {
+      "epoch": 0.17098423336340593,
+      "grad_norm": 0.5941748660659438,
+      "learning_rate": 1.8723394198181333e-05,
+      "loss": 0.7995,
+      "step": 7960
+    },
+    {
+      "epoch": 0.17119903767667655,
+      "grad_norm": 0.5742787477493947,
+      "learning_rate": 1.8720059575485462e-05,
+      "loss": 0.8093,
+      "step": 7970
+    },
+    {
+      "epoch": 0.17141384198994716,
+      "grad_norm": 0.5862196213050433,
+      "learning_rate": 1.8716720901031983e-05,
+      "loss": 0.82,
+      "step": 7980
+    },
+    {
+      "epoch": 0.17162864630321778,
+      "grad_norm": 0.595590568398061,
+      "learning_rate": 1.8713378176372198e-05,
+      "loss": 0.8208,
+      "step": 7990
+    },
+    {
+      "epoch": 0.17184345061648837,
+      "grad_norm": 0.6014060853390284,
+      "learning_rate": 1.871003140305931e-05,
+      "loss": 0.8243,
+      "step": 8000
+    },
+    {
+      "epoch": 0.17205825492975899,
+      "grad_norm": 0.5885951381362904,
+      "learning_rate": 1.870668058264838e-05,
+      "loss": 0.8184,
+      "step": 8010
+    },
+    {
+      "epoch": 0.1722730592430296,
+      "grad_norm": 0.5675767542282781,
+      "learning_rate": 1.870332571669637e-05,
+      "loss": 0.8353,
+      "step": 8020
+    },
+    {
+      "epoch": 0.17248786355630022,
+      "grad_norm": 0.5885374361788073,
+      "learning_rate": 1.86999668067621e-05,
+      "loss": 0.8067,
+      "step": 8030
+    },
+    {
+      "epoch": 0.1727026678695708,
+      "grad_norm": 0.5725047507144028,
+      "learning_rate": 1.869660385440629e-05,
+      "loss": 0.809,
+      "step": 8040
+    },
+    {
+      "epoch": 0.17291747218284143,
+      "grad_norm": 0.5669767417801742,
+      "learning_rate": 1.8693236861191524e-05,
+      "loss": 0.7987,
+      "step": 8050
+    },
+    {
+      "epoch": 0.17313227649611204,
+      "grad_norm": 0.6015883976560725,
+      "learning_rate": 1.8689865828682266e-05,
+      "loss": 0.8249,
+      "step": 8060
+    },
+    {
+      "epoch": 0.17334708080938266,
+      "grad_norm": 0.5761246694287212,
+      "learning_rate": 1.868649075844487e-05,
+      "loss": 0.8161,
+      "step": 8070
+    },
+    {
+      "epoch": 0.17356188512265328,
+      "grad_norm": 0.5969047559623099,
+      "learning_rate": 1.8683111652047543e-05,
+      "loss": 0.8361,
+      "step": 8080
+    },
+    {
+      "epoch": 0.17377668943592386,
+      "grad_norm": 0.5929471728979538,
+      "learning_rate": 1.8679728511060385e-05,
+      "loss": 0.8157,
+      "step": 8090
+    },
+    {
+      "epoch": 0.17399149374919448,
+      "grad_norm": 0.5709547291713754,
+      "learning_rate": 1.8676341337055367e-05,
+      "loss": 0.7972,
+      "step": 8100
+    },
+    {
+      "epoch": 0.1742062980624651,
+      "grad_norm": 0.5792794867839971,
+      "learning_rate": 1.867295013160633e-05,
+      "loss": 0.8236,
+      "step": 8110
+    },
+    {
+      "epoch": 0.17442110237573571,
+      "grad_norm": 0.580543068320473,
+      "learning_rate": 1.866955489628899e-05,
+      "loss": 0.8322,
+      "step": 8120
+    },
+    {
+      "epoch": 0.1746359066890063,
+      "grad_norm": 0.5795792932547941,
+      "learning_rate": 1.8666155632680938e-05,
+      "loss": 0.8128,
+      "step": 8130
+    },
+    {
+      "epoch": 0.17485071100227692,
+      "grad_norm": 0.5800071115326852,
+      "learning_rate": 1.8662752342361633e-05,
+      "loss": 0.8164,
+      "step": 8140
+    },
+    {
+      "epoch": 0.17506551531554754,
+      "grad_norm": 0.5845246581742584,
+      "learning_rate": 1.865934502691241e-05,
+      "loss": 0.8117,
+      "step": 8150
+    },
+    {
+      "epoch": 0.17528031962881815,
+      "grad_norm": 0.5894935518444326,
+      "learning_rate": 1.865593368791647e-05,
+      "loss": 0.8063,
+      "step": 8160
+    },
+    {
+      "epoch": 0.17549512394208877,
+      "grad_norm": 0.5877616770415447,
+      "learning_rate": 1.8652518326958882e-05,
+      "loss": 0.8112,
+      "step": 8170
+    },
+    {
+      "epoch": 0.17570992825535936,
+      "grad_norm": 0.6072757728168394,
+      "learning_rate": 1.8649098945626588e-05,
+      "loss": 0.8029,
+      "step": 8180
+    },
+    {
+      "epoch": 0.17592473256862998,
+      "grad_norm": 0.6010925475932754,
+      "learning_rate": 1.8645675545508397e-05,
+      "loss": 0.8162,
+      "step": 8190
+    },
+    {
+      "epoch": 0.1761395368819006,
+      "grad_norm": 0.5718419491303891,
+      "learning_rate": 1.864224812819498e-05,
+      "loss": 0.8124,
+      "step": 8200
+    },
+    {
+      "epoch": 0.1763543411951712,
+      "grad_norm": 0.5527383739528845,
+      "learning_rate": 1.863881669527889e-05,
+      "loss": 0.808,
+      "step": 8210
+    },
+    {
+      "epoch": 0.1765691455084418,
+      "grad_norm": 0.573823793660002,
+      "learning_rate": 1.8635381248354525e-05,
+      "loss": 0.8089,
+      "step": 8220
+    },
+    {
+      "epoch": 0.17678394982171242,
+      "grad_norm": 0.5638340296138096,
+      "learning_rate": 1.8631941789018162e-05,
+      "loss": 0.7961,
+      "step": 8230
+    },
+    {
+      "epoch": 0.17699875413498303,
+      "grad_norm": 0.5646934644189842,
+      "learning_rate": 1.862849831886794e-05,
+      "loss": 0.8187,
+      "step": 8240
+    },
+    {
+      "epoch": 0.17721355844825365,
+      "grad_norm": 0.5729549283565889,
+      "learning_rate": 1.862505083950386e-05,
+      "loss": 0.7946,
+      "step": 8250
+    },
+    {
+      "epoch": 0.17742836276152424,
+      "grad_norm": 0.5697299724643171,
+      "learning_rate": 1.8621599352527783e-05,
+      "loss": 0.7989,
+      "step": 8260
+    },
+    {
+      "epoch": 0.17764316707479486,
+      "grad_norm": 0.5742935431980787,
+      "learning_rate": 1.8618143859543436e-05,
+      "loss": 0.8068,
+      "step": 8270
+    },
+    {
+      "epoch": 0.17785797138806547,
+      "grad_norm": 0.6022944664204131,
+      "learning_rate": 1.861468436215641e-05,
+      "loss": 0.8316,
+      "step": 8280
+    },
+    {
+      "epoch": 0.1780727757013361,
+      "grad_norm": 0.6181521941813161,
+      "learning_rate": 1.861122086197415e-05,
+      "loss": 0.8297,
+      "step": 8290
+    },
+    {
+      "epoch": 0.1782875800146067,
+      "grad_norm": 0.5854718908960673,
+      "learning_rate": 1.8607753360605968e-05,
+      "loss": 0.8139,
+      "step": 8300
+    },
+    {
+      "epoch": 0.1785023843278773,
+      "grad_norm": 0.5677926977124813,
+      "learning_rate": 1.8604281859663027e-05,
+      "loss": 0.8067,
+      "step": 8310
+    },
+    {
+      "epoch": 0.1787171886411479,
+      "grad_norm": 1.2822660953314349,
+      "learning_rate": 1.8600806360758355e-05,
+      "loss": 0.7901,
+      "step": 8320
+    },
+    {
+      "epoch": 0.17893199295441853,
+      "grad_norm": 0.561871053213264,
+      "learning_rate": 1.8597326865506838e-05,
+      "loss": 0.8073,
+      "step": 8330
+    },
+    {
+      "epoch": 0.17914679726768915,
+      "grad_norm": 0.5900981652272886,
+      "learning_rate": 1.8593843375525205e-05,
+      "loss": 0.8019,
+      "step": 8340
+    },
+    {
+      "epoch": 0.17936160158095973,
+      "grad_norm": 0.6033017106283523,
+      "learning_rate": 1.8590355892432068e-05,
+      "loss": 0.8129,
+      "step": 8350
+    },
+    {
+      "epoch": 0.17957640589423035,
+      "grad_norm": 0.5755568064961115,
+      "learning_rate": 1.858686441784787e-05,
+      "loss": 0.825,
+      "step": 8360
+    },
+    {
+      "epoch": 0.17979121020750097,
+      "grad_norm": 0.5837961244746128,
+      "learning_rate": 1.8583368953394917e-05,
+      "loss": 0.8152,
+      "step": 8370
+    },
+    {
+      "epoch": 0.18000601452077158,
+      "grad_norm": 0.5830801987993122,
+      "learning_rate": 1.857986950069737e-05,
+      "loss": 0.7978,
+      "step": 8380
+    },
+    {
+      "epoch": 0.1802208188340422,
+      "grad_norm": 0.6071324798915776,
+      "learning_rate": 1.8576366061381246e-05,
+      "loss": 0.8173,
+      "step": 8390
+    },
+    {
+      "epoch": 0.1804356231473128,
+      "grad_norm": 0.5787242342053142,
+      "learning_rate": 1.857285863707441e-05,
+      "loss": 0.8266,
+      "step": 8400
+    },
+    {
+      "epoch": 0.1806504274605834,
+      "grad_norm": 0.5830266332748674,
+      "learning_rate": 1.8569347229406575e-05,
+      "loss": 0.8196,
+      "step": 8410
+    },
+    {
+      "epoch": 0.18086523177385402,
+      "grad_norm": 0.5826514559974909,
+      "learning_rate": 1.8565831840009312e-05,
+      "loss": 0.7978,
+      "step": 8420
+    },
+    {
+      "epoch": 0.18108003608712464,
+      "grad_norm": 0.5647693782228277,
+      "learning_rate": 1.856231247051604e-05,
+      "loss": 0.8152,
+      "step": 8430
+    },
+    {
+      "epoch": 0.18129484040039523,
+      "grad_norm": 0.6171161964315975,
+      "learning_rate": 1.8558789122562024e-05,
+      "loss": 0.805,
+      "step": 8440
+    },
+    {
+      "epoch": 0.18150964471366585,
+      "grad_norm": 0.574924762582272,
+      "learning_rate": 1.8555261797784387e-05,
+      "loss": 0.825,
+      "step": 8450
+    },
+    {
+      "epoch": 0.18172444902693646,
+      "grad_norm": 0.5849907741058937,
+      "learning_rate": 1.8551730497822086e-05,
+      "loss": 0.813,
+      "step": 8460
+    },
+    {
+      "epoch": 0.18193925334020708,
+      "grad_norm": 0.6006674682429172,
+      "learning_rate": 1.8548195224315933e-05,
+      "loss": 0.8058,
+      "step": 8470
+    },
+    {
+      "epoch": 0.18215405765347767,
+      "grad_norm": 0.5694938669138162,
+      "learning_rate": 1.854465597890859e-05,
+      "loss": 0.8128,
+      "step": 8480
+    },
+    {
+      "epoch": 0.18236886196674829,
+      "grad_norm": 0.5800620824875963,
+      "learning_rate": 1.8541112763244554e-05,
+      "loss": 0.8,
+      "step": 8490
+    },
+    {
+      "epoch": 0.1825836662800189,
+      "grad_norm": 0.6003822667618681,
+      "learning_rate": 1.8537565578970182e-05,
+      "loss": 0.8125,
+      "step": 8500
+    },
+    {
+      "epoch": 0.18279847059328952,
+      "grad_norm": 0.5805075835998008,
+      "learning_rate": 1.8534014427733655e-05,
+      "loss": 0.814,
+      "step": 8510
+    },
+    {
+      "epoch": 0.18301327490656014,
+      "grad_norm": 0.608140890949641,
+      "learning_rate": 1.8530459311185017e-05,
+      "loss": 0.8074,
+      "step": 8520
+    },
+    {
+      "epoch": 0.18322807921983073,
+      "grad_norm": 0.5939316987741534,
+      "learning_rate": 1.852690023097614e-05,
+      "loss": 0.8083,
+      "step": 8530
+    },
+    {
+      "epoch": 0.18344288353310134,
+      "grad_norm": 0.5634001132722806,
+      "learning_rate": 1.8523337188760747e-05,
+      "loss": 0.814,
+      "step": 8540
+    },
+    {
+      "epoch": 0.18365768784637196,
+      "grad_norm": 0.6076400115964997,
+      "learning_rate": 1.8519770186194393e-05,
+      "loss": 0.8131,
+      "step": 8550
+    },
+    {
+      "epoch": 0.18387249215964258,
+      "grad_norm": 0.5804582235616338,
+      "learning_rate": 1.851619922493448e-05,
+      "loss": 0.8026,
+      "step": 8560
+    },
+    {
+      "epoch": 0.18408729647291316,
+      "grad_norm": 0.5516633970987541,
+      "learning_rate": 1.8512624306640254e-05,
+      "loss": 0.8008,
+      "step": 8570
+    },
+    {
+      "epoch": 0.18430210078618378,
+      "grad_norm": 0.5977108054692924,
+      "learning_rate": 1.850904543297278e-05,
+      "loss": 0.8093,
+      "step": 8580
+    },
+    {
+      "epoch": 0.1845169050994544,
+      "grad_norm": 0.5751584053301783,
+      "learning_rate": 1.850546260559499e-05,
+      "loss": 0.8069,
+      "step": 8590
+    },
+    {
+      "epoch": 0.18473170941272501,
+      "grad_norm": 0.5683030910307522,
+      "learning_rate": 1.8501875826171632e-05,
+      "loss": 0.8088,
+      "step": 8600
+    },
+    {
+      "epoch": 0.18494651372599563,
+      "grad_norm": 0.5981914621652948,
+      "learning_rate": 1.8498285096369287e-05,
+      "loss": 0.815,
+      "step": 8610
+    },
+    {
+      "epoch": 0.18516131803926622,
+      "grad_norm": 0.5987116882619,
+      "learning_rate": 1.849469041785639e-05,
+      "loss": 0.8247,
+      "step": 8620
+    },
+    {
+      "epoch": 0.18537612235253684,
+      "grad_norm": 0.598340119849919,
+      "learning_rate": 1.8491091792303203e-05,
+      "loss": 0.7891,
+      "step": 8630
+    },
+    {
+      "epoch": 0.18559092666580745,
+      "grad_norm": 0.5874259069859069,
+      "learning_rate": 1.8487489221381807e-05,
+      "loss": 0.7963,
+      "step": 8640
+    },
+    {
+      "epoch": 0.18580573097907807,
+      "grad_norm": 0.60338871634641,
+      "learning_rate": 1.848388270676614e-05,
+      "loss": 0.8176,
+      "step": 8650
+    },
+    {
+      "epoch": 0.18602053529234866,
+      "grad_norm": 0.6382989451380708,
+      "learning_rate": 1.848027225013196e-05,
+      "loss": 0.8034,
+      "step": 8660
+    },
+    {
+      "epoch": 0.18623533960561928,
+      "grad_norm": 0.6100978833111902,
+      "learning_rate": 1.8476657853156854e-05,
+      "loss": 0.8295,
+      "step": 8670
+    },
+    {
+      "epoch": 0.1864501439188899,
+      "grad_norm": 0.5847210050409527,
+      "learning_rate": 1.8473039517520245e-05,
+      "loss": 0.8093,
+      "step": 8680
+    },
+    {
+      "epoch": 0.1866649482321605,
+      "grad_norm": 0.5720679964643604,
+      "learning_rate": 1.846941724490339e-05,
+      "loss": 0.7988,
+      "step": 8690
+    },
+    {
+      "epoch": 0.1868797525454311,
+      "grad_norm": 0.6251350978467913,
+      "learning_rate": 1.8465791036989367e-05,
+      "loss": 0.7998,
+      "step": 8700
+    },
+    {
+      "epoch": 0.18709455685870172,
+      "grad_norm": 0.5765117453954964,
+      "learning_rate": 1.8462160895463087e-05,
+      "loss": 0.8121,
+      "step": 8710
+    },
+    {
+      "epoch": 0.18730936117197233,
+      "grad_norm": 0.5652462786030213,
+      "learning_rate": 1.8458526822011283e-05,
+      "loss": 0.8102,
+      "step": 8720
+    },
+    {
+      "epoch": 0.18752416548524295,
+      "grad_norm": 0.5618404540951464,
+      "learning_rate": 1.8454888818322527e-05,
+      "loss": 0.7841,
+      "step": 8730
+    },
+    {
+      "epoch": 0.18773896979851357,
+      "grad_norm": 0.5715079105467191,
+      "learning_rate": 1.8451246886087207e-05,
+      "loss": 0.8019,
+      "step": 8740
+    },
+    {
+      "epoch": 0.18795377411178416,
+      "grad_norm": 0.5786067429900265,
+      "learning_rate": 1.8447601026997534e-05,
+      "loss": 0.8185,
+      "step": 8750
+    },
+    {
+      "epoch": 0.18816857842505477,
+      "grad_norm": 0.5618942537054571,
+      "learning_rate": 1.8443951242747558e-05,
+      "loss": 0.8122,
+      "step": 8760
+    },
+    {
+      "epoch": 0.1883833827383254,
+      "grad_norm": 0.5602005480295104,
+      "learning_rate": 1.8440297535033137e-05,
+      "loss": 0.8044,
+      "step": 8770
+    },
+    {
+      "epoch": 0.188598187051596,
+      "grad_norm": 0.5526135225587552,
+      "learning_rate": 1.8436639905551966e-05,
+      "loss": 0.8132,
+      "step": 8780
+    },
+    {
+      "epoch": 0.1888129913648666,
+      "grad_norm": 0.5708051150082964,
+      "learning_rate": 1.8432978356003544e-05,
+      "loss": 0.7984,
+      "step": 8790
+    },
+    {
+      "epoch": 0.1890277956781372,
+      "grad_norm": 0.5845241545642078,
+      "learning_rate": 1.842931288808921e-05,
+      "loss": 0.8215,
+      "step": 8800
+    },
+    {
+      "epoch": 0.18924259999140783,
+      "grad_norm": 0.5747369107161527,
+      "learning_rate": 1.8425643503512115e-05,
+      "loss": 0.8027,
+      "step": 8810
+    },
+    {
+      "epoch": 0.18945740430467845,
+      "grad_norm": 0.5726755466246907,
+      "learning_rate": 1.842197020397723e-05,
+      "loss": 0.7891,
+      "step": 8820
+    },
+    {
+      "epoch": 0.18967220861794906,
+      "grad_norm": 0.5757944229048639,
+      "learning_rate": 1.841829299119135e-05,
+      "loss": 0.8064,
+      "step": 8830
+    },
+    {
+      "epoch": 0.18988701293121965,
+      "grad_norm": 0.5911903584887567,
+      "learning_rate": 1.8414611866863078e-05,
+      "loss": 0.8239,
+      "step": 8840
+    },
+    {
+      "epoch": 0.19010181724449027,
+      "grad_norm": 0.5784649631481444,
+      "learning_rate": 1.8410926832702842e-05,
+      "loss": 0.8074,
+      "step": 8850
+    },
+    {
+      "epoch": 0.19031662155776088,
+      "grad_norm": 0.5930932814488833,
+      "learning_rate": 1.8407237890422888e-05,
+      "loss": 0.8226,
+      "step": 8860
+    },
+    {
+      "epoch": 0.1905314258710315,
+      "grad_norm": 0.5579023925128831,
+      "learning_rate": 1.8403545041737275e-05,
+      "loss": 0.8008,
+      "step": 8870
+    },
+    {
+      "epoch": 0.1907462301843021,
+      "grad_norm": 0.5988868685379616,
+      "learning_rate": 1.8399848288361878e-05,
+      "loss": 0.8202,
+      "step": 8880
+    },
+    {
+      "epoch": 0.1909610344975727,
+      "grad_norm": 0.5633872393293222,
+      "learning_rate": 1.8396147632014383e-05,
+      "loss": 0.7865,
+      "step": 8890
+    },
+    {
+      "epoch": 0.19117583881084332,
+      "grad_norm": 0.579775369575011,
+      "learning_rate": 1.8392443074414292e-05,
+      "loss": 0.8081,
+      "step": 8900
+    },
+    {
+      "epoch": 0.19139064312411394,
+      "grad_norm": 0.5761284131901169,
+      "learning_rate": 1.8388734617282926e-05,
+      "loss": 0.7957,
+      "step": 8910
+    },
+    {
+      "epoch": 0.19160544743738453,
+      "grad_norm": 0.5790288426152821,
+      "learning_rate": 1.8385022262343405e-05,
+      "loss": 0.8016,
+      "step": 8920
+    },
+    {
+      "epoch": 0.19182025175065515,
+      "grad_norm": 0.5835056469508209,
+      "learning_rate": 1.8381306011320673e-05,
+      "loss": 0.8192,
+      "step": 8930
+    },
+    {
+      "epoch": 0.19203505606392576,
+      "grad_norm": 0.5600680666644086,
+      "learning_rate": 1.837758586594147e-05,
+      "loss": 0.8041,
+      "step": 8940
+    },
+    {
+      "epoch": 0.19224986037719638,
+      "grad_norm": 0.6130002119585894,
+      "learning_rate": 1.8373861827934364e-05,
+      "loss": 0.8173,
+      "step": 8950
+    },
+    {
+      "epoch": 0.192464664690467,
+      "grad_norm": 0.5673926783038583,
+      "learning_rate": 1.8370133899029717e-05,
+      "loss": 0.8119,
+      "step": 8960
+    },
+    {
+      "epoch": 0.19267946900373759,
+      "grad_norm": 0.5917000843066154,
+      "learning_rate": 1.83664020809597e-05,
+      "loss": 0.8172,
+      "step": 8970
+    },
+    {
+      "epoch": 0.1928942733170082,
+      "grad_norm": 0.5733143994810274,
+      "learning_rate": 1.83626663754583e-05,
+      "loss": 0.8032,
+      "step": 8980
+    },
+    {
+      "epoch": 0.19310907763027882,
+      "grad_norm": 0.5592658911509888,
+      "learning_rate": 1.8358926784261303e-05,
+      "loss": 0.7981,
+      "step": 8990
+    },
+    {
+      "epoch": 0.19332388194354944,
+      "grad_norm": 0.562962728797357,
+      "learning_rate": 1.8355183309106297e-05,
+      "loss": 0.8156,
+      "step": 9000
+    },
+    {
+      "epoch": 0.19353868625682003,
+      "grad_norm": 0.569803487699369,
+      "learning_rate": 1.8351435951732692e-05,
+      "loss": 0.8001,
+      "step": 9010
+    },
+    {
+      "epoch": 0.19375349057009064,
+      "grad_norm": 0.5789202611510278,
+      "learning_rate": 1.8347684713881675e-05,
+      "loss": 0.8116,
+      "step": 9020
+    },
+    {
+      "epoch": 0.19396829488336126,
+      "grad_norm": 0.5768841779522369,
+      "learning_rate": 1.8343929597296265e-05,
+      "loss": 0.8075,
+      "step": 9030
+    },
+    {
+      "epoch": 0.19418309919663188,
+      "grad_norm": 0.5643884618141694,
+      "learning_rate": 1.8340170603721258e-05,
+      "loss": 0.8005,
+      "step": 9040
+    },
+    {
+      "epoch": 0.1943979035099025,
+      "grad_norm": 0.6040260449303918,
+      "learning_rate": 1.8336407734903266e-05,
+      "loss": 0.8037,
+      "step": 9050
+    },
+    {
+      "epoch": 0.19461270782317308,
+      "grad_norm": 0.547010947358847,
+      "learning_rate": 1.83326409925907e-05,
+      "loss": 0.8086,
+      "step": 9060
+    },
+    {
+      "epoch": 0.1948275121364437,
+      "grad_norm": 0.5791718088895571,
+      "learning_rate": 1.8328870378533774e-05,
+      "loss": 0.8221,
+      "step": 9070
+    },
+    {
+      "epoch": 0.19504231644971431,
+      "grad_norm": 0.5929678639479692,
+      "learning_rate": 1.8325095894484487e-05,
+      "loss": 0.8037,
+      "step": 9080
+    },
+    {
+      "epoch": 0.19525712076298493,
+      "grad_norm": 0.5837325290517947,
+      "learning_rate": 1.8321317542196645e-05,
+      "loss": 0.8214,
+      "step": 9090
+    },
+    {
+      "epoch": 0.19547192507625552,
+      "grad_norm": 0.6021341746404556,
+      "learning_rate": 1.831753532342586e-05,
+      "loss": 0.808,
+      "step": 9100
+    },
+    {
+      "epoch": 0.19568672938952614,
+      "grad_norm": 0.5851595859581337,
+      "learning_rate": 1.8313749239929527e-05,
+      "loss": 0.8027,
+      "step": 9110
+    },
+    {
+      "epoch": 0.19590153370279675,
+      "grad_norm": 0.6429053021754854,
+      "learning_rate": 1.830995929346684e-05,
+      "loss": 0.8165,
+      "step": 9120
+    },
+    {
+      "epoch": 0.19611633801606737,
+      "grad_norm": 0.6030014532201955,
+      "learning_rate": 1.830616548579879e-05,
+      "loss": 0.8246,
+      "step": 9130
+    },
+    {
+      "epoch": 0.19633114232933796,
+      "grad_norm": 0.6323373448001347,
+      "learning_rate": 1.8302367818688168e-05,
+      "loss": 0.8079,
+      "step": 9140
+    },
+    {
+      "epoch": 0.19654594664260858,
+      "grad_norm": 0.5846336022969297,
+      "learning_rate": 1.8298566293899543e-05,
+      "loss": 0.8269,
+      "step": 9150
+    },
+    {
+      "epoch": 0.1967607509558792,
+      "grad_norm": 0.5607765403444699,
+      "learning_rate": 1.8294760913199296e-05,
+      "loss": 0.7932,
+      "step": 9160
+    },
+    {
+      "epoch": 0.1969755552691498,
+      "grad_norm": 0.5738811351652273,
+      "learning_rate": 1.8290951678355583e-05,
+      "loss": 0.8121,
+      "step": 9170
+    },
+    {
+      "epoch": 0.19719035958242043,
+      "grad_norm": 0.5893344214505519,
+      "learning_rate": 1.8287138591138355e-05,
+      "loss": 0.8227,
+      "step": 9180
+    },
+    {
+      "epoch": 0.19740516389569102,
+      "grad_norm": 0.5849184704039714,
+      "learning_rate": 1.828332165331936e-05,
+      "loss": 0.8037,
+      "step": 9190
+    },
+    {
+      "epoch": 0.19761996820896163,
+      "grad_norm": 0.5844113484112203,
+      "learning_rate": 1.8279500866672124e-05,
+      "loss": 0.814,
+      "step": 9200
+    },
+    {
+      "epoch": 0.19783477252223225,
+      "grad_norm": 0.5652668990930761,
+      "learning_rate": 1.8275676232971977e-05,
+      "loss": 0.8119,
+      "step": 9210
+    },
+    {
+      "epoch": 0.19804957683550287,
+      "grad_norm": 0.597730632354087,
+      "learning_rate": 1.827184775399602e-05,
+      "loss": 0.8053,
+      "step": 9220
+    },
+    {
+      "epoch": 0.19826438114877346,
+      "grad_norm": 0.5888675343237012,
+      "learning_rate": 1.8268015431523147e-05,
+      "loss": 0.801,
+      "step": 9230
+    },
+    {
+      "epoch": 0.19847918546204407,
+      "grad_norm": 0.5663470763977005,
+      "learning_rate": 1.8264179267334043e-05,
+      "loss": 0.8135,
+      "step": 9240
+    },
+    {
+      "epoch": 0.1986939897753147,
+      "grad_norm": 0.5533581191534693,
+      "learning_rate": 1.8260339263211174e-05,
+      "loss": 0.79,
+      "step": 9250
+    },
+    {
+      "epoch": 0.1989087940885853,
+      "grad_norm": 0.5979441715439602,
+      "learning_rate": 1.8256495420938786e-05,
+      "loss": 0.7929,
+      "step": 9260
+    },
+    {
+      "epoch": 0.19912359840185592,
+      "grad_norm": 0.5825141443552068,
+      "learning_rate": 1.8252647742302914e-05,
+      "loss": 0.8056,
+      "step": 9270
+    },
+    {
+      "epoch": 0.1993384027151265,
+      "grad_norm": 0.5757431007161539,
+      "learning_rate": 1.8248796229091373e-05,
+      "loss": 0.8138,
+      "step": 9280
+    },
+    {
+      "epoch": 0.19955320702839713,
+      "grad_norm": 0.6259334303275133,
+      "learning_rate": 1.8244940883093767e-05,
+      "loss": 0.813,
+      "step": 9290
+    },
+    {
+      "epoch": 0.19976801134166774,
+      "grad_norm": 0.549857706300012,
+      "learning_rate": 1.824108170610147e-05,
+      "loss": 0.8077,
+      "step": 9300
+    },
+    {
+      "epoch": 0.19998281565493836,
+      "grad_norm": 0.56236031168535,
+      "learning_rate": 1.8237218699907635e-05,
+      "loss": 0.8045,
+      "step": 9310
+    },
+    {
+      "epoch": 0.20019761996820895,
+      "grad_norm": 0.6015575274536221,
+      "learning_rate": 1.8233351866307206e-05,
+      "loss": 0.8061,
+      "step": 9320
+    },
+    {
+      "epoch": 0.20041242428147957,
+      "grad_norm": 0.6085710236514931,
+      "learning_rate": 1.8229481207096896e-05,
+      "loss": 0.8018,
+      "step": 9330
+    },
+    {
+      "epoch": 0.20062722859475018,
+      "grad_norm": 0.597465062355765,
+      "learning_rate": 1.8225606724075205e-05,
+      "loss": 0.8122,
+      "step": 9340
+    },
+    {
+      "epoch": 0.2008420329080208,
+      "grad_norm": 0.5653695698060156,
+      "learning_rate": 1.82217284190424e-05,
+      "loss": 0.7998,
+      "step": 9350
+    },
+    {
+      "epoch": 0.2010568372212914,
+      "grad_norm": 0.5803499948562468,
+      "learning_rate": 1.8217846293800523e-05,
+      "loss": 0.8171,
+      "step": 9360
+    },
+    {
+      "epoch": 0.201271641534562,
+      "grad_norm": 0.5914980494379427,
+      "learning_rate": 1.82139603501534e-05,
+      "loss": 0.8078,
+      "step": 9370
+    },
+    {
+      "epoch": 0.20148644584783262,
+      "grad_norm": 0.5753469591052403,
+      "learning_rate": 1.8210070589906628e-05,
+      "loss": 0.8144,
+      "step": 9380
+    },
+    {
+      "epoch": 0.20170125016110324,
+      "grad_norm": 0.5917347916307549,
+      "learning_rate": 1.8206177014867572e-05,
+      "loss": 0.8078,
+      "step": 9390
+    },
+    {
+      "epoch": 0.20191605447437386,
+      "grad_norm": 0.5524089908557045,
+      "learning_rate": 1.8202279626845377e-05,
+      "loss": 0.8028,
+      "step": 9400
+    },
+    {
+      "epoch": 0.20213085878764445,
+      "grad_norm": 0.5529213134018454,
+      "learning_rate": 1.8198378427650955e-05,
+      "loss": 0.7994,
+      "step": 9410
+    },
+    {
+      "epoch": 0.20234566310091506,
+      "grad_norm": 0.567892445973469,
+      "learning_rate": 1.8194473419096987e-05,
+      "loss": 0.8094,
+      "step": 9420
+    },
+    {
+      "epoch": 0.20256046741418568,
+      "grad_norm": 0.5858778724872428,
+      "learning_rate": 1.819056460299793e-05,
+      "loss": 0.7983,
+      "step": 9430
+    },
+    {
+      "epoch": 0.2027752717274563,
+      "grad_norm": 0.573018383200498,
+      "learning_rate": 1.818665198117001e-05,
+      "loss": 0.8113,
+      "step": 9440
+    },
+    {
+      "epoch": 0.20299007604072689,
+      "grad_norm": 0.5815406945956942,
+      "learning_rate": 1.8182735555431213e-05,
+      "loss": 0.8131,
+      "step": 9450
+    },
+    {
+      "epoch": 0.2032048803539975,
+      "grad_norm": 0.5712680885146894,
+      "learning_rate": 1.8178815327601306e-05,
+      "loss": 0.8,
+      "step": 9460
+    },
+    {
+      "epoch": 0.20341968466726812,
+      "grad_norm": 0.7090531423336768,
+      "learning_rate": 1.8174891299501807e-05,
+      "loss": 0.8117,
+      "step": 9470
+    },
+    {
+      "epoch": 0.20363448898053874,
+      "grad_norm": 0.6104073346551036,
+      "learning_rate": 1.8170963472956006e-05,
+      "loss": 0.8203,
+      "step": 9480
+    },
+    {
+      "epoch": 0.20384929329380935,
+      "grad_norm": 0.5768361886322763,
+      "learning_rate": 1.816703184978897e-05,
+      "loss": 0.8079,
+      "step": 9490
+    },
+    {
+      "epoch": 0.20406409760707994,
+      "grad_norm": 0.5351105176871881,
+      "learning_rate": 1.816309643182751e-05,
+      "loss": 0.7961,
+      "step": 9500
+    },
+    {
+      "epoch": 0.20427890192035056,
+      "grad_norm": 0.5636835938070731,
+      "learning_rate": 1.8159157220900216e-05,
+      "loss": 0.7832,
+      "step": 9510
+    },
+    {
+      "epoch": 0.20449370623362118,
+      "grad_norm": 0.5730516941042331,
+      "learning_rate": 1.815521421883743e-05,
+      "loss": 0.7937,
+      "step": 9520
+    },
+    {
+      "epoch": 0.2047085105468918,
+      "grad_norm": 0.5733890782084365,
+      "learning_rate": 1.815126742747126e-05,
+      "loss": 0.8011,
+      "step": 9530
+    },
+    {
+      "epoch": 0.20492331486016238,
+      "grad_norm": 0.5616802718550227,
+      "learning_rate": 1.8147316848635576e-05,
+      "loss": 0.8038,
+      "step": 9540
+    },
+    {
+      "epoch": 0.205138119173433,
+      "grad_norm": 0.5889286620725945,
+      "learning_rate": 1.8143362484166004e-05,
+      "loss": 0.8021,
+      "step": 9550
+    },
+    {
+      "epoch": 0.20535292348670361,
+      "grad_norm": 0.5585779556583447,
+      "learning_rate": 1.8139404335899937e-05,
+      "loss": 0.8058,
+      "step": 9560
+    },
+    {
+      "epoch": 0.20556772779997423,
+      "grad_norm": 0.571453117672738,
+      "learning_rate": 1.813544240567651e-05,
+      "loss": 0.804,
+      "step": 9570
+    },
+    {
+      "epoch": 0.20578253211324482,
+      "grad_norm": 0.5654185920520682,
+      "learning_rate": 1.8131476695336636e-05,
+      "loss": 0.802,
+      "step": 9580
+    },
+    {
+      "epoch": 0.20599733642651544,
+      "grad_norm": 0.5844027397396927,
+      "learning_rate": 1.8127507206722973e-05,
+      "loss": 0.8019,
+      "step": 9590
+    },
+    {
+      "epoch": 0.20621214073978605,
+      "grad_norm": 0.5689638203458927,
+      "learning_rate": 1.8123533941679928e-05,
+      "loss": 0.8094,
+      "step": 9600
+    },
+    {
+      "epoch": 0.20642694505305667,
+      "grad_norm": 0.577234661880537,
+      "learning_rate": 1.8119556902053678e-05,
+      "loss": 0.8246,
+      "step": 9610
+    },
+    {
+      "epoch": 0.2066417493663273,
+      "grad_norm": 0.5763434534447465,
+      "learning_rate": 1.811557608969214e-05,
+      "loss": 0.8158,
+      "step": 9620
+    },
+    {
+      "epoch": 0.20685655367959788,
+      "grad_norm": 0.5778688135687731,
+      "learning_rate": 1.8111591506444997e-05,
+      "loss": 0.7955,
+      "step": 9630
+    },
+    {
+      "epoch": 0.2070713579928685,
+      "grad_norm": 0.5813830588006104,
+      "learning_rate": 1.810760315416367e-05,
+      "loss": 0.7891,
+      "step": 9640
+    },
+    {
+      "epoch": 0.2072861623061391,
+      "grad_norm": 0.576252297984537,
+      "learning_rate": 1.8103611034701348e-05,
+      "loss": 0.8085,
+      "step": 9650
+    },
+    {
+      "epoch": 0.20750096661940973,
+      "grad_norm": 0.5837430318834449,
+      "learning_rate": 1.8099615149912953e-05,
+      "loss": 0.807,
+      "step": 9660
+    },
+    {
+      "epoch": 0.20771577093268032,
+      "grad_norm": 0.5890346965782849,
+      "learning_rate": 1.8095615501655166e-05,
+      "loss": 0.7944,
+      "step": 9670
+    },
+    {
+      "epoch": 0.20793057524595093,
+      "grad_norm": 0.5852131021460842,
+      "learning_rate": 1.8091612091786416e-05,
+      "loss": 0.8088,
+      "step": 9680
+    },
+    {
+      "epoch": 0.20814537955922155,
+      "grad_norm": 0.5589031639536682,
+      "learning_rate": 1.8087604922166884e-05,
+      "loss": 0.8121,
+      "step": 9690
+    },
+    {
+      "epoch": 0.20836018387249217,
+      "grad_norm": 0.5837838090958873,
+      "learning_rate": 1.8083593994658483e-05,
+      "loss": 0.8031,
+      "step": 9700
+    },
+    {
+      "epoch": 0.20857498818576278,
+      "grad_norm": 0.5759528081913161,
+      "learning_rate": 1.807957931112489e-05,
+      "loss": 0.7909,
+      "step": 9710
+    },
+    {
+      "epoch": 0.20878979249903337,
+      "grad_norm": 0.6143282377901799,
+      "learning_rate": 1.807556087343152e-05,
+      "loss": 0.8115,
+      "step": 9720
+    },
+    {
+      "epoch": 0.209004596812304,
+      "grad_norm": 0.578534081939633,
+      "learning_rate": 1.8071538683445524e-05,
+      "loss": 0.8073,
+      "step": 9730
+    },
+    {
+      "epoch": 0.2092194011255746,
+      "grad_norm": 0.5990944031783112,
+      "learning_rate": 1.806751274303581e-05,
+      "loss": 0.8145,
+      "step": 9740
+    },
+    {
+      "epoch": 0.20943420543884522,
+      "grad_norm": 0.5374932915323056,
+      "learning_rate": 1.8063483054073026e-05,
+      "loss": 0.8186,
+      "step": 9750
+    },
+    {
+      "epoch": 0.2096490097521158,
+      "grad_norm": 0.573621801810147,
+      "learning_rate": 1.805944961842955e-05,
+      "loss": 0.7858,
+      "step": 9760
+    },
+    {
+      "epoch": 0.20986381406538643,
+      "grad_norm": 0.5667771093016414,
+      "learning_rate": 1.8055412437979517e-05,
+      "loss": 0.815,
+      "step": 9770
+    },
+    {
+      "epoch": 0.21007861837865704,
+      "grad_norm": 0.5777392591556257,
+      "learning_rate": 1.805137151459879e-05,
+      "loss": 0.8099,
+      "step": 9780
+    },
+    {
+      "epoch": 0.21029342269192766,
+      "grad_norm": 0.5716970868389603,
+      "learning_rate": 1.804732685016498e-05,
+      "loss": 0.7963,
+      "step": 9790
+    },
+    {
+      "epoch": 0.21050822700519825,
+      "grad_norm": 0.5784937064403081,
+      "learning_rate": 1.8043278446557424e-05,
+      "loss": 0.7985,
+      "step": 9800
+    },
+    {
+      "epoch": 0.21072303131846887,
+      "grad_norm": 0.5883045627604655,
+      "learning_rate": 1.8039226305657212e-05,
+      "loss": 0.8074,
+      "step": 9810
+    },
+    {
+      "epoch": 0.21093783563173948,
+      "grad_norm": 0.5514729284429919,
+      "learning_rate": 1.803517042934716e-05,
+      "loss": 0.8004,
+      "step": 9820
+    },
+    {
+      "epoch": 0.2111526399450101,
+      "grad_norm": 0.5816019682238576,
+      "learning_rate": 1.8031110819511825e-05,
+      "loss": 0.8012,
+      "step": 9830
+    },
+    {
+      "epoch": 0.21136744425828072,
+      "grad_norm": 0.5870070746603354,
+      "learning_rate": 1.8027047478037495e-05,
+      "loss": 0.8057,
+      "step": 9840
+    },
+    {
+      "epoch": 0.2115822485715513,
+      "grad_norm": 0.6076535851376632,
+      "learning_rate": 1.802298040681219e-05,
+      "loss": 0.8269,
+      "step": 9850
+    },
+    {
+      "epoch": 0.21179705288482192,
+      "grad_norm": 0.5748747580649357,
+      "learning_rate": 1.801890960772567e-05,
+      "loss": 0.8158,
+      "step": 9860
+    },
+    {
+      "epoch": 0.21201185719809254,
+      "grad_norm": 0.5786976018669034,
+      "learning_rate": 1.801483508266942e-05,
+      "loss": 0.7988,
+      "step": 9870
+    },
+    {
+      "epoch": 0.21222666151136316,
+      "grad_norm": 0.540488437271146,
+      "learning_rate": 1.8010756833536663e-05,
+      "loss": 0.8049,
+      "step": 9880
+    },
+    {
+      "epoch": 0.21244146582463375,
+      "grad_norm": 0.6221560289316503,
+      "learning_rate": 1.800667486222235e-05,
+      "loss": 0.7987,
+      "step": 9890
+    },
+    {
+      "epoch": 0.21265627013790436,
+      "grad_norm": 0.5961921503665486,
+      "learning_rate": 1.8002589170623155e-05,
+      "loss": 0.8264,
+      "step": 9900
+    },
+    {
+      "epoch": 0.21287107445117498,
+      "grad_norm": 0.5741672373117042,
+      "learning_rate": 1.7998499760637492e-05,
+      "loss": 0.798,
+      "step": 9910
+    },
+    {
+      "epoch": 0.2130858787644456,
+      "grad_norm": 0.5615151845280278,
+      "learning_rate": 1.7994406634165492e-05,
+      "loss": 0.8067,
+      "step": 9920
+    },
+    {
+      "epoch": 0.2133006830777162,
+      "grad_norm": 0.5836885054107523,
+      "learning_rate": 1.799030979310902e-05,
+      "loss": 0.7914,
+      "step": 9930
+    },
+    {
+      "epoch": 0.2135154873909868,
+      "grad_norm": 0.5764046639257419,
+      "learning_rate": 1.7986209239371665e-05,
+      "loss": 0.7975,
+      "step": 9940
+    },
+    {
+      "epoch": 0.21373029170425742,
+      "grad_norm": 0.5848553181136704,
+      "learning_rate": 1.798210497485874e-05,
+      "loss": 0.802,
+      "step": 9950
+    },
+    {
+      "epoch": 0.21394509601752804,
+      "grad_norm": 0.5739518953663529,
+      "learning_rate": 1.7977997001477285e-05,
+      "loss": 0.8157,
+      "step": 9960
+    },
+    {
+      "epoch": 0.21415990033079865,
+      "grad_norm": 0.5929025643332227,
+      "learning_rate": 1.797388532113606e-05,
+      "loss": 0.7859,
+      "step": 9970
+    },
+    {
+      "epoch": 0.21437470464406924,
+      "grad_norm": 0.5661521287114709,
+      "learning_rate": 1.7969769935745544e-05,
+      "loss": 0.7999,
+      "step": 9980
+    },
+    {
+      "epoch": 0.21458950895733986,
+      "grad_norm": 0.5422378316949167,
+      "learning_rate": 1.796565084721795e-05,
+      "loss": 0.7842,
+      "step": 9990
+    },
+    {
+      "epoch": 0.21480431327061048,
+      "grad_norm": 0.6043847597224838,
+      "learning_rate": 1.79615280574672e-05,
+      "loss": 0.8107,
+      "step": 10000
+    },
+    {
+      "epoch": 0.2150191175838811,
+      "grad_norm": 0.588943789547453,
+      "learning_rate": 1.795740156840894e-05,
+      "loss": 0.8095,
+      "step": 10010
+    },
+    {
+      "epoch": 0.21523392189715168,
+      "grad_norm": 0.5773036210661655,
+      "learning_rate": 1.7953271381960536e-05,
+      "loss": 0.8043,
+      "step": 10020
+    },
+    {
+      "epoch": 0.2154487262104223,
+      "grad_norm": 0.5758207275091537,
+      "learning_rate": 1.794913750004107e-05,
+      "loss": 0.793,
+      "step": 10030
+    },
+    {
+      "epoch": 0.21566353052369291,
+      "grad_norm": 0.5566484791876923,
+      "learning_rate": 1.7944999924571345e-05,
+      "loss": 0.7877,
+      "step": 10040
+    },
+    {
+      "epoch": 0.21587833483696353,
+      "grad_norm": 0.5695977896565292,
+      "learning_rate": 1.7940858657473867e-05,
+      "loss": 0.7956,
+      "step": 10050
+    },
+    {
+      "epoch": 0.21609313915023415,
+      "grad_norm": 0.5792438417645663,
+      "learning_rate": 1.7936713700672874e-05,
+      "loss": 0.7905,
+      "step": 10060
+    },
+    {
+      "epoch": 0.21630794346350474,
+      "grad_norm": 0.5702659128860479,
+      "learning_rate": 1.7932565056094312e-05,
+      "loss": 0.8087,
+      "step": 10070
+    },
+    {
+      "epoch": 0.21652274777677535,
+      "grad_norm": 0.6188318728871035,
+      "learning_rate": 1.7928412725665844e-05,
+      "loss": 0.8039,
+      "step": 10080
+    },
+    {
+      "epoch": 0.21673755209004597,
+      "grad_norm": 0.5606872536463133,
+      "learning_rate": 1.792425671131683e-05,
+      "loss": 0.7881,
+      "step": 10090
+    },
+    {
+      "epoch": 0.2169523564033166,
+      "grad_norm": 0.5722706317917913,
+      "learning_rate": 1.792009701497836e-05,
+      "loss": 0.7996,
+      "step": 10100
+    },
+    {
+      "epoch": 0.21716716071658718,
+      "grad_norm": 0.7107815425462004,
+      "learning_rate": 1.791593363858323e-05,
+      "loss": 0.8148,
+      "step": 10110
+    },
+    {
+      "epoch": 0.2173819650298578,
+      "grad_norm": 0.5883477981610786,
+      "learning_rate": 1.7911766584065945e-05,
+      "loss": 0.7909,
+      "step": 10120
+    },
+    {
+      "epoch": 0.2175967693431284,
+      "grad_norm": 0.5798116807331893,
+      "learning_rate": 1.7907595853362713e-05,
+      "loss": 0.7983,
+      "step": 10130
+    },
+    {
+      "epoch": 0.21781157365639903,
+      "grad_norm": 0.5609531537994755,
+      "learning_rate": 1.790342144841146e-05,
+      "loss": 0.7961,
+      "step": 10140
+    },
+    {
+      "epoch": 0.21802637796966964,
+      "grad_norm": 0.6517122093198296,
+      "learning_rate": 1.7899243371151813e-05,
+      "loss": 0.7861,
+      "step": 10150
+    },
+    {
+      "epoch": 0.21824118228294023,
+      "grad_norm": 0.5516309036102172,
+      "learning_rate": 1.7895061623525104e-05,
+      "loss": 0.8067,
+      "step": 10160
+    },
+    {
+      "epoch": 0.21845598659621085,
+      "grad_norm": 0.5613471363551992,
+      "learning_rate": 1.789087620747438e-05,
+      "loss": 0.7923,
+      "step": 10170
+    },
+    {
+      "epoch": 0.21867079090948147,
+      "grad_norm": 0.9188778029201942,
+      "learning_rate": 1.788668712494438e-05,
+      "loss": 0.8096,
+      "step": 10180
+    },
+    {
+      "epoch": 0.21888559522275208,
+      "grad_norm": 0.5617547681562008,
+      "learning_rate": 1.7882494377881558e-05,
+      "loss": 0.7964,
+      "step": 10190
+    },
+    {
+      "epoch": 0.21910039953602267,
+      "grad_norm": 0.61534749263247,
+      "learning_rate": 1.7878297968234055e-05,
+      "loss": 0.7946,
+      "step": 10200
+    },
+    {
+      "epoch": 0.2193152038492933,
+      "grad_norm": 0.5782211052832341,
+      "learning_rate": 1.7874097897951737e-05,
+      "loss": 0.7873,
+      "step": 10210
+    },
+    {
+      "epoch": 0.2195300081625639,
+      "grad_norm": 0.5514543385854074,
+      "learning_rate": 1.786989416898615e-05,
+      "loss": 0.7975,
+      "step": 10220
+    },
+    {
+      "epoch": 0.21974481247583452,
+      "grad_norm": 0.5480749170558831,
+      "learning_rate": 1.786568678329055e-05,
+      "loss": 0.8018,
+      "step": 10230
+    },
+    {
+      "epoch": 0.21995961678910514,
+      "grad_norm": 0.5783322467533664,
+      "learning_rate": 1.7861475742819885e-05,
+      "loss": 0.7848,
+      "step": 10240
+    },
+    {
+      "epoch": 0.22017442110237573,
+      "grad_norm": 0.5737996720368215,
+      "learning_rate": 1.7857261049530817e-05,
+      "loss": 0.7953,
+      "step": 10250
+    },
+    {
+      "epoch": 0.22038922541564634,
+      "grad_norm": 0.5763402239431955,
+      "learning_rate": 1.7853042705381684e-05,
+      "loss": 0.7851,
+      "step": 10260
+    },
+    {
+      "epoch": 0.22060402972891696,
+      "grad_norm": 0.5731754369723199,
+      "learning_rate": 1.7848820712332542e-05,
+      "loss": 0.8126,
+      "step": 10270
+    },
+    {
+      "epoch": 0.22081883404218758,
+      "grad_norm": 0.5772583690128964,
+      "learning_rate": 1.784459507234512e-05,
+      "loss": 0.82,
+      "step": 10280
+    },
+    {
+      "epoch": 0.22103363835545817,
+      "grad_norm": 0.5411569999708086,
+      "learning_rate": 1.7840365787382858e-05,
+      "loss": 0.7931,
+      "step": 10290
+    },
+    {
+      "epoch": 0.22124844266872878,
+      "grad_norm": 0.5653187339476065,
+      "learning_rate": 1.7836132859410885e-05,
+      "loss": 0.8067,
+      "step": 10300
+    },
+    {
+      "epoch": 0.2214632469819994,
+      "grad_norm": 0.565092919613001,
+      "learning_rate": 1.783189629039602e-05,
+      "loss": 0.7973,
+      "step": 10310
+    },
+    {
+      "epoch": 0.22167805129527002,
+      "grad_norm": 0.5520905710174285,
+      "learning_rate": 1.782765608230678e-05,
+      "loss": 0.8074,
+      "step": 10320
+    },
+    {
+      "epoch": 0.2218928556085406,
+      "grad_norm": 0.5499770181938648,
+      "learning_rate": 1.782341223711336e-05,
+      "loss": 0.8047,
+      "step": 10330
+    },
+    {
+      "epoch": 0.22210765992181122,
+      "grad_norm": 0.5770398453057336,
+      "learning_rate": 1.7819164756787667e-05,
+      "loss": 0.8024,
+      "step": 10340
+    },
+    {
+      "epoch": 0.22232246423508184,
+      "grad_norm": 0.6005922987428989,
+      "learning_rate": 1.781491364330327e-05,
+      "loss": 0.8023,
+      "step": 10350
+    },
+    {
+      "epoch": 0.22253726854835246,
+      "grad_norm": 0.5692299117115992,
+      "learning_rate": 1.7810658898635455e-05,
+      "loss": 0.7944,
+      "step": 10360
+    },
+    {
+      "epoch": 0.22275207286162307,
+      "grad_norm": 0.5657236504416966,
+      "learning_rate": 1.780640052476117e-05,
+      "loss": 0.8146,
+      "step": 10370
+    },
+    {
+      "epoch": 0.22296687717489366,
+      "grad_norm": 0.5561699306834719,
+      "learning_rate": 1.780213852365906e-05,
+      "loss": 0.785,
+      "step": 10380
+    },
+    {
+      "epoch": 0.22318168148816428,
+      "grad_norm": 0.7081144347430791,
+      "learning_rate": 1.779787289730946e-05,
+      "loss": 0.8007,
+      "step": 10390
+    },
+    {
+      "epoch": 0.2233964858014349,
+      "grad_norm": 0.5625449988527748,
+      "learning_rate": 1.779360364769438e-05,
+      "loss": 0.7903,
+      "step": 10400
+    },
+    {
+      "epoch": 0.2236112901147055,
+      "grad_norm": 0.5443207117710023,
+      "learning_rate": 1.7789330776797515e-05,
+      "loss": 0.8029,
+      "step": 10410
+    },
+    {
+      "epoch": 0.2238260944279761,
+      "grad_norm": 0.5625349389867544,
+      "learning_rate": 1.7785054286604254e-05,
+      "loss": 0.8169,
+      "step": 10420
+    },
+    {
+      "epoch": 0.22404089874124672,
+      "grad_norm": 0.5476014487186761,
+      "learning_rate": 1.7780774179101654e-05,
+      "loss": 0.7917,
+      "step": 10430
+    },
+    {
+      "epoch": 0.22425570305451734,
+      "grad_norm": 0.5800827416013181,
+      "learning_rate": 1.7776490456278462e-05,
+      "loss": 0.8073,
+      "step": 10440
+    },
+    {
+      "epoch": 0.22447050736778795,
+      "grad_norm": 0.5542113664451737,
+      "learning_rate": 1.7772203120125095e-05,
+      "loss": 0.7947,
+      "step": 10450
+    },
+    {
+      "epoch": 0.22468531168105857,
+      "grad_norm": 0.5404026820286727,
+      "learning_rate": 1.776791217263366e-05,
+      "loss": 0.8024,
+      "step": 10460
+    },
+    {
+      "epoch": 0.22490011599432916,
+      "grad_norm": 0.5538986415644934,
+      "learning_rate": 1.7763617615797934e-05,
+      "loss": 0.7828,
+      "step": 10470
+    },
+    {
+      "epoch": 0.22511492030759978,
+      "grad_norm": 0.5585745280123449,
+      "learning_rate": 1.7759319451613376e-05,
+      "loss": 0.801,
+      "step": 10480
+    },
+    {
+      "epoch": 0.2253297246208704,
+      "grad_norm": 0.5622099418484012,
+      "learning_rate": 1.7755017682077118e-05,
+      "loss": 0.7945,
+      "step": 10490
+    },
+    {
+      "epoch": 0.225544528934141,
+      "grad_norm": 0.5370985490496776,
+      "learning_rate": 1.7750712309187967e-05,
+      "loss": 0.7863,
+      "step": 10500
+    },
+    {
+      "epoch": 0.2257593332474116,
+      "grad_norm": 0.5372310056648028,
+      "learning_rate": 1.7746403334946407e-05,
+      "loss": 0.7946,
+      "step": 10510
+    },
+    {
+      "epoch": 0.22597413756068221,
+      "grad_norm": 0.556076402437103,
+      "learning_rate": 1.7742090761354596e-05,
+      "loss": 0.7881,
+      "step": 10520
+    },
+    {
+      "epoch": 0.22618894187395283,
+      "grad_norm": 0.5580749453853192,
+      "learning_rate": 1.7737774590416358e-05,
+      "loss": 0.7938,
+      "step": 10530
+    },
+    {
+      "epoch": 0.22640374618722345,
+      "grad_norm": 0.5957795192444436,
+      "learning_rate": 1.7733454824137196e-05,
+      "loss": 0.7955,
+      "step": 10540
+    },
+    {
+      "epoch": 0.22661855050049404,
+      "grad_norm": 0.5712477046163247,
+      "learning_rate": 1.772913146452428e-05,
+      "loss": 0.8038,
+      "step": 10550
+    },
+    {
+      "epoch": 0.22683335481376465,
+      "grad_norm": 0.548160058046596,
+      "learning_rate": 1.7724804513586448e-05,
+      "loss": 0.7939,
+      "step": 10560
+    },
+    {
+      "epoch": 0.22704815912703527,
+      "grad_norm": 0.5369592474615921,
+      "learning_rate": 1.7720473973334213e-05,
+      "loss": 0.7993,
+      "step": 10570
+    },
+    {
+      "epoch": 0.2272629634403059,
+      "grad_norm": 0.5423611002909383,
+      "learning_rate": 1.7716139845779746e-05,
+      "loss": 0.8045,
+      "step": 10580
+    },
+    {
+      "epoch": 0.2274777677535765,
+      "grad_norm": 0.5550989146015426,
+      "learning_rate": 1.7711802132936896e-05,
+      "loss": 0.7892,
+      "step": 10590
+    },
+    {
+      "epoch": 0.2276925720668471,
+      "grad_norm": 0.5703653518043215,
+      "learning_rate": 1.770746083682117e-05,
+      "loss": 0.8055,
+      "step": 10600
+    },
+    {
+      "epoch": 0.2279073763801177,
+      "grad_norm": 0.5615344431225333,
+      "learning_rate": 1.7703115959449738e-05,
+      "loss": 0.7979,
+      "step": 10610
+    },
+    {
+      "epoch": 0.22812218069338833,
+      "grad_norm": 0.553731045220489,
+      "learning_rate": 1.7698767502841445e-05,
+      "loss": 0.811,
+      "step": 10620
+    },
+    {
+      "epoch": 0.22833698500665894,
+      "grad_norm": 0.5501876227065152,
+      "learning_rate": 1.769441546901679e-05,
+      "loss": 0.7964,
+      "step": 10630
+    },
+    {
+      "epoch": 0.22855178931992953,
+      "grad_norm": 0.5412824361191674,
+      "learning_rate": 1.7690059859997935e-05,
+      "loss": 0.7859,
+      "step": 10640
+    },
+    {
+      "epoch": 0.22876659363320015,
+      "grad_norm": 0.5751057669815162,
+      "learning_rate": 1.7685700677808703e-05,
+      "loss": 0.8021,
+      "step": 10650
+    },
+    {
+      "epoch": 0.22898139794647077,
+      "grad_norm": 0.5413904254802253,
+      "learning_rate": 1.7681337924474585e-05,
+      "loss": 0.7911,
+      "step": 10660
+    },
+    {
+      "epoch": 0.22919620225974138,
+      "grad_norm": 0.5443592022724559,
+      "learning_rate": 1.7676971602022722e-05,
+      "loss": 0.804,
+      "step": 10670
+    },
+    {
+      "epoch": 0.229411006573012,
+      "grad_norm": 0.5963768922972459,
+      "learning_rate": 1.7672601712481916e-05,
+      "loss": 0.7838,
+      "step": 10680
+    },
+    {
+      "epoch": 0.2296258108862826,
+      "grad_norm": 0.5814546759073268,
+      "learning_rate": 1.7668228257882628e-05,
+      "loss": 0.7934,
+      "step": 10690
+    },
+    {
+      "epoch": 0.2298406151995532,
+      "grad_norm": 0.5513099385655584,
+      "learning_rate": 1.7663851240256973e-05,
+      "loss": 0.7901,
+      "step": 10700
+    },
+    {
+      "epoch": 0.23005541951282382,
+      "grad_norm": 0.5531620819292025,
+      "learning_rate": 1.7659470661638727e-05,
+      "loss": 0.8093,
+      "step": 10710
+    },
+    {
+      "epoch": 0.23027022382609444,
+      "grad_norm": 0.5656118475573507,
+      "learning_rate": 1.7655086524063314e-05,
+      "loss": 0.793,
+      "step": 10720
+    },
+    {
+      "epoch": 0.23048502813936503,
+      "grad_norm": 0.5575833476944946,
+      "learning_rate": 1.765069882956781e-05,
+      "loss": 0.7976,
+      "step": 10730
+    },
+    {
+      "epoch": 0.23069983245263564,
+      "grad_norm": 0.5564123290228586,
+      "learning_rate": 1.764630758019096e-05,
+      "loss": 0.8022,
+      "step": 10740
+    },
+    {
+      "epoch": 0.23091463676590626,
+      "grad_norm": 0.5534405369439489,
+      "learning_rate": 1.7641912777973136e-05,
+      "loss": 0.8079,
+      "step": 10750
+    },
+    {
+      "epoch": 0.23112944107917688,
+      "grad_norm": 0.5473270189074221,
+      "learning_rate": 1.7637514424956386e-05,
+      "loss": 0.7938,
+      "step": 10760
+    },
+    {
+      "epoch": 0.23134424539244747,
+      "grad_norm": 0.5357674731887498,
+      "learning_rate": 1.7633112523184383e-05,
+      "loss": 0.7846,
+      "step": 10770
+    },
+    {
+      "epoch": 0.23155904970571808,
+      "grad_norm": 0.5552201424591323,
+      "learning_rate": 1.762870707470247e-05,
+      "loss": 0.7803,
+      "step": 10780
+    },
+    {
+      "epoch": 0.2317738540189887,
+      "grad_norm": 0.558152601578086,
+      "learning_rate": 1.7624298081557626e-05,
+      "loss": 0.7902,
+      "step": 10790
+    },
+    {
+      "epoch": 0.23198865833225932,
+      "grad_norm": 0.5595014924866345,
+      "learning_rate": 1.7619885545798486e-05,
+      "loss": 0.7891,
+      "step": 10800
+    },
+    {
+      "epoch": 0.23220346264552993,
+      "grad_norm": 0.5495730986900741,
+      "learning_rate": 1.7615469469475315e-05,
+      "loss": 0.8026,
+      "step": 10810
+    },
+    {
+      "epoch": 0.23241826695880052,
+      "grad_norm": 0.5760660001958988,
+      "learning_rate": 1.7611049854640044e-05,
+      "loss": 0.8077,
+      "step": 10820
+    },
+    {
+      "epoch": 0.23263307127207114,
+      "grad_norm": 0.5550120424472271,
+      "learning_rate": 1.7606626703346235e-05,
+      "loss": 0.7965,
+      "step": 10830
+    },
+    {
+      "epoch": 0.23284787558534176,
+      "grad_norm": 0.5534917812163986,
+      "learning_rate": 1.7602200017649093e-05,
+      "loss": 0.7862,
+      "step": 10840
+    },
+    {
+      "epoch": 0.23306267989861237,
+      "grad_norm": 0.567697162536355,
+      "learning_rate": 1.759776979960547e-05,
+      "loss": 0.7882,
+      "step": 10850
+    },
+    {
+      "epoch": 0.23327748421188296,
+      "grad_norm": 0.5544878261968137,
+      "learning_rate": 1.7593336051273857e-05,
+      "loss": 0.7894,
+      "step": 10860
+    },
+    {
+      "epoch": 0.23349228852515358,
+      "grad_norm": 0.5522547428302197,
+      "learning_rate": 1.7588898774714387e-05,
+      "loss": 0.792,
+      "step": 10870
+    },
+    {
+      "epoch": 0.2337070928384242,
+      "grad_norm": 0.5561521408221043,
+      "learning_rate": 1.7584457971988836e-05,
+      "loss": 0.7984,
+      "step": 10880
+    },
+    {
+      "epoch": 0.2339218971516948,
+      "grad_norm": 0.5701789587662502,
+      "learning_rate": 1.75800136451606e-05,
+      "loss": 0.7891,
+      "step": 10890
+    },
+    {
+      "epoch": 0.23413670146496543,
+      "grad_norm": 0.5512069625729566,
+      "learning_rate": 1.7575565796294745e-05,
+      "loss": 0.8033,
+      "step": 10900
+    },
+    {
+      "epoch": 0.23435150577823602,
+      "grad_norm": 0.5671925282466402,
+      "learning_rate": 1.7571114427457942e-05,
+      "loss": 0.7914,
+      "step": 10910
+    },
+    {
+      "epoch": 0.23456631009150664,
+      "grad_norm": 0.5987674977748154,
+      "learning_rate": 1.7566659540718512e-05,
+      "loss": 0.8081,
+      "step": 10920
+    },
+    {
+      "epoch": 0.23478111440477725,
+      "grad_norm": 0.5576016505511755,
+      "learning_rate": 1.7562201138146407e-05,
+      "loss": 0.7881,
+      "step": 10930
+    },
+    {
+      "epoch": 0.23499591871804787,
+      "grad_norm": 0.5606224459108454,
+      "learning_rate": 1.7557739221813226e-05,
+      "loss": 0.7999,
+      "step": 10940
+    },
+    {
+      "epoch": 0.23521072303131846,
+      "grad_norm": 0.5778821827576899,
+      "learning_rate": 1.7553273793792176e-05,
+      "loss": 0.8156,
+      "step": 10950
+    },
+    {
+      "epoch": 0.23542552734458908,
+      "grad_norm": 0.5512352918888542,
+      "learning_rate": 1.7548804856158113e-05,
+      "loss": 0.7987,
+      "step": 10960
+    },
+    {
+      "epoch": 0.2356403316578597,
+      "grad_norm": 0.5566762408624399,
+      "learning_rate": 1.7544332410987523e-05,
+      "loss": 0.7976,
+      "step": 10970
+    },
+    {
+      "epoch": 0.2358551359711303,
+      "grad_norm": 0.5645656364576122,
+      "learning_rate": 1.7539856460358515e-05,
+      "loss": 0.7869,
+      "step": 10980
+    },
+    {
+      "epoch": 0.2360699402844009,
+      "grad_norm": 0.5445877178654857,
+      "learning_rate": 1.753537700635083e-05,
+      "loss": 0.7869,
+      "step": 10990
+    },
+    {
+      "epoch": 0.23628474459767151,
+      "grad_norm": 0.5590715894295356,
+      "learning_rate": 1.753089405104584e-05,
+      "loss": 0.8091,
+      "step": 11000
+    },
+    {
+      "epoch": 0.23649954891094213,
+      "grad_norm": 0.5671948914432249,
+      "learning_rate": 1.7526407596526536e-05,
+      "loss": 0.8074,
+      "step": 11010
+    },
+    {
+      "epoch": 0.23671435322421275,
+      "grad_norm": 0.5623507718873683,
+      "learning_rate": 1.7521917644877546e-05,
+      "loss": 0.7854,
+      "step": 11020
+    },
+    {
+      "epoch": 0.23692915753748336,
+      "grad_norm": 0.5932218934601358,
+      "learning_rate": 1.7517424198185108e-05,
+      "loss": 0.8094,
+      "step": 11030
+    },
+    {
+      "epoch": 0.23714396185075395,
+      "grad_norm": 0.6112761900916727,
+      "learning_rate": 1.7512927258537105e-05,
+      "loss": 0.7942,
+      "step": 11040
+    },
+    {
+      "epoch": 0.23735876616402457,
+      "grad_norm": 0.566759308520504,
+      "learning_rate": 1.750842682802302e-05,
+      "loss": 0.7797,
+      "step": 11050
+    },
+    {
+      "epoch": 0.2375735704772952,
+      "grad_norm": 0.536263668019581,
+      "learning_rate": 1.7503922908733972e-05,
+      "loss": 0.7899,
+      "step": 11060
+    },
+    {
+      "epoch": 0.2377883747905658,
+      "grad_norm": 0.5570104505775965,
+      "learning_rate": 1.74994155027627e-05,
+      "loss": 0.7971,
+      "step": 11070
+    },
+    {
+      "epoch": 0.2380031791038364,
+      "grad_norm": 0.5609858905324858,
+      "learning_rate": 1.7494904612203557e-05,
+      "loss": 0.7866,
+      "step": 11080
+    },
+    {
+      "epoch": 0.238217983417107,
+      "grad_norm": 0.5571287784622291,
+      "learning_rate": 1.7490390239152523e-05,
+      "loss": 0.7964,
+      "step": 11090
+    },
+    {
+      "epoch": 0.23843278773037763,
+      "grad_norm": 0.5618669494447492,
+      "learning_rate": 1.748587238570719e-05,
+      "loss": 0.8074,
+      "step": 11100
+    },
+    {
+      "epoch": 0.23864759204364824,
+      "grad_norm": 0.5955821944742522,
+      "learning_rate": 1.748135105396677e-05,
+      "loss": 0.8106,
+      "step": 11110
+    },
+    {
+      "epoch": 0.23886239635691886,
+      "grad_norm": 0.5507627007193192,
+      "learning_rate": 1.747682624603209e-05,
+      "loss": 0.7828,
+      "step": 11120
+    },
+    {
+      "epoch": 0.23907720067018945,
+      "grad_norm": 0.5861560768919388,
+      "learning_rate": 1.747229796400559e-05,
+      "loss": 0.8241,
+      "step": 11130
+    },
+    {
+      "epoch": 0.23929200498346007,
+      "grad_norm": 0.5718220707403837,
+      "learning_rate": 1.7467766209991332e-05,
+      "loss": 0.794,
+      "step": 11140
+    },
+    {
+      "epoch": 0.23950680929673068,
+      "grad_norm": 0.6034932739087517,
+      "learning_rate": 1.7463230986094982e-05,
+      "loss": 0.801,
+      "step": 11150
+    },
+    {
+      "epoch": 0.2397216136100013,
+      "grad_norm": 0.5581638199123323,
+      "learning_rate": 1.7458692294423825e-05,
+      "loss": 0.8061,
+      "step": 11160
+    },
+    {
+      "epoch": 0.2399364179232719,
+      "grad_norm": 0.5395247610317755,
+      "learning_rate": 1.7454150137086753e-05,
+      "loss": 0.7864,
+      "step": 11170
+    },
+    {
+      "epoch": 0.2401512222365425,
+      "grad_norm": 0.5931900048885603,
+      "learning_rate": 1.7449604516194266e-05,
+      "loss": 0.7947,
+      "step": 11180
+    },
+    {
+      "epoch": 0.24036602654981312,
+      "grad_norm": 0.5574150201939502,
+      "learning_rate": 1.7445055433858485e-05,
+      "loss": 0.7992,
+      "step": 11190
+    },
+    {
+      "epoch": 0.24058083086308374,
+      "grad_norm": 0.558048596510596,
+      "learning_rate": 1.744050289219313e-05,
+      "loss": 0.8036,
+      "step": 11200
+    },
+    {
+      "epoch": 0.24079563517635433,
+      "grad_norm": 0.5544503116923134,
+      "learning_rate": 1.743594689331353e-05,
+      "loss": 0.7892,
+      "step": 11210
+    },
+    {
+      "epoch": 0.24101043948962494,
+      "grad_norm": 0.5913966197755747,
+      "learning_rate": 1.7431387439336615e-05,
+      "loss": 0.7882,
+      "step": 11220
+    },
+    {
+      "epoch": 0.24122524380289556,
+      "grad_norm": 0.5609044365761472,
+      "learning_rate": 1.7426824532380932e-05,
+      "loss": 0.8006,
+      "step": 11230
+    },
+    {
+      "epoch": 0.24144004811616618,
+      "grad_norm": 0.5598501767608501,
+      "learning_rate": 1.7422258174566625e-05,
+      "loss": 0.8034,
+      "step": 11240
+    },
+    {
+      "epoch": 0.2416548524294368,
+      "grad_norm": 0.5598467174087662,
+      "learning_rate": 1.741768836801544e-05,
+      "loss": 0.808,
+      "step": 11250
+    },
+    {
+      "epoch": 0.24186965674270738,
+      "grad_norm": 0.5571089293330055,
+      "learning_rate": 1.741311511485073e-05,
+      "loss": 0.7828,
+      "step": 11260
+    },
+    {
+      "epoch": 0.242084461055978,
+      "grad_norm": 0.5858030780897929,
+      "learning_rate": 1.7408538417197447e-05,
+      "loss": 0.7881,
+      "step": 11270
+    },
+    {
+      "epoch": 0.24229926536924862,
+      "grad_norm": 0.5664884004513725,
+      "learning_rate": 1.7403958277182143e-05,
+      "loss": 0.792,
+      "step": 11280
+    },
+    {
+      "epoch": 0.24251406968251923,
+      "grad_norm": 0.5523972315190862,
+      "learning_rate": 1.7399374696932977e-05,
+      "loss": 0.795,
+      "step": 11290
+    },
+    {
+      "epoch": 0.24272887399578982,
+      "grad_norm": 0.5801081531529861,
+      "learning_rate": 1.7394787678579693e-05,
+      "loss": 0.8056,
+      "step": 11300
+    },
+    {
+      "epoch": 0.24294367830906044,
+      "grad_norm": 0.5851376223053917,
+      "learning_rate": 1.739019722425364e-05,
+      "loss": 0.797,
+      "step": 11310
+    },
+    {
+      "epoch": 0.24315848262233106,
+      "grad_norm": 0.5826239567427932,
+      "learning_rate": 1.7385603336087766e-05,
+      "loss": 0.7967,
+      "step": 11320
+    },
+    {
+      "epoch": 0.24337328693560167,
+      "grad_norm": 0.5671104979630887,
+      "learning_rate": 1.738100601621661e-05,
+      "loss": 0.8042,
+      "step": 11330
+    },
+    {
+      "epoch": 0.2435880912488723,
+      "grad_norm": 0.5526309623916124,
+      "learning_rate": 1.737640526677631e-05,
+      "loss": 0.7974,
+      "step": 11340
+    },
+    {
+      "epoch": 0.24380289556214288,
+      "grad_norm": 0.5952220328485525,
+      "learning_rate": 1.737180108990459e-05,
+      "loss": 0.7988,
+      "step": 11350
+    },
+    {
+      "epoch": 0.2440176998754135,
+      "grad_norm": 0.5576900075243258,
+      "learning_rate": 1.736719348774077e-05,
+      "loss": 0.792,
+      "step": 11360
+    },
+    {
+      "epoch": 0.2442325041886841,
+      "grad_norm": 0.5646590624675798,
+      "learning_rate": 1.7362582462425775e-05,
+      "loss": 0.7918,
+      "step": 11370
+    },
+    {
+      "epoch": 0.24444730850195473,
+      "grad_norm": 0.5650066999862957,
+      "learning_rate": 1.7357968016102094e-05,
+      "loss": 0.8026,
+      "step": 11380
+    },
+    {
+      "epoch": 0.24466211281522532,
+      "grad_norm": 0.5675843616193625,
+      "learning_rate": 1.7353350150913826e-05,
+      "loss": 0.7983,
+      "step": 11390
+    },
+    {
+      "epoch": 0.24487691712849594,
+      "grad_norm": 0.516651808309515,
+      "learning_rate": 1.734872886900665e-05,
+      "loss": 0.7707,
+      "step": 11400
+    },
+    {
+      "epoch": 0.24509172144176655,
+      "grad_norm": 0.5480548050237413,
+      "learning_rate": 1.7344104172527845e-05,
+      "loss": 0.7901,
+      "step": 11410
+    },
+    {
+      "epoch": 0.24530652575503717,
+      "grad_norm": 0.5467792895993273,
+      "learning_rate": 1.7339476063626252e-05,
+      "loss": 0.805,
+      "step": 11420
+    },
+    {
+      "epoch": 0.24552133006830776,
+      "grad_norm": 0.5675774870807921,
+      "learning_rate": 1.733484454445232e-05,
+      "loss": 0.7789,
+      "step": 11430
+    },
+    {
+      "epoch": 0.24573613438157837,
+      "grad_norm": 0.5430330356783232,
+      "learning_rate": 1.7330209617158075e-05,
+      "loss": 0.7893,
+      "step": 11440
+    },
+    {
+      "epoch": 0.245950938694849,
+      "grad_norm": 0.5408803807446219,
+      "learning_rate": 1.7325571283897126e-05,
+      "loss": 0.7897,
+      "step": 11450
+    },
+    {
+      "epoch": 0.2461657430081196,
+      "grad_norm": 0.5536255557245513,
+      "learning_rate": 1.7320929546824662e-05,
+      "loss": 0.793,
+      "step": 11460
+    },
+    {
+      "epoch": 0.24638054732139023,
+      "grad_norm": 0.5669589146457302,
+      "learning_rate": 1.731628440809746e-05,
+      "loss": 0.7802,
+      "step": 11470
+    },
+    {
+      "epoch": 0.24659535163466081,
+      "grad_norm": 0.5574302178274012,
+      "learning_rate": 1.731163586987387e-05,
+      "loss": 0.7847,
+      "step": 11480
+    },
+    {
+      "epoch": 0.24681015594793143,
+      "grad_norm": 0.5563586377765385,
+      "learning_rate": 1.7306983934313833e-05,
+      "loss": 0.7973,
+      "step": 11490
+    },
+    {
+      "epoch": 0.24702496026120205,
+      "grad_norm": 0.5541682433080475,
+      "learning_rate": 1.730232860357885e-05,
+      "loss": 0.7972,
+      "step": 11500
+    },
+    {
+      "epoch": 0.24723976457447266,
+      "grad_norm": 0.5487057694995032,
+      "learning_rate": 1.7297669879832025e-05,
+      "loss": 0.783,
+      "step": 11510
+    },
+    {
+      "epoch": 0.24745456888774325,
+      "grad_norm": 0.5407120966552204,
+      "learning_rate": 1.7293007765238012e-05,
+      "loss": 0.7858,
+      "step": 11520
+    },
+    {
+      "epoch": 0.24766937320101387,
+      "grad_norm": 0.6189329723987206,
+      "learning_rate": 1.728834226196306e-05,
+      "loss": 0.7928,
+      "step": 11530
+    },
+    {
+      "epoch": 0.2478841775142845,
+      "grad_norm": 0.5365545625639055,
+      "learning_rate": 1.728367337217498e-05,
+      "loss": 0.7922,
+      "step": 11540
+    },
+    {
+      "epoch": 0.2480989818275551,
+      "grad_norm": 0.5541237894376929,
+      "learning_rate": 1.7279001098043164e-05,
+      "loss": 0.7864,
+      "step": 11550
+    },
+    {
+      "epoch": 0.24831378614082572,
+      "grad_norm": 0.5518328452834319,
+      "learning_rate": 1.7274325441738578e-05,
+      "loss": 0.7993,
+      "step": 11560
+    },
+    {
+      "epoch": 0.2485285904540963,
+      "grad_norm": 0.5476209039060521,
+      "learning_rate": 1.7269646405433754e-05,
+      "loss": 0.7845,
+      "step": 11570
+    },
+    {
+      "epoch": 0.24874339476736693,
+      "grad_norm": 0.5374788890443903,
+      "learning_rate": 1.7264963991302798e-05,
+      "loss": 0.7736,
+      "step": 11580
+    },
+    {
+      "epoch": 0.24895819908063754,
+      "grad_norm": 0.5528441478075925,
+      "learning_rate": 1.7260278201521377e-05,
+      "loss": 0.8082,
+      "step": 11590
+    },
+    {
+      "epoch": 0.24917300339390816,
+      "grad_norm": 0.5517627198236814,
+      "learning_rate": 1.725558903826674e-05,
+      "loss": 0.7964,
+      "step": 11600
+    },
+    {
+      "epoch": 0.24938780770717875,
+      "grad_norm": 0.526708104255714,
+      "learning_rate": 1.7250896503717697e-05,
+      "loss": 0.7846,
+      "step": 11610
+    },
+    {
+      "epoch": 0.24960261202044937,
+      "grad_norm": 0.5677760683591515,
+      "learning_rate": 1.724620060005462e-05,
+      "loss": 0.8056,
+      "step": 11620
+    },
+    {
+      "epoch": 0.24981741633371998,
+      "grad_norm": 0.5445293154923263,
+      "learning_rate": 1.724150132945946e-05,
+      "loss": 0.7925,
+      "step": 11630
+    },
+    {
+      "epoch": 0.2500322206469906,
+      "grad_norm": 0.5253609554392292,
+      "learning_rate": 1.723679869411571e-05,
+      "loss": 0.795,
+      "step": 11640
+    },
+    {
+      "epoch": 0.2502470249602612,
+      "grad_norm": 0.5717202718646365,
+      "learning_rate": 1.723209269620845e-05,
+      "loss": 0.776,
+      "step": 11650
+    },
+    {
+      "epoch": 0.25046182927353183,
+      "grad_norm": 0.5795012398020939,
+      "learning_rate": 1.722738333792431e-05,
+      "loss": 0.7879,
+      "step": 11660
+    },
+    {
+      "epoch": 0.2506766335868024,
+      "grad_norm": 0.5572184542777283,
+      "learning_rate": 1.722267062145148e-05,
+      "loss": 0.7822,
+      "step": 11670
+    },
+    {
+      "epoch": 0.250891437900073,
+      "grad_norm": 0.5617363405871534,
+      "learning_rate": 1.721795454897972e-05,
+      "loss": 0.7905,
+      "step": 11680
+    },
+    {
+      "epoch": 0.25110624221334366,
+      "grad_norm": 0.5445190221346468,
+      "learning_rate": 1.7213235122700334e-05,
+      "loss": 0.8082,
+      "step": 11690
+    },
+    {
+      "epoch": 0.25132104652661424,
+      "grad_norm": 0.5395663294268565,
+      "learning_rate": 1.7208512344806204e-05,
+      "loss": 0.776,
+      "step": 11700
+    },
+    {
+      "epoch": 0.2515358508398849,
+      "grad_norm": 0.5454264355323872,
+      "learning_rate": 1.7203786217491757e-05,
+      "loss": 0.7828,
+      "step": 11710
+    },
+    {
+      "epoch": 0.2517506551531555,
+      "grad_norm": 0.5778897096941694,
+      "learning_rate": 1.7199056742952973e-05,
+      "loss": 0.7861,
+      "step": 11720
+    },
+    {
+      "epoch": 0.25196545946642607,
+      "grad_norm": 0.5436108360977332,
+      "learning_rate": 1.7194323923387396e-05,
+      "loss": 0.7887,
+      "step": 11730
+    },
+    {
+      "epoch": 0.2521802637796967,
+      "grad_norm": 0.5480998527967708,
+      "learning_rate": 1.718958776099412e-05,
+      "loss": 0.8009,
+      "step": 11740
+    },
+    {
+      "epoch": 0.2523950680929673,
+      "grad_norm": 0.5553968599080923,
+      "learning_rate": 1.718484825797379e-05,
+      "loss": 0.7953,
+      "step": 11750
+    },
+    {
+      "epoch": 0.2526098724062379,
+      "grad_norm": 0.5633121625376932,
+      "learning_rate": 1.7180105416528613e-05,
+      "loss": 0.7836,
+      "step": 11760
+    },
+    {
+      "epoch": 0.25282467671950853,
+      "grad_norm": 0.5706756900855683,
+      "learning_rate": 1.7175359238862335e-05,
+      "loss": 0.7963,
+      "step": 11770
+    },
+    {
+      "epoch": 0.2530394810327791,
+      "grad_norm": 0.5594666956751811,
+      "learning_rate": 1.7170609727180258e-05,
+      "loss": 0.7881,
+      "step": 11780
+    },
+    {
+      "epoch": 0.25325428534604977,
+      "grad_norm": 0.5537400526042484,
+      "learning_rate": 1.7165856883689237e-05,
+      "loss": 0.7862,
+      "step": 11790
+    },
+    {
+      "epoch": 0.25346908965932036,
+      "grad_norm": 0.5561566681420613,
+      "learning_rate": 1.7161100710597668e-05,
+      "loss": 0.7856,
+      "step": 11800
+    },
+    {
+      "epoch": 0.25368389397259095,
+      "grad_norm": 0.575034466737522,
+      "learning_rate": 1.71563412101155e-05,
+      "loss": 0.7872,
+      "step": 11810
+    },
+    {
+      "epoch": 0.2538986982858616,
+      "grad_norm": 0.5674721717746797,
+      "learning_rate": 1.7151578384454218e-05,
+      "loss": 0.8035,
+      "step": 11820
+    },
+    {
+      "epoch": 0.2541135025991322,
+      "grad_norm": 0.5471441442602709,
+      "learning_rate": 1.714681223582686e-05,
+      "loss": 0.7835,
+      "step": 11830
+    },
+    {
+      "epoch": 0.2543283069124028,
+      "grad_norm": 0.5349338344027432,
+      "learning_rate": 1.7142042766448016e-05,
+      "loss": 0.7704,
+      "step": 11840
+    },
+    {
+      "epoch": 0.2545431112256734,
+      "grad_norm": 0.5289756077207555,
+      "learning_rate": 1.7137269978533804e-05,
+      "loss": 0.7881,
+      "step": 11850
+    },
+    {
+      "epoch": 0.254757915538944,
+      "grad_norm": 0.5654604443984577,
+      "learning_rate": 1.713249387430189e-05,
+      "loss": 0.7873,
+      "step": 11860
+    },
+    {
+      "epoch": 0.25497271985221465,
+      "grad_norm": 0.5479301372643116,
+      "learning_rate": 1.7127714455971476e-05,
+      "loss": 0.7853,
+      "step": 11870
+    },
+    {
+      "epoch": 0.25518752416548524,
+      "grad_norm": 0.5642448334289781,
+      "learning_rate": 1.7122931725763318e-05,
+      "loss": 0.7955,
+      "step": 11880
+    },
+    {
+      "epoch": 0.2554023284787559,
+      "grad_norm": 0.5561951319477092,
+      "learning_rate": 1.71181456858997e-05,
+      "loss": 0.7998,
+      "step": 11890
+    },
+    {
+      "epoch": 0.25561713279202647,
+      "grad_norm": 0.5551743423950759,
+      "learning_rate": 1.711335633860444e-05,
+      "loss": 0.7868,
+      "step": 11900
+    },
+    {
+      "epoch": 0.25583193710529706,
+      "grad_norm": 0.575663087645707,
+      "learning_rate": 1.71085636861029e-05,
+      "loss": 0.7946,
+      "step": 11910
+    },
+    {
+      "epoch": 0.2560467414185677,
+      "grad_norm": 0.5705702380544423,
+      "learning_rate": 1.710376773062198e-05,
+      "loss": 0.8077,
+      "step": 11920
+    },
+    {
+      "epoch": 0.2562615457318383,
+      "grad_norm": 0.5517950323778168,
+      "learning_rate": 1.709896847439011e-05,
+      "loss": 0.7798,
+      "step": 11930
+    },
+    {
+      "epoch": 0.2564763500451089,
+      "grad_norm": 0.5654242254525084,
+      "learning_rate": 1.709416591963725e-05,
+      "loss": 0.7854,
+      "step": 11940
+    },
+    {
+      "epoch": 0.2566911543583795,
+      "grad_norm": 0.556721343917251,
+      "learning_rate": 1.7089360068594903e-05,
+      "loss": 0.7845,
+      "step": 11950
+    },
+    {
+      "epoch": 0.2569059586716501,
+      "grad_norm": 0.5647470558063254,
+      "learning_rate": 1.7084550923496094e-05,
+      "loss": 0.7827,
+      "step": 11960
+    },
+    {
+      "epoch": 0.25712076298492076,
+      "grad_norm": 0.5580955635924464,
+      "learning_rate": 1.7079738486575382e-05,
+      "loss": 0.7857,
+      "step": 11970
+    },
+    {
+      "epoch": 0.25733556729819135,
+      "grad_norm": 0.5655207694822407,
+      "learning_rate": 1.7074922760068855e-05,
+      "loss": 0.7751,
+      "step": 11980
+    },
+    {
+      "epoch": 0.25755037161146194,
+      "grad_norm": 0.5700568764310818,
+      "learning_rate": 1.7070103746214135e-05,
+      "loss": 0.796,
+      "step": 11990
+    },
+    {
+      "epoch": 0.2577651759247326,
+      "grad_norm": 0.530803254919797,
+      "learning_rate": 1.7065281447250363e-05,
+      "loss": 0.7799,
+      "step": 12000
+    },
+    {
+      "epoch": 0.25797998023800317,
+      "grad_norm": 0.559034831132991,
+      "learning_rate": 1.706045586541821e-05,
+      "loss": 0.7957,
+      "step": 12010
+    },
+    {
+      "epoch": 0.2581947845512738,
+      "grad_norm": 0.5597546125165273,
+      "learning_rate": 1.7055627002959872e-05,
+      "loss": 0.7845,
+      "step": 12020
+    },
+    {
+      "epoch": 0.2584095888645444,
+      "grad_norm": 0.6058840929391112,
+      "learning_rate": 1.7050794862119078e-05,
+      "loss": 0.7895,
+      "step": 12030
+    },
+    {
+      "epoch": 0.258624393177815,
+      "grad_norm": 0.5651994162157234,
+      "learning_rate": 1.704595944514106e-05,
+      "loss": 0.8003,
+      "step": 12040
+    },
+    {
+      "epoch": 0.25883919749108564,
+      "grad_norm": 0.5596861708905858,
+      "learning_rate": 1.7041120754272594e-05,
+      "loss": 0.8011,
+      "step": 12050
+    },
+    {
+      "epoch": 0.2590540018043562,
+      "grad_norm": 0.5505706888687052,
+      "learning_rate": 1.7036278791761965e-05,
+      "loss": 0.791,
+      "step": 12060
+    },
+    {
+      "epoch": 0.2592688061176268,
+      "grad_norm": 0.5545931326572373,
+      "learning_rate": 1.7031433559858977e-05,
+      "loss": 0.7804,
+      "step": 12070
+    },
+    {
+      "epoch": 0.25948361043089746,
+      "grad_norm": 3.102401423632908,
+      "learning_rate": 1.702658506081496e-05,
+      "loss": 0.8037,
+      "step": 12080
+    },
+    {
+      "epoch": 0.25969841474416805,
+      "grad_norm": 0.5450387286613136,
+      "learning_rate": 1.7021733296882758e-05,
+      "loss": 0.8026,
+      "step": 12090
+    },
+    {
+      "epoch": 0.2599132190574387,
+      "grad_norm": 0.541274551933326,
+      "learning_rate": 1.7016878270316738e-05,
+      "loss": 0.7963,
+      "step": 12100
+    },
+    {
+      "epoch": 0.2601280233707093,
+      "grad_norm": 0.5316129158445448,
+      "learning_rate": 1.701201998337277e-05,
+      "loss": 0.7797,
+      "step": 12110
+    },
+    {
+      "epoch": 0.26034282768397987,
+      "grad_norm": 0.5332911250510366,
+      "learning_rate": 1.700715843830825e-05,
+      "loss": 0.788,
+      "step": 12120
+    },
+    {
+      "epoch": 0.2605576319972505,
+      "grad_norm": 0.5385176920383756,
+      "learning_rate": 1.700229363738209e-05,
+      "loss": 0.7766,
+      "step": 12130
+    },
+    {
+      "epoch": 0.2607724363105211,
+      "grad_norm": 0.5598906242851724,
+      "learning_rate": 1.6997425582854704e-05,
+      "loss": 0.777,
+      "step": 12140
+    },
+    {
+      "epoch": 0.26098724062379175,
+      "grad_norm": 0.5487776354269147,
+      "learning_rate": 1.6992554276988022e-05,
+      "loss": 0.7737,
+      "step": 12150
+    },
+    {
+      "epoch": 0.26120204493706234,
+      "grad_norm": 0.5330929016698871,
+      "learning_rate": 1.6987679722045493e-05,
+      "loss": 0.7907,
+      "step": 12160
+    },
+    {
+      "epoch": 0.2614168492503329,
+      "grad_norm": 0.5382294593191113,
+      "learning_rate": 1.6982801920292063e-05,
+      "loss": 0.792,
+      "step": 12170
+    },
+    {
+      "epoch": 0.2616316535636036,
+      "grad_norm": 0.5458332411357165,
+      "learning_rate": 1.6977920873994196e-05,
+      "loss": 0.8109,
+      "step": 12180
+    },
+    {
+      "epoch": 0.26184645787687416,
+      "grad_norm": 0.5588172559612458,
+      "learning_rate": 1.697303658541986e-05,
+      "loss": 0.7976,
+      "step": 12190
+    },
+    {
+      "epoch": 0.2620612621901448,
+      "grad_norm": 0.5291579474271565,
+      "learning_rate": 1.6968149056838525e-05,
+      "loss": 0.7796,
+      "step": 12200
+    },
+    {
+      "epoch": 0.2622760665034154,
+      "grad_norm": 0.5477937857011975,
+      "learning_rate": 1.6963258290521173e-05,
+      "loss": 0.7831,
+      "step": 12210
+    },
+    {
+      "epoch": 0.262490870816686,
+      "grad_norm": 0.5311847854529211,
+      "learning_rate": 1.6958364288740293e-05,
+      "loss": 0.7584,
+      "step": 12220
+    },
+    {
+      "epoch": 0.26270567512995663,
+      "grad_norm": 0.5360611480627258,
+      "learning_rate": 1.6953467053769864e-05,
+      "loss": 0.7846,
+      "step": 12230
+    },
+    {
+      "epoch": 0.2629204794432272,
+      "grad_norm": 0.5704482862900305,
+      "learning_rate": 1.6948566587885388e-05,
+      "loss": 0.8022,
+      "step": 12240
+    },
+    {
+      "epoch": 0.2631352837564978,
+      "grad_norm": 0.5362588968839999,
+      "learning_rate": 1.6943662893363845e-05,
+      "loss": 0.7916,
+      "step": 12250
+    },
+    {
+      "epoch": 0.26335008806976845,
+      "grad_norm": 0.5675401491595801,
+      "learning_rate": 1.6938755972483732e-05,
+      "loss": 0.7959,
+      "step": 12260
+    },
+    {
+      "epoch": 0.26356489238303904,
+      "grad_norm": 0.5540442999120004,
+      "learning_rate": 1.6933845827525037e-05,
+      "loss": 0.7906,
+      "step": 12270
+    },
+    {
+      "epoch": 0.2637796966963097,
+      "grad_norm": 0.5414329042014562,
+      "learning_rate": 1.6928932460769254e-05,
+      "loss": 0.7819,
+      "step": 12280
+    },
+    {
+      "epoch": 0.2639945010095803,
+      "grad_norm": 0.5410748013073313,
+      "learning_rate": 1.6924015874499363e-05,
+      "loss": 0.7669,
+      "step": 12290
+    },
+    {
+      "epoch": 0.26420930532285086,
+      "grad_norm": 0.5472743067105443,
+      "learning_rate": 1.691909607099985e-05,
+      "loss": 0.7821,
+      "step": 12300
+    },
+    {
+      "epoch": 0.2644241096361215,
+      "grad_norm": 0.5475456006044419,
+      "learning_rate": 1.6914173052556688e-05,
+      "loss": 0.7811,
+      "step": 12310
+    },
+    {
+      "epoch": 0.2646389139493921,
+      "grad_norm": 0.5507328978978161,
+      "learning_rate": 1.6909246821457346e-05,
+      "loss": 0.7774,
+      "step": 12320
+    },
+    {
+      "epoch": 0.26485371826266274,
+      "grad_norm": 0.5599737303976604,
+      "learning_rate": 1.690431737999079e-05,
+      "loss": 0.7921,
+      "step": 12330
+    },
+    {
+      "epoch": 0.26506852257593333,
+      "grad_norm": 0.5280366450787144,
+      "learning_rate": 1.6899384730447477e-05,
+      "loss": 0.7847,
+      "step": 12340
+    },
+    {
+      "epoch": 0.2652833268892039,
+      "grad_norm": 0.5435632022539839,
+      "learning_rate": 1.689444887511935e-05,
+      "loss": 0.7854,
+      "step": 12350
+    },
+    {
+      "epoch": 0.26549813120247456,
+      "grad_norm": 0.5647493787301932,
+      "learning_rate": 1.6889509816299844e-05,
+      "loss": 0.7899,
+      "step": 12360
+    },
+    {
+      "epoch": 0.26571293551574515,
+      "grad_norm": 0.5610876101815535,
+      "learning_rate": 1.688456755628388e-05,
+      "loss": 0.7874,
+      "step": 12370
+    },
+    {
+      "epoch": 0.26592773982901574,
+      "grad_norm": 0.5498086456378252,
+      "learning_rate": 1.6879622097367874e-05,
+      "loss": 0.7941,
+      "step": 12380
+    },
+    {
+      "epoch": 0.2661425441422864,
+      "grad_norm": 0.5622568962892611,
+      "learning_rate": 1.6874673441849715e-05,
+      "loss": 0.7856,
+      "step": 12390
+    },
+    {
+      "epoch": 0.266357348455557,
+      "grad_norm": 0.5279820352190503,
+      "learning_rate": 1.6869721592028792e-05,
+      "loss": 0.7965,
+      "step": 12400
+    },
+    {
+      "epoch": 0.2665721527688276,
+      "grad_norm": 0.5316923893826371,
+      "learning_rate": 1.6864766550205977e-05,
+      "loss": 0.8056,
+      "step": 12410
+    },
+    {
+      "epoch": 0.2667869570820982,
+      "grad_norm": 0.5847422188711642,
+      "learning_rate": 1.6859808318683606e-05,
+      "loss": 0.7845,
+      "step": 12420
+    },
+    {
+      "epoch": 0.2670017613953688,
+      "grad_norm": 0.5183382490604992,
+      "learning_rate": 1.685484689976552e-05,
+      "loss": 0.7954,
+      "step": 12430
+    },
+    {
+      "epoch": 0.26721656570863944,
+      "grad_norm": 0.548534083313305,
+      "learning_rate": 1.6849882295757037e-05,
+      "loss": 0.7834,
+      "step": 12440
+    },
+    {
+      "epoch": 0.26743137002191003,
+      "grad_norm": 0.555857167141612,
+      "learning_rate": 1.6844914508964937e-05,
+      "loss": 0.7885,
+      "step": 12450
+    },
+    {
+      "epoch": 0.2676461743351807,
+      "grad_norm": 0.5609532665687568,
+      "learning_rate": 1.6839943541697498e-05,
+      "loss": 0.7804,
+      "step": 12460
+    },
+    {
+      "epoch": 0.26786097864845126,
+      "grad_norm": 0.5601428417687324,
+      "learning_rate": 1.6834969396264472e-05,
+      "loss": 0.7858,
+      "step": 12470
+    },
+    {
+      "epoch": 0.26807578296172185,
+      "grad_norm": 0.5476848469778967,
+      "learning_rate": 1.682999207497708e-05,
+      "loss": 0.7938,
+      "step": 12480
+    },
+    {
+      "epoch": 0.2682905872749925,
+      "grad_norm": 0.5638917271864747,
+      "learning_rate": 1.6825011580148033e-05,
+      "loss": 0.7837,
+      "step": 12490
+    },
+    {
+      "epoch": 0.2685053915882631,
+      "grad_norm": 0.5894522331907057,
+      "learning_rate": 1.6820027914091497e-05,
+      "loss": 0.7844,
+      "step": 12500
+    },
+    {
+      "epoch": 0.2687201959015337,
+      "grad_norm": 0.5610666708837045,
+      "learning_rate": 1.681504107912313e-05,
+      "loss": 0.7843,
+      "step": 12510
+    },
+    {
+      "epoch": 0.2689350002148043,
+      "grad_norm": 0.5446880622878415,
+      "learning_rate": 1.681005107756005e-05,
+      "loss": 0.7921,
+      "step": 12520
+    },
+    {
+      "epoch": 0.2691498045280749,
+      "grad_norm": 0.542926637375099,
+      "learning_rate": 1.6805057911720852e-05,
+      "loss": 0.7906,
+      "step": 12530
+    },
+    {
+      "epoch": 0.26936460884134555,
+      "grad_norm": 0.5750689040548754,
+      "learning_rate": 1.6800061583925603e-05,
+      "loss": 0.785,
+      "step": 12540
+    },
+    {
+      "epoch": 0.26957941315461614,
+      "grad_norm": 0.5144723555135201,
+      "learning_rate": 1.679506209649583e-05,
+      "loss": 0.7801,
+      "step": 12550
+    },
+    {
+      "epoch": 0.26979421746788673,
+      "grad_norm": 0.5368301892072017,
+      "learning_rate": 1.6790059451754545e-05,
+      "loss": 0.7764,
+      "step": 12560
+    },
+    {
+      "epoch": 0.2700090217811574,
+      "grad_norm": 0.5642425304345098,
+      "learning_rate": 1.6785053652026204e-05,
+      "loss": 0.8061,
+      "step": 12570
+    },
+    {
+      "epoch": 0.27022382609442797,
+      "grad_norm": 0.5539198208515579,
+      "learning_rate": 1.678004469963675e-05,
+      "loss": 0.7863,
+      "step": 12580
+    },
+    {
+      "epoch": 0.2704386304076986,
+      "grad_norm": 0.5512816690327085,
+      "learning_rate": 1.6775032596913576e-05,
+      "loss": 0.7911,
+      "step": 12590
+    },
+    {
+      "epoch": 0.2706534347209692,
+      "grad_norm": 0.5815790405385224,
+      "learning_rate": 1.677001734618555e-05,
+      "loss": 0.7852,
+      "step": 12600
+    },
+    {
+      "epoch": 0.2708682390342398,
+      "grad_norm": 0.5612088205707622,
+      "learning_rate": 1.6764998949783e-05,
+      "loss": 0.7828,
+      "step": 12610
+    },
+    {
+      "epoch": 0.27108304334751043,
+      "grad_norm": 0.530172464989036,
+      "learning_rate": 1.675997741003771e-05,
+      "loss": 0.7698,
+      "step": 12620
+    },
+    {
+      "epoch": 0.271297847660781,
+      "grad_norm": 0.552005993228374,
+      "learning_rate": 1.6754952729282926e-05,
+      "loss": 0.7745,
+      "step": 12630
+    },
+    {
+      "epoch": 0.27151265197405167,
+      "grad_norm": 0.5435122213473302,
+      "learning_rate": 1.6749924909853355e-05,
+      "loss": 0.7854,
+      "step": 12640
+    },
+    {
+      "epoch": 0.27172745628732226,
+      "grad_norm": 0.5530374063509907,
+      "learning_rate": 1.6744893954085173e-05,
+      "loss": 0.7958,
+      "step": 12650
+    },
+    {
+      "epoch": 0.27194226060059284,
+      "grad_norm": 0.5611802591773057,
+      "learning_rate": 1.673985986431599e-05,
+      "loss": 0.7966,
+      "step": 12660
+    },
+    {
+      "epoch": 0.2721570649138635,
+      "grad_norm": 0.5568671271875136,
+      "learning_rate": 1.6734822642884893e-05,
+      "loss": 0.7764,
+      "step": 12670
+    },
+    {
+      "epoch": 0.2723718692271341,
+      "grad_norm": 0.5572155300702621,
+      "learning_rate": 1.6729782292132415e-05,
+      "loss": 0.7757,
+      "step": 12680
+    },
+    {
+      "epoch": 0.27258667354040467,
+      "grad_norm": 0.5615138514243179,
+      "learning_rate": 1.6724738814400545e-05,
+      "loss": 0.798,
+      "step": 12690
+    },
+    {
+      "epoch": 0.2728014778536753,
+      "grad_norm": 0.5339476128169535,
+      "learning_rate": 1.671969221203272e-05,
+      "loss": 0.7828,
+      "step": 12700
+    },
+    {
+      "epoch": 0.2730162821669459,
+      "grad_norm": 0.5438064983467911,
+      "learning_rate": 1.671464248737384e-05,
+      "loss": 0.7792,
+      "step": 12710
+    },
+    {
+      "epoch": 0.27323108648021655,
+      "grad_norm": 0.5349876744589488,
+      "learning_rate": 1.6709589642770247e-05,
+      "loss": 0.7855,
+      "step": 12720
+    },
+    {
+      "epoch": 0.27344589079348713,
+      "grad_norm": 0.553863217138617,
+      "learning_rate": 1.6704533680569732e-05,
+      "loss": 0.7868,
+      "step": 12730
+    },
+    {
+      "epoch": 0.2736606951067577,
+      "grad_norm": 0.5390862263215795,
+      "learning_rate": 1.669947460312154e-05,
+      "loss": 0.7754,
+      "step": 12740
+    },
+    {
+      "epoch": 0.27387549942002837,
+      "grad_norm": 0.583549319445325,
+      "learning_rate": 1.669441241277636e-05,
+      "loss": 0.7917,
+      "step": 12750
+    },
+    {
+      "epoch": 0.27409030373329896,
+      "grad_norm": 0.5293298289589918,
+      "learning_rate": 1.668934711188633e-05,
+      "loss": 0.7909,
+      "step": 12760
+    },
+    {
+      "epoch": 0.2743051080465696,
+      "grad_norm": 0.569877867106919,
+      "learning_rate": 1.6684278702805024e-05,
+      "loss": 0.7767,
+      "step": 12770
+    },
+    {
+      "epoch": 0.2745199123598402,
+      "grad_norm": 0.5773582415344543,
+      "learning_rate": 1.667920718788748e-05,
+      "loss": 0.7856,
+      "step": 12780
+    },
+    {
+      "epoch": 0.2747347166731108,
+      "grad_norm": 0.5208586202102348,
+      "learning_rate": 1.667413256949016e-05,
+      "loss": 0.7805,
+      "step": 12790
+    },
+    {
+      "epoch": 0.2749495209863814,
+      "grad_norm": 0.5353938318257099,
+      "learning_rate": 1.666905484997097e-05,
+      "loss": 0.7876,
+      "step": 12800
+    },
+    {
+      "epoch": 0.275164325299652,
+      "grad_norm": 0.5424525339376945,
+      "learning_rate": 1.6663974031689272e-05,
+      "loss": 0.803,
+      "step": 12810
+    },
+    {
+      "epoch": 0.2753791296129226,
+      "grad_norm": 0.5505344864467518,
+      "learning_rate": 1.665889011700585e-05,
+      "loss": 0.7839,
+      "step": 12820
+    },
+    {
+      "epoch": 0.27559393392619325,
+      "grad_norm": 0.5688231819530734,
+      "learning_rate": 1.665380310828294e-05,
+      "loss": 0.8044,
+      "step": 12830
+    },
+    {
+      "epoch": 0.27580873823946384,
+      "grad_norm": 0.5423707638785459,
+      "learning_rate": 1.6648713007884207e-05,
+      "loss": 0.7813,
+      "step": 12840
+    },
+    {
+      "epoch": 0.2760235425527345,
+      "grad_norm": 0.5622431891874606,
+      "learning_rate": 1.6643619818174755e-05,
+      "loss": 0.792,
+      "step": 12850
+    },
+    {
+      "epoch": 0.27623834686600507,
+      "grad_norm": 0.5348635272783925,
+      "learning_rate": 1.6638523541521126e-05,
+      "loss": 0.8017,
+      "step": 12860
+    },
+    {
+      "epoch": 0.27645315117927566,
+      "grad_norm": 0.5539184848623694,
+      "learning_rate": 1.6633424180291293e-05,
+      "loss": 0.7935,
+      "step": 12870
+    },
+    {
+      "epoch": 0.2766679554925463,
+      "grad_norm": 0.531258076725924,
+      "learning_rate": 1.6628321736854664e-05,
+      "loss": 0.7771,
+      "step": 12880
+    },
+    {
+      "epoch": 0.2768827598058169,
+      "grad_norm": 0.5429903770907508,
+      "learning_rate": 1.662321621358208e-05,
+      "loss": 0.7682,
+      "step": 12890
+    },
+    {
+      "epoch": 0.27709756411908754,
+      "grad_norm": 0.5338561005284871,
+      "learning_rate": 1.6618107612845812e-05,
+      "loss": 0.7562,
+      "step": 12900
+    },
+    {
+      "epoch": 0.2773123684323581,
+      "grad_norm": 0.563699713244769,
+      "learning_rate": 1.6612995937019557e-05,
+      "loss": 0.7831,
+      "step": 12910
+    },
+    {
+      "epoch": 0.2775271727456287,
+      "grad_norm": 0.5974035990189168,
+      "learning_rate": 1.6607881188478446e-05,
+      "loss": 0.793,
+      "step": 12920
+    },
+    {
+      "epoch": 0.27774197705889936,
+      "grad_norm": 0.5786108959234898,
+      "learning_rate": 1.6602763369599037e-05,
+      "loss": 0.7701,
+      "step": 12930
+    },
+    {
+      "epoch": 0.27795678137216995,
+      "grad_norm": 0.5244598945283411,
+      "learning_rate": 1.659764248275932e-05,
+      "loss": 0.7752,
+      "step": 12940
+    },
+    {
+      "epoch": 0.27817158568544054,
+      "grad_norm": 0.5392263527270091,
+      "learning_rate": 1.6592518530338692e-05,
+      "loss": 0.789,
+      "step": 12950
+    },
+    {
+      "epoch": 0.2783863899987112,
+      "grad_norm": 0.5572545049836365,
+      "learning_rate": 1.6587391514718e-05,
+      "loss": 0.8154,
+      "step": 12960
+    },
+    {
+      "epoch": 0.27860119431198177,
+      "grad_norm": 0.5549009307377286,
+      "learning_rate": 1.6582261438279488e-05,
+      "loss": 0.7904,
+      "step": 12970
+    },
+    {
+      "epoch": 0.2788159986252524,
+      "grad_norm": 0.5736525203694448,
+      "learning_rate": 1.6577128303406843e-05,
+      "loss": 0.7838,
+      "step": 12980
+    },
+    {
+      "epoch": 0.279030802938523,
+      "grad_norm": 0.5422061227803314,
+      "learning_rate": 1.657199211248517e-05,
+      "loss": 0.7775,
+      "step": 12990
+    },
+    {
+      "epoch": 0.2792456072517936,
+      "grad_norm": 0.5552087862142333,
+      "learning_rate": 1.656685286790098e-05,
+      "loss": 0.7918,
+      "step": 13000
+    },
+    {
+      "epoch": 0.27946041156506424,
+      "grad_norm": 0.5769438755666713,
+      "learning_rate": 1.656171057204222e-05,
+      "loss": 0.773,
+      "step": 13010
+    },
+    {
+      "epoch": 0.2796752158783348,
+      "grad_norm": 0.544310206877259,
+      "learning_rate": 1.655656522729824e-05,
+      "loss": 0.787,
+      "step": 13020
+    },
+    {
+      "epoch": 0.27989002019160547,
+      "grad_norm": 0.5380675260235818,
+      "learning_rate": 1.6551416836059818e-05,
+      "loss": 0.7874,
+      "step": 13030
+    },
+    {
+      "epoch": 0.28010482450487606,
+      "grad_norm": 0.5340600751027904,
+      "learning_rate": 1.6546265400719143e-05,
+      "loss": 0.7858,
+      "step": 13040
+    },
+    {
+      "epoch": 0.28031962881814665,
+      "grad_norm": 0.547403375047864,
+      "learning_rate": 1.6541110923669816e-05,
+      "loss": 0.7847,
+      "step": 13050
+    },
+    {
+      "epoch": 0.2805344331314173,
+      "grad_norm": 0.533402580926284,
+      "learning_rate": 1.653595340730686e-05,
+      "loss": 0.8004,
+      "step": 13060
+    },
+    {
+      "epoch": 0.2807492374446879,
+      "grad_norm": 0.5426741999184683,
+      "learning_rate": 1.65307928540267e-05,
+      "loss": 0.7797,
+      "step": 13070
+    },
+    {
+      "epoch": 0.2809640417579585,
+      "grad_norm": 0.5342611404015724,
+      "learning_rate": 1.6525629266227177e-05,
+      "loss": 0.7787,
+      "step": 13080
+    },
+    {
+      "epoch": 0.2811788460712291,
+      "grad_norm": 1.7861965467084178,
+      "learning_rate": 1.6520462646307543e-05,
+      "loss": 0.8042,
+      "step": 13090
+    },
+    {
+      "epoch": 0.2813936503844997,
+      "grad_norm": 0.5453771703119942,
+      "learning_rate": 1.6515292996668452e-05,
+      "loss": 0.7896,
+      "step": 13100
+    },
+    {
+      "epoch": 0.28160845469777035,
+      "grad_norm": 0.5393581439017914,
+      "learning_rate": 1.6510120319711974e-05,
+      "loss": 0.7636,
+      "step": 13110
+    },
+    {
+      "epoch": 0.28182325901104094,
+      "grad_norm": 0.566923191821844,
+      "learning_rate": 1.650494461784159e-05,
+      "loss": 0.7824,
+      "step": 13120
+    },
+    {
+      "epoch": 0.2820380633243115,
+      "grad_norm": 0.5909033333766904,
+      "learning_rate": 1.649976589346217e-05,
+      "loss": 0.7791,
+      "step": 13130
+    },
+    {
+      "epoch": 0.28225286763758217,
+      "grad_norm": 0.5459678033195983,
+      "learning_rate": 1.6494584148979996e-05,
+      "loss": 0.7861,
+      "step": 13140
+    },
+    {
+      "epoch": 0.28246767195085276,
+      "grad_norm": 0.5763291162585454,
+      "learning_rate": 1.648939938680276e-05,
+      "loss": 0.7864,
+      "step": 13150
+    },
+    {
+      "epoch": 0.2826824762641234,
+      "grad_norm": 0.5341209524759871,
+      "learning_rate": 1.6484211609339555e-05,
+      "loss": 0.7803,
+      "step": 13160
+    },
+    {
+      "epoch": 0.282897280577394,
+      "grad_norm": 0.5736855699742509,
+      "learning_rate": 1.647902081900086e-05,
+      "loss": 0.7897,
+      "step": 13170
+    },
+    {
+      "epoch": 0.2831120848906646,
+      "grad_norm": 0.5531932895913767,
+      "learning_rate": 1.6473827018198573e-05,
+      "loss": 0.7928,
+      "step": 13180
+    },
+    {
+      "epoch": 0.28332688920393523,
+      "grad_norm": 0.5640369977192191,
+      "learning_rate": 1.6468630209345984e-05,
+      "loss": 0.8076,
+      "step": 13190
+    },
+    {
+      "epoch": 0.2835416935172058,
+      "grad_norm": 0.5609123963099327,
+      "learning_rate": 1.6463430394857772e-05,
+      "loss": 0.7912,
+      "step": 13200
+    },
+    {
+      "epoch": 0.28375649783047646,
+      "grad_norm": 0.5557233716140605,
+      "learning_rate": 1.6458227577150024e-05,
+      "loss": 0.788,
+      "step": 13210
+    },
+    {
+      "epoch": 0.28397130214374705,
+      "grad_norm": 0.5662855198896333,
+      "learning_rate": 1.6453021758640218e-05,
+      "loss": 0.7884,
+      "step": 13220
+    },
+    {
+      "epoch": 0.28418610645701764,
+      "grad_norm": 0.5412381516020387,
+      "learning_rate": 1.644781294174723e-05,
+      "loss": 0.7956,
+      "step": 13230
+    },
+    {
+      "epoch": 0.2844009107702883,
+      "grad_norm": 0.5385533947113791,
+      "learning_rate": 1.6442601128891323e-05,
+      "loss": 0.7864,
+      "step": 13240
+    },
+    {
+      "epoch": 0.2846157150835589,
+      "grad_norm": 0.531510421245019,
+      "learning_rate": 1.6437386322494155e-05,
+      "loss": 0.7705,
+      "step": 13250
+    },
+    {
+      "epoch": 0.28483051939682946,
+      "grad_norm": 0.5426033523747722,
+      "learning_rate": 1.6432168524978777e-05,
+      "loss": 0.7787,
+      "step": 13260
+    },
+    {
+      "epoch": 0.2850453237101001,
+      "grad_norm": 0.5623236256089558,
+      "learning_rate": 1.6426947738769623e-05,
+      "loss": 0.7813,
+      "step": 13270
+    },
+    {
+      "epoch": 0.2852601280233707,
+      "grad_norm": 0.5394725869592929,
+      "learning_rate": 1.6421723966292525e-05,
+      "loss": 0.7738,
+      "step": 13280
+    },
+    {
+      "epoch": 0.28547493233664134,
+      "grad_norm": 0.5465469510807215,
+      "learning_rate": 1.6416497209974697e-05,
+      "loss": 0.7557,
+      "step": 13290
+    },
+    {
+      "epoch": 0.28568973664991193,
+      "grad_norm": 0.5371708781271813,
+      "learning_rate": 1.641126747224474e-05,
+      "loss": 0.7871,
+      "step": 13300
+    },
+    {
+      "epoch": 0.2859045409631825,
+      "grad_norm": 0.5400513136671761,
+      "learning_rate": 1.640603475553264e-05,
+      "loss": 0.7835,
+      "step": 13310
+    },
+    {
+      "epoch": 0.28611934527645316,
+      "grad_norm": 0.5394221231389933,
+      "learning_rate": 1.640079906226977e-05,
+      "loss": 0.765,
+      "step": 13320
+    },
+    {
+      "epoch": 0.28633414958972375,
+      "grad_norm": 0.5222441443064099,
+      "learning_rate": 1.6395560394888888e-05,
+      "loss": 0.7757,
+      "step": 13330
+    },
+    {
+      "epoch": 0.2865489539029944,
+      "grad_norm": 0.5606502765218887,
+      "learning_rate": 1.6390318755824118e-05,
+      "loss": 0.7868,
+      "step": 13340
+    },
+    {
+      "epoch": 0.286763758216265,
+      "grad_norm": 0.609652591588806,
+      "learning_rate": 1.6385074147510987e-05,
+      "loss": 0.7809,
+      "step": 13350
+    },
+    {
+      "epoch": 0.2869785625295356,
+      "grad_norm": 0.526740208124346,
+      "learning_rate": 1.6379826572386386e-05,
+      "loss": 0.7898,
+      "step": 13360
+    },
+    {
+      "epoch": 0.2871933668428062,
+      "grad_norm": 0.5305922224130004,
+      "learning_rate": 1.6374576032888594e-05,
+      "loss": 0.7807,
+      "step": 13370
+    },
+    {
+      "epoch": 0.2874081711560768,
+      "grad_norm": 0.5540486723971066,
+      "learning_rate": 1.6369322531457263e-05,
+      "loss": 0.7788,
+      "step": 13380
+    },
+    {
+      "epoch": 0.2876229754693474,
+      "grad_norm": 0.52462212158824,
+      "learning_rate": 1.6364066070533414e-05,
+      "loss": 0.7933,
+      "step": 13390
+    },
+    {
+      "epoch": 0.28783777978261804,
+      "grad_norm": 0.531247543108683,
+      "learning_rate": 1.635880665255946e-05,
+      "loss": 0.776,
+      "step": 13400
+    },
+    {
+      "epoch": 0.28805258409588863,
+      "grad_norm": 0.5732783680452849,
+      "learning_rate": 1.6353544279979177e-05,
+      "loss": 0.7931,
+      "step": 13410
+    },
+    {
+      "epoch": 0.2882673884091593,
+      "grad_norm": 0.542974827627677,
+      "learning_rate": 1.634827895523771e-05,
+      "loss": 0.7782,
+      "step": 13420
+    },
+    {
+      "epoch": 0.28848219272242986,
+      "grad_norm": 0.5876492011137021,
+      "learning_rate": 1.6343010680781586e-05,
+      "loss": 0.7774,
+      "step": 13430
+    },
+    {
+      "epoch": 0.28869699703570045,
+      "grad_norm": 0.5333223558241071,
+      "learning_rate": 1.63377394590587e-05,
+      "loss": 0.7908,
+      "step": 13440
+    },
+    {
+      "epoch": 0.2889118013489711,
+      "grad_norm": 0.5435404704513007,
+      "learning_rate": 1.6332465292518306e-05,
+      "loss": 0.784,
+      "step": 13450
+    },
+    {
+      "epoch": 0.2891266056622417,
+      "grad_norm": 0.5372220581463112,
+      "learning_rate": 1.6327188183611043e-05,
+      "loss": 0.7877,
+      "step": 13460
+    },
+    {
+      "epoch": 0.28934140997551233,
+      "grad_norm": 0.5517875641980315,
+      "learning_rate": 1.6321908134788904e-05,
+      "loss": 0.7755,
+      "step": 13470
+    },
+    {
+      "epoch": 0.2895562142887829,
+      "grad_norm": 0.5624183224646342,
+      "learning_rate": 1.6316625148505253e-05,
+      "loss": 0.7697,
+      "step": 13480
+    },
+    {
+      "epoch": 0.2897710186020535,
+      "grad_norm": 0.5190475908632274,
+      "learning_rate": 1.631133922721482e-05,
+      "loss": 0.7768,
+      "step": 13490
+    },
+    {
+      "epoch": 0.28998582291532415,
+      "grad_norm": 0.5513772019738882,
+      "learning_rate": 1.6306050373373698e-05,
+      "loss": 0.7742,
+      "step": 13500
+    },
+    {
+      "epoch": 0.29020062722859474,
+      "grad_norm": 0.552607695073294,
+      "learning_rate": 1.6300758589439342e-05,
+      "loss": 0.787,
+      "step": 13510
+    },
+    {
+      "epoch": 0.2904154315418654,
+      "grad_norm": 0.5521440595481499,
+      "learning_rate": 1.6295463877870566e-05,
+      "loss": 0.7812,
+      "step": 13520
+    },
+    {
+      "epoch": 0.290630235855136,
+      "grad_norm": 0.541378746109039,
+      "learning_rate": 1.6290166241127545e-05,
+      "loss": 0.7813,
+      "step": 13530
+    },
+    {
+      "epoch": 0.29084504016840657,
+      "grad_norm": 0.5837702840940312,
+      "learning_rate": 1.6284865681671826e-05,
+      "loss": 0.7875,
+      "step": 13540
+    },
+    {
+      "epoch": 0.2910598444816772,
+      "grad_norm": 0.531456315705228,
+      "learning_rate": 1.6279562201966287e-05,
+      "loss": 0.7744,
+      "step": 13550
+    },
+    {
+      "epoch": 0.2912746487949478,
+      "grad_norm": 0.5409752871762843,
+      "learning_rate": 1.627425580447519e-05,
+      "loss": 0.7848,
+      "step": 13560
+    },
+    {
+      "epoch": 0.2914894531082184,
+      "grad_norm": 0.5685739203049746,
+      "learning_rate": 1.626894649166414e-05,
+      "loss": 0.7902,
+      "step": 13570
+    },
+    {
+      "epoch": 0.29170425742148903,
+      "grad_norm": 0.5290562340681636,
+      "learning_rate": 1.6263634266000093e-05,
+      "loss": 0.7748,
+      "step": 13580
+    },
+    {
+      "epoch": 0.2919190617347596,
+      "grad_norm": 0.5526529473322866,
+      "learning_rate": 1.6258319129951366e-05,
+      "loss": 0.7911,
+      "step": 13590
+    },
+    {
+      "epoch": 0.29213386604803027,
+      "grad_norm": 0.5379467854648743,
+      "learning_rate": 1.6253001085987635e-05,
+      "loss": 0.7837,
+      "step": 13600
+    },
+    {
+      "epoch": 0.29234867036130086,
+      "grad_norm": 0.5629511775533564,
+      "learning_rate": 1.6247680136579904e-05,
+      "loss": 0.7904,
+      "step": 13610
+    },
+    {
+      "epoch": 0.29256347467457144,
+      "grad_norm": 0.5637137982010608,
+      "learning_rate": 1.624235628420055e-05,
+      "loss": 0.7789,
+      "step": 13620
+    },
+    {
+      "epoch": 0.2927782789878421,
+      "grad_norm": 0.5552139861154454,
+      "learning_rate": 1.6237029531323286e-05,
+      "loss": 0.7766,
+      "step": 13630
+    },
+    {
+      "epoch": 0.2929930833011127,
+      "grad_norm": 0.5311375889317766,
+      "learning_rate": 1.6231699880423182e-05,
+      "loss": 0.7878,
+      "step": 13640
+    },
+    {
+      "epoch": 0.2932078876143833,
+      "grad_norm": 0.5438973923343646,
+      "learning_rate": 1.6226367333976642e-05,
+      "loss": 0.7665,
+      "step": 13650
+    },
+    {
+      "epoch": 0.2934226919276539,
+      "grad_norm": 0.5471337589971532,
+      "learning_rate": 1.6221031894461426e-05,
+      "loss": 0.7712,
+      "step": 13660
+    },
+    {
+      "epoch": 0.2936374962409245,
+      "grad_norm": 0.5478152109462807,
+      "learning_rate": 1.6215693564356635e-05,
+      "loss": 0.7734,
+      "step": 13670
+    },
+    {
+      "epoch": 0.29385230055419514,
+      "grad_norm": 0.5523511877357635,
+      "learning_rate": 1.6210352346142713e-05,
+      "loss": 0.7882,
+      "step": 13680
+    },
+    {
+      "epoch": 0.29406710486746573,
+      "grad_norm": 0.5405593748173753,
+      "learning_rate": 1.6205008242301445e-05,
+      "loss": 0.7967,
+      "step": 13690
+    },
+    {
+      "epoch": 0.2942819091807363,
+      "grad_norm": 0.5393904185913311,
+      "learning_rate": 1.619966125531596e-05,
+      "loss": 0.7975,
+      "step": 13700
+    },
+    {
+      "epoch": 0.29449671349400697,
+      "grad_norm": 0.5459167139610831,
+      "learning_rate": 1.6194311387670726e-05,
+      "loss": 0.7802,
+      "step": 13710
+    },
+    {
+      "epoch": 0.29471151780727756,
+      "grad_norm": 0.5734671415245739,
+      "learning_rate": 1.618895864185154e-05,
+      "loss": 0.7805,
+      "step": 13720
+    },
+    {
+      "epoch": 0.2949263221205482,
+      "grad_norm": 0.5200409133048494,
+      "learning_rate": 1.6183603020345552e-05,
+      "loss": 0.7684,
+      "step": 13730
+    },
+    {
+      "epoch": 0.2951411264338188,
+      "grad_norm": 0.532032766630498,
+      "learning_rate": 1.6178244525641238e-05,
+      "loss": 0.7607,
+      "step": 13740
+    },
+    {
+      "epoch": 0.2953559307470894,
+      "grad_norm": 0.5450386213659163,
+      "learning_rate": 1.617288316022841e-05,
+      "loss": 0.7853,
+      "step": 13750
+    },
+    {
+      "epoch": 0.29557073506036,
+      "grad_norm": 0.5412583231864979,
+      "learning_rate": 1.6167518926598215e-05,
+      "loss": 0.7878,
+      "step": 13760
+    },
+    {
+      "epoch": 0.2957855393736306,
+      "grad_norm": 0.5495103389055368,
+      "learning_rate": 1.616215182724314e-05,
+      "loss": 0.7768,
+      "step": 13770
+    },
+    {
+      "epoch": 0.29600034368690126,
+      "grad_norm": 0.5735150374820255,
+      "learning_rate": 1.6156781864656984e-05,
+      "loss": 0.7826,
+      "step": 13780
+    },
+    {
+      "epoch": 0.29621514800017185,
+      "grad_norm": 0.535148317937703,
+      "learning_rate": 1.6151409041334903e-05,
+      "loss": 0.7871,
+      "step": 13790
+    },
+    {
+      "epoch": 0.29642995231344244,
+      "grad_norm": 0.5237498381145003,
+      "learning_rate": 1.6146033359773356e-05,
+      "loss": 0.7759,
+      "step": 13800
+    },
+    {
+      "epoch": 0.2966447566267131,
+      "grad_norm": 0.5342689644277192,
+      "learning_rate": 1.614065482247015e-05,
+      "loss": 0.7736,
+      "step": 13810
+    },
+    {
+      "epoch": 0.29685956093998367,
+      "grad_norm": 0.5322183986533855,
+      "learning_rate": 1.6135273431924408e-05,
+      "loss": 0.7913,
+      "step": 13820
+    },
+    {
+      "epoch": 0.29707436525325426,
+      "grad_norm": 0.5536125865349972,
+      "learning_rate": 1.6129889190636582e-05,
+      "loss": 0.7836,
+      "step": 13830
+    },
+    {
+      "epoch": 0.2972891695665249,
+      "grad_norm": 0.5320481447634432,
+      "learning_rate": 1.6124502101108453e-05,
+      "loss": 0.7802,
+      "step": 13840
+    },
+    {
+      "epoch": 0.2975039738797955,
+      "grad_norm": 0.5337884474675023,
+      "learning_rate": 1.6119112165843114e-05,
+      "loss": 0.7711,
+      "step": 13850
+    },
+    {
+      "epoch": 0.29771877819306614,
+      "grad_norm": 0.5448873311868312,
+      "learning_rate": 1.6113719387344992e-05,
+      "loss": 0.7763,
+      "step": 13860
+    },
+    {
+      "epoch": 0.2979335825063367,
+      "grad_norm": 0.5390758922033408,
+      "learning_rate": 1.6108323768119827e-05,
+      "loss": 0.7739,
+      "step": 13870
+    },
+    {
+      "epoch": 0.2981483868196073,
+      "grad_norm": 0.5365705766236541,
+      "learning_rate": 1.6102925310674686e-05,
+      "loss": 0.7847,
+      "step": 13880
+    },
+    {
+      "epoch": 0.29836319113287796,
+      "grad_norm": 0.5590376051193626,
+      "learning_rate": 1.6097524017517948e-05,
+      "loss": 0.7854,
+      "step": 13890
+    },
+    {
+      "epoch": 0.29857799544614855,
+      "grad_norm": 0.5309793859058803,
+      "learning_rate": 1.609211989115932e-05,
+      "loss": 0.7692,
+      "step": 13900
+    },
+    {
+      "epoch": 0.2987927997594192,
+      "grad_norm": 0.5308210455003953,
+      "learning_rate": 1.6086712934109804e-05,
+      "loss": 0.785,
+      "step": 13910
+    },
+    {
+      "epoch": 0.2990076040726898,
+      "grad_norm": 0.5490327798250314,
+      "learning_rate": 1.608130314888174e-05,
+      "loss": 0.7735,
+      "step": 13920
+    },
+    {
+      "epoch": 0.29922240838596037,
+      "grad_norm": 0.5343320632700467,
+      "learning_rate": 1.6075890537988778e-05,
+      "loss": 0.7903,
+      "step": 13930
+    },
+    {
+      "epoch": 0.299437212699231,
+      "grad_norm": 0.5479636639748584,
+      "learning_rate": 1.6070475103945867e-05,
+      "loss": 0.7889,
+      "step": 13940
+    },
+    {
+      "epoch": 0.2996520170125016,
+      "grad_norm": 0.5527723222499173,
+      "learning_rate": 1.6065056849269286e-05,
+      "loss": 0.7697,
+      "step": 13950
+    },
+    {
+      "epoch": 0.29986682132577225,
+      "grad_norm": 0.5576773833757296,
+      "learning_rate": 1.6059635776476607e-05,
+      "loss": 0.7796,
+      "step": 13960
+    },
+    {
+      "epoch": 0.30008162563904284,
+      "grad_norm": 0.540017123946042,
+      "learning_rate": 1.6054211888086726e-05,
+      "loss": 0.7789,
+      "step": 13970
+    },
+    {
+      "epoch": 0.3002964299523134,
+      "grad_norm": 0.5382534676965747,
+      "learning_rate": 1.604878518661984e-05,
+      "loss": 0.7864,
+      "step": 13980
+    },
+    {
+      "epoch": 0.30051123426558407,
+      "grad_norm": 0.5620971765430572,
+      "learning_rate": 1.6043355674597456e-05,
+      "loss": 0.7724,
+      "step": 13990
+    },
+    {
+      "epoch": 0.30072603857885466,
+      "grad_norm": 0.5512920688522195,
+      "learning_rate": 1.603792335454238e-05,
+      "loss": 0.7686,
+      "step": 14000
+    },
+    {
+      "epoch": 0.30094084289212525,
+      "grad_norm": 0.5681571893413786,
+      "learning_rate": 1.603248822897874e-05,
+      "loss": 0.7796,
+      "step": 14010
+    },
+    {
+      "epoch": 0.3011556472053959,
+      "grad_norm": 0.5466486772113822,
+      "learning_rate": 1.6027050300431945e-05,
+      "loss": 0.7809,
+      "step": 14020
+    },
+    {
+      "epoch": 0.3013704515186665,
+      "grad_norm": 0.5235657333916433,
+      "learning_rate": 1.6021609571428718e-05,
+      "loss": 0.7868,
+      "step": 14030
+    },
+    {
+      "epoch": 0.3015852558319371,
+      "grad_norm": 0.5456016921796866,
+      "learning_rate": 1.6016166044497085e-05,
+      "loss": 0.7682,
+      "step": 14040
+    },
+    {
+      "epoch": 0.3018000601452077,
+      "grad_norm": 0.5352992818298888,
+      "learning_rate": 1.6010719722166373e-05,
+      "loss": 0.7809,
+      "step": 14050
+    },
+    {
+      "epoch": 0.3020148644584783,
+      "grad_norm": 0.5282682341595342,
+      "learning_rate": 1.6005270606967197e-05,
+      "loss": 0.7844,
+      "step": 14060
+    },
+    {
+      "epoch": 0.30222966877174895,
+      "grad_norm": 0.5139635209270281,
+      "learning_rate": 1.5999818701431485e-05,
+      "loss": 0.7667,
+      "step": 14070
+    },
+    {
+      "epoch": 0.30244447308501954,
+      "grad_norm": 0.5381885807038028,
+      "learning_rate": 1.599436400809245e-05,
+      "loss": 0.7911,
+      "step": 14080
+    },
+    {
+      "epoch": 0.3026592773982902,
+      "grad_norm": 0.5155734950185481,
+      "learning_rate": 1.59889065294846e-05,
+      "loss": 0.7656,
+      "step": 14090
+    },
+    {
+      "epoch": 0.30287408171156077,
+      "grad_norm": 0.5564242549239296,
+      "learning_rate": 1.5983446268143746e-05,
+      "loss": 0.7822,
+      "step": 14100
+    },
+    {
+      "epoch": 0.30308888602483136,
+      "grad_norm": 0.5187487100642706,
+      "learning_rate": 1.5977983226606987e-05,
+      "loss": 0.7908,
+      "step": 14110
+    },
+    {
+      "epoch": 0.303303690338102,
+      "grad_norm": 0.5642387470463246,
+      "learning_rate": 1.597251740741271e-05,
+      "loss": 0.7885,
+      "step": 14120
+    },
+    {
+      "epoch": 0.3035184946513726,
+      "grad_norm": 0.536223279687209,
+      "learning_rate": 1.59670488131006e-05,
+      "loss": 0.7718,
+      "step": 14130
+    },
+    {
+      "epoch": 0.3037332989646432,
+      "grad_norm": 0.5600475408070319,
+      "learning_rate": 1.5961577446211627e-05,
+      "loss": 0.7776,
+      "step": 14140
+    },
+    {
+      "epoch": 0.30394810327791383,
+      "grad_norm": 0.5437843190566551,
+      "learning_rate": 1.5956103309288053e-05,
+      "loss": 0.7685,
+      "step": 14150
+    },
+    {
+      "epoch": 0.3041629075911844,
+      "grad_norm": 0.5628863721101852,
+      "learning_rate": 1.5950626404873418e-05,
+      "loss": 0.7918,
+      "step": 14160
+    },
+    {
+      "epoch": 0.30437771190445506,
+      "grad_norm": 0.5149513964803338,
+      "learning_rate": 1.594514673551256e-05,
+      "loss": 0.7751,
+      "step": 14170
+    },
+    {
+      "epoch": 0.30459251621772565,
+      "grad_norm": 0.5563402316389365,
+      "learning_rate": 1.5939664303751596e-05,
+      "loss": 0.761,
+      "step": 14180
+    },
+    {
+      "epoch": 0.30480732053099624,
+      "grad_norm": 0.55820340770972,
+      "learning_rate": 1.5934179112137923e-05,
+      "loss": 0.8045,
+      "step": 14190
+    },
+    {
+      "epoch": 0.3050221248442669,
+      "grad_norm": 0.5265505329965681,
+      "learning_rate": 1.5928691163220228e-05,
+      "loss": 0.7685,
+      "step": 14200
+    },
+    {
+      "epoch": 0.3052369291575375,
+      "grad_norm": 0.5362485863281781,
+      "learning_rate": 1.592320045954847e-05,
+      "loss": 0.7765,
+      "step": 14210
+    },
+    {
+      "epoch": 0.3054517334708081,
+      "grad_norm": 0.5529028127823579,
+      "learning_rate": 1.5917707003673895e-05,
+      "loss": 0.7899,
+      "step": 14220
+    },
+    {
+      "epoch": 0.3056665377840787,
+      "grad_norm": 0.5546518668601578,
+      "learning_rate": 1.591221079814903e-05,
+      "loss": 0.7863,
+      "step": 14230
+    },
+    {
+      "epoch": 0.3058813420973493,
+      "grad_norm": 0.5542533511738308,
+      "learning_rate": 1.590671184552767e-05,
+      "loss": 0.7713,
+      "step": 14240
+    },
+    {
+      "epoch": 0.30609614641061994,
+      "grad_norm": 0.5234620656060907,
+      "learning_rate": 1.5901210148364895e-05,
+      "loss": 0.7673,
+      "step": 14250
+    },
+    {
+      "epoch": 0.30631095072389053,
+      "grad_norm": 0.5687517220863817,
+      "learning_rate": 1.5895705709217056e-05,
+      "loss": 0.796,
+      "step": 14260
+    },
+    {
+      "epoch": 0.3065257550371612,
+      "grad_norm": 0.5396488135156106,
+      "learning_rate": 1.589019853064178e-05,
+      "loss": 0.7718,
+      "step": 14270
+    },
+    {
+      "epoch": 0.30674055935043176,
+      "grad_norm": 0.538832149688794,
+      "learning_rate": 1.5884688615197964e-05,
+      "loss": 0.7881,
+      "step": 14280
+    },
+    {
+      "epoch": 0.30695536366370235,
+      "grad_norm": 0.5560141103067452,
+      "learning_rate": 1.5879175965445782e-05,
+      "loss": 0.7714,
+      "step": 14290
+    },
+    {
+      "epoch": 0.307170167976973,
+      "grad_norm": 0.5565327133097089,
+      "learning_rate": 1.5873660583946673e-05,
+      "loss": 0.781,
+      "step": 14300
+    },
+    {
+      "epoch": 0.3073849722902436,
+      "grad_norm": 0.561419912796845,
+      "learning_rate": 1.5868142473263352e-05,
+      "loss": 0.7683,
+      "step": 14310
+    },
+    {
+      "epoch": 0.3075997766035142,
+      "grad_norm": 0.5469680337173222,
+      "learning_rate": 1.5862621635959788e-05,
+      "loss": 0.7952,
+      "step": 14320
+    },
+    {
+      "epoch": 0.3078145809167848,
+      "grad_norm": 0.6073248751313165,
+      "learning_rate": 1.5857098074601236e-05,
+      "loss": 0.7978,
+      "step": 14330
+    },
+    {
+      "epoch": 0.3080293852300554,
+      "grad_norm": 0.5435528930279098,
+      "learning_rate": 1.5851571791754205e-05,
+      "loss": 0.7731,
+      "step": 14340
+    },
+    {
+      "epoch": 0.30824418954332605,
+      "grad_norm": 0.5201227548391855,
+      "learning_rate": 1.584604278998647e-05,
+      "loss": 0.7774,
+      "step": 14350
+    },
+    {
+      "epoch": 0.30845899385659664,
+      "grad_norm": 0.5494541529358633,
+      "learning_rate": 1.5840511071867065e-05,
+      "loss": 0.7668,
+      "step": 14360
+    },
+    {
+      "epoch": 0.30867379816986723,
+      "grad_norm": 0.5795936450755371,
+      "learning_rate": 1.58349766399663e-05,
+      "loss": 0.7727,
+      "step": 14370
+    },
+    {
+      "epoch": 0.3088886024831379,
+      "grad_norm": 0.569527328062543,
+      "learning_rate": 1.5829439496855735e-05,
+      "loss": 0.7942,
+      "step": 14380
+    },
+    {
+      "epoch": 0.30910340679640846,
+      "grad_norm": 0.5347014125524017,
+      "learning_rate": 1.582389964510819e-05,
+      "loss": 0.763,
+      "step": 14390
+    },
+    {
+      "epoch": 0.3093182111096791,
+      "grad_norm": 0.5292356683082766,
+      "learning_rate": 1.5818357087297746e-05,
+      "loss": 0.7733,
+      "step": 14400
+    },
+    {
+      "epoch": 0.3095330154229497,
+      "grad_norm": 0.5564018120086363,
+      "learning_rate": 1.5812811825999742e-05,
+      "loss": 0.7832,
+      "step": 14410
+    },
+    {
+      "epoch": 0.3097478197362203,
+      "grad_norm": 0.5406172696967337,
+      "learning_rate": 1.580726386379077e-05,
+      "loss": 0.7755,
+      "step": 14420
+    },
+    {
+      "epoch": 0.30996262404949093,
+      "grad_norm": 0.5461638054581808,
+      "learning_rate": 1.5801713203248683e-05,
+      "loss": 0.7625,
+      "step": 14430
+    },
+    {
+      "epoch": 0.3101774283627615,
+      "grad_norm": 0.5454736211561317,
+      "learning_rate": 1.5796159846952578e-05,
+      "loss": 0.7815,
+      "step": 14440
+    },
+    {
+      "epoch": 0.3103922326760321,
+      "grad_norm": 0.5474695031583966,
+      "learning_rate": 1.579060379748282e-05,
+      "loss": 0.793,
+      "step": 14450
+    },
+    {
+      "epoch": 0.31060703698930275,
+      "grad_norm": 0.5312856498486378,
+      "learning_rate": 1.5785045057421006e-05,
+      "loss": 0.7645,
+      "step": 14460
+    },
+    {
+      "epoch": 0.31082184130257334,
+      "grad_norm": 0.5165648723128465,
+      "learning_rate": 1.5779483629349997e-05,
+      "loss": 0.7595,
+      "step": 14470
+    },
+    {
+      "epoch": 0.311036645615844,
+      "grad_norm": 0.5559441987275184,
+      "learning_rate": 1.57739195158539e-05,
+      "loss": 0.7839,
+      "step": 14480
+    },
+    {
+      "epoch": 0.3112514499291146,
+      "grad_norm": 0.5769441262612243,
+      "learning_rate": 1.5768352719518068e-05,
+      "loss": 0.7738,
+      "step": 14490
+    },
+    {
+      "epoch": 0.31146625424238517,
+      "grad_norm": 0.5473900907836559,
+      "learning_rate": 1.57627832429291e-05,
+      "loss": 0.7865,
+      "step": 14500
+    },
+    {
+      "epoch": 0.3116810585556558,
+      "grad_norm": 0.529556628584659,
+      "learning_rate": 1.5757211088674845e-05,
+      "loss": 0.7745,
+      "step": 14510
+    },
+    {
+      "epoch": 0.3118958628689264,
+      "grad_norm": 0.5361987622170035,
+      "learning_rate": 1.575163625934439e-05,
+      "loss": 0.7777,
+      "step": 14520
+    },
+    {
+      "epoch": 0.31211066718219704,
+      "grad_norm": 0.5262241261547033,
+      "learning_rate": 1.574605875752807e-05,
+      "loss": 0.778,
+      "step": 14530
+    },
+    {
+      "epoch": 0.31232547149546763,
+      "grad_norm": 0.5232473606874927,
+      "learning_rate": 1.5740478585817455e-05,
+      "loss": 0.7606,
+      "step": 14540
+    },
+    {
+      "epoch": 0.3125402758087382,
+      "grad_norm": 0.5302585690675944,
+      "learning_rate": 1.573489574680537e-05,
+      "loss": 0.7738,
+      "step": 14550
+    },
+    {
+      "epoch": 0.31275508012200887,
+      "grad_norm": 0.5338970296170283,
+      "learning_rate": 1.5729310243085858e-05,
+      "loss": 0.7769,
+      "step": 14560
+    },
+    {
+      "epoch": 0.31296988443527946,
+      "grad_norm": 0.5337153071748394,
+      "learning_rate": 1.5723722077254216e-05,
+      "loss": 0.7738,
+      "step": 14570
+    },
+    {
+      "epoch": 0.31318468874855004,
+      "grad_norm": 0.5408069906876969,
+      "learning_rate": 1.5718131251906978e-05,
+      "loss": 0.7837,
+      "step": 14580
+    },
+    {
+      "epoch": 0.3133994930618207,
+      "grad_norm": 0.5317712434488392,
+      "learning_rate": 1.5712537769641905e-05,
+      "loss": 0.7823,
+      "step": 14590
+    },
+    {
+      "epoch": 0.3136142973750913,
+      "grad_norm": 0.5312973847806638,
+      "learning_rate": 1.5706941633058e-05,
+      "loss": 0.7712,
+      "step": 14600
+    },
+    {
+      "epoch": 0.3138291016883619,
+      "grad_norm": 0.5465034121886095,
+      "learning_rate": 1.5701342844755492e-05,
+      "loss": 0.7767,
+      "step": 14610
+    },
+    {
+      "epoch": 0.3140439060016325,
+      "grad_norm": 0.5301591392732491,
+      "learning_rate": 1.5695741407335852e-05,
+      "loss": 0.7629,
+      "step": 14620
+    },
+    {
+      "epoch": 0.3142587103149031,
+      "grad_norm": 0.5474541814548766,
+      "learning_rate": 1.569013732340177e-05,
+      "loss": 0.7772,
+      "step": 14630
+    },
+    {
+      "epoch": 0.31447351462817374,
+      "grad_norm": 0.5168423966046714,
+      "learning_rate": 1.5684530595557174e-05,
+      "loss": 0.7595,
+      "step": 14640
+    },
+    {
+      "epoch": 0.31468831894144433,
+      "grad_norm": 0.5322937512665464,
+      "learning_rate": 1.5678921226407222e-05,
+      "loss": 0.7806,
+      "step": 14650
+    },
+    {
+      "epoch": 0.314903123254715,
+      "grad_norm": 0.5418401693011264,
+      "learning_rate": 1.567330921855829e-05,
+      "loss": 0.7729,
+      "step": 14660
+    },
+    {
+      "epoch": 0.31511792756798557,
+      "grad_norm": 0.5301368247497126,
+      "learning_rate": 1.566769457461799e-05,
+      "loss": 0.7683,
+      "step": 14670
+    },
+    {
+      "epoch": 0.31533273188125616,
+      "grad_norm": 0.5273233080040064,
+      "learning_rate": 1.5662077297195154e-05,
+      "loss": 0.7711,
+      "step": 14680
+    },
+    {
+      "epoch": 0.3155475361945268,
+      "grad_norm": 0.5162157276165048,
+      "learning_rate": 1.565645738889984e-05,
+      "loss": 0.7688,
+      "step": 14690
+    },
+    {
+      "epoch": 0.3157623405077974,
+      "grad_norm": 0.5571046212802432,
+      "learning_rate": 1.5650834852343324e-05,
+      "loss": 0.7767,
+      "step": 14700
+    },
+    {
+      "epoch": 0.31597714482106803,
+      "grad_norm": 0.5174571005771366,
+      "learning_rate": 1.5645209690138107e-05,
+      "loss": 0.7717,
+      "step": 14710
+    },
+    {
+      "epoch": 0.3161919491343386,
+      "grad_norm": 0.5250948275688863,
+      "learning_rate": 1.5639581904897908e-05,
+      "loss": 0.7875,
+      "step": 14720
+    },
+    {
+      "epoch": 0.3164067534476092,
+      "grad_norm": 0.5449756805014006,
+      "learning_rate": 1.5633951499237667e-05,
+      "loss": 0.7702,
+      "step": 14730
+    },
+    {
+      "epoch": 0.31662155776087986,
+      "grad_norm": 0.5249159478890357,
+      "learning_rate": 1.5628318475773538e-05,
+      "loss": 0.7704,
+      "step": 14740
+    },
+    {
+      "epoch": 0.31683636207415045,
+      "grad_norm": 0.5363351915331944,
+      "learning_rate": 1.5622682837122895e-05,
+      "loss": 0.7795,
+      "step": 14750
+    },
+    {
+      "epoch": 0.31705116638742104,
+      "grad_norm": 0.5366684931300841,
+      "learning_rate": 1.5617044585904328e-05,
+      "loss": 0.7788,
+      "step": 14760
+    },
+    {
+      "epoch": 0.3172659707006917,
+      "grad_norm": 0.5176930213448244,
+      "learning_rate": 1.5611403724737635e-05,
+      "loss": 0.7729,
+      "step": 14770
+    },
+    {
+      "epoch": 0.31748077501396227,
+      "grad_norm": 0.53731576710967,
+      "learning_rate": 1.5605760256243834e-05,
+      "loss": 0.7673,
+      "step": 14780
+    },
+    {
+      "epoch": 0.3176955793272329,
+      "grad_norm": 0.5530960087868758,
+      "learning_rate": 1.560011418304515e-05,
+      "loss": 0.7719,
+      "step": 14790
+    },
+    {
+      "epoch": 0.3179103836405035,
+      "grad_norm": 0.5325204498819602,
+      "learning_rate": 1.559446550776502e-05,
+      "loss": 0.7777,
+      "step": 14800
+    },
+    {
+      "epoch": 0.3181251879537741,
+      "grad_norm": 0.5199725220262916,
+      "learning_rate": 1.558881423302808e-05,
+      "loss": 0.7752,
+      "step": 14810
+    },
+    {
+      "epoch": 0.31833999226704474,
+      "grad_norm": 0.5370286678853574,
+      "learning_rate": 1.5583160361460198e-05,
+      "loss": 0.7681,
+      "step": 14820
+    },
+    {
+      "epoch": 0.3185547965803153,
+      "grad_norm": 0.5257072462311427,
+      "learning_rate": 1.5577503895688427e-05,
+      "loss": 0.7619,
+      "step": 14830
+    },
+    {
+      "epoch": 0.31876960089358597,
+      "grad_norm": 0.5343487750052494,
+      "learning_rate": 1.557184483834103e-05,
+      "loss": 0.7703,
+      "step": 14840
+    },
+    {
+      "epoch": 0.31898440520685656,
+      "grad_norm": 0.5251831277823487,
+      "learning_rate": 1.5566183192047476e-05,
+      "loss": 0.7682,
+      "step": 14850
+    },
+    {
+      "epoch": 0.31919920952012715,
+      "grad_norm": 0.5458229759413251,
+      "learning_rate": 1.556051895943844e-05,
+      "loss": 0.7716,
+      "step": 14860
+    },
+    {
+      "epoch": 0.3194140138333978,
+      "grad_norm": 0.5272161429140526,
+      "learning_rate": 1.5554852143145794e-05,
+      "loss": 0.7759,
+      "step": 14870
+    },
+    {
+      "epoch": 0.3196288181466684,
+      "grad_norm": 0.5344254727616319,
+      "learning_rate": 1.5549182745802614e-05,
+      "loss": 0.7777,
+      "step": 14880
+    },
+    {
+      "epoch": 0.31984362245993897,
+      "grad_norm": 0.5548010647048678,
+      "learning_rate": 1.5543510770043177e-05,
+      "loss": 0.769,
+      "step": 14890
+    },
+    {
+      "epoch": 0.3200584267732096,
+      "grad_norm": 0.5412970475079941,
+      "learning_rate": 1.553783621850295e-05,
+      "loss": 0.7703,
+      "step": 14900
+    },
+    {
+      "epoch": 0.3202732310864802,
+      "grad_norm": 0.5351285928852882,
+      "learning_rate": 1.55321590938186e-05,
+      "loss": 0.7715,
+      "step": 14910
+    },
+    {
+      "epoch": 0.32048803539975085,
+      "grad_norm": 0.5223639221501771,
+      "learning_rate": 1.5526479398627997e-05,
+      "loss": 0.767,
+      "step": 14920
+    },
+    {
+      "epoch": 0.32070283971302144,
+      "grad_norm": 0.5335867628748313,
+      "learning_rate": 1.5520797135570192e-05,
+      "loss": 0.7703,
+      "step": 14930
+    },
+    {
+      "epoch": 0.320917644026292,
+      "grad_norm": 0.5441683074016587,
+      "learning_rate": 1.5515112307285446e-05,
+      "loss": 0.7753,
+      "step": 14940
+    },
+    {
+      "epoch": 0.32113244833956267,
+      "grad_norm": 0.5601042025539121,
+      "learning_rate": 1.5509424916415198e-05,
+      "loss": 0.7777,
+      "step": 14950
+    },
+    {
+      "epoch": 0.32134725265283326,
+      "grad_norm": 0.538515402379057,
+      "learning_rate": 1.5503734965602078e-05,
+      "loss": 0.7808,
+      "step": 14960
+    },
+    {
+      "epoch": 0.3215620569661039,
+      "grad_norm": 0.5436748120429375,
+      "learning_rate": 1.5498042457489916e-05,
+      "loss": 0.782,
+      "step": 14970
+    },
+    {
+      "epoch": 0.3217768612793745,
+      "grad_norm": 0.5158949131886207,
+      "learning_rate": 1.5492347394723726e-05,
+      "loss": 0.7631,
+      "step": 14980
+    },
+    {
+      "epoch": 0.3219916655926451,
+      "grad_norm": 0.5328439457730709,
+      "learning_rate": 1.5486649779949695e-05,
+      "loss": 0.7774,
+      "step": 14990
+    },
+    {
+      "epoch": 0.3222064699059157,
+      "grad_norm": 0.5433132241965292,
+      "learning_rate": 1.5480949615815223e-05,
+      "loss": 0.7834,
+      "step": 15000
+    },
+    {
+      "epoch": 0.3224212742191863,
+      "grad_norm": 0.5470052170551871,
+      "learning_rate": 1.547524690496887e-05,
+      "loss": 0.7727,
+      "step": 15010
+    },
+    {
+      "epoch": 0.3226360785324569,
+      "grad_norm": 0.5466443586756153,
+      "learning_rate": 1.5469541650060396e-05,
+      "loss": 0.7803,
+      "step": 15020
+    },
+    {
+      "epoch": 0.32285088284572755,
+      "grad_norm": 0.5351421951971927,
+      "learning_rate": 1.5463833853740723e-05,
+      "loss": 0.7955,
+      "step": 15030
+    },
+    {
+      "epoch": 0.32306568715899814,
+      "grad_norm": 0.5236071642893849,
+      "learning_rate": 1.5458123518661984e-05,
+      "loss": 0.774,
+      "step": 15040
+    },
+    {
+      "epoch": 0.3232804914722688,
+      "grad_norm": 0.5201113478881045,
+      "learning_rate": 1.5452410647477462e-05,
+      "loss": 0.7765,
+      "step": 15050
+    },
+    {
+      "epoch": 0.32349529578553937,
+      "grad_norm": 0.551550727831456,
+      "learning_rate": 1.544669524284163e-05,
+      "loss": 0.7749,
+      "step": 15060
+    },
+    {
+      "epoch": 0.32371010009880996,
+      "grad_norm": 0.5396886090448376,
+      "learning_rate": 1.5440977307410148e-05,
+      "loss": 0.7748,
+      "step": 15070
+    },
+    {
+      "epoch": 0.3239249044120806,
+      "grad_norm": 0.5490265215897929,
+      "learning_rate": 1.5435256843839837e-05,
+      "loss": 0.7782,
+      "step": 15080
+    },
+    {
+      "epoch": 0.3241397087253512,
+      "grad_norm": 0.5680139354556848,
+      "learning_rate": 1.5429533854788698e-05,
+      "loss": 0.769,
+      "step": 15090
+    },
+    {
+      "epoch": 0.32435451303862184,
+      "grad_norm": 0.5724567681997974,
+      "learning_rate": 1.542380834291591e-05,
+      "loss": 0.7743,
+      "step": 15100
+    },
+    {
+      "epoch": 0.32456931735189243,
+      "grad_norm": 0.5551606358878742,
+      "learning_rate": 1.5418080310881816e-05,
+      "loss": 0.7874,
+      "step": 15110
+    },
+    {
+      "epoch": 0.324784121665163,
+      "grad_norm": 0.5273126015719424,
+      "learning_rate": 1.5412349761347937e-05,
+      "loss": 0.7681,
+      "step": 15120
+    },
+    {
+      "epoch": 0.32499892597843366,
+      "grad_norm": 0.5347341788619094,
+      "learning_rate": 1.5406616696976957e-05,
+      "loss": 0.7616,
+      "step": 15130
+    },
+    {
+      "epoch": 0.32521373029170425,
+      "grad_norm": 0.5239352669850086,
+      "learning_rate": 1.540088112043274e-05,
+      "loss": 0.7626,
+      "step": 15140
+    },
+    {
+      "epoch": 0.3254285346049749,
+      "grad_norm": 0.5303037869360442,
+      "learning_rate": 1.53951430343803e-05,
+      "loss": 0.7695,
+      "step": 15150
+    },
+    {
+      "epoch": 0.3256433389182455,
+      "grad_norm": 0.5396447269485543,
+      "learning_rate": 1.5389402441485835e-05,
+      "loss": 0.7534,
+      "step": 15160
+    },
+    {
+      "epoch": 0.3258581432315161,
+      "grad_norm": 0.5288994376628046,
+      "learning_rate": 1.5383659344416696e-05,
+      "loss": 0.7661,
+      "step": 15170
+    },
+    {
+      "epoch": 0.3260729475447867,
+      "grad_norm": 0.5366154432296876,
+      "learning_rate": 1.5377913745841404e-05,
+      "loss": 0.7796,
+      "step": 15180
+    },
+    {
+      "epoch": 0.3262877518580573,
+      "grad_norm": 0.5221692922283262,
+      "learning_rate": 1.5372165648429633e-05,
+      "loss": 0.7726,
+      "step": 15190
+    },
+    {
+      "epoch": 0.3265025561713279,
+      "grad_norm": 0.5422324565519508,
+      "learning_rate": 1.536641505485223e-05,
+      "loss": 0.7633,
+      "step": 15200
+    },
+    {
+      "epoch": 0.32671736048459854,
+      "grad_norm": 0.5408071342677044,
+      "learning_rate": 1.5360661967781194e-05,
+      "loss": 0.7689,
+      "step": 15210
+    },
+    {
+      "epoch": 0.32693216479786913,
+      "grad_norm": 0.5454745974106339,
+      "learning_rate": 1.5354906389889686e-05,
+      "loss": 0.7741,
+      "step": 15220
+    },
+    {
+      "epoch": 0.3271469691111398,
+      "grad_norm": 0.5218591885366504,
+      "learning_rate": 1.534914832385203e-05,
+      "loss": 0.7909,
+      "step": 15230
+    },
+    {
+      "epoch": 0.32736177342441036,
+      "grad_norm": 0.5519162115511386,
+      "learning_rate": 1.5343387772343687e-05,
+      "loss": 0.7809,
+      "step": 15240
+    },
+    {
+      "epoch": 0.32757657773768095,
+      "grad_norm": 0.5327394461244905,
+      "learning_rate": 1.533762473804129e-05,
+      "loss": 0.773,
+      "step": 15250
+    },
+    {
+      "epoch": 0.3277913820509516,
+      "grad_norm": 0.5584463854444557,
+      "learning_rate": 1.5331859223622623e-05,
+      "loss": 0.7735,
+      "step": 15260
+    },
+    {
+      "epoch": 0.3280061863642222,
+      "grad_norm": 0.5390984045147804,
+      "learning_rate": 1.532609123176662e-05,
+      "loss": 0.7757,
+      "step": 15270
+    },
+    {
+      "epoch": 0.32822099067749283,
+      "grad_norm": 0.543837161370247,
+      "learning_rate": 1.5320320765153367e-05,
+      "loss": 0.762,
+      "step": 15280
+    },
+    {
+      "epoch": 0.3284357949907634,
+      "grad_norm": 0.5352816574606879,
+      "learning_rate": 1.53145478264641e-05,
+      "loss": 0.7693,
+      "step": 15290
+    },
+    {
+      "epoch": 0.328650599304034,
+      "grad_norm": 0.5605556044481591,
+      "learning_rate": 1.5308772418381196e-05,
+      "loss": 0.7623,
+      "step": 15300
+    },
+    {
+      "epoch": 0.32886540361730465,
+      "grad_norm": 0.5446921802277784,
+      "learning_rate": 1.5302994543588194e-05,
+      "loss": 0.774,
+      "step": 15310
+    },
+    {
+      "epoch": 0.32908020793057524,
+      "grad_norm": 0.5151829148535916,
+      "learning_rate": 1.529721420476977e-05,
+      "loss": 0.7826,
+      "step": 15320
+    },
+    {
+      "epoch": 0.32929501224384583,
+      "grad_norm": 0.5223909324949548,
+      "learning_rate": 1.5291431404611744e-05,
+      "loss": 0.7593,
+      "step": 15330
+    },
+    {
+      "epoch": 0.3295098165571165,
+      "grad_norm": 0.5267824527110152,
+      "learning_rate": 1.5285646145801086e-05,
+      "loss": 0.7665,
+      "step": 15340
+    },
+    {
+      "epoch": 0.32972462087038706,
+      "grad_norm": 0.5466419782290777,
+      "learning_rate": 1.5279858431025903e-05,
+      "loss": 0.7704,
+      "step": 15350
+    },
+    {
+      "epoch": 0.3299394251836577,
+      "grad_norm": 0.5557083967709183,
+      "learning_rate": 1.527406826297544e-05,
+      "loss": 0.7661,
+      "step": 15360
+    },
+    {
+      "epoch": 0.3301542294969283,
+      "grad_norm": 0.5337371028909507,
+      "learning_rate": 1.526827564434009e-05,
+      "loss": 0.7604,
+      "step": 15370
+    },
+    {
+      "epoch": 0.3303690338101989,
+      "grad_norm": 0.5703493813264273,
+      "learning_rate": 1.5262480577811386e-05,
+      "loss": 0.7672,
+      "step": 15380
+    },
+    {
+      "epoch": 0.33058383812346953,
+      "grad_norm": 0.5472412709147122,
+      "learning_rate": 1.5256683066081986e-05,
+      "loss": 0.7709,
+      "step": 15390
+    },
+    {
+      "epoch": 0.3307986424367401,
+      "grad_norm": 0.5457023754017395,
+      "learning_rate": 1.5250883111845697e-05,
+      "loss": 0.787,
+      "step": 15400
+    },
+    {
+      "epoch": 0.33101344675001076,
+      "grad_norm": 0.5416955760823877,
+      "learning_rate": 1.5245080717797454e-05,
+      "loss": 0.759,
+      "step": 15410
+    },
+    {
+      "epoch": 0.33122825106328135,
+      "grad_norm": 0.5138493444141494,
+      "learning_rate": 1.5239275886633326e-05,
+      "loss": 0.7735,
+      "step": 15420
+    },
+    {
+      "epoch": 0.33144305537655194,
+      "grad_norm": 0.5211626703386022,
+      "learning_rate": 1.5233468621050518e-05,
+      "loss": 0.7784,
+      "step": 15430
+    },
+    {
+      "epoch": 0.3316578596898226,
+      "grad_norm": 0.5275918206293356,
+      "learning_rate": 1.5227658923747364e-05,
+      "loss": 0.7833,
+      "step": 15440
+    },
+    {
+      "epoch": 0.3318726640030932,
+      "grad_norm": 0.6206866088264538,
+      "learning_rate": 1.5221846797423325e-05,
+      "loss": 0.7708,
+      "step": 15450
+    },
+    {
+      "epoch": 0.33208746831636377,
+      "grad_norm": 0.5261380809398264,
+      "learning_rate": 1.5216032244778993e-05,
+      "loss": 0.7782,
+      "step": 15460
+    },
+    {
+      "epoch": 0.3323022726296344,
+      "grad_norm": 0.541782059211599,
+      "learning_rate": 1.5210215268516093e-05,
+      "loss": 0.7608,
+      "step": 15470
+    },
+    {
+      "epoch": 0.332517076942905,
+      "grad_norm": 0.5240473732673129,
+      "learning_rate": 1.520439587133747e-05,
+      "loss": 0.7726,
+      "step": 15480
+    },
+    {
+      "epoch": 0.33273188125617564,
+      "grad_norm": 0.5191076944218657,
+      "learning_rate": 1.5198574055947091e-05,
+      "loss": 0.7642,
+      "step": 15490
+    },
+    {
+      "epoch": 0.33294668556944623,
+      "grad_norm": 0.539873653252131,
+      "learning_rate": 1.5192749825050052e-05,
+      "loss": 0.7636,
+      "step": 15500
+    },
+    {
+      "epoch": 0.3331614898827168,
+      "grad_norm": 0.531660226924968,
+      "learning_rate": 1.518692318135257e-05,
+      "loss": 0.7796,
+      "step": 15510
+    },
+    {
+      "epoch": 0.33337629419598747,
+      "grad_norm": 0.5173762361506008,
+      "learning_rate": 1.5181094127561982e-05,
+      "loss": 0.7615,
+      "step": 15520
+    },
+    {
+      "epoch": 0.33359109850925805,
+      "grad_norm": 0.5114268089145959,
+      "learning_rate": 1.5175262666386745e-05,
+      "loss": 0.7559,
+      "step": 15530
+    },
+    {
+      "epoch": 0.3338059028225287,
+      "grad_norm": 0.5271733156368176,
+      "learning_rate": 1.5169428800536441e-05,
+      "loss": 0.7711,
+      "step": 15540
+    },
+    {
+      "epoch": 0.3340207071357993,
+      "grad_norm": 0.5135401558436904,
+      "learning_rate": 1.5163592532721763e-05,
+      "loss": 0.7776,
+      "step": 15550
+    },
+    {
+      "epoch": 0.3342355114490699,
+      "grad_norm": 0.53726076093529,
+      "learning_rate": 1.515775386565451e-05,
+      "loss": 0.7594,
+      "step": 15560
+    },
+    {
+      "epoch": 0.3344503157623405,
+      "grad_norm": 0.5184174928971732,
+      "learning_rate": 1.515191280204762e-05,
+      "loss": 0.7774,
+      "step": 15570
+    },
+    {
+      "epoch": 0.3346651200756111,
+      "grad_norm": 0.5225627226289906,
+      "learning_rate": 1.514606934461512e-05,
+      "loss": 0.7653,
+      "step": 15580
+    },
+    {
+      "epoch": 0.33487992438888176,
+      "grad_norm": 0.521185908731502,
+      "learning_rate": 1.5140223496072168e-05,
+      "loss": 0.7702,
+      "step": 15590
+    },
+    {
+      "epoch": 0.33509472870215234,
+      "grad_norm": 0.5605456313685063,
+      "learning_rate": 1.5134375259135024e-05,
+      "loss": 0.7725,
+      "step": 15600
+    },
+    {
+      "epoch": 0.33530953301542293,
+      "grad_norm": 0.5504337164442501,
+      "learning_rate": 1.5128524636521058e-05,
+      "loss": 0.7502,
+      "step": 15610
+    },
+    {
+      "epoch": 0.3355243373286936,
+      "grad_norm": 0.5304219453129593,
+      "learning_rate": 1.5122671630948748e-05,
+      "loss": 0.7731,
+      "step": 15620
+    },
+    {
+      "epoch": 0.33573914164196417,
+      "grad_norm": 0.5405431462257487,
+      "learning_rate": 1.5116816245137684e-05,
+      "loss": 0.7787,
+      "step": 15630
+    },
+    {
+      "epoch": 0.33595394595523476,
+      "grad_norm": 0.540626596116053,
+      "learning_rate": 1.5110958481808558e-05,
+      "loss": 0.7707,
+      "step": 15640
+    },
+    {
+      "epoch": 0.3361687502685054,
+      "grad_norm": 0.5107167602298724,
+      "learning_rate": 1.5105098343683166e-05,
+      "loss": 0.7853,
+      "step": 15650
+    },
+    {
+      "epoch": 0.336383554581776,
+      "grad_norm": 0.5466449645523599,
+      "learning_rate": 1.5099235833484411e-05,
+      "loss": 0.7756,
+      "step": 15660
+    },
+    {
+      "epoch": 0.33659835889504663,
+      "grad_norm": 0.5123157159149067,
+      "learning_rate": 1.5093370953936298e-05,
+      "loss": 0.7635,
+      "step": 15670
+    },
+    {
+      "epoch": 0.3368131632083172,
+      "grad_norm": 0.5561787006388946,
+      "learning_rate": 1.5087503707763925e-05,
+      "loss": 0.7777,
+      "step": 15680
+    },
+    {
+      "epoch": 0.3370279675215878,
+      "grad_norm": 0.5423194514303097,
+      "learning_rate": 1.5081634097693498e-05,
+      "loss": 0.7709,
+      "step": 15690
+    },
+    {
+      "epoch": 0.33724277183485846,
+      "grad_norm": 0.5131134543368485,
+      "learning_rate": 1.5075762126452324e-05,
+      "loss": 0.7536,
+      "step": 15700
+    },
+    {
+      "epoch": 0.33745757614812905,
+      "grad_norm": 0.5263701017681474,
+      "learning_rate": 1.5069887796768798e-05,
+      "loss": 0.7646,
+      "step": 15710
+    },
+    {
+      "epoch": 0.3376723804613997,
+      "grad_norm": 0.5335909937404967,
+      "learning_rate": 1.5064011111372417e-05,
+      "loss": 0.774,
+      "step": 15720
+    },
+    {
+      "epoch": 0.3378871847746703,
+      "grad_norm": 0.5205555068375117,
+      "learning_rate": 1.5058132072993767e-05,
+      "loss": 0.7729,
+      "step": 15730
+    },
+    {
+      "epoch": 0.33810198908794087,
+      "grad_norm": 0.529162741413939,
+      "learning_rate": 1.5052250684364535e-05,
+      "loss": 0.7823,
+      "step": 15740
+    },
+    {
+      "epoch": 0.3383167934012115,
+      "grad_norm": 0.5335725684320926,
+      "learning_rate": 1.5046366948217495e-05,
+      "loss": 0.7881,
+      "step": 15750
+    },
+    {
+      "epoch": 0.3385315977144821,
+      "grad_norm": 0.5422608148576844,
+      "learning_rate": 1.5040480867286511e-05,
+      "loss": 0.7723,
+      "step": 15760
+    },
+    {
+      "epoch": 0.3387464020277527,
+      "grad_norm": 0.525108601464528,
+      "learning_rate": 1.5034592444306541e-05,
+      "loss": 0.7661,
+      "step": 15770
+    },
+    {
+      "epoch": 0.33896120634102334,
+      "grad_norm": 0.5212632664656071,
+      "learning_rate": 1.5028701682013626e-05,
+      "loss": 0.7722,
+      "step": 15780
+    },
+    {
+      "epoch": 0.3391760106542939,
+      "grad_norm": 0.523147333913358,
+      "learning_rate": 1.50228085831449e-05,
+      "loss": 0.766,
+      "step": 15790
+    },
+    {
+      "epoch": 0.33939081496756457,
+      "grad_norm": 0.5155900387298158,
+      "learning_rate": 1.5016913150438575e-05,
+      "loss": 0.7544,
+      "step": 15800
+    },
+    {
+      "epoch": 0.33960561928083516,
+      "grad_norm": 2.57094670571956,
+      "learning_rate": 1.5011015386633954e-05,
+      "loss": 0.7738,
+      "step": 15810
+    },
+    {
+      "epoch": 0.33982042359410575,
+      "grad_norm": 0.5163335640756341,
+      "learning_rate": 1.5005115294471422e-05,
+      "loss": 0.755,
+      "step": 15820
+    },
+    {
+      "epoch": 0.3400352279073764,
+      "grad_norm": 0.5597200416389281,
+      "learning_rate": 1.499921287669244e-05,
+      "loss": 0.7905,
+      "step": 15830
+    },
+    {
+      "epoch": 0.340250032220647,
+      "grad_norm": 0.5137889894949109,
+      "learning_rate": 1.4993308136039557e-05,
+      "loss": 0.7632,
+      "step": 15840
+    },
+    {
+      "epoch": 0.3404648365339176,
+      "grad_norm": 0.5245555303836951,
+      "learning_rate": 1.4987401075256398e-05,
+      "loss": 0.7573,
+      "step": 15850
+    },
+    {
+      "epoch": 0.3406796408471882,
+      "grad_norm": 0.5466373460073314,
+      "learning_rate": 1.498149169708767e-05,
+      "loss": 0.7738,
+      "step": 15860
+    },
+    {
+      "epoch": 0.3408944451604588,
+      "grad_norm": 0.5229866422016913,
+      "learning_rate": 1.4975580004279149e-05,
+      "loss": 0.7583,
+      "step": 15870
+    },
+    {
+      "epoch": 0.34110924947372945,
+      "grad_norm": 0.5114015357336622,
+      "learning_rate": 1.4969665999577693e-05,
+      "loss": 0.7522,
+      "step": 15880
+    },
+    {
+      "epoch": 0.34132405378700004,
+      "grad_norm": 0.5125004960033485,
+      "learning_rate": 1.4963749685731231e-05,
+      "loss": 0.7739,
+      "step": 15890
+    },
+    {
+      "epoch": 0.3415388581002706,
+      "grad_norm": 0.5365066975799684,
+      "learning_rate": 1.4957831065488763e-05,
+      "loss": 0.7767,
+      "step": 15900
+    },
+    {
+      "epoch": 0.34175366241354127,
+      "grad_norm": 0.5224206864785478,
+      "learning_rate": 1.495191014160037e-05,
+      "loss": 0.7736,
+      "step": 15910
+    },
+    {
+      "epoch": 0.34196846672681186,
+      "grad_norm": 0.5138735687941413,
+      "learning_rate": 1.4945986916817194e-05,
+      "loss": 0.7558,
+      "step": 15920
+    },
+    {
+      "epoch": 0.3421832710400825,
+      "grad_norm": 0.5156382777636157,
+      "learning_rate": 1.494006139389145e-05,
+      "loss": 0.755,
+      "step": 15930
+    },
+    {
+      "epoch": 0.3423980753533531,
+      "grad_norm": 0.5173997349464476,
+      "learning_rate": 1.4934133575576418e-05,
+      "loss": 0.7644,
+      "step": 15940
+    },
+    {
+      "epoch": 0.3426128796666237,
+      "grad_norm": 0.5773347686325697,
+      "learning_rate": 1.4928203464626446e-05,
+      "loss": 0.7803,
+      "step": 15950
+    },
+    {
+      "epoch": 0.3428276839798943,
+      "grad_norm": 0.5238727000577228,
+      "learning_rate": 1.4922271063796946e-05,
+      "loss": 0.7688,
+      "step": 15960
+    },
+    {
+      "epoch": 0.3430424882931649,
+      "grad_norm": 0.49544133793048006,
+      "learning_rate": 1.4916336375844402e-05,
+      "loss": 0.765,
+      "step": 15970
+    },
+    {
+      "epoch": 0.34325729260643556,
+      "grad_norm": 0.5559128491079445,
+      "learning_rate": 1.4910399403526355e-05,
+      "loss": 0.7643,
+      "step": 15980
+    },
+    {
+      "epoch": 0.34347209691970615,
+      "grad_norm": 0.5224474824510437,
+      "learning_rate": 1.4904460149601399e-05,
+      "loss": 0.7605,
+      "step": 15990
+    },
+    {
+      "epoch": 0.34368690123297674,
+      "grad_norm": 0.5084160847363332,
+      "learning_rate": 1.48985186168292e-05,
+      "loss": 0.7525,
+      "step": 16000
+    },
+    {
+      "epoch": 0.3439017055462474,
+      "grad_norm": 0.5178973898547171,
+      "learning_rate": 1.489257480797048e-05,
+      "loss": 0.7623,
+      "step": 16010
+    },
+    {
+      "epoch": 0.34411650985951797,
+      "grad_norm": 0.5480129624627523,
+      "learning_rate": 1.4886628725787017e-05,
+      "loss": 0.7831,
+      "step": 16020
+    },
+    {
+      "epoch": 0.3443313141727886,
+      "grad_norm": 0.524153874187991,
+      "learning_rate": 1.4880680373041646e-05,
+      "loss": 0.7659,
+      "step": 16030
+    },
+    {
+      "epoch": 0.3445461184860592,
+      "grad_norm": 0.5488165779868663,
+      "learning_rate": 1.4874729752498256e-05,
+      "loss": 0.7508,
+      "step": 16040
+    },
+    {
+      "epoch": 0.3447609227993298,
+      "grad_norm": 0.6146964896934437,
+      "learning_rate": 1.4868776866921792e-05,
+      "loss": 0.7708,
+      "step": 16050
+    },
+    {
+      "epoch": 0.34497572711260044,
+      "grad_norm": 0.5367282750533525,
+      "learning_rate": 1.4862821719078246e-05,
+      "loss": 0.7647,
+      "step": 16060
+    },
+    {
+      "epoch": 0.34519053142587103,
+      "grad_norm": 0.5370082911271288,
+      "learning_rate": 1.4856864311734667e-05,
+      "loss": 0.7716,
+      "step": 16070
+    },
+    {
+      "epoch": 0.3454053357391416,
+      "grad_norm": 0.5210318070953477,
+      "learning_rate": 1.4850904647659155e-05,
+      "loss": 0.7562,
+      "step": 16080
+    },
+    {
+      "epoch": 0.34562014005241226,
+      "grad_norm": 0.5426172156914827,
+      "learning_rate": 1.4844942729620853e-05,
+      "loss": 0.7517,
+      "step": 16090
+    },
+    {
+      "epoch": 0.34583494436568285,
+      "grad_norm": 0.5353756178515557,
+      "learning_rate": 1.4838978560389952e-05,
+      "loss": 0.777,
+      "step": 16100
+    },
+    {
+      "epoch": 0.3460497486789535,
+      "grad_norm": 0.5458690566171005,
+      "learning_rate": 1.4833012142737696e-05,
+      "loss": 0.7596,
+      "step": 16110
+    },
+    {
+      "epoch": 0.3462645529922241,
+      "grad_norm": 0.5222605704575792,
+      "learning_rate": 1.4827043479436362e-05,
+      "loss": 0.7641,
+      "step": 16120
+    },
+    {
+      "epoch": 0.3464793573054947,
+      "grad_norm": 0.517515304030216,
+      "learning_rate": 1.4821072573259283e-05,
+      "loss": 0.7735,
+      "step": 16130
+    },
+    {
+      "epoch": 0.3466941616187653,
+      "grad_norm": 0.5619458732612833,
+      "learning_rate": 1.4815099426980825e-05,
+      "loss": 0.7646,
+      "step": 16140
+    },
+    {
+      "epoch": 0.3469089659320359,
+      "grad_norm": 0.5308436572477332,
+      "learning_rate": 1.4809124043376399e-05,
+      "loss": 0.7767,
+      "step": 16150
+    },
+    {
+      "epoch": 0.34712377024530655,
+      "grad_norm": 0.5135551609727562,
+      "learning_rate": 1.4803146425222457e-05,
+      "loss": 0.7611,
+      "step": 16160
+    },
+    {
+      "epoch": 0.34733857455857714,
+      "grad_norm": 0.5330663686858607,
+      "learning_rate": 1.479716657529648e-05,
+      "loss": 0.7702,
+      "step": 16170
+    },
+    {
+      "epoch": 0.34755337887184773,
+      "grad_norm": 0.5359601127061994,
+      "learning_rate": 1.4791184496377e-05,
+      "loss": 0.776,
+      "step": 16180
+    },
+    {
+      "epoch": 0.3477681831851184,
+      "grad_norm": 0.5140711185809557,
+      "learning_rate": 1.4785200191243574e-05,
+      "loss": 0.7595,
+      "step": 16190
+    },
+    {
+      "epoch": 0.34798298749838896,
+      "grad_norm": 0.5406658844636193,
+      "learning_rate": 1.4779213662676797e-05,
+      "loss": 0.7729,
+      "step": 16200
+    },
+    {
+      "epoch": 0.34819779181165955,
+      "grad_norm": 0.5104834415117768,
+      "learning_rate": 1.4773224913458298e-05,
+      "loss": 0.7811,
+      "step": 16210
+    },
+    {
+      "epoch": 0.3484125961249302,
+      "grad_norm": 0.5595639311867889,
+      "learning_rate": 1.4767233946370735e-05,
+      "loss": 0.7691,
+      "step": 16220
+    },
+    {
+      "epoch": 0.3486274004382008,
+      "grad_norm": 0.5453224484229146,
+      "learning_rate": 1.4761240764197804e-05,
+      "loss": 0.7711,
+      "step": 16230
+    },
+    {
+      "epoch": 0.34884220475147143,
+      "grad_norm": 0.5533371724805077,
+      "learning_rate": 1.4755245369724219e-05,
+      "loss": 0.7647,
+      "step": 16240
+    },
+    {
+      "epoch": 0.349057009064742,
+      "grad_norm": 0.5029137026412714,
+      "learning_rate": 1.4749247765735727e-05,
+      "loss": 0.7625,
+      "step": 16250
+    },
+    {
+      "epoch": 0.3492718133780126,
+      "grad_norm": 0.5407424760271811,
+      "learning_rate": 1.474324795501911e-05,
+      "loss": 0.7777,
+      "step": 16260
+    },
+    {
+      "epoch": 0.34948661769128325,
+      "grad_norm": 0.5360642655145464,
+      "learning_rate": 1.4737245940362158e-05,
+      "loss": 0.7644,
+      "step": 16270
+    },
+    {
+      "epoch": 0.34970142200455384,
+      "grad_norm": 0.5431989168889423,
+      "learning_rate": 1.47312417245537e-05,
+      "loss": 0.7725,
+      "step": 16280
+    },
+    {
+      "epoch": 0.3499162263178245,
+      "grad_norm": 0.5212803701855959,
+      "learning_rate": 1.472523531038358e-05,
+      "loss": 0.7503,
+      "step": 16290
+    },
+    {
+      "epoch": 0.3501310306310951,
+      "grad_norm": 0.5375323551611733,
+      "learning_rate": 1.471922670064267e-05,
+      "loss": 0.7599,
+      "step": 16300
+    },
+    {
+      "epoch": 0.35034583494436566,
+      "grad_norm": 0.5239870812277655,
+      "learning_rate": 1.4713215898122857e-05,
+      "loss": 0.7863,
+      "step": 16310
+    },
+    {
+      "epoch": 0.3505606392576363,
+      "grad_norm": 0.5180937115766086,
+      "learning_rate": 1.4707202905617042e-05,
+      "loss": 0.7689,
+      "step": 16320
+    },
+    {
+      "epoch": 0.3507754435709069,
+      "grad_norm": 0.5417292224014176,
+      "learning_rate": 1.4701187725919157e-05,
+      "loss": 0.7527,
+      "step": 16330
+    },
+    {
+      "epoch": 0.35099024788417754,
+      "grad_norm": 0.5357164040418824,
+      "learning_rate": 1.4695170361824136e-05,
+      "loss": 0.7642,
+      "step": 16340
+    },
+    {
+      "epoch": 0.35120505219744813,
+      "grad_norm": 0.5303332337613179,
+      "learning_rate": 1.468915081612794e-05,
+      "loss": 0.7566,
+      "step": 16350
+    },
+    {
+      "epoch": 0.3514198565107187,
+      "grad_norm": 0.5382188260863363,
+      "learning_rate": 1.468312909162754e-05,
+      "loss": 0.7672,
+      "step": 16360
+    },
+    {
+      "epoch": 0.35163466082398936,
+      "grad_norm": 0.5385128353648729,
+      "learning_rate": 1.467710519112091e-05,
+      "loss": 0.7809,
+      "step": 16370
+    },
+    {
+      "epoch": 0.35184946513725995,
+      "grad_norm": 0.5310058917606305,
+      "learning_rate": 1.467107911740705e-05,
+      "loss": 0.7654,
+      "step": 16380
+    },
+    {
+      "epoch": 0.35206426945053054,
+      "grad_norm": 0.5031221816084535,
+      "learning_rate": 1.4665050873285957e-05,
+      "loss": 0.7507,
+      "step": 16390
+    },
+    {
+      "epoch": 0.3522790737638012,
+      "grad_norm": 0.54140785390428,
+      "learning_rate": 1.4659020461558649e-05,
+      "loss": 0.7545,
+      "step": 16400
+    },
+    {
+      "epoch": 0.3524938780770718,
+      "grad_norm": 0.5370610309213006,
+      "learning_rate": 1.4652987885027142e-05,
+      "loss": 0.7658,
+      "step": 16410
+    },
+    {
+      "epoch": 0.3527086823903424,
+      "grad_norm": 0.5490842521616487,
+      "learning_rate": 1.4646953146494454e-05,
+      "loss": 0.7606,
+      "step": 16420
+    },
+    {
+      "epoch": 0.352923486703613,
+      "grad_norm": 0.5131178943599031,
+      "learning_rate": 1.4640916248764621e-05,
+      "loss": 0.7738,
+      "step": 16430
+    },
+    {
+      "epoch": 0.3531382910168836,
+      "grad_norm": 0.5328970310455506,
+      "learning_rate": 1.4634877194642672e-05,
+      "loss": 0.7774,
+      "step": 16440
+    },
+    {
+      "epoch": 0.35335309533015424,
+      "grad_norm": 0.5137803015241591,
+      "learning_rate": 1.462883598693464e-05,
+      "loss": 0.7728,
+      "step": 16450
+    },
+    {
+      "epoch": 0.35356789964342483,
+      "grad_norm": 0.5199243192373794,
+      "learning_rate": 1.4622792628447562e-05,
+      "loss": 0.77,
+      "step": 16460
+    },
+    {
+      "epoch": 0.3537827039566955,
+      "grad_norm": 0.5425545285150617,
+      "learning_rate": 1.4616747121989474e-05,
+      "loss": 0.777,
+      "step": 16470
+    },
+    {
+      "epoch": 0.35399750826996607,
+      "grad_norm": 0.5395199010657613,
+      "learning_rate": 1.4610699470369401e-05,
+      "loss": 0.7688,
+      "step": 16480
+    },
+    {
+      "epoch": 0.35421231258323665,
+      "grad_norm": 0.5475850285247497,
+      "learning_rate": 1.4604649676397377e-05,
+      "loss": 0.769,
+      "step": 16490
+    },
+    {
+      "epoch": 0.3544271168965073,
+      "grad_norm": 0.5170429327520523,
+      "learning_rate": 1.4598597742884429e-05,
+      "loss": 0.7693,
+      "step": 16500
+    },
+    {
+      "epoch": 0.3546419212097779,
+      "grad_norm": 0.5402757575811128,
+      "learning_rate": 1.4592543672642567e-05,
+      "loss": 0.7663,
+      "step": 16510
+    },
+    {
+      "epoch": 0.3548567255230485,
+      "grad_norm": 0.536139194934942,
+      "learning_rate": 1.4586487468484809e-05,
+      "loss": 0.777,
+      "step": 16520
+    },
+    {
+      "epoch": 0.3550715298363191,
+      "grad_norm": 0.5276298531695289,
+      "learning_rate": 1.4580429133225153e-05,
+      "loss": 0.7669,
+      "step": 16530
+    },
+    {
+      "epoch": 0.3552863341495897,
+      "grad_norm": 0.5245685397779214,
+      "learning_rate": 1.4574368669678598e-05,
+      "loss": 0.7905,
+      "step": 16540
+    },
+    {
+      "epoch": 0.35550113846286036,
+      "grad_norm": 0.5281075002913366,
+      "learning_rate": 1.4568306080661118e-05,
+      "loss": 0.778,
+      "step": 16550
+    },
+    {
+      "epoch": 0.35571594277613094,
+      "grad_norm": 0.5113618186605529,
+      "learning_rate": 1.4562241368989691e-05,
+      "loss": 0.7669,
+      "step": 16560
+    },
+    {
+      "epoch": 0.35593074708940153,
+      "grad_norm": 0.506708948383745,
+      "learning_rate": 1.4556174537482267e-05,
+      "loss": 0.7476,
+      "step": 16570
+    },
+    {
+      "epoch": 0.3561455514026722,
+      "grad_norm": 0.5508043321880731,
+      "learning_rate": 1.4550105588957789e-05,
+      "loss": 0.7609,
+      "step": 16580
+    },
+    {
+      "epoch": 0.35636035571594277,
+      "grad_norm": 1.4314950449809294,
+      "learning_rate": 1.4544034526236174e-05,
+      "loss": 0.7623,
+      "step": 16590
+    },
+    {
+      "epoch": 0.3565751600292134,
+      "grad_norm": 0.5244372331321259,
+      "learning_rate": 1.453796135213834e-05,
+      "loss": 0.7669,
+      "step": 16600
+    },
+    {
+      "epoch": 0.356789964342484,
+      "grad_norm": 0.5224396045025858,
+      "learning_rate": 1.4531886069486169e-05,
+      "loss": 0.7605,
+      "step": 16610
+    },
+    {
+      "epoch": 0.3570047686557546,
+      "grad_norm": 0.5070138150623654,
+      "learning_rate": 1.4525808681102533e-05,
+      "loss": 0.7668,
+      "step": 16620
+    },
+    {
+      "epoch": 0.35721957296902523,
+      "grad_norm": 0.5288742137487699,
+      "learning_rate": 1.4519729189811271e-05,
+      "loss": 0.7554,
+      "step": 16630
+    },
+    {
+      "epoch": 0.3574343772822958,
+      "grad_norm": 0.5677338154821888,
+      "learning_rate": 1.4513647598437208e-05,
+      "loss": 0.7609,
+      "step": 16640
+    },
+    {
+      "epoch": 0.3576491815955664,
+      "grad_norm": 0.5357651788636495,
+      "learning_rate": 1.4507563909806146e-05,
+      "loss": 0.7742,
+      "step": 16650
+    },
+    {
+      "epoch": 0.35786398590883706,
+      "grad_norm": 0.5179083963651603,
+      "learning_rate": 1.4501478126744855e-05,
+      "loss": 0.7726,
+      "step": 16660
+    },
+    {
+      "epoch": 0.35807879022210765,
+      "grad_norm": 0.5295056672371503,
+      "learning_rate": 1.4495390252081082e-05,
+      "loss": 0.7458,
+      "step": 16670
+    },
+    {
+      "epoch": 0.3582935945353783,
+      "grad_norm": 0.5214891906282768,
+      "learning_rate": 1.448930028864355e-05,
+      "loss": 0.7619,
+      "step": 16680
+    },
+    {
+      "epoch": 0.3585083988486489,
+      "grad_norm": 0.5578244478041603,
+      "learning_rate": 1.4483208239261943e-05,
+      "loss": 0.7703,
+      "step": 16690
+    },
+    {
+      "epoch": 0.35872320316191947,
+      "grad_norm": 0.5376124885496097,
+      "learning_rate": 1.4477114106766921e-05,
+      "loss": 0.7546,
+      "step": 16700
+    },
+    {
+      "epoch": 0.3589380074751901,
+      "grad_norm": 0.5177377132348334,
+      "learning_rate": 1.4471017893990107e-05,
+      "loss": 0.7662,
+      "step": 16710
+    },
+    {
+      "epoch": 0.3591528117884607,
+      "grad_norm": 0.5134062584490465,
+      "learning_rate": 1.4464919603764097e-05,
+      "loss": 0.7667,
+      "step": 16720
+    },
+    {
+      "epoch": 0.35936761610173135,
+      "grad_norm": 0.5054504951310723,
+      "learning_rate": 1.4458819238922446e-05,
+      "loss": 0.7467,
+      "step": 16730
+    },
+    {
+      "epoch": 0.35958242041500194,
+      "grad_norm": 0.5344939536393053,
+      "learning_rate": 1.445271680229968e-05,
+      "loss": 0.7705,
+      "step": 16740
+    },
+    {
+      "epoch": 0.3597972247282725,
+      "grad_norm": 0.5470998293542991,
+      "learning_rate": 1.4446612296731282e-05,
+      "loss": 0.7564,
+      "step": 16750
+    },
+    {
+      "epoch": 0.36001202904154317,
+      "grad_norm": 0.5743146214103152,
+      "learning_rate": 1.4440505725053693e-05,
+      "loss": 0.7623,
+      "step": 16760
+    },
+    {
+      "epoch": 0.36022683335481376,
+      "grad_norm": 0.5333319615482879,
+      "learning_rate": 1.4434397090104324e-05,
+      "loss": 0.7714,
+      "step": 16770
+    },
+    {
+      "epoch": 0.3604416376680844,
+      "grad_norm": 0.5202228507285751,
+      "learning_rate": 1.442828639472154e-05,
+      "loss": 0.7515,
+      "step": 16780
+    },
+    {
+      "epoch": 0.360656441981355,
+      "grad_norm": 0.5277143639598636,
+      "learning_rate": 1.4422173641744663e-05,
+      "loss": 0.7633,
+      "step": 16790
+    },
+    {
+      "epoch": 0.3608712462946256,
+      "grad_norm": 0.5468528449571223,
+      "learning_rate": 1.4416058834013967e-05,
+      "loss": 0.7629,
+      "step": 16800
+    },
+    {
+      "epoch": 0.3610860506078962,
+      "grad_norm": 0.5389600692166813,
+      "learning_rate": 1.440994197437069e-05,
+      "loss": 0.7681,
+      "step": 16810
+    },
+    {
+      "epoch": 0.3613008549211668,
+      "grad_norm": 0.5335455178234667,
+      "learning_rate": 1.4403823065657012e-05,
+      "loss": 0.7607,
+      "step": 16820
+    },
+    {
+      "epoch": 0.3615156592344374,
+      "grad_norm": 0.5400771216289987,
+      "learning_rate": 1.4397702110716076e-05,
+      "loss": 0.7659,
+      "step": 16830
+    },
+    {
+      "epoch": 0.36173046354770805,
+      "grad_norm": 0.5602481821229619,
+      "learning_rate": 1.4391579112391969e-05,
+      "loss": 0.7765,
+      "step": 16840
+    },
+    {
+      "epoch": 0.36194526786097864,
+      "grad_norm": 0.5272902803012298,
+      "learning_rate": 1.4385454073529731e-05,
+      "loss": 0.7649,
+      "step": 16850
+    },
+    {
+      "epoch": 0.3621600721742493,
+      "grad_norm": 0.5187058176818714,
+      "learning_rate": 1.4379326996975347e-05,
+      "loss": 0.7648,
+      "step": 16860
+    },
+    {
+      "epoch": 0.36237487648751987,
+      "grad_norm": 0.5331452093710046,
+      "learning_rate": 1.4373197885575752e-05,
+      "loss": 0.7713,
+      "step": 16870
+    },
+    {
+      "epoch": 0.36258968080079046,
+      "grad_norm": 0.5302443268053321,
+      "learning_rate": 1.4367066742178824e-05,
+      "loss": 0.7568,
+      "step": 16880
+    },
+    {
+      "epoch": 0.3628044851140611,
+      "grad_norm": 0.5114903800320675,
+      "learning_rate": 1.4360933569633386e-05,
+      "loss": 0.7532,
+      "step": 16890
+    },
+    {
+      "epoch": 0.3630192894273317,
+      "grad_norm": 0.5240676752585789,
+      "learning_rate": 1.4354798370789204e-05,
+      "loss": 0.7726,
+      "step": 16900
+    },
+    {
+      "epoch": 0.36323409374060234,
+      "grad_norm": 0.5477577408032249,
+      "learning_rate": 1.4348661148496985e-05,
+      "loss": 0.773,
+      "step": 16910
+    },
+    {
+      "epoch": 0.3634488980538729,
+      "grad_norm": 0.5276323223923417,
+      "learning_rate": 1.4342521905608377e-05,
+      "loss": 0.7604,
+      "step": 16920
+    },
+    {
+      "epoch": 0.3636637023671435,
+      "grad_norm": 0.5170851911406245,
+      "learning_rate": 1.4336380644975964e-05,
+      "loss": 0.7491,
+      "step": 16930
+    },
+    {
+      "epoch": 0.36387850668041416,
+      "grad_norm": 0.5356549892783816,
+      "learning_rate": 1.433023736945328e-05,
+      "loss": 0.7657,
+      "step": 16940
+    },
+    {
+      "epoch": 0.36409331099368475,
+      "grad_norm": 0.5428984566871172,
+      "learning_rate": 1.4324092081894771e-05,
+      "loss": 0.7661,
+      "step": 16950
+    },
+    {
+      "epoch": 0.36430811530695534,
+      "grad_norm": 0.5324713712005715,
+      "learning_rate": 1.4317944785155841e-05,
+      "loss": 0.7628,
+      "step": 16960
+    },
+    {
+      "epoch": 0.364522919620226,
+      "grad_norm": 0.5162388083041777,
+      "learning_rate": 1.4311795482092819e-05,
+      "loss": 0.7667,
+      "step": 16970
+    },
+    {
+      "epoch": 0.36473772393349657,
+      "grad_norm": 0.5143878924802234,
+      "learning_rate": 1.430564417556296e-05,
+      "loss": 0.7567,
+      "step": 16980
+    },
+    {
+      "epoch": 0.3649525282467672,
+      "grad_norm": 0.5152886688089272,
+      "learning_rate": 1.429949086842446e-05,
+      "loss": 0.7579,
+      "step": 16990
+    },
+    {
+      "epoch": 0.3651673325600378,
+      "grad_norm": 0.5232658905035079,
+      "learning_rate": 1.4293335563536444e-05,
+      "loss": 0.773,
+      "step": 17000
+    },
+    {
+      "epoch": 0.3653821368733084,
+      "grad_norm": 0.5351086541332937,
+      "learning_rate": 1.4287178263758954e-05,
+      "loss": 0.7743,
+      "step": 17010
+    },
+    {
+      "epoch": 0.36559694118657904,
+      "grad_norm": 0.534471003893465,
+      "learning_rate": 1.4281018971952968e-05,
+      "loss": 0.7766,
+      "step": 17020
+    },
+    {
+      "epoch": 0.3658117454998496,
+      "grad_norm": 0.5285689768103836,
+      "learning_rate": 1.4274857690980393e-05,
+      "loss": 0.7609,
+      "step": 17030
+    },
+    {
+      "epoch": 0.36602654981312027,
+      "grad_norm": 0.5247525596746352,
+      "learning_rate": 1.426869442370405e-05,
+      "loss": 0.7584,
+      "step": 17040
+    },
+    {
+      "epoch": 0.36624135412639086,
+      "grad_norm": 0.5343814259733137,
+      "learning_rate": 1.4262529172987694e-05,
+      "loss": 0.7715,
+      "step": 17050
+    },
+    {
+      "epoch": 0.36645615843966145,
+      "grad_norm": 1.5564508451969874,
+      "learning_rate": 1.4256361941695994e-05,
+      "loss": 0.7641,
+      "step": 17060
+    },
+    {
+      "epoch": 0.3666709627529321,
+      "grad_norm": 0.5184034090359657,
+      "learning_rate": 1.4250192732694539e-05,
+      "loss": 0.7544,
+      "step": 17070
+    },
+    {
+      "epoch": 0.3668857670662027,
+      "grad_norm": 0.9518622956465803,
+      "learning_rate": 1.4244021548849838e-05,
+      "loss": 0.7681,
+      "step": 17080
+    },
+    {
+      "epoch": 0.3671005713794733,
+      "grad_norm": 0.5468495559267383,
+      "learning_rate": 1.4237848393029321e-05,
+      "loss": 0.761,
+      "step": 17090
+    },
+    {
+      "epoch": 0.3673153756927439,
+      "grad_norm": 0.5213501404226146,
+      "learning_rate": 1.4231673268101334e-05,
+      "loss": 0.7667,
+      "step": 17100
+    },
+    {
+      "epoch": 0.3675301800060145,
+      "grad_norm": 0.5088259296767228,
+      "learning_rate": 1.4225496176935135e-05,
+      "loss": 0.7781,
+      "step": 17110
+    },
+    {
+      "epoch": 0.36774498431928515,
+      "grad_norm": 0.5236538699356668,
+      "learning_rate": 1.4219317122400895e-05,
+      "loss": 0.771,
+      "step": 17120
+    },
+    {
+      "epoch": 0.36795978863255574,
+      "grad_norm": 0.5211467299731635,
+      "learning_rate": 1.4213136107369701e-05,
+      "loss": 0.7793,
+      "step": 17130
+    },
+    {
+      "epoch": 0.36817459294582633,
+      "grad_norm": 0.5472140162284628,
+      "learning_rate": 1.4206953134713546e-05,
+      "loss": 0.7692,
+      "step": 17140
+    },
+    {
+      "epoch": 0.368389397259097,
+      "grad_norm": 0.5189807730044983,
+      "learning_rate": 1.4200768207305338e-05,
+      "loss": 0.7814,
+      "step": 17150
+    },
+    {
+      "epoch": 0.36860420157236756,
+      "grad_norm": 0.5355282252957151,
+      "learning_rate": 1.4194581328018887e-05,
+      "loss": 0.7701,
+      "step": 17160
+    },
+    {
+      "epoch": 0.3688190058856382,
+      "grad_norm": 0.5178079624135835,
+      "learning_rate": 1.418839249972892e-05,
+      "loss": 0.7692,
+      "step": 17170
+    },
+    {
+      "epoch": 0.3690338101989088,
+      "grad_norm": 0.5294877879343505,
+      "learning_rate": 1.4182201725311056e-05,
+      "loss": 0.7493,
+      "step": 17180
+    },
+    {
+      "epoch": 0.3692486145121794,
+      "grad_norm": 0.506889416207276,
+      "learning_rate": 1.417600900764183e-05,
+      "loss": 0.7575,
+      "step": 17190
+    },
+    {
+      "epoch": 0.36946341882545003,
+      "grad_norm": 0.5285238726260074,
+      "learning_rate": 1.4169814349598676e-05,
+      "loss": 0.761,
+      "step": 17200
+    },
+    {
+      "epoch": 0.3696782231387206,
+      "grad_norm": 0.5263966505400081,
+      "learning_rate": 1.4163617754059927e-05,
+      "loss": 0.7574,
+      "step": 17210
+    },
+    {
+      "epoch": 0.36989302745199126,
+      "grad_norm": 0.5444318728567634,
+      "learning_rate": 1.4157419223904816e-05,
+      "loss": 0.7836,
+      "step": 17220
+    },
+    {
+      "epoch": 0.37010783176526185,
+      "grad_norm": 0.504161618679252,
+      "learning_rate": 1.4151218762013486e-05,
+      "loss": 0.7667,
+      "step": 17230
+    },
+    {
+      "epoch": 0.37032263607853244,
+      "grad_norm": 0.5367134980237321,
+      "learning_rate": 1.4145016371266958e-05,
+      "loss": 0.7658,
+      "step": 17240
+    },
+    {
+      "epoch": 0.3705374403918031,
+      "grad_norm": 0.5437915470695406,
+      "learning_rate": 1.4138812054547164e-05,
+      "loss": 0.7642,
+      "step": 17250
+    },
+    {
+      "epoch": 0.3707522447050737,
+      "grad_norm": 0.5250865582583174,
+      "learning_rate": 1.413260581473693e-05,
+      "loss": 0.7629,
+      "step": 17260
+    },
+    {
+      "epoch": 0.37096704901834426,
+      "grad_norm": 0.5219131992470888,
+      "learning_rate": 1.412639765471997e-05,
+      "loss": 0.7492,
+      "step": 17270
+    },
+    {
+      "epoch": 0.3711818533316149,
+      "grad_norm": 0.5434175810188079,
+      "learning_rate": 1.4120187577380893e-05,
+      "loss": 0.7721,
+      "step": 17280
+    },
+    {
+      "epoch": 0.3713966576448855,
+      "grad_norm": 0.5297492710247524,
+      "learning_rate": 1.4113975585605197e-05,
+      "loss": 0.7627,
+      "step": 17290
+    },
+    {
+      "epoch": 0.37161146195815614,
+      "grad_norm": 0.5172129186389717,
+      "learning_rate": 1.4107761682279273e-05,
+      "loss": 0.7824,
+      "step": 17300
+    },
+    {
+      "epoch": 0.37182626627142673,
+      "grad_norm": 0.5530724301179245,
+      "learning_rate": 1.4101545870290397e-05,
+      "loss": 0.7623,
+      "step": 17310
+    },
+    {
+      "epoch": 0.3720410705846973,
+      "grad_norm": 0.5228165330027325,
+      "learning_rate": 1.4095328152526734e-05,
+      "loss": 0.7603,
+      "step": 17320
+    },
+    {
+      "epoch": 0.37225587489796796,
+      "grad_norm": 0.5134940227470199,
+      "learning_rate": 1.4089108531877336e-05,
+      "loss": 0.7396,
+      "step": 17330
+    },
+    {
+      "epoch": 0.37247067921123855,
+      "grad_norm": 0.5259963794438499,
+      "learning_rate": 1.4082887011232131e-05,
+      "loss": 0.7694,
+      "step": 17340
+    },
+    {
+      "epoch": 0.3726854835245092,
+      "grad_norm": 0.5189340620745447,
+      "learning_rate": 1.4076663593481943e-05,
+      "loss": 0.7759,
+      "step": 17350
+    },
+    {
+      "epoch": 0.3729002878377798,
+      "grad_norm": 0.5035877483790274,
+      "learning_rate": 1.4070438281518462e-05,
+      "loss": 0.7516,
+      "step": 17360
+    },
+    {
+      "epoch": 0.3731150921510504,
+      "grad_norm": 0.5274597389686024,
+      "learning_rate": 1.406421107823427e-05,
+      "loss": 0.7705,
+      "step": 17370
+    },
+    {
+      "epoch": 0.373329896464321,
+      "grad_norm": 0.5817576244957883,
+      "learning_rate": 1.405798198652283e-05,
+      "loss": 0.7664,
+      "step": 17380
+    },
+    {
+      "epoch": 0.3735447007775916,
+      "grad_norm": 0.5517499638677766,
+      "learning_rate": 1.405175100927847e-05,
+      "loss": 0.775,
+      "step": 17390
+    },
+    {
+      "epoch": 0.3737595050908622,
+      "grad_norm": 0.5527733297913258,
+      "learning_rate": 1.40455181493964e-05,
+      "loss": 0.7621,
+      "step": 17400
+    },
+    {
+      "epoch": 0.37397430940413284,
+      "grad_norm": 0.5363364768618227,
+      "learning_rate": 1.4039283409772706e-05,
+      "loss": 0.755,
+      "step": 17410
+    },
+    {
+      "epoch": 0.37418911371740343,
+      "grad_norm": 0.5375291344235176,
+      "learning_rate": 1.4033046793304348e-05,
+      "loss": 0.7594,
+      "step": 17420
+    },
+    {
+      "epoch": 0.3744039180306741,
+      "grad_norm": 0.5402648263612099,
+      "learning_rate": 1.402680830288916e-05,
+      "loss": 0.7605,
+      "step": 17430
+    },
+    {
+      "epoch": 0.37461872234394467,
+      "grad_norm": 0.538129594884827,
+      "learning_rate": 1.4020567941425837e-05,
+      "loss": 0.7753,
+      "step": 17440
+    },
+    {
+      "epoch": 0.37483352665721525,
+      "grad_norm": 0.5261936791371908,
+      "learning_rate": 1.4014325711813958e-05,
+      "loss": 0.7571,
+      "step": 17450
+    },
+    {
+      "epoch": 0.3750483309704859,
+      "grad_norm": 0.534378842346243,
+      "learning_rate": 1.4008081616953954e-05,
+      "loss": 0.7813,
+      "step": 17460
+    },
+    {
+      "epoch": 0.3752631352837565,
+      "grad_norm": 0.5320321584083502,
+      "learning_rate": 1.4001835659747137e-05,
+      "loss": 0.7602,
+      "step": 17470
+    },
+    {
+      "epoch": 0.37547793959702713,
+      "grad_norm": 0.5315203643805748,
+      "learning_rate": 1.3995587843095675e-05,
+      "loss": 0.7575,
+      "step": 17480
+    },
+    {
+      "epoch": 0.3756927439102977,
+      "grad_norm": 0.5301936608790958,
+      "learning_rate": 1.3989338169902604e-05,
+      "loss": 0.7676,
+      "step": 17490
+    },
+    {
+      "epoch": 0.3759075482235683,
+      "grad_norm": 0.5577865205877498,
+      "learning_rate": 1.398308664307182e-05,
+      "loss": 0.7614,
+      "step": 17500
+    },
+    {
+      "epoch": 0.37612235253683896,
+      "grad_norm": 0.5179881999767708,
+      "learning_rate": 1.3976833265508085e-05,
+      "loss": 0.7601,
+      "step": 17510
+    },
+    {
+      "epoch": 0.37633715685010954,
+      "grad_norm": 0.5208749334999229,
+      "learning_rate": 1.3970578040117013e-05,
+      "loss": 0.7525,
+      "step": 17520
+    },
+    {
+      "epoch": 0.37655196116338013,
+      "grad_norm": 0.5352630545243395,
+      "learning_rate": 1.3964320969805085e-05,
+      "loss": 0.7647,
+      "step": 17530
+    },
+    {
+      "epoch": 0.3767667654766508,
+      "grad_norm": 0.5193667340738525,
+      "learning_rate": 1.3958062057479638e-05,
+      "loss": 0.7527,
+      "step": 17540
+    },
+    {
+      "epoch": 0.37698156978992137,
+      "grad_norm": 0.5361696072953572,
+      "learning_rate": 1.395180130604886e-05,
+      "loss": 0.7672,
+      "step": 17550
+    },
+    {
+      "epoch": 0.377196374103192,
+      "grad_norm": 0.5518592474145112,
+      "learning_rate": 1.3945538718421797e-05,
+      "loss": 0.769,
+      "step": 17560
+    },
+    {
+      "epoch": 0.3774111784164626,
+      "grad_norm": 0.5148045343692099,
+      "learning_rate": 1.3939274297508343e-05,
+      "loss": 0.7542,
+      "step": 17570
+    },
+    {
+      "epoch": 0.3776259827297332,
+      "grad_norm": 0.5228184172172592,
+      "learning_rate": 1.3933008046219256e-05,
+      "loss": 0.7722,
+      "step": 17580
+    },
+    {
+      "epoch": 0.37784078704300383,
+      "grad_norm": 0.5606699310701568,
+      "learning_rate": 1.392673996746613e-05,
+      "loss": 0.7608,
+      "step": 17590
+    },
+    {
+      "epoch": 0.3780555913562744,
+      "grad_norm": 0.5155322885195631,
+      "learning_rate": 1.3920470064161417e-05,
+      "loss": 0.7648,
+      "step": 17600
+    },
+    {
+      "epoch": 0.37827039566954507,
+      "grad_norm": 0.5295832405444447,
+      "learning_rate": 1.3914198339218417e-05,
+      "loss": 0.7529,
+      "step": 17610
+    },
+    {
+      "epoch": 0.37848519998281566,
+      "grad_norm": 0.5379568307902901,
+      "learning_rate": 1.3907924795551269e-05,
+      "loss": 0.7638,
+      "step": 17620
+    },
+    {
+      "epoch": 0.37870000429608625,
+      "grad_norm": 0.5202511460977277,
+      "learning_rate": 1.3901649436074967e-05,
+      "loss": 0.747,
+      "step": 17630
+    },
+    {
+      "epoch": 0.3789148086093569,
+      "grad_norm": 0.5082094007910867,
+      "learning_rate": 1.3895372263705342e-05,
+      "loss": 0.7553,
+      "step": 17640
+    },
+    {
+      "epoch": 0.3791296129226275,
+      "grad_norm": 0.5019905535533106,
+      "learning_rate": 1.3889093281359068e-05,
+      "loss": 0.7723,
+      "step": 17650
+    },
+    {
+      "epoch": 0.3793444172358981,
+      "grad_norm": 0.5289787148328216,
+      "learning_rate": 1.388281249195366e-05,
+      "loss": 0.7538,
+      "step": 17660
+    },
+    {
+      "epoch": 0.3795592215491687,
+      "grad_norm": 0.5138445337600223,
+      "learning_rate": 1.3876529898407479e-05,
+      "loss": 0.7703,
+      "step": 17670
+    },
+    {
+      "epoch": 0.3797740258624393,
+      "grad_norm": 0.4950419489829235,
+      "learning_rate": 1.3870245503639715e-05,
+      "loss": 0.7576,
+      "step": 17680
+    },
+    {
+      "epoch": 0.37998883017570995,
+      "grad_norm": 0.5254250563222473,
+      "learning_rate": 1.3863959310570398e-05,
+      "loss": 0.7667,
+      "step": 17690
+    },
+    {
+      "epoch": 0.38020363448898054,
+      "grad_norm": 0.520664787706605,
+      "learning_rate": 1.38576713221204e-05,
+      "loss": 0.7635,
+      "step": 17700
+    },
+    {
+      "epoch": 0.3804184388022511,
+      "grad_norm": 0.5313493460126173,
+      "learning_rate": 1.3851381541211418e-05,
+      "loss": 0.7751,
+      "step": 17710
+    },
+    {
+      "epoch": 0.38063324311552177,
+      "grad_norm": 0.5398410905297825,
+      "learning_rate": 1.384508997076598e-05,
+      "loss": 0.7624,
+      "step": 17720
+    },
+    {
+      "epoch": 0.38084804742879236,
+      "grad_norm": 0.5263114278035196,
+      "learning_rate": 1.3838796613707462e-05,
+      "loss": 0.7613,
+      "step": 17730
+    },
+    {
+      "epoch": 0.381062851742063,
+      "grad_norm": 0.5283488770743783,
+      "learning_rate": 1.3832501472960051e-05,
+      "loss": 0.7544,
+      "step": 17740
+    },
+    {
+      "epoch": 0.3812776560553336,
+      "grad_norm": 0.543513524216455,
+      "learning_rate": 1.3826204551448777e-05,
+      "loss": 0.7504,
+      "step": 17750
+    },
+    {
+      "epoch": 0.3814924603686042,
+      "grad_norm": 0.5214420491236335,
+      "learning_rate": 1.3819905852099492e-05,
+      "loss": 0.7651,
+      "step": 17760
+    },
+    {
+      "epoch": 0.3817072646818748,
+      "grad_norm": 0.5184770230999961,
+      "learning_rate": 1.3813605377838866e-05,
+      "loss": 0.752,
+      "step": 17770
+    },
+    {
+      "epoch": 0.3819220689951454,
+      "grad_norm": 0.5283519182814274,
+      "learning_rate": 1.3807303131594407e-05,
+      "loss": 0.7547,
+      "step": 17780
+    },
+    {
+      "epoch": 0.38213687330841606,
+      "grad_norm": 0.5385698106036354,
+      "learning_rate": 1.380099911629444e-05,
+      "loss": 0.7593,
+      "step": 17790
+    },
+    {
+      "epoch": 0.38235167762168665,
+      "grad_norm": 0.5402749618995843,
+      "learning_rate": 1.379469333486811e-05,
+      "loss": 0.7516,
+      "step": 17800
+    },
+    {
+      "epoch": 0.38256648193495724,
+      "grad_norm": 0.5247063198184665,
+      "learning_rate": 1.378838579024539e-05,
+      "loss": 0.7652,
+      "step": 17810
+    },
+    {
+      "epoch": 0.3827812862482279,
+      "grad_norm": 0.5378877903900204,
+      "learning_rate": 1.3782076485357062e-05,
+      "loss": 0.7635,
+      "step": 17820
+    },
+    {
+      "epoch": 0.38299609056149847,
+      "grad_norm": 0.5338589078746822,
+      "learning_rate": 1.377576542313474e-05,
+      "loss": 0.7582,
+      "step": 17830
+    },
+    {
+      "epoch": 0.38321089487476906,
+      "grad_norm": 0.5103257254373668,
+      "learning_rate": 1.3769452606510837e-05,
+      "loss": 0.76,
+      "step": 17840
+    },
+    {
+      "epoch": 0.3834256991880397,
+      "grad_norm": 0.5109691457275021,
+      "learning_rate": 1.3763138038418592e-05,
+      "loss": 0.7556,
+      "step": 17850
+    },
+    {
+      "epoch": 0.3836405035013103,
+      "grad_norm": 0.5358067276980497,
+      "learning_rate": 1.375682172179206e-05,
+      "loss": 0.7708,
+      "step": 17860
+    },
+    {
+      "epoch": 0.38385530781458094,
+      "grad_norm": 0.518350530032854,
+      "learning_rate": 1.37505036595661e-05,
+      "loss": 0.7784,
+      "step": 17870
+    },
+    {
+      "epoch": 0.3840701121278515,
+      "grad_norm": 0.5255617991273894,
+      "learning_rate": 1.374418385467639e-05,
+      "loss": 0.751,
+      "step": 17880
+    },
+    {
+      "epoch": 0.3842849164411221,
+      "grad_norm": 0.5491809745075493,
+      "learning_rate": 1.3737862310059413e-05,
+      "loss": 0.7485,
+      "step": 17890
+    },
+    {
+      "epoch": 0.38449972075439276,
+      "grad_norm": 0.5437186078897788,
+      "learning_rate": 1.373153902865246e-05,
+      "loss": 0.7716,
+      "step": 17900
+    },
+    {
+      "epoch": 0.38471452506766335,
+      "grad_norm": 0.553356354641266,
+      "learning_rate": 1.372521401339363e-05,
+      "loss": 0.7578,
+      "step": 17910
+    },
+    {
+      "epoch": 0.384929329380934,
+      "grad_norm": 0.531549500667387,
+      "learning_rate": 1.3718887267221835e-05,
+      "loss": 0.759,
+      "step": 17920
+    },
+    {
+      "epoch": 0.3851441336942046,
+      "grad_norm": 0.5164949440136289,
+      "learning_rate": 1.3712558793076777e-05,
+      "loss": 0.7706,
+      "step": 17930
+    },
+    {
+      "epoch": 0.38535893800747517,
+      "grad_norm": 0.508363864448989,
+      "learning_rate": 1.3706228593898971e-05,
+      "loss": 0.761,
+      "step": 17940
+    },
+    {
+      "epoch": 0.3855737423207458,
+      "grad_norm": 0.5645367042304027,
+      "learning_rate": 1.369989667262973e-05,
+      "loss": 0.7529,
+      "step": 17950
+    },
+    {
+      "epoch": 0.3857885466340164,
+      "grad_norm": 0.5276474618348226,
+      "learning_rate": 1.3693563032211173e-05,
+      "loss": 0.7619,
+      "step": 17960
+    },
+    {
+      "epoch": 0.386003350947287,
+      "grad_norm": 0.5219439134743928,
+      "learning_rate": 1.3687227675586205e-05,
+      "loss": 0.7602,
+      "step": 17970
+    },
+    {
+      "epoch": 0.38621815526055764,
+      "grad_norm": 0.4996278152443501,
+      "learning_rate": 1.3680890605698543e-05,
+      "loss": 0.7644,
+      "step": 17980
+    },
+    {
+      "epoch": 0.3864329595738282,
+      "grad_norm": 0.522998743172047,
+      "learning_rate": 1.3674551825492688e-05,
+      "loss": 0.7701,
+      "step": 17990
+    },
+    {
+      "epoch": 0.38664776388709887,
+      "grad_norm": 0.5374567407341325,
+      "learning_rate": 1.3668211337913944e-05,
+      "loss": 0.7746,
+      "step": 18000
+    },
+    {
+      "epoch": 0.38686256820036946,
+      "grad_norm": 0.5196875770806691,
+      "learning_rate": 1.3661869145908407e-05,
+      "loss": 0.7478,
+      "step": 18010
+    },
+    {
+      "epoch": 0.38707737251364005,
+      "grad_norm": 0.5292498035951604,
+      "learning_rate": 1.365552525242296e-05,
+      "loss": 0.7645,
+      "step": 18020
+    },
+    {
+      "epoch": 0.3872921768269107,
+      "grad_norm": 0.5248382854490489,
+      "learning_rate": 1.3649179660405282e-05,
+      "loss": 0.7602,
+      "step": 18030
+    },
+    {
+      "epoch": 0.3875069811401813,
+      "grad_norm": 0.5313710119749431,
+      "learning_rate": 1.364283237280384e-05,
+      "loss": 0.7786,
+      "step": 18040
+    },
+    {
+      "epoch": 0.38772178545345193,
+      "grad_norm": 0.5307887108904168,
+      "learning_rate": 1.3636483392567884e-05,
+      "loss": 0.7738,
+      "step": 18050
+    },
+    {
+      "epoch": 0.3879365897667225,
+      "grad_norm": 0.540751829854641,
+      "learning_rate": 1.363013272264746e-05,
+      "loss": 0.7731,
+      "step": 18060
+    },
+    {
+      "epoch": 0.3881513940799931,
+      "grad_norm": 0.5156037186616597,
+      "learning_rate": 1.3623780365993389e-05,
+      "loss": 0.7552,
+      "step": 18070
+    },
+    {
+      "epoch": 0.38836619839326375,
+      "grad_norm": 0.5141088635988575,
+      "learning_rate": 1.3617426325557283e-05,
+      "loss": 0.7547,
+      "step": 18080
+    },
+    {
+      "epoch": 0.38858100270653434,
+      "grad_norm": 0.5318673688677644,
+      "learning_rate": 1.3611070604291535e-05,
+      "loss": 0.7689,
+      "step": 18090
+    },
+    {
+      "epoch": 0.388795807019805,
+      "grad_norm": 0.5436642226772909,
+      "learning_rate": 1.360471320514931e-05,
+      "loss": 0.7598,
+      "step": 18100
+    },
+    {
+      "epoch": 0.3890106113330756,
+      "grad_norm": 0.5438843302772419,
+      "learning_rate": 1.359835413108457e-05,
+      "loss": 0.7589,
+      "step": 18110
+    },
+    {
+      "epoch": 0.38922541564634616,
+      "grad_norm": 0.5257581273173912,
+      "learning_rate": 1.359199338505204e-05,
+      "loss": 0.7497,
+      "step": 18120
+    },
+    {
+      "epoch": 0.3894402199596168,
+      "grad_norm": 0.5190745673753684,
+      "learning_rate": 1.3585630970007228e-05,
+      "loss": 0.757,
+      "step": 18130
+    },
+    {
+      "epoch": 0.3896550242728874,
+      "grad_norm": 0.5299254643622509,
+      "learning_rate": 1.3579266888906422e-05,
+      "loss": 0.7564,
+      "step": 18140
+    },
+    {
+      "epoch": 0.389869828586158,
+      "grad_norm": 0.5305677893941887,
+      "learning_rate": 1.3572901144706675e-05,
+      "loss": 0.743,
+      "step": 18150
+    },
+    {
+      "epoch": 0.39008463289942863,
+      "grad_norm": 0.5173856069530872,
+      "learning_rate": 1.3566533740365812e-05,
+      "loss": 0.7552,
+      "step": 18160
+    },
+    {
+      "epoch": 0.3902994372126992,
+      "grad_norm": 0.5148950374087616,
+      "learning_rate": 1.3560164678842442e-05,
+      "loss": 0.7568,
+      "step": 18170
+    },
+    {
+      "epoch": 0.39051424152596986,
+      "grad_norm": 0.5407560251354756,
+      "learning_rate": 1.355379396309593e-05,
+      "loss": 0.7546,
+      "step": 18180
+    },
+    {
+      "epoch": 0.39072904583924045,
+      "grad_norm": 0.5124304727359418,
+      "learning_rate": 1.3547421596086425e-05,
+      "loss": 0.7565,
+      "step": 18190
+    },
+    {
+      "epoch": 0.39094385015251104,
+      "grad_norm": 0.5431987417174345,
+      "learning_rate": 1.3541047580774827e-05,
+      "loss": 0.756,
+      "step": 18200
+    },
+    {
+      "epoch": 0.3911586544657817,
+      "grad_norm": 0.5073011287145005,
+      "learning_rate": 1.3534671920122809e-05,
+      "loss": 0.7464,
+      "step": 18210
+    },
+    {
+      "epoch": 0.3913734587790523,
+      "grad_norm": 0.5325329739414308,
+      "learning_rate": 1.3528294617092807e-05,
+      "loss": 0.7605,
+      "step": 18220
+    },
+    {
+      "epoch": 0.3915882630923229,
+      "grad_norm": 0.5369634007500615,
+      "learning_rate": 1.3521915674648027e-05,
+      "loss": 0.7621,
+      "step": 18230
+    },
+    {
+      "epoch": 0.3918030674055935,
+      "grad_norm": 0.5167864671283865,
+      "learning_rate": 1.351553509575243e-05,
+      "loss": 0.7719,
+      "step": 18240
+    },
+    {
+      "epoch": 0.3920178717188641,
+      "grad_norm": 0.5135322995315039,
+      "learning_rate": 1.3509152883370738e-05,
+      "loss": 0.7687,
+      "step": 18250
+    },
+    {
+      "epoch": 0.39223267603213474,
+      "grad_norm": 0.5320096412814102,
+      "learning_rate": 1.3502769040468428e-05,
+      "loss": 0.7483,
+      "step": 18260
+    },
+    {
+      "epoch": 0.39244748034540533,
+      "grad_norm": 0.5413014307472465,
+      "learning_rate": 1.3496383570011749e-05,
+      "loss": 0.7631,
+      "step": 18270
+    },
+    {
+      "epoch": 0.3926622846586759,
+      "grad_norm": 0.5335344669447194,
+      "learning_rate": 1.3489996474967688e-05,
+      "loss": 0.7688,
+      "step": 18280
+    },
+    {
+      "epoch": 0.39287708897194656,
+      "grad_norm": 0.5155463587590748,
+      "learning_rate": 1.3483607758304e-05,
+      "loss": 0.7603,
+      "step": 18290
+    },
+    {
+      "epoch": 0.39309189328521715,
+      "grad_norm": 0.5144797987402185,
+      "learning_rate": 1.3477217422989186e-05,
+      "loss": 0.7516,
+      "step": 18300
+    },
+    {
+      "epoch": 0.3933066975984878,
+      "grad_norm": 0.5138996863954206,
+      "learning_rate": 1.3470825471992508e-05,
+      "loss": 0.743,
+      "step": 18310
+    },
+    {
+      "epoch": 0.3935215019117584,
+      "grad_norm": 0.5084404683155206,
+      "learning_rate": 1.3464431908283966e-05,
+      "loss": 0.748,
+      "step": 18320
+    },
+    {
+      "epoch": 0.393736306225029,
+      "grad_norm": 0.5145070791156562,
+      "learning_rate": 1.3458036734834317e-05,
+      "loss": 0.7624,
+      "step": 18330
+    },
+    {
+      "epoch": 0.3939511105382996,
+      "grad_norm": 0.5228286492426271,
+      "learning_rate": 1.345163995461507e-05,
+      "loss": 0.7453,
+      "step": 18340
+    },
+    {
+      "epoch": 0.3941659148515702,
+      "grad_norm": 0.5234868717582319,
+      "learning_rate": 1.3445241570598471e-05,
+      "loss": 0.7755,
+      "step": 18350
+    },
+    {
+      "epoch": 0.39438071916484085,
+      "grad_norm": 0.5157437356324394,
+      "learning_rate": 1.3438841585757518e-05,
+      "loss": 0.7495,
+      "step": 18360
+    },
+    {
+      "epoch": 0.39459552347811144,
+      "grad_norm": 0.5390531614993764,
+      "learning_rate": 1.3432440003065949e-05,
+      "loss": 0.76,
+      "step": 18370
+    },
+    {
+      "epoch": 0.39481032779138203,
+      "grad_norm": 0.5384002860850948,
+      "learning_rate": 1.3426036825498248e-05,
+      "loss": 0.7666,
+      "step": 18380
+    },
+    {
+      "epoch": 0.3950251321046527,
+      "grad_norm": 0.5121988305761532,
+      "learning_rate": 1.3419632056029637e-05,
+      "loss": 0.7597,
+      "step": 18390
+    },
+    {
+      "epoch": 0.39523993641792327,
+      "grad_norm": 0.5104755486106719,
+      "learning_rate": 1.3413225697636079e-05,
+      "loss": 0.7483,
+      "step": 18400
+    },
+    {
+      "epoch": 0.3954547407311939,
+      "grad_norm": 0.5032637298232104,
+      "learning_rate": 1.3406817753294277e-05,
+      "loss": 0.7607,
+      "step": 18410
+    },
+    {
+      "epoch": 0.3956695450444645,
+      "grad_norm": 0.510737282878729,
+      "learning_rate": 1.3400408225981666e-05,
+      "loss": 0.7487,
+      "step": 18420
+    },
+    {
+      "epoch": 0.3958843493577351,
+      "grad_norm": 0.5527046441176208,
+      "learning_rate": 1.339399711867642e-05,
+      "loss": 0.76,
+      "step": 18430
+    },
+    {
+      "epoch": 0.39609915367100573,
+      "grad_norm": 0.5226806691957405,
+      "learning_rate": 1.338758443435745e-05,
+      "loss": 0.7537,
+      "step": 18440
+    },
+    {
+      "epoch": 0.3963139579842763,
+      "grad_norm": 0.5235233310915026,
+      "learning_rate": 1.3381170176004393e-05,
+      "loss": 0.7559,
+      "step": 18450
+    },
+    {
+      "epoch": 0.3965287622975469,
+      "grad_norm": 0.5168608712748513,
+      "learning_rate": 1.3374754346597622e-05,
+      "loss": 0.7523,
+      "step": 18460
+    },
+    {
+      "epoch": 0.39674356661081756,
+      "grad_norm": 0.5506026541978102,
+      "learning_rate": 1.3368336949118241e-05,
+      "loss": 0.7521,
+      "step": 18470
+    },
+    {
+      "epoch": 0.39695837092408814,
+      "grad_norm": 0.5268087236613826,
+      "learning_rate": 1.3361917986548076e-05,
+      "loss": 0.7693,
+      "step": 18480
+    },
+    {
+      "epoch": 0.3971731752373588,
+      "grad_norm": 0.5299142550170114,
+      "learning_rate": 1.3355497461869686e-05,
+      "loss": 0.762,
+      "step": 18490
+    },
+    {
+      "epoch": 0.3973879795506294,
+      "grad_norm": 0.49923625368152,
+      "learning_rate": 1.3349075378066358e-05,
+      "loss": 0.7538,
+      "step": 18500
+    },
+    {
+      "epoch": 0.39760278386389997,
+      "grad_norm": 0.526770181891027,
+      "learning_rate": 1.3342651738122096e-05,
+      "loss": 0.7692,
+      "step": 18510
+    },
+    {
+      "epoch": 0.3978175881771706,
+      "grad_norm": 0.5278097868657259,
+      "learning_rate": 1.3336226545021633e-05,
+      "loss": 0.7674,
+      "step": 18520
+    },
+    {
+      "epoch": 0.3980323924904412,
+      "grad_norm": 0.5253722249102108,
+      "learning_rate": 1.3329799801750425e-05,
+      "loss": 0.768,
+      "step": 18530
+    },
+    {
+      "epoch": 0.39824719680371184,
+      "grad_norm": 0.536105570764583,
+      "learning_rate": 1.3323371511294636e-05,
+      "loss": 0.762,
+      "step": 18540
+    },
+    {
+      "epoch": 0.39846200111698243,
+      "grad_norm": 0.5193573833212372,
+      "learning_rate": 1.3316941676641162e-05,
+      "loss": 0.7589,
+      "step": 18550
+    },
+    {
+      "epoch": 0.398676805430253,
+      "grad_norm": 0.5070127139643907,
+      "learning_rate": 1.3310510300777615e-05,
+      "loss": 0.7628,
+      "step": 18560
+    },
+    {
+      "epoch": 0.39889160974352367,
+      "grad_norm": 0.5191238299803989,
+      "learning_rate": 1.3304077386692318e-05,
+      "loss": 0.751,
+      "step": 18570
+    },
+    {
+      "epoch": 0.39910641405679426,
+      "grad_norm": 0.5205364024981556,
+      "learning_rate": 1.3297642937374313e-05,
+      "loss": 0.7545,
+      "step": 18580
+    },
+    {
+      "epoch": 0.39932121837006485,
+      "grad_norm": 0.5187329471387797,
+      "learning_rate": 1.3291206955813354e-05,
+      "loss": 0.7464,
+      "step": 18590
+    },
+    {
+      "epoch": 0.3995360226833355,
+      "grad_norm": 0.5049021637761884,
+      "learning_rate": 1.3284769444999901e-05,
+      "loss": 0.7595,
+      "step": 18600
+    },
+    {
+      "epoch": 0.3997508269966061,
+      "grad_norm": 0.5029990992742273,
+      "learning_rate": 1.3278330407925135e-05,
+      "loss": 0.7543,
+      "step": 18610
+    },
+    {
+      "epoch": 0.3999656313098767,
+      "grad_norm": 0.5173702160515916,
+      "learning_rate": 1.327188984758094e-05,
+      "loss": 0.7619,
+      "step": 18620
+    },
+    {
+      "epoch": 0.4001804356231473,
+      "grad_norm": 0.5192562549904551,
+      "learning_rate": 1.3265447766959911e-05,
+      "loss": 0.7561,
+      "step": 18630
+    },
+    {
+      "epoch": 0.4003952399364179,
+      "grad_norm": 0.5312951137431535,
+      "learning_rate": 1.3259004169055346e-05,
+      "loss": 0.7694,
+      "step": 18640
+    },
+    {
+      "epoch": 0.40061004424968855,
+      "grad_norm": 0.5266364154510266,
+      "learning_rate": 1.3252559056861246e-05,
+      "loss": 0.7543,
+      "step": 18650
+    },
+    {
+      "epoch": 0.40082484856295914,
+      "grad_norm": 0.533316964707828,
+      "learning_rate": 1.3246112433372322e-05,
+      "loss": 0.7554,
+      "step": 18660
+    },
+    {
+      "epoch": 0.4010396528762298,
+      "grad_norm": 0.5482477597760441,
+      "learning_rate": 1.3239664301583988e-05,
+      "loss": 0.7656,
+      "step": 18670
+    },
+    {
+      "epoch": 0.40125445718950037,
+      "grad_norm": 0.5206451326269249,
+      "learning_rate": 1.3233214664492349e-05,
+      "loss": 0.7543,
+      "step": 18680
+    },
+    {
+      "epoch": 0.40146926150277096,
+      "grad_norm": 0.5418016042557484,
+      "learning_rate": 1.3226763525094217e-05,
+      "loss": 0.7611,
+      "step": 18690
+    },
+    {
+      "epoch": 0.4016840658160416,
+      "grad_norm": 0.5558615173058707,
+      "learning_rate": 1.3220310886387103e-05,
+      "loss": 0.7627,
+      "step": 18700
+    },
+    {
+      "epoch": 0.4018988701293122,
+      "grad_norm": 0.5143709640231149,
+      "learning_rate": 1.3213856751369207e-05,
+      "loss": 0.7714,
+      "step": 18710
+    },
+    {
+      "epoch": 0.4021136744425828,
+      "grad_norm": 0.5376677743737073,
+      "learning_rate": 1.3207401123039438e-05,
+      "loss": 0.7691,
+      "step": 18720
+    },
+    {
+      "epoch": 0.4023284787558534,
+      "grad_norm": 0.5399761390877215,
+      "learning_rate": 1.3200944004397383e-05,
+      "loss": 0.7456,
+      "step": 18730
+    },
+    {
+      "epoch": 0.402543283069124,
+      "grad_norm": 0.5249199592562105,
+      "learning_rate": 1.3194485398443332e-05,
+      "loss": 0.7505,
+      "step": 18740
+    },
+    {
+      "epoch": 0.40275808738239466,
+      "grad_norm": 0.5155915016561029,
+      "learning_rate": 1.318802530817826e-05,
+      "loss": 0.7306,
+      "step": 18750
+    },
+    {
+      "epoch": 0.40297289169566525,
+      "grad_norm": 0.5318691830490915,
+      "learning_rate": 1.3181563736603837e-05,
+      "loss": 0.748,
+      "step": 18760
+    },
+    {
+      "epoch": 0.40318769600893584,
+      "grad_norm": 0.515075633683763,
+      "learning_rate": 1.317510068672242e-05,
+      "loss": 0.7597,
+      "step": 18770
+    },
+    {
+      "epoch": 0.4034025003222065,
+      "grad_norm": 0.5179116497827865,
+      "learning_rate": 1.316863616153705e-05,
+      "loss": 0.7613,
+      "step": 18780
+    },
+    {
+      "epoch": 0.40361730463547707,
+      "grad_norm": 0.5332150745577201,
+      "learning_rate": 1.3162170164051456e-05,
+      "loss": 0.7521,
+      "step": 18790
+    },
+    {
+      "epoch": 0.4038321089487477,
+      "grad_norm": 0.5253427224156337,
+      "learning_rate": 1.3155702697270047e-05,
+      "loss": 0.7733,
+      "step": 18800
+    },
+    {
+      "epoch": 0.4040469132620183,
+      "grad_norm": 0.49768785272010546,
+      "learning_rate": 1.3149233764197922e-05,
+      "loss": 0.7566,
+      "step": 18810
+    },
+    {
+      "epoch": 0.4042617175752889,
+      "grad_norm": 0.5267923414911647,
+      "learning_rate": 1.3142763367840857e-05,
+      "loss": 0.7667,
+      "step": 18820
+    },
+    {
+      "epoch": 0.40447652188855954,
+      "grad_norm": 0.5282610764687665,
+      "learning_rate": 1.3136291511205306e-05,
+      "loss": 0.755,
+      "step": 18830
+    },
+    {
+      "epoch": 0.4046913262018301,
+      "grad_norm": 0.5223892713619213,
+      "learning_rate": 1.3129818197298405e-05,
+      "loss": 0.7738,
+      "step": 18840
+    },
+    {
+      "epoch": 0.40490613051510077,
+      "grad_norm": 0.5096764777578232,
+      "learning_rate": 1.3123343429127968e-05,
+      "loss": 0.7588,
+      "step": 18850
+    },
+    {
+      "epoch": 0.40512093482837136,
+      "grad_norm": 0.5467891192906261,
+      "learning_rate": 1.3116867209702479e-05,
+      "loss": 0.7592,
+      "step": 18860
+    },
+    {
+      "epoch": 0.40533573914164195,
+      "grad_norm": 1.134506804779343,
+      "learning_rate": 1.3110389542031102e-05,
+      "loss": 0.7478,
+      "step": 18870
+    },
+    {
+      "epoch": 0.4055505434549126,
+      "grad_norm": 0.5097103865358088,
+      "learning_rate": 1.310391042912367e-05,
+      "loss": 0.7512,
+      "step": 18880
+    },
+    {
+      "epoch": 0.4057653477681832,
+      "grad_norm": 0.5393124410906558,
+      "learning_rate": 1.3097429873990693e-05,
+      "loss": 0.7485,
+      "step": 18890
+    },
+    {
+      "epoch": 0.40598015208145377,
+      "grad_norm": 0.5190274805087902,
+      "learning_rate": 1.3090947879643344e-05,
+      "loss": 0.7406,
+      "step": 18900
+    },
+    {
+      "epoch": 0.4061949563947244,
+      "grad_norm": 0.5440017480426464,
+      "learning_rate": 1.308446444909347e-05,
+      "loss": 0.7711,
+      "step": 18910
+    },
+    {
+      "epoch": 0.406409760707995,
+      "grad_norm": 0.5154904534059127,
+      "learning_rate": 1.3077979585353582e-05,
+      "loss": 0.75,
+      "step": 18920
+    },
+    {
+      "epoch": 0.40662456502126565,
+      "grad_norm": 0.5278477316199867,
+      "learning_rate": 1.307149329143686e-05,
+      "loss": 0.7612,
+      "step": 18930
+    },
+    {
+      "epoch": 0.40683936933453624,
+      "grad_norm": 0.5322812965577661,
+      "learning_rate": 1.3065005570357148e-05,
+      "loss": 0.7624,
+      "step": 18940
+    },
+    {
+      "epoch": 0.4070541736478068,
+      "grad_norm": 0.5160265527970386,
+      "learning_rate": 1.305851642512895e-05,
+      "loss": 0.7472,
+      "step": 18950
+    },
+    {
+      "epoch": 0.40726897796107747,
+      "grad_norm": 0.5318911272420563,
+      "learning_rate": 1.3052025858767435e-05,
+      "loss": 0.7703,
+      "step": 18960
+    },
+    {
+      "epoch": 0.40748378227434806,
+      "grad_norm": 0.5055369294794778,
+      "learning_rate": 1.3045533874288429e-05,
+      "loss": 0.7505,
+      "step": 18970
+    },
+    {
+      "epoch": 0.4076985865876187,
+      "grad_norm": 0.5298819562145214,
+      "learning_rate": 1.3039040474708422e-05,
+      "loss": 0.7645,
+      "step": 18980
+    },
+    {
+      "epoch": 0.4079133909008893,
+      "grad_norm": 0.552924739228477,
+      "learning_rate": 1.3032545663044558e-05,
+      "loss": 0.7734,
+      "step": 18990
+    },
+    {
+      "epoch": 0.4081281952141599,
+      "grad_norm": 0.5346123293209584,
+      "learning_rate": 1.3026049442314636e-05,
+      "loss": 0.7553,
+      "step": 19000
+    },
+    {
+      "epoch": 0.40834299952743053,
+      "grad_norm": 0.5246739179110647,
+      "learning_rate": 1.3019551815537117e-05,
+      "loss": 0.7505,
+      "step": 19010
+    },
+    {
+      "epoch": 0.4085578038407011,
+      "grad_norm": 0.5338166445740993,
+      "learning_rate": 1.3013052785731102e-05,
+      "loss": 0.7414,
+      "step": 19020
+    },
+    {
+      "epoch": 0.4087726081539717,
+      "grad_norm": 0.5109686263652257,
+      "learning_rate": 1.3006552355916355e-05,
+      "loss": 0.7477,
+      "step": 19030
+    },
+    {
+      "epoch": 0.40898741246724235,
+      "grad_norm": 0.5045191677684051,
+      "learning_rate": 1.300005052911329e-05,
+      "loss": 0.7546,
+      "step": 19040
+    },
+    {
+      "epoch": 0.40920221678051294,
+      "grad_norm": 0.5374579208608851,
+      "learning_rate": 1.2993547308342965e-05,
+      "loss": 0.7639,
+      "step": 19050
+    },
+    {
+      "epoch": 0.4094170210937836,
+      "grad_norm": 0.5316853043110944,
+      "learning_rate": 1.2987042696627085e-05,
+      "loss": 0.7595,
+      "step": 19060
+    },
+    {
+      "epoch": 0.4096318254070542,
+      "grad_norm": 0.5278988439756973,
+      "learning_rate": 1.298053669698801e-05,
+      "loss": 0.735,
+      "step": 19070
+    },
+    {
+      "epoch": 0.40984662972032476,
+      "grad_norm": 0.5654598465285191,
+      "learning_rate": 1.2974029312448737e-05,
+      "loss": 0.7684,
+      "step": 19080
+    },
+    {
+      "epoch": 0.4100614340335954,
+      "grad_norm": 0.5162837541623254,
+      "learning_rate": 1.2967520546032905e-05,
+      "loss": 0.7579,
+      "step": 19090
+    },
+    {
+      "epoch": 0.410276238346866,
+      "grad_norm": 0.5224389671190344,
+      "learning_rate": 1.2961010400764807e-05,
+      "loss": 0.7513,
+      "step": 19100
+    },
+    {
+      "epoch": 0.41049104266013664,
+      "grad_norm": 0.5176219223001366,
+      "learning_rate": 1.2954498879669361e-05,
+      "loss": 0.7638,
+      "step": 19110
+    },
+    {
+      "epoch": 0.41070584697340723,
+      "grad_norm": 0.534813023204064,
+      "learning_rate": 1.2947985985772137e-05,
+      "loss": 0.762,
+      "step": 19120
+    },
+    {
+      "epoch": 0.4109206512866778,
+      "grad_norm": 0.5272695180224509,
+      "learning_rate": 1.2941471722099332e-05,
+      "loss": 0.7592,
+      "step": 19130
+    },
+    {
+      "epoch": 0.41113545559994846,
+      "grad_norm": 0.5378612978101573,
+      "learning_rate": 1.2934956091677788e-05,
+      "loss": 0.7554,
+      "step": 19140
+    },
+    {
+      "epoch": 0.41135025991321905,
+      "grad_norm": 0.5258722653546618,
+      "learning_rate": 1.2928439097534981e-05,
+      "loss": 0.7635,
+      "step": 19150
+    },
+    {
+      "epoch": 0.41156506422648964,
+      "grad_norm": 0.5351602497774721,
+      "learning_rate": 1.2921920742699019e-05,
+      "loss": 0.7483,
+      "step": 19160
+    },
+    {
+      "epoch": 0.4117798685397603,
+      "grad_norm": 0.5367356185774689,
+      "learning_rate": 1.291540103019864e-05,
+      "loss": 0.755,
+      "step": 19170
+    },
+    {
+      "epoch": 0.4119946728530309,
+      "grad_norm": 0.5255579867546255,
+      "learning_rate": 1.2908879963063212e-05,
+      "loss": 0.7603,
+      "step": 19180
+    },
+    {
+      "epoch": 0.4122094771663015,
+      "grad_norm": 0.5058299224660839,
+      "learning_rate": 1.2902357544322741e-05,
+      "loss": 0.7623,
+      "step": 19190
+    },
+    {
+      "epoch": 0.4124242814795721,
+      "grad_norm": 0.5257377107858072,
+      "learning_rate": 1.289583377700785e-05,
+      "loss": 0.7433,
+      "step": 19200
+    },
+    {
+      "epoch": 0.4126390857928427,
+      "grad_norm": 0.5202414578374462,
+      "learning_rate": 1.28893086641498e-05,
+      "loss": 0.7522,
+      "step": 19210
+    },
+    {
+      "epoch": 0.41285389010611334,
+      "grad_norm": 0.514742134271601,
+      "learning_rate": 1.288278220878047e-05,
+      "loss": 0.7478,
+      "step": 19220
+    },
+    {
+      "epoch": 0.41306869441938393,
+      "grad_norm": 0.498665233670508,
+      "learning_rate": 1.287625441393236e-05,
+      "loss": 0.7732,
+      "step": 19230
+    },
+    {
+      "epoch": 0.4132834987326546,
+      "grad_norm": 0.5152993923389213,
+      "learning_rate": 1.2869725282638596e-05,
+      "loss": 0.7389,
+      "step": 19240
+    },
+    {
+      "epoch": 0.41349830304592516,
+      "grad_norm": 0.5369968074928185,
+      "learning_rate": 1.286319481793293e-05,
+      "loss": 0.7559,
+      "step": 19250
+    },
+    {
+      "epoch": 0.41371310735919575,
+      "grad_norm": 0.5168991961504903,
+      "learning_rate": 1.2856663022849724e-05,
+      "loss": 0.7425,
+      "step": 19260
+    },
+    {
+      "epoch": 0.4139279116724664,
+      "grad_norm": 0.5226064009439039,
+      "learning_rate": 1.2850129900423972e-05,
+      "loss": 0.7476,
+      "step": 19270
+    },
+    {
+      "epoch": 0.414142715985737,
+      "grad_norm": 0.5338894520277626,
+      "learning_rate": 1.2843595453691262e-05,
+      "loss": 0.7701,
+      "step": 19280
+    },
+    {
+      "epoch": 0.41435752029900763,
+      "grad_norm": 0.5143229406464946,
+      "learning_rate": 1.2837059685687823e-05,
+      "loss": 0.7565,
+      "step": 19290
+    },
+    {
+      "epoch": 0.4145723246122782,
+      "grad_norm": 0.5162616207638232,
+      "learning_rate": 1.2830522599450479e-05,
+      "loss": 0.765,
+      "step": 19300
+    },
+    {
+      "epoch": 0.4147871289255488,
+      "grad_norm": 0.5006245141202402,
+      "learning_rate": 1.2823984198016676e-05,
+      "loss": 0.7403,
+      "step": 19310
+    },
+    {
+      "epoch": 0.41500193323881945,
+      "grad_norm": 0.5284695404789536,
+      "learning_rate": 1.281744448442447e-05,
+      "loss": 0.7545,
+      "step": 19320
+    },
+    {
+      "epoch": 0.41521673755209004,
+      "grad_norm": 0.5214124159977402,
+      "learning_rate": 1.2810903461712524e-05,
+      "loss": 0.761,
+      "step": 19330
+    },
+    {
+      "epoch": 0.41543154186536063,
+      "grad_norm": 0.5254239866013033,
+      "learning_rate": 1.2804361132920114e-05,
+      "loss": 0.7446,
+      "step": 19340
+    },
+    {
+      "epoch": 0.4156463461786313,
+      "grad_norm": 0.5278037952638479,
+      "learning_rate": 1.2797817501087113e-05,
+      "loss": 0.7572,
+      "step": 19350
+    },
+    {
+      "epoch": 0.41586115049190187,
+      "grad_norm": 0.5367094234076616,
+      "learning_rate": 1.2791272569254009e-05,
+      "loss": 0.7462,
+      "step": 19360
+    },
+    {
+      "epoch": 0.4160759548051725,
+      "grad_norm": 0.5252418860683364,
+      "learning_rate": 1.2784726340461892e-05,
+      "loss": 0.7693,
+      "step": 19370
+    },
+    {
+      "epoch": 0.4162907591184431,
+      "grad_norm": 0.5055520616899003,
+      "learning_rate": 1.2778178817752454e-05,
+      "loss": 0.7481,
+      "step": 19380
+    },
+    {
+      "epoch": 0.4165055634317137,
+      "grad_norm": 0.5204456122832675,
+      "learning_rate": 1.2771630004167985e-05,
+      "loss": 0.7419,
+      "step": 19390
+    },
+    {
+      "epoch": 0.41672036774498433,
+      "grad_norm": 0.5217514946799076,
+      "learning_rate": 1.2765079902751381e-05,
+      "loss": 0.7459,
+      "step": 19400
+    },
+    {
+      "epoch": 0.4169351720582549,
+      "grad_norm": 0.53197721768162,
+      "learning_rate": 1.275852851654613e-05,
+      "loss": 0.7451,
+      "step": 19410
+    },
+    {
+      "epoch": 0.41714997637152557,
+      "grad_norm": 0.5345894761871923,
+      "learning_rate": 1.2751975848596324e-05,
+      "loss": 0.7545,
+      "step": 19420
+    },
+    {
+      "epoch": 0.41736478068479615,
+      "grad_norm": 0.5172473442706328,
+      "learning_rate": 1.274542190194664e-05,
+      "loss": 0.7633,
+      "step": 19430
+    },
+    {
+      "epoch": 0.41757958499806674,
+      "grad_norm": 0.5325296172080332,
+      "learning_rate": 1.2738866679642365e-05,
+      "loss": 0.749,
+      "step": 19440
+    },
+    {
+      "epoch": 0.4177943893113374,
+      "grad_norm": 0.5270837129930179,
+      "learning_rate": 1.2732310184729362e-05,
+      "loss": 0.7563,
+      "step": 19450
+    },
+    {
+      "epoch": 0.418009193624608,
+      "grad_norm": 0.514258180577775,
+      "learning_rate": 1.2725752420254094e-05,
+      "loss": 0.7591,
+      "step": 19460
+    },
+    {
+      "epoch": 0.41822399793787857,
+      "grad_norm": 0.5187969540802417,
+      "learning_rate": 1.2719193389263613e-05,
+      "loss": 0.7482,
+      "step": 19470
+    },
+    {
+      "epoch": 0.4184388022511492,
+      "grad_norm": 0.5303861691448711,
+      "learning_rate": 1.2712633094805561e-05,
+      "loss": 0.7547,
+      "step": 19480
+    },
+    {
+      "epoch": 0.4186536065644198,
+      "grad_norm": 0.5105008679720247,
+      "learning_rate": 1.2706071539928166e-05,
+      "loss": 0.7591,
+      "step": 19490
+    },
+    {
+      "epoch": 0.41886841087769044,
+      "grad_norm": 0.5277024751757646,
+      "learning_rate": 1.2699508727680238e-05,
+      "loss": 0.7563,
+      "step": 19500
+    },
+    {
+      "epoch": 0.41908321519096103,
+      "grad_norm": 0.5341955327167359,
+      "learning_rate": 1.2692944661111176e-05,
+      "loss": 0.7616,
+      "step": 19510
+    },
+    {
+      "epoch": 0.4192980195042316,
+      "grad_norm": 0.5321207219044839,
+      "learning_rate": 1.2686379343270956e-05,
+      "loss": 0.7487,
+      "step": 19520
+    },
+    {
+      "epoch": 0.41951282381750227,
+      "grad_norm": 0.5028973933951222,
+      "learning_rate": 1.2679812777210142e-05,
+      "loss": 0.7589,
+      "step": 19530
+    },
+    {
+      "epoch": 0.41972762813077286,
+      "grad_norm": 0.5333649392140662,
+      "learning_rate": 1.2673244965979881e-05,
+      "loss": 0.7499,
+      "step": 19540
+    },
+    {
+      "epoch": 0.4199424324440435,
+      "grad_norm": 0.8809161047806207,
+      "learning_rate": 1.2666675912631885e-05,
+      "loss": 0.7431,
+      "step": 19550
+    },
+    {
+      "epoch": 0.4201572367573141,
+      "grad_norm": 0.547365448998943,
+      "learning_rate": 1.2660105620218452e-05,
+      "loss": 0.7516,
+      "step": 19560
+    },
+    {
+      "epoch": 0.4203720410705847,
+      "grad_norm": 0.5278558806116719,
+      "learning_rate": 1.2653534091792459e-05,
+      "loss": 0.7572,
+      "step": 19570
+    },
+    {
+      "epoch": 0.4205868453838553,
+      "grad_norm": 0.5242417730519813,
+      "learning_rate": 1.2646961330407349e-05,
+      "loss": 0.7561,
+      "step": 19580
+    },
+    {
+      "epoch": 0.4208016496971259,
+      "grad_norm": 0.5145264148757996,
+      "learning_rate": 1.264038733911714e-05,
+      "loss": 0.7526,
+      "step": 19590
+    },
+    {
+      "epoch": 0.4210164540103965,
+      "grad_norm": 0.5226054611704246,
+      "learning_rate": 1.2633812120976432e-05,
+      "loss": 0.7585,
+      "step": 19600
+    },
+    {
+      "epoch": 0.42123125832366715,
+      "grad_norm": 0.5094656107452357,
+      "learning_rate": 1.262723567904038e-05,
+      "loss": 0.7431,
+      "step": 19610
+    },
+    {
+      "epoch": 0.42144606263693773,
+      "grad_norm": 0.5208004120644633,
+      "learning_rate": 1.2620658016364713e-05,
+      "loss": 0.7481,
+      "step": 19620
+    },
+    {
+      "epoch": 0.4216608669502084,
+      "grad_norm": 0.5368286486289077,
+      "learning_rate": 1.2614079136005732e-05,
+      "loss": 0.7659,
+      "step": 19630
+    },
+    {
+      "epoch": 0.42187567126347897,
+      "grad_norm": 0.5470952446979304,
+      "learning_rate": 1.26074990410203e-05,
+      "loss": 0.7511,
+      "step": 19640
+    },
+    {
+      "epoch": 0.42209047557674956,
+      "grad_norm": 0.5132946022336456,
+      "learning_rate": 1.2600917734465843e-05,
+      "loss": 0.7725,
+      "step": 19650
+    },
+    {
+      "epoch": 0.4223052798900202,
+      "grad_norm": 0.5597310989128308,
+      "learning_rate": 1.2594335219400349e-05,
+      "loss": 0.7431,
+      "step": 19660
+    },
+    {
+      "epoch": 0.4225200842032908,
+      "grad_norm": 0.5260208167606223,
+      "learning_rate": 1.2587751498882376e-05,
+      "loss": 0.7533,
+      "step": 19670
+    },
+    {
+      "epoch": 0.42273488851656144,
+      "grad_norm": 0.5029129520256707,
+      "learning_rate": 1.2581166575971031e-05,
+      "loss": 0.7506,
+      "step": 19680
+    },
+    {
+      "epoch": 0.422949692829832,
+      "grad_norm": 0.5202725382006363,
+      "learning_rate": 1.2574580453725987e-05,
+      "loss": 0.7541,
+      "step": 19690
+    },
+    {
+      "epoch": 0.4231644971431026,
+      "grad_norm": 0.5134917201548623,
+      "learning_rate": 1.256799313520747e-05,
+      "loss": 0.7564,
+      "step": 19700
+    },
+    {
+      "epoch": 0.42337930145637326,
+      "grad_norm": 0.5354376298898013,
+      "learning_rate": 1.256140462347627e-05,
+      "loss": 0.7594,
+      "step": 19710
+    },
+    {
+      "epoch": 0.42359410576964385,
+      "grad_norm": 0.5211973403768663,
+      "learning_rate": 1.2554814921593716e-05,
+      "loss": 0.7584,
+      "step": 19720
+    },
+    {
+      "epoch": 0.4238089100829145,
+      "grad_norm": 0.5520144836915775,
+      "learning_rate": 1.2548224032621707e-05,
+      "loss": 0.7486,
+      "step": 19730
+    },
+    {
+      "epoch": 0.4240237143961851,
+      "grad_norm": 0.4998308570756764,
+      "learning_rate": 1.2541631959622684e-05,
+      "loss": 0.7462,
+      "step": 19740
+    },
+    {
+      "epoch": 0.42423851870945567,
+      "grad_norm": 0.496745741982448,
+      "learning_rate": 1.2535038705659637e-05,
+      "loss": 0.7557,
+      "step": 19750
+    },
+    {
+      "epoch": 0.4244533230227263,
+      "grad_norm": 0.5179608442643614,
+      "learning_rate": 1.2528444273796113e-05,
+      "loss": 0.7559,
+      "step": 19760
+    },
+    {
+      "epoch": 0.4246681273359969,
+      "grad_norm": 0.5237889562247441,
+      "learning_rate": 1.2521848667096196e-05,
+      "loss": 0.759,
+      "step": 19770
+    },
+    {
+      "epoch": 0.4248829316492675,
+      "grad_norm": 0.5448250284811663,
+      "learning_rate": 1.251525188862452e-05,
+      "loss": 0.7612,
+      "step": 19780
+    },
+    {
+      "epoch": 0.42509773596253814,
+      "grad_norm": 0.5060894163492381,
+      "learning_rate": 1.250865394144627e-05,
+      "loss": 0.7593,
+      "step": 19790
+    },
+    {
+      "epoch": 0.4253125402758087,
+      "grad_norm": 0.49819573581917886,
+      "learning_rate": 1.2502054828627168e-05,
+      "loss": 0.7429,
+      "step": 19800
+    },
+    {
+      "epoch": 0.42552734458907937,
+      "grad_norm": 0.5213140703162686,
+      "learning_rate": 1.2495454553233473e-05,
+      "loss": 0.7549,
+      "step": 19810
+    },
+    {
+      "epoch": 0.42574214890234996,
+      "grad_norm": 0.5205619160973527,
+      "learning_rate": 1.2488853118331993e-05,
+      "loss": 0.7381,
+      "step": 19820
+    },
+    {
+      "epoch": 0.42595695321562055,
+      "grad_norm": 0.5411772964173832,
+      "learning_rate": 1.2482250526990072e-05,
+      "loss": 0.7563,
+      "step": 19830
+    },
+    {
+      "epoch": 0.4261717575288912,
+      "grad_norm": 0.5070043547298757,
+      "learning_rate": 1.2475646782275588e-05,
+      "loss": 0.7345,
+      "step": 19840
+    },
+    {
+      "epoch": 0.4263865618421618,
+      "grad_norm": 0.5446352761437885,
+      "learning_rate": 1.2469041887256955e-05,
+      "loss": 0.7575,
+      "step": 19850
+    },
+    {
+      "epoch": 0.4266013661554324,
+      "grad_norm": 0.5071987431688498,
+      "learning_rate": 1.2462435845003131e-05,
+      "loss": 0.7446,
+      "step": 19860
+    },
+    {
+      "epoch": 0.426816170468703,
+      "grad_norm": 0.5207497470969883,
+      "learning_rate": 1.2455828658583595e-05,
+      "loss": 0.7656,
+      "step": 19870
+    },
+    {
+      "epoch": 0.4270309747819736,
+      "grad_norm": 0.5117446569849975,
+      "learning_rate": 1.2449220331068363e-05,
+      "loss": 0.7582,
+      "step": 19880
+    },
+    {
+      "epoch": 0.42724577909524425,
+      "grad_norm": 0.5264990840912481,
+      "learning_rate": 1.2442610865527977e-05,
+      "loss": 0.7442,
+      "step": 19890
+    },
+    {
+      "epoch": 0.42746058340851484,
+      "grad_norm": 0.5258670236773598,
+      "learning_rate": 1.2436000265033518e-05,
+      "loss": 0.7531,
+      "step": 19900
+    },
+    {
+      "epoch": 0.4276753877217854,
+      "grad_norm": 0.538439943387158,
+      "learning_rate": 1.2429388532656586e-05,
+      "loss": 0.764,
+      "step": 19910
+    },
+    {
+      "epoch": 0.42789019203505607,
+      "grad_norm": 0.5177931014614214,
+      "learning_rate": 1.242277567146931e-05,
+      "loss": 0.7528,
+      "step": 19920
+    },
+    {
+      "epoch": 0.42810499634832666,
+      "grad_norm": 0.5194066108047999,
+      "learning_rate": 1.2416161684544337e-05,
+      "loss": 0.7478,
+      "step": 19930
+    },
+    {
+      "epoch": 0.4283198006615973,
+      "grad_norm": 0.5223576429675216,
+      "learning_rate": 1.2409546574954844e-05,
+      "loss": 0.7436,
+      "step": 19940
+    },
+    {
+      "epoch": 0.4285346049748679,
+      "grad_norm": 0.5331311220294771,
+      "learning_rate": 1.2402930345774533e-05,
+      "loss": 0.7626,
+      "step": 19950
+    },
+    {
+      "epoch": 0.4287494092881385,
+      "grad_norm": 0.5321622654012298,
+      "learning_rate": 1.2396313000077613e-05,
+      "loss": 0.7528,
+      "step": 19960
+    },
+    {
+      "epoch": 0.42896421360140913,
+      "grad_norm": 0.5213440299088972,
+      "learning_rate": 1.2389694540938828e-05,
+      "loss": 0.7607,
+      "step": 19970
+    },
+    {
+      "epoch": 0.4291790179146797,
+      "grad_norm": 0.5451420730102323,
+      "learning_rate": 1.238307497143343e-05,
+      "loss": 0.7514,
+      "step": 19980
+    },
+    {
+      "epoch": 0.42939382222795036,
+      "grad_norm": 0.5396602450861735,
+      "learning_rate": 1.2376454294637182e-05,
+      "loss": 0.754,
+      "step": 19990
+    },
+    {
+      "epoch": 0.42960862654122095,
+      "grad_norm": 0.5294320493844773,
+      "learning_rate": 1.2369832513626372e-05,
+      "loss": 0.7417,
+      "step": 20000
+    },
+    {
+      "epoch": 0.42982343085449154,
+      "grad_norm": 0.5229398240238433,
+      "learning_rate": 1.2363209631477797e-05,
+      "loss": 0.7516,
+      "step": 20010
+    },
+    {
+      "epoch": 0.4300382351677622,
+      "grad_norm": 0.5169446602861949,
+      "learning_rate": 1.2356585651268765e-05,
+      "loss": 0.7475,
+      "step": 20020
+    },
+    {
+      "epoch": 0.4302530394810328,
+      "grad_norm": 0.5471875007778104,
+      "learning_rate": 1.2349960576077097e-05,
+      "loss": 0.7589,
+      "step": 20030
+    },
+    {
+      "epoch": 0.43046784379430336,
+      "grad_norm": 0.5125110507107423,
+      "learning_rate": 1.2343334408981116e-05,
+      "loss": 0.7572,
+      "step": 20040
+    },
+    {
+      "epoch": 0.430682648107574,
+      "grad_norm": 0.5036197114282005,
+      "learning_rate": 1.2336707153059662e-05,
+      "loss": 0.7507,
+      "step": 20050
+    },
+    {
+      "epoch": 0.4308974524208446,
+      "grad_norm": 0.5169855576103094,
+      "learning_rate": 1.2330078811392068e-05,
+      "loss": 0.7567,
+      "step": 20060
+    },
+    {
+      "epoch": 0.43111225673411524,
+      "grad_norm": 0.5229836121047303,
+      "learning_rate": 1.2323449387058185e-05,
+      "loss": 0.7433,
+      "step": 20070
+    },
+    {
+      "epoch": 0.43132706104738583,
+      "grad_norm": 0.5272331493503679,
+      "learning_rate": 1.2316818883138362e-05,
+      "loss": 0.7618,
+      "step": 20080
+    },
+    {
+      "epoch": 0.4315418653606564,
+      "grad_norm": 0.5127750810913362,
+      "learning_rate": 1.2310187302713448e-05,
+      "loss": 0.7557,
+      "step": 20090
+    },
+    {
+      "epoch": 0.43175666967392706,
+      "grad_norm": 0.5046998338723229,
+      "learning_rate": 1.2303554648864791e-05,
+      "loss": 0.7406,
+      "step": 20100
+    },
+    {
+      "epoch": 0.43197147398719765,
+      "grad_norm": 0.5256594356973203,
+      "learning_rate": 1.229692092467424e-05,
+      "loss": 0.7516,
+      "step": 20110
+    },
+    {
+      "epoch": 0.4321862783004683,
+      "grad_norm": 0.5171058923668201,
+      "learning_rate": 1.2290286133224146e-05,
+      "loss": 0.7513,
+      "step": 20120
+    },
+    {
+      "epoch": 0.4324010826137389,
+      "grad_norm": 0.5104680411343472,
+      "learning_rate": 1.2283650277597343e-05,
+      "loss": 0.7378,
+      "step": 20130
+    },
+    {
+      "epoch": 0.4326158869270095,
+      "grad_norm": 0.5133259060250314,
+      "learning_rate": 1.2277013360877174e-05,
+      "loss": 0.7497,
+      "step": 20140
+    },
+    {
+      "epoch": 0.4328306912402801,
+      "grad_norm": 0.5068993271066068,
+      "learning_rate": 1.2270375386147466e-05,
+      "loss": 0.7562,
+      "step": 20150
+    },
+    {
+      "epoch": 0.4330454955535507,
+      "grad_norm": 0.5411116105681395,
+      "learning_rate": 1.2263736356492541e-05,
+      "loss": 0.7513,
+      "step": 20160
+    },
+    {
+      "epoch": 0.43326029986682135,
+      "grad_norm": 0.49754370415948745,
+      "learning_rate": 1.2257096274997209e-05,
+      "loss": 0.7427,
+      "step": 20170
+    },
+    {
+      "epoch": 0.43347510418009194,
+      "grad_norm": 0.5163307527085924,
+      "learning_rate": 1.2250455144746776e-05,
+      "loss": 0.7383,
+      "step": 20180
+    },
+    {
+      "epoch": 0.43368990849336253,
+      "grad_norm": 0.5302499060158602,
+      "learning_rate": 1.2243812968827021e-05,
+      "loss": 0.7304,
+      "step": 20190
+    },
+    {
+      "epoch": 0.4339047128066332,
+      "grad_norm": 0.5089583346915815,
+      "learning_rate": 1.223716975032422e-05,
+      "loss": 0.7482,
+      "step": 20200
+    },
+    {
+      "epoch": 0.43411951711990376,
+      "grad_norm": 0.5302791890282693,
+      "learning_rate": 1.2230525492325133e-05,
+      "loss": 0.7587,
+      "step": 20210
+    },
+    {
+      "epoch": 0.43433432143317435,
+      "grad_norm": 0.5183200687604282,
+      "learning_rate": 1.2223880197916997e-05,
+      "loss": 0.7385,
+      "step": 20220
+    },
+    {
+      "epoch": 0.434549125746445,
+      "grad_norm": 0.5283129948501645,
+      "learning_rate": 1.2217233870187539e-05,
+      "loss": 0.7618,
+      "step": 20230
+    },
+    {
+      "epoch": 0.4347639300597156,
+      "grad_norm": 0.5334500046605607,
+      "learning_rate": 1.221058651222496e-05,
+      "loss": 0.7574,
+      "step": 20240
+    },
+    {
+      "epoch": 0.43497873437298623,
+      "grad_norm": 0.5397388701861974,
+      "learning_rate": 1.220393812711794e-05,
+      "loss": 0.7544,
+      "step": 20250
+    },
+    {
+      "epoch": 0.4351935386862568,
+      "grad_norm": 0.5557994197082688,
+      "learning_rate": 1.2197288717955636e-05,
+      "loss": 0.7515,
+      "step": 20260
+    },
+    {
+      "epoch": 0.4354083429995274,
+      "grad_norm": 0.5214757763867583,
+      "learning_rate": 1.2190638287827683e-05,
+      "loss": 0.7562,
+      "step": 20270
+    },
+    {
+      "epoch": 0.43562314731279805,
+      "grad_norm": 0.5363420017553637,
+      "learning_rate": 1.2183986839824191e-05,
+      "loss": 0.7552,
+      "step": 20280
+    },
+    {
+      "epoch": 0.43583795162606864,
+      "grad_norm": 0.5173925675212045,
+      "learning_rate": 1.217733437703574e-05,
+      "loss": 0.7454,
+      "step": 20290
+    },
+    {
+      "epoch": 0.4360527559393393,
+      "grad_norm": 0.5052520562415731,
+      "learning_rate": 1.2170680902553388e-05,
+      "loss": 0.7335,
+      "step": 20300
+    },
+    {
+      "epoch": 0.4362675602526099,
+      "grad_norm": 0.5214947525173893,
+      "learning_rate": 1.216402641946865e-05,
+      "loss": 0.7533,
+      "step": 20310
+    },
+    {
+      "epoch": 0.43648236456588047,
+      "grad_norm": 0.5066915832021032,
+      "learning_rate": 1.2157370930873521e-05,
+      "loss": 0.7466,
+      "step": 20320
+    },
+    {
+      "epoch": 0.4366971688791511,
+      "grad_norm": 0.544927092725932,
+      "learning_rate": 1.2150714439860463e-05,
+      "loss": 0.7422,
+      "step": 20330
+    },
+    {
+      "epoch": 0.4369119731924217,
+      "grad_norm": 0.5232539415746384,
+      "learning_rate": 1.2144056949522396e-05,
+      "loss": 0.7456,
+      "step": 20340
+    },
+    {
+      "epoch": 0.4371267775056923,
+      "grad_norm": 0.5358017048774394,
+      "learning_rate": 1.2137398462952714e-05,
+      "loss": 0.761,
+      "step": 20350
+    },
+    {
+      "epoch": 0.43734158181896293,
+      "grad_norm": 0.514296276827805,
+      "learning_rate": 1.2130738983245265e-05,
+      "loss": 0.7406,
+      "step": 20360
+    },
+    {
+      "epoch": 0.4375563861322335,
+      "grad_norm": 0.5274917168112747,
+      "learning_rate": 1.2124078513494366e-05,
+      "loss": 0.7498,
+      "step": 20370
+    },
+    {
+      "epoch": 0.43777119044550417,
+      "grad_norm": 0.5091641073845611,
+      "learning_rate": 1.2117417056794787e-05,
+      "loss": 0.7669,
+      "step": 20380
+    },
+    {
+      "epoch": 0.43798599475877475,
+      "grad_norm": 0.5238531386173619,
+      "learning_rate": 1.211075461624176e-05,
+      "loss": 0.7454,
+      "step": 20390
+    },
+    {
+      "epoch": 0.43820079907204534,
+      "grad_norm": 0.5263012775160619,
+      "learning_rate": 1.2104091194930976e-05,
+      "loss": 0.7546,
+      "step": 20400
+    },
+    {
+      "epoch": 0.438415603385316,
+      "grad_norm": 0.5206973036533034,
+      "learning_rate": 1.2097426795958582e-05,
+      "loss": 0.7633,
+      "step": 20410
+    },
+    {
+      "epoch": 0.4386304076985866,
+      "grad_norm": 0.5329927051305835,
+      "learning_rate": 1.2090761422421173e-05,
+      "loss": 0.7492,
+      "step": 20420
+    },
+    {
+      "epoch": 0.4388452120118572,
+      "grad_norm": 0.5248347095576246,
+      "learning_rate": 1.2084095077415802e-05,
+      "loss": 0.7525,
+      "step": 20430
+    },
+    {
+      "epoch": 0.4390600163251278,
+      "grad_norm": 0.5763619430364404,
+      "learning_rate": 1.2077427764039976e-05,
+      "loss": 0.763,
+      "step": 20440
+    },
+    {
+      "epoch": 0.4392748206383984,
+      "grad_norm": 0.5277721502587376,
+      "learning_rate": 1.2070759485391642e-05,
+      "loss": 0.7728,
+      "step": 20450
+    },
+    {
+      "epoch": 0.43948962495166904,
+      "grad_norm": 0.5349602899110175,
+      "learning_rate": 1.206409024456921e-05,
+      "loss": 0.7529,
+      "step": 20460
+    },
+    {
+      "epoch": 0.43970442926493963,
+      "grad_norm": 0.5322814748856364,
+      "learning_rate": 1.2057420044671517e-05,
+      "loss": 0.7403,
+      "step": 20470
+    },
+    {
+      "epoch": 0.4399192335782103,
+      "grad_norm": 0.5118893319232871,
+      "learning_rate": 1.205074888879787e-05,
+      "loss": 0.7409,
+      "step": 20480
+    },
+    {
+      "epoch": 0.44013403789148087,
+      "grad_norm": 0.5455136550456627,
+      "learning_rate": 1.2044076780048e-05,
+      "loss": 0.7471,
+      "step": 20490
+    },
+    {
+      "epoch": 0.44034884220475146,
+      "grad_norm": 0.5299308092265677,
+      "learning_rate": 1.2037403721522095e-05,
+      "loss": 0.7577,
+      "step": 20500
+    },
+    {
+      "epoch": 0.4405636465180221,
+      "grad_norm": 0.5121296470133871,
+      "learning_rate": 1.2030729716320768e-05,
+      "loss": 0.747,
+      "step": 20510
+    },
+    {
+      "epoch": 0.4407784508312927,
+      "grad_norm": 0.5386378414121707,
+      "learning_rate": 1.2024054767545092e-05,
+      "loss": 0.7534,
+      "step": 20520
+    },
+    {
+      "epoch": 0.4409932551445633,
+      "grad_norm": 0.5433674923886992,
+      "learning_rate": 1.2017378878296562e-05,
+      "loss": 0.7503,
+      "step": 20530
+    },
+    {
+      "epoch": 0.4412080594578339,
+      "grad_norm": 0.5016141269307198,
+      "learning_rate": 1.2010702051677118e-05,
+      "loss": 0.7502,
+      "step": 20540
+    },
+    {
+      "epoch": 0.4414228637711045,
+      "grad_norm": 0.504046182876863,
+      "learning_rate": 1.2004024290789133e-05,
+      "loss": 0.7516,
+      "step": 20550
+    },
+    {
+      "epoch": 0.44163766808437516,
+      "grad_norm": 0.5226732086112348,
+      "learning_rate": 1.1997345598735418e-05,
+      "loss": 0.7494,
+      "step": 20560
+    },
+    {
+      "epoch": 0.44185247239764575,
+      "grad_norm": 0.5114919367543062,
+      "learning_rate": 1.1990665978619207e-05,
+      "loss": 0.7402,
+      "step": 20570
+    },
+    {
+      "epoch": 0.44206727671091633,
+      "grad_norm": 0.5440210462668759,
+      "learning_rate": 1.1983985433544176e-05,
+      "loss": 0.7533,
+      "step": 20580
+    },
+    {
+      "epoch": 0.442282081024187,
+      "grad_norm": 0.4942048477137153,
+      "learning_rate": 1.1977303966614426e-05,
+      "loss": 0.7581,
+      "step": 20590
+    },
+    {
+      "epoch": 0.44249688533745757,
+      "grad_norm": 0.5320405941370843,
+      "learning_rate": 1.1970621580934487e-05,
+      "loss": 0.7624,
+      "step": 20600
+    },
+    {
+      "epoch": 0.4427116896507282,
+      "grad_norm": 0.5185257384736662,
+      "learning_rate": 1.1963938279609313e-05,
+      "loss": 0.7406,
+      "step": 20610
+    },
+    {
+      "epoch": 0.4429264939639988,
+      "grad_norm": 0.5263134282281284,
+      "learning_rate": 1.195725406574429e-05,
+      "loss": 0.7621,
+      "step": 20620
+    },
+    {
+      "epoch": 0.4431412982772694,
+      "grad_norm": 0.5089836853097602,
+      "learning_rate": 1.1950568942445225e-05,
+      "loss": 0.7457,
+      "step": 20630
+    },
+    {
+      "epoch": 0.44335610259054004,
+      "grad_norm": 0.5201727085179321,
+      "learning_rate": 1.1943882912818339e-05,
+      "loss": 0.7349,
+      "step": 20640
+    },
+    {
+      "epoch": 0.4435709069038106,
+      "grad_norm": 0.5595981699934405,
+      "learning_rate": 1.1937195979970289e-05,
+      "loss": 0.7588,
+      "step": 20650
+    },
+    {
+      "epoch": 0.4437857112170812,
+      "grad_norm": 0.5276981692403241,
+      "learning_rate": 1.193050814700814e-05,
+      "loss": 0.7736,
+      "step": 20660
+    },
+    {
+      "epoch": 0.44400051553035186,
+      "grad_norm": 0.5236268232985871,
+      "learning_rate": 1.1923819417039383e-05,
+      "loss": 0.7595,
+      "step": 20670
+    },
+    {
+      "epoch": 0.44421531984362245,
+      "grad_norm": 0.49905338520313625,
+      "learning_rate": 1.1917129793171924e-05,
+      "loss": 0.7413,
+      "step": 20680
+    },
+    {
+      "epoch": 0.4444301241568931,
+      "grad_norm": 0.5164080638755711,
+      "learning_rate": 1.1910439278514081e-05,
+      "loss": 0.752,
+      "step": 20690
+    },
+    {
+      "epoch": 0.4446449284701637,
+      "grad_norm": 0.49930489760055635,
+      "learning_rate": 1.1903747876174583e-05,
+      "loss": 0.7514,
+      "step": 20700
+    },
+    {
+      "epoch": 0.44485973278343427,
+      "grad_norm": 0.5300452626534115,
+      "learning_rate": 1.1897055589262583e-05,
+      "loss": 0.7396,
+      "step": 20710
+    },
+    {
+      "epoch": 0.4450745370967049,
+      "grad_norm": 0.5321479894953371,
+      "learning_rate": 1.1890362420887634e-05,
+      "loss": 0.7508,
+      "step": 20720
+    },
+    {
+      "epoch": 0.4452893414099755,
+      "grad_norm": 0.5308819147481129,
+      "learning_rate": 1.1883668374159705e-05,
+      "loss": 0.7538,
+      "step": 20730
+    },
+    {
+      "epoch": 0.44550414572324615,
+      "grad_norm": 0.5069966537674646,
+      "learning_rate": 1.1876973452189172e-05,
+      "loss": 0.7629,
+      "step": 20740
+    },
+    {
+      "epoch": 0.44571895003651674,
+      "grad_norm": 0.5585367021939927,
+      "learning_rate": 1.1870277658086813e-05,
+      "loss": 0.7432,
+      "step": 20750
+    },
+    {
+      "epoch": 0.4459337543497873,
+      "grad_norm": 0.5223674054406362,
+      "learning_rate": 1.1863580994963817e-05,
+      "loss": 0.7439,
+      "step": 20760
+    },
+    {
+      "epoch": 0.44614855866305797,
+      "grad_norm": 0.5141402859358744,
+      "learning_rate": 1.1856883465931772e-05,
+      "loss": 0.746,
+      "step": 20770
+    },
+    {
+      "epoch": 0.44636336297632856,
+      "grad_norm": 0.5461400382474692,
+      "learning_rate": 1.1850185074102675e-05,
+      "loss": 0.7681,
+      "step": 20780
+    },
+    {
+      "epoch": 0.44657816728959915,
+      "grad_norm": 0.5190276865079356,
+      "learning_rate": 1.1843485822588923e-05,
+      "loss": 0.746,
+      "step": 20790
+    },
+    {
+      "epoch": 0.4467929716028698,
+      "grad_norm": 0.5149818719484064,
+      "learning_rate": 1.18367857145033e-05,
+      "loss": 0.7511,
+      "step": 20800
+    },
+    {
+      "epoch": 0.4470077759161404,
+      "grad_norm": 0.5038594272188172,
+      "learning_rate": 1.1830084752959002e-05,
+      "loss": 0.7601,
+      "step": 20810
+    },
+    {
+      "epoch": 0.447222580229411,
+      "grad_norm": 0.5057367274782396,
+      "learning_rate": 1.1823382941069618e-05,
+      "loss": 0.7504,
+      "step": 20820
+    },
+    {
+      "epoch": 0.4474373845426816,
+      "grad_norm": 0.5278200169568622,
+      "learning_rate": 1.1816680281949132e-05,
+      "loss": 0.734,
+      "step": 20830
+    },
+    {
+      "epoch": 0.4476521888559522,
+      "grad_norm": 0.4932889583978271,
+      "learning_rate": 1.1809976778711915e-05,
+      "loss": 0.7427,
+      "step": 20840
+    },
+    {
+      "epoch": 0.44786699316922285,
+      "grad_norm": 0.5254976142469129,
+      "learning_rate": 1.180327243447274e-05,
+      "loss": 0.7495,
+      "step": 20850
+    },
+    {
+      "epoch": 0.44808179748249344,
+      "grad_norm": 0.5258601024362219,
+      "learning_rate": 1.1796567252346766e-05,
+      "loss": 0.751,
+      "step": 20860
+    },
+    {
+      "epoch": 0.4482966017957641,
+      "grad_norm": 0.5361224414625771,
+      "learning_rate": 1.1789861235449542e-05,
+      "loss": 0.7354,
+      "step": 20870
+    },
+    {
+      "epoch": 0.44851140610903467,
+      "grad_norm": 0.508201235290661,
+      "learning_rate": 1.1783154386897008e-05,
+      "loss": 0.7453,
+      "step": 20880
+    },
+    {
+      "epoch": 0.44872621042230526,
+      "grad_norm": 0.5054092342839664,
+      "learning_rate": 1.1776446709805482e-05,
+      "loss": 0.7541,
+      "step": 20890
+    },
+    {
+      "epoch": 0.4489410147355759,
+      "grad_norm": 0.5178562956293307,
+      "learning_rate": 1.1769738207291674e-05,
+      "loss": 0.7564,
+      "step": 20900
+    },
+    {
+      "epoch": 0.4491558190488465,
+      "grad_norm": 0.5160643686715897,
+      "learning_rate": 1.1763028882472678e-05,
+      "loss": 0.7482,
+      "step": 20910
+    },
+    {
+      "epoch": 0.44937062336211714,
+      "grad_norm": 0.5200060584249153,
+      "learning_rate": 1.1756318738465963e-05,
+      "loss": 0.7432,
+      "step": 20920
+    },
+    {
+      "epoch": 0.4495854276753877,
+      "grad_norm": 0.5059948102109856,
+      "learning_rate": 1.1749607778389386e-05,
+      "loss": 0.7501,
+      "step": 20930
+    },
+    {
+      "epoch": 0.4498002319886583,
+      "grad_norm": 0.5186746353091075,
+      "learning_rate": 1.1742896005361186e-05,
+      "loss": 0.7333,
+      "step": 20940
+    },
+    {
+      "epoch": 0.45001503630192896,
+      "grad_norm": 0.5600841006515612,
+      "learning_rate": 1.173618342249997e-05,
+      "loss": 0.7478,
+      "step": 20950
+    },
+    {
+      "epoch": 0.45022984061519955,
+      "grad_norm": 0.5766313324253101,
+      "learning_rate": 1.1729470032924721e-05,
+      "loss": 0.7432,
+      "step": 20960
+    },
+    {
+      "epoch": 0.45044464492847014,
+      "grad_norm": 0.5166006082552747,
+      "learning_rate": 1.1722755839754807e-05,
+      "loss": 0.7473,
+      "step": 20970
+    },
+    {
+      "epoch": 0.4506594492417408,
+      "grad_norm": 0.49578934781253853,
+      "learning_rate": 1.1716040846109965e-05,
+      "loss": 0.7479,
+      "step": 20980
+    },
+    {
+      "epoch": 0.4508742535550114,
+      "grad_norm": 0.5016517079821076,
+      "learning_rate": 1.1709325055110296e-05,
+      "loss": 0.7349,
+      "step": 20990
+    },
+    {
+      "epoch": 0.451089057868282,
+      "grad_norm": 0.5210845740181136,
+      "learning_rate": 1.1702608469876288e-05,
+      "loss": 0.7338,
+      "step": 21000
+    },
+    {
+      "epoch": 0.4513038621815526,
+      "grad_norm": 0.5096278925929334,
+      "learning_rate": 1.1695891093528785e-05,
+      "loss": 0.7563,
+      "step": 21010
+    },
+    {
+      "epoch": 0.4515186664948232,
+      "grad_norm": 0.5086441029859898,
+      "learning_rate": 1.1689172929188997e-05,
+      "loss": 0.748,
+      "step": 21020
+    },
+    {
+      "epoch": 0.45173347080809384,
+      "grad_norm": 0.5387202509164648,
+      "learning_rate": 1.1682453979978507e-05,
+      "loss": 0.7562,
+      "step": 21030
+    },
+    {
+      "epoch": 0.45194827512136443,
+      "grad_norm": 0.5167975539045492,
+      "learning_rate": 1.1675734249019264e-05,
+      "loss": 0.7459,
+      "step": 21040
+    },
+    {
+      "epoch": 0.4521630794346351,
+      "grad_norm": 0.5094712446059221,
+      "learning_rate": 1.1669013739433576e-05,
+      "loss": 0.7372,
+      "step": 21050
+    },
+    {
+      "epoch": 0.45237788374790566,
+      "grad_norm": 0.5467898326634167,
+      "learning_rate": 1.1662292454344116e-05,
+      "loss": 0.7407,
+      "step": 21060
+    },
+    {
+      "epoch": 0.45259268806117625,
+      "grad_norm": 0.4989509970894457,
+      "learning_rate": 1.1655570396873911e-05,
+      "loss": 0.7469,
+      "step": 21070
+    },
+    {
+      "epoch": 0.4528074923744469,
+      "grad_norm": 0.5500057146231173,
+      "learning_rate": 1.1648847570146353e-05,
+      "loss": 0.7385,
+      "step": 21080
+    },
+    {
+      "epoch": 0.4530222966877175,
+      "grad_norm": 0.5119213457731346,
+      "learning_rate": 1.1642123977285187e-05,
+      "loss": 0.7535,
+      "step": 21090
+    },
+    {
+      "epoch": 0.4532371010009881,
+      "grad_norm": 0.5347652595873134,
+      "learning_rate": 1.163539962141452e-05,
+      "loss": 0.7514,
+      "step": 21100
+    },
+    {
+      "epoch": 0.4534519053142587,
+      "grad_norm": 0.5181244057297967,
+      "learning_rate": 1.162867450565881e-05,
+      "loss": 0.7448,
+      "step": 21110
+    },
+    {
+      "epoch": 0.4536667096275293,
+      "grad_norm": 0.5097669753456439,
+      "learning_rate": 1.1621948633142863e-05,
+      "loss": 0.7512,
+      "step": 21120
+    },
+    {
+      "epoch": 0.45388151394079995,
+      "grad_norm": 0.521551942346145,
+      "learning_rate": 1.161522200699185e-05,
+      "loss": 0.7335,
+      "step": 21130
+    },
+    {
+      "epoch": 0.45409631825407054,
+      "grad_norm": 0.5183982050773465,
+      "learning_rate": 1.1608494630331278e-05,
+      "loss": 0.7389,
+      "step": 21140
+    },
+    {
+      "epoch": 0.45431112256734113,
+      "grad_norm": 0.5005973180005574,
+      "learning_rate": 1.1601766506287009e-05,
+      "loss": 0.7244,
+      "step": 21150
+    },
+    {
+      "epoch": 0.4545259268806118,
+      "grad_norm": 0.5510297098838621,
+      "learning_rate": 1.1595037637985254e-05,
+      "loss": 0.7363,
+      "step": 21160
+    },
+    {
+      "epoch": 0.45474073119388236,
+      "grad_norm": 0.5172681651878976,
+      "learning_rate": 1.1588308028552567e-05,
+      "loss": 0.7469,
+      "step": 21170
+    },
+    {
+      "epoch": 0.454955535507153,
+      "grad_norm": 0.5101036668126001,
+      "learning_rate": 1.1581577681115844e-05,
+      "loss": 0.7403,
+      "step": 21180
+    },
+    {
+      "epoch": 0.4551703398204236,
+      "grad_norm": 0.5128096662612829,
+      "learning_rate": 1.1574846598802331e-05,
+      "loss": 0.7546,
+      "step": 21190
+    },
+    {
+      "epoch": 0.4553851441336942,
+      "grad_norm": 0.500612279234514,
+      "learning_rate": 1.1568114784739612e-05,
+      "loss": 0.748,
+      "step": 21200
+    },
+    {
+      "epoch": 0.45559994844696483,
+      "grad_norm": 0.5376520924082702,
+      "learning_rate": 1.1561382242055607e-05,
+      "loss": 0.7583,
+      "step": 21210
+    },
+    {
+      "epoch": 0.4558147527602354,
+      "grad_norm": 0.536096912372685,
+      "learning_rate": 1.1554648973878582e-05,
+      "loss": 0.7516,
+      "step": 21220
+    },
+    {
+      "epoch": 0.456029557073506,
+      "grad_norm": 0.5174865347901014,
+      "learning_rate": 1.154791498333713e-05,
+      "loss": 0.7433,
+      "step": 21230
+    },
+    {
+      "epoch": 0.45624436138677665,
+      "grad_norm": 0.5344982319266313,
+      "learning_rate": 1.154118027356019e-05,
+      "loss": 0.7348,
+      "step": 21240
+    },
+    {
+      "epoch": 0.45645916570004724,
+      "grad_norm": 0.49211553724020624,
+      "learning_rate": 1.153444484767703e-05,
+      "loss": 0.7431,
+      "step": 21250
+    },
+    {
+      "epoch": 0.4566739700133179,
+      "grad_norm": 0.5309412807127188,
+      "learning_rate": 1.1527708708817255e-05,
+      "loss": 0.7592,
+      "step": 21260
+    },
+    {
+      "epoch": 0.4568887743265885,
+      "grad_norm": 0.5266934403871943,
+      "learning_rate": 1.1520971860110795e-05,
+      "loss": 0.7467,
+      "step": 21270
+    },
+    {
+      "epoch": 0.45710357863985907,
+      "grad_norm": 0.5113799966183405,
+      "learning_rate": 1.1514234304687912e-05,
+      "loss": 0.7492,
+      "step": 21280
+    },
+    {
+      "epoch": 0.4573183829531297,
+      "grad_norm": 0.5265597304769074,
+      "learning_rate": 1.1507496045679196e-05,
+      "loss": 0.7577,
+      "step": 21290
+    },
+    {
+      "epoch": 0.4575331872664003,
+      "grad_norm": 0.5011207398210541,
+      "learning_rate": 1.1500757086215568e-05,
+      "loss": 0.7358,
+      "step": 21300
+    },
+    {
+      "epoch": 0.45774799157967094,
+      "grad_norm": 0.5197831944252082,
+      "learning_rate": 1.1494017429428271e-05,
+      "loss": 0.7402,
+      "step": 21310
+    },
+    {
+      "epoch": 0.45796279589294153,
+      "grad_norm": 0.5051983633873826,
+      "learning_rate": 1.1487277078448873e-05,
+      "loss": 0.7386,
+      "step": 21320
+    },
+    {
+      "epoch": 0.4581776002062121,
+      "grad_norm": 0.5177898532161163,
+      "learning_rate": 1.1480536036409262e-05,
+      "loss": 0.74,
+      "step": 21330
+    },
+    {
+      "epoch": 0.45839240451948277,
+      "grad_norm": 0.5150685367159163,
+      "learning_rate": 1.1473794306441652e-05,
+      "loss": 0.7379,
+      "step": 21340
+    },
+    {
+      "epoch": 0.45860720883275335,
+      "grad_norm": 0.5074160095947531,
+      "learning_rate": 1.1467051891678568e-05,
+      "loss": 0.7314,
+      "step": 21350
+    },
+    {
+      "epoch": 0.458822013146024,
+      "grad_norm": 0.5145616285881711,
+      "learning_rate": 1.1460308795252867e-05,
+      "loss": 0.7399,
+      "step": 21360
+    },
+    {
+      "epoch": 0.4590368174592946,
+      "grad_norm": 0.5160410203461876,
+      "learning_rate": 1.145356502029771e-05,
+      "loss": 0.7495,
+      "step": 21370
+    },
+    {
+      "epoch": 0.4592516217725652,
+      "grad_norm": 0.5315723943002935,
+      "learning_rate": 1.1446820569946581e-05,
+      "loss": 0.75,
+      "step": 21380
+    },
+    {
+      "epoch": 0.4594664260858358,
+      "grad_norm": 0.5106116756563384,
+      "learning_rate": 1.1440075447333274e-05,
+      "loss": 0.7417,
+      "step": 21390
+    },
+    {
+      "epoch": 0.4596812303991064,
+      "grad_norm": 0.5122130397254849,
+      "learning_rate": 1.1433329655591894e-05,
+      "loss": 0.7449,
+      "step": 21400
+    },
+    {
+      "epoch": 0.459896034712377,
+      "grad_norm": 0.5054378223152889,
+      "learning_rate": 1.1426583197856858e-05,
+      "loss": 0.7418,
+      "step": 21410
+    },
+    {
+      "epoch": 0.46011083902564764,
+      "grad_norm": 0.5244487700127819,
+      "learning_rate": 1.1419836077262899e-05,
+      "loss": 0.7419,
+      "step": 21420
+    },
+    {
+      "epoch": 0.46032564333891823,
+      "grad_norm": 0.5252805536342786,
+      "learning_rate": 1.1413088296945048e-05,
+      "loss": 0.7474,
+      "step": 21430
+    },
+    {
+      "epoch": 0.4605404476521889,
+      "grad_norm": 0.5098516964749148,
+      "learning_rate": 1.1406339860038648e-05,
+      "loss": 0.7566,
+      "step": 21440
+    },
+    {
+      "epoch": 0.46075525196545947,
+      "grad_norm": 0.53055505223306,
+      "learning_rate": 1.139959076967935e-05,
+      "loss": 0.7448,
+      "step": 21450
+    },
+    {
+      "epoch": 0.46097005627873006,
+      "grad_norm": 0.5135560710510545,
+      "learning_rate": 1.1392841029003102e-05,
+      "loss": 0.74,
+      "step": 21460
+    },
+    {
+      "epoch": 0.4611848605920007,
+      "grad_norm": 0.5214337220451569,
+      "learning_rate": 1.1386090641146152e-05,
+      "loss": 0.745,
+      "step": 21470
+    },
+    {
+      "epoch": 0.4613996649052713,
+      "grad_norm": 0.5375224268150335,
+      "learning_rate": 1.137933960924506e-05,
+      "loss": 0.7442,
+      "step": 21480
+    },
+    {
+      "epoch": 0.46161446921854193,
+      "grad_norm": 0.5139666564481621,
+      "learning_rate": 1.1372587936436683e-05,
+      "loss": 0.7419,
+      "step": 21490
+    },
+    {
+      "epoch": 0.4618292735318125,
+      "grad_norm": 0.5559740898339312,
+      "learning_rate": 1.1365835625858162e-05,
+      "loss": 0.7544,
+      "step": 21500
+    },
+    {
+      "epoch": 0.4620440778450831,
+      "grad_norm": 0.49830043641658905,
+      "learning_rate": 1.1359082680646952e-05,
+      "loss": 0.7341,
+      "step": 21510
+    },
+    {
+      "epoch": 0.46225888215835376,
+      "grad_norm": 0.5125992039187067,
+      "learning_rate": 1.1352329103940788e-05,
+      "loss": 0.7559,
+      "step": 21520
+    },
+    {
+      "epoch": 0.46247368647162435,
+      "grad_norm": 0.5194082716862066,
+      "learning_rate": 1.1345574898877707e-05,
+      "loss": 0.7439,
+      "step": 21530
+    },
+    {
+      "epoch": 0.46268849078489493,
+      "grad_norm": 0.5147396997044464,
+      "learning_rate": 1.1338820068596044e-05,
+      "loss": 0.7511,
+      "step": 21540
+    },
+    {
+      "epoch": 0.4629032950981656,
+      "grad_norm": 0.5392671660722685,
+      "learning_rate": 1.1332064616234407e-05,
+      "loss": 0.7653,
+      "step": 21550
+    },
+    {
+      "epoch": 0.46311809941143617,
+      "grad_norm": 0.5201496726087629,
+      "learning_rate": 1.1325308544931706e-05,
+      "loss": 0.7381,
+      "step": 21560
+    },
+    {
+      "epoch": 0.4633329037247068,
+      "grad_norm": 0.5180175465707748,
+      "learning_rate": 1.131855185782714e-05,
+      "loss": 0.7391,
+      "step": 21570
+    },
+    {
+      "epoch": 0.4635477080379774,
+      "grad_norm": 0.5336962808784633,
+      "learning_rate": 1.1311794558060186e-05,
+      "loss": 0.7469,
+      "step": 21580
+    },
+    {
+      "epoch": 0.463762512351248,
+      "grad_norm": 0.49650035979570856,
+      "learning_rate": 1.1305036648770608e-05,
+      "loss": 0.732,
+      "step": 21590
+    },
+    {
+      "epoch": 0.46397731666451864,
+      "grad_norm": 0.5281758487503815,
+      "learning_rate": 1.129827813309846e-05,
+      "loss": 0.7528,
+      "step": 21600
+    },
+    {
+      "epoch": 0.4641921209777892,
+      "grad_norm": 0.5234542589841056,
+      "learning_rate": 1.1291519014184062e-05,
+      "loss": 0.7247,
+      "step": 21610
+    },
+    {
+      "epoch": 0.46440692529105987,
+      "grad_norm": 0.517114689702638,
+      "learning_rate": 1.1284759295168035e-05,
+      "loss": 0.7511,
+      "step": 21620
+    },
+    {
+      "epoch": 0.46462172960433046,
+      "grad_norm": 0.5054831713408943,
+      "learning_rate": 1.1277998979191262e-05,
+      "loss": 0.7348,
+      "step": 21630
+    },
+    {
+      "epoch": 0.46483653391760105,
+      "grad_norm": 0.5169749307688332,
+      "learning_rate": 1.1271238069394916e-05,
+      "loss": 0.7517,
+      "step": 21640
+    },
+    {
+      "epoch": 0.4650513382308717,
+      "grad_norm": 0.5287808498540633,
+      "learning_rate": 1.1264476568920434e-05,
+      "loss": 0.7437,
+      "step": 21650
+    },
+    {
+      "epoch": 0.4652661425441423,
+      "grad_norm": 0.5133827362263325,
+      "learning_rate": 1.1257714480909538e-05,
+      "loss": 0.7256,
+      "step": 21660
+    },
+    {
+      "epoch": 0.46548094685741287,
+      "grad_norm": 0.4986667333468926,
+      "learning_rate": 1.125095180850421e-05,
+      "loss": 0.733,
+      "step": 21670
+    },
+    {
+      "epoch": 0.4656957511706835,
+      "grad_norm": 0.5041449057994531,
+      "learning_rate": 1.1244188554846722e-05,
+      "loss": 0.739,
+      "step": 21680
+    },
+    {
+      "epoch": 0.4659105554839541,
+      "grad_norm": 0.5091046285922776,
+      "learning_rate": 1.1237424723079597e-05,
+      "loss": 0.7352,
+      "step": 21690
+    },
+    {
+      "epoch": 0.46612535979722475,
+      "grad_norm": 0.5135885820320195,
+      "learning_rate": 1.1230660316345643e-05,
+      "loss": 0.7329,
+      "step": 21700
+    },
+    {
+      "epoch": 0.46634016411049534,
+      "grad_norm": 0.5200190247573033,
+      "learning_rate": 1.1223895337787924e-05,
+      "loss": 0.743,
+      "step": 21710
+    },
+    {
+      "epoch": 0.4665549684237659,
+      "grad_norm": 0.5353216172156325,
+      "learning_rate": 1.121712979054977e-05,
+      "loss": 0.7483,
+      "step": 21720
+    },
+    {
+      "epoch": 0.46676977273703657,
+      "grad_norm": 0.5082835412156764,
+      "learning_rate": 1.1210363677774782e-05,
+      "loss": 0.7464,
+      "step": 21730
+    },
+    {
+      "epoch": 0.46698457705030716,
+      "grad_norm": 0.5280559944508908,
+      "learning_rate": 1.1203597002606821e-05,
+      "loss": 0.7537,
+      "step": 21740
+    },
+    {
+      "epoch": 0.4671993813635778,
+      "grad_norm": 0.5260182340412108,
+      "learning_rate": 1.1196829768190008e-05,
+      "loss": 0.7474,
+      "step": 21750
+    },
+    {
+      "epoch": 0.4674141856768484,
+      "grad_norm": 0.5318302840617868,
+      "learning_rate": 1.1190061977668723e-05,
+      "loss": 0.752,
+      "step": 21760
+    },
+    {
+      "epoch": 0.467628989990119,
+      "grad_norm": 0.5202925069615598,
+      "learning_rate": 1.1183293634187609e-05,
+      "loss": 0.7395,
+      "step": 21770
+    },
+    {
+      "epoch": 0.4678437943033896,
+      "grad_norm": 0.5420558362059565,
+      "learning_rate": 1.1176524740891558e-05,
+      "loss": 0.746,
+      "step": 21780
+    },
+    {
+      "epoch": 0.4680585986166602,
+      "grad_norm": 0.5209989958223947,
+      "learning_rate": 1.1169755300925723e-05,
+      "loss": 0.7528,
+      "step": 21790
+    },
+    {
+      "epoch": 0.46827340292993086,
+      "grad_norm": 0.5323335344870374,
+      "learning_rate": 1.1162985317435514e-05,
+      "loss": 0.7419,
+      "step": 21800
+    },
+    {
+      "epoch": 0.46848820724320145,
+      "grad_norm": 0.5268811768109316,
+      "learning_rate": 1.1156214793566591e-05,
+      "loss": 0.7413,
+      "step": 21810
+    },
+    {
+      "epoch": 0.46870301155647204,
+      "grad_norm": 0.5526549348147461,
+      "learning_rate": 1.1149443732464858e-05,
+      "loss": 0.7423,
+      "step": 21820
+    },
+    {
+      "epoch": 0.4689178158697427,
+      "grad_norm": 0.5428671229053159,
+      "learning_rate": 1.1142672137276478e-05,
+      "loss": 0.7478,
+      "step": 21830
+    },
+    {
+      "epoch": 0.46913262018301327,
+      "grad_norm": 0.513399124764161,
+      "learning_rate": 1.1135900011147858e-05,
+      "loss": 0.7479,
+      "step": 21840
+    },
+    {
+      "epoch": 0.46934742449628386,
+      "grad_norm": 0.536835542346075,
+      "learning_rate": 1.1129127357225648e-05,
+      "loss": 0.7415,
+      "step": 21850
+    },
+    {
+      "epoch": 0.4695622288095545,
+      "grad_norm": 0.517525876955086,
+      "learning_rate": 1.1122354178656756e-05,
+      "loss": 0.7454,
+      "step": 21860
+    },
+    {
+      "epoch": 0.4697770331228251,
+      "grad_norm": 0.5016054637504687,
+      "learning_rate": 1.111558047858832e-05,
+      "loss": 0.7571,
+      "step": 21870
+    },
+    {
+      "epoch": 0.46999183743609574,
+      "grad_norm": 0.5114598800325706,
+      "learning_rate": 1.1108806260167727e-05,
+      "loss": 0.728,
+      "step": 21880
+    },
+    {
+      "epoch": 0.4702066417493663,
+      "grad_norm": 0.5092042135062025,
+      "learning_rate": 1.1102031526542605e-05,
+      "loss": 0.7356,
+      "step": 21890
+    },
+    {
+      "epoch": 0.4704214460626369,
+      "grad_norm": 0.5223633189785332,
+      "learning_rate": 1.109525628086082e-05,
+      "loss": 0.7477,
+      "step": 21900
+    },
+    {
+      "epoch": 0.47063625037590756,
+      "grad_norm": 0.5137541232678928,
+      "learning_rate": 1.1088480526270472e-05,
+      "loss": 0.7384,
+      "step": 21910
+    },
+    {
+      "epoch": 0.47085105468917815,
+      "grad_norm": 0.5268764811281718,
+      "learning_rate": 1.1081704265919904e-05,
+      "loss": 0.7434,
+      "step": 21920
+    },
+    {
+      "epoch": 0.4710658590024488,
+      "grad_norm": 0.5214178282194654,
+      "learning_rate": 1.1074927502957688e-05,
+      "loss": 0.747,
+      "step": 21930
+    },
+    {
+      "epoch": 0.4712806633157194,
+      "grad_norm": 0.5150008031537309,
+      "learning_rate": 1.1068150240532637e-05,
+      "loss": 0.7427,
+      "step": 21940
+    },
+    {
+      "epoch": 0.47149546762899,
+      "grad_norm": 0.5155181176356146,
+      "learning_rate": 1.1061372481793793e-05,
+      "loss": 0.735,
+      "step": 21950
+    },
+    {
+      "epoch": 0.4717102719422606,
+      "grad_norm": 0.528603646425843,
+      "learning_rate": 1.1054594229890425e-05,
+      "loss": 0.7206,
+      "step": 21960
+    },
+    {
+      "epoch": 0.4719250762555312,
+      "grad_norm": 0.5119688474719511,
+      "learning_rate": 1.1047815487972034e-05,
+      "loss": 0.7467,
+      "step": 21970
+    },
+    {
+      "epoch": 0.4721398805688018,
+      "grad_norm": 0.5252997267209152,
+      "learning_rate": 1.104103625918835e-05,
+      "loss": 0.7466,
+      "step": 21980
+    },
+    {
+      "epoch": 0.47235468488207244,
+      "grad_norm": 0.5257720369841751,
+      "learning_rate": 1.1034256546689321e-05,
+      "loss": 0.7479,
+      "step": 21990
+    },
+    {
+      "epoch": 0.47256948919534303,
+      "grad_norm": 0.536029357234098,
+      "learning_rate": 1.1027476353625132e-05,
+      "loss": 0.7447,
+      "step": 22000
+    },
+    {
+      "epoch": 0.4727842935086137,
+      "grad_norm": 0.5050633427695835,
+      "learning_rate": 1.1020695683146188e-05,
+      "loss": 0.7504,
+      "step": 22010
+    },
+    {
+      "epoch": 0.47299909782188426,
+      "grad_norm": 0.514261629466442,
+      "learning_rate": 1.101391453840311e-05,
+      "loss": 0.7397,
+      "step": 22020
+    },
+    {
+      "epoch": 0.47321390213515485,
+      "grad_norm": 0.5198813281021253,
+      "learning_rate": 1.1007132922546743e-05,
+      "loss": 0.7506,
+      "step": 22030
+    },
+    {
+      "epoch": 0.4734287064484255,
+      "grad_norm": 0.5181411136523617,
+      "learning_rate": 1.100035083872815e-05,
+      "loss": 0.746,
+      "step": 22040
+    },
+    {
+      "epoch": 0.4736435107616961,
+      "grad_norm": 0.5147323905400899,
+      "learning_rate": 1.0993568290098615e-05,
+      "loss": 0.7356,
+      "step": 22050
+    },
+    {
+      "epoch": 0.47385831507496673,
+      "grad_norm": 0.526477837321041,
+      "learning_rate": 1.0986785279809631e-05,
+      "loss": 0.7412,
+      "step": 22060
+    },
+    {
+      "epoch": 0.4740731193882373,
+      "grad_norm": 0.5313979721453921,
+      "learning_rate": 1.0980001811012915e-05,
+      "loss": 0.7336,
+      "step": 22070
+    },
+    {
+      "epoch": 0.4742879237015079,
+      "grad_norm": 0.5170598479629712,
+      "learning_rate": 1.0973217886860387e-05,
+      "loss": 0.7238,
+      "step": 22080
+    },
+    {
+      "epoch": 0.47450272801477855,
+      "grad_norm": 0.5221293133998406,
+      "learning_rate": 1.0966433510504188e-05,
+      "loss": 0.7344,
+      "step": 22090
+    },
+    {
+      "epoch": 0.47471753232804914,
+      "grad_norm": 0.5190509203550013,
+      "learning_rate": 1.0959648685096657e-05,
+      "loss": 0.7536,
+      "step": 22100
+    },
+    {
+      "epoch": 0.47493233664131973,
+      "grad_norm": 0.5103636926313977,
+      "learning_rate": 1.0952863413790355e-05,
+      "loss": 0.7493,
+      "step": 22110
+    },
+    {
+      "epoch": 0.4751471409545904,
+      "grad_norm": 0.5017701329508603,
+      "learning_rate": 1.0946077699738045e-05,
+      "loss": 0.7218,
+      "step": 22120
+    },
+    {
+      "epoch": 0.47536194526786096,
+      "grad_norm": 0.5334534885238904,
+      "learning_rate": 1.093929154609269e-05,
+      "loss": 0.7355,
+      "step": 22130
+    },
+    {
+      "epoch": 0.4755767495811316,
+      "grad_norm": 0.5347297096552461,
+      "learning_rate": 1.0932504956007468e-05,
+      "loss": 0.7414,
+      "step": 22140
+    },
+    {
+      "epoch": 0.4757915538944022,
+      "grad_norm": 0.5361415400554347,
+      "learning_rate": 1.0925717932635751e-05,
+      "loss": 0.7429,
+      "step": 22150
+    },
+    {
+      "epoch": 0.4760063582076728,
+      "grad_norm": 0.5276150896460355,
+      "learning_rate": 1.0918930479131114e-05,
+      "loss": 0.7343,
+      "step": 22160
+    },
+    {
+      "epoch": 0.47622116252094343,
+      "grad_norm": 0.5242096273960569,
+      "learning_rate": 1.0912142598647332e-05,
+      "loss": 0.7352,
+      "step": 22170
+    },
+    {
+      "epoch": 0.476435966834214,
+      "grad_norm": 0.5124773180856481,
+      "learning_rate": 1.0905354294338384e-05,
+      "loss": 0.7458,
+      "step": 22180
+    },
+    {
+      "epoch": 0.47665077114748466,
+      "grad_norm": 0.5009918414284297,
+      "learning_rate": 1.089856556935844e-05,
+      "loss": 0.7338,
+      "step": 22190
+    },
+    {
+      "epoch": 0.47686557546075525,
+      "grad_norm": 0.5108245418067267,
+      "learning_rate": 1.0891776426861868e-05,
+      "loss": 0.7421,
+      "step": 22200
+    },
+    {
+      "epoch": 0.47708037977402584,
+      "grad_norm": 0.509138834615506,
+      "learning_rate": 1.0884986870003229e-05,
+      "loss": 0.7431,
+      "step": 22210
+    },
+    {
+      "epoch": 0.4772951840872965,
+      "grad_norm": 0.5272156276514611,
+      "learning_rate": 1.0878196901937272e-05,
+      "loss": 0.7522,
+      "step": 22220
+    },
+    {
+      "epoch": 0.4775099884005671,
+      "grad_norm": 0.5053807372710165,
+      "learning_rate": 1.0871406525818947e-05,
+      "loss": 0.735,
+      "step": 22230
+    },
+    {
+      "epoch": 0.4777247927138377,
+      "grad_norm": 0.5166733531977794,
+      "learning_rate": 1.0864615744803382e-05,
+      "loss": 0.7278,
+      "step": 22240
+    },
+    {
+      "epoch": 0.4779395970271083,
+      "grad_norm": 0.5117417119746009,
+      "learning_rate": 1.0857824562045907e-05,
+      "loss": 0.7521,
+      "step": 22250
+    },
+    {
+      "epoch": 0.4781544013403789,
+      "grad_norm": 0.5256764124583501,
+      "learning_rate": 1.0851032980702025e-05,
+      "loss": 0.7392,
+      "step": 22260
+    },
+    {
+      "epoch": 0.47836920565364954,
+      "grad_norm": 0.5169468372261654,
+      "learning_rate": 1.0844241003927433e-05,
+      "loss": 0.7442,
+      "step": 22270
+    },
+    {
+      "epoch": 0.47858400996692013,
+      "grad_norm": 0.5075813885039875,
+      "learning_rate": 1.0837448634878011e-05,
+      "loss": 0.7498,
+      "step": 22280
+    },
+    {
+      "epoch": 0.4787988142801907,
+      "grad_norm": 0.49867951346910183,
+      "learning_rate": 1.0830655876709817e-05,
+      "loss": 0.7405,
+      "step": 22290
+    },
+    {
+      "epoch": 0.47901361859346137,
+      "grad_norm": 0.549353416061091,
+      "learning_rate": 1.0823862732579088e-05,
+      "loss": 0.7356,
+      "step": 22300
+    },
+    {
+      "epoch": 0.47922842290673195,
+      "grad_norm": 0.5233565980832593,
+      "learning_rate": 1.081706920564225e-05,
+      "loss": 0.7393,
+      "step": 22310
+    },
+    {
+      "epoch": 0.4794432272200026,
+      "grad_norm": 0.5166833901430437,
+      "learning_rate": 1.0810275299055899e-05,
+      "loss": 0.7399,
+      "step": 22320
+    },
+    {
+      "epoch": 0.4796580315332732,
+      "grad_norm": 0.5156376883464526,
+      "learning_rate": 1.0803481015976809e-05,
+      "loss": 0.7356,
+      "step": 22330
+    },
+    {
+      "epoch": 0.4798728358465438,
+      "grad_norm": 0.5361312416973215,
+      "learning_rate": 1.0796686359561931e-05,
+      "loss": 0.7575,
+      "step": 22340
+    },
+    {
+      "epoch": 0.4800876401598144,
+      "grad_norm": 0.52020980112163,
+      "learning_rate": 1.0789891332968387e-05,
+      "loss": 0.7236,
+      "step": 22350
+    },
+    {
+      "epoch": 0.480302444473085,
+      "grad_norm": 0.5250129443878427,
+      "learning_rate": 1.0783095939353474e-05,
+      "loss": 0.7297,
+      "step": 22360
+    },
+    {
+      "epoch": 0.48051724878635566,
+      "grad_norm": 0.5134420144953245,
+      "learning_rate": 1.0776300181874654e-05,
+      "loss": 0.7355,
+      "step": 22370
+    },
+    {
+      "epoch": 0.48073205309962624,
+      "grad_norm": 0.5291422486339019,
+      "learning_rate": 1.0769504063689564e-05,
+      "loss": 0.7422,
+      "step": 22380
+    },
+    {
+      "epoch": 0.48094685741289683,
+      "grad_norm": 0.5115283701598856,
+      "learning_rate": 1.0762707587956005e-05,
+      "loss": 0.7521,
+      "step": 22390
+    },
+    {
+      "epoch": 0.4811616617261675,
+      "grad_norm": 0.5148143798916653,
+      "learning_rate": 1.0755910757831949e-05,
+      "loss": 0.7351,
+      "step": 22400
+    },
+    {
+      "epoch": 0.48137646603943807,
+      "grad_norm": 0.5291173579659362,
+      "learning_rate": 1.0749113576475525e-05,
+      "loss": 0.7431,
+      "step": 22410
+    },
+    {
+      "epoch": 0.48159127035270866,
+      "grad_norm": 0.5391435806265806,
+      "learning_rate": 1.0742316047045029e-05,
+      "loss": 0.7435,
+      "step": 22420
+    },
+    {
+      "epoch": 0.4818060746659793,
+      "grad_norm": 0.5092459862659836,
+      "learning_rate": 1.0735518172698922e-05,
+      "loss": 0.7368,
+      "step": 22430
+    },
+    {
+      "epoch": 0.4820208789792499,
+      "grad_norm": 0.5267935560236964,
+      "learning_rate": 1.0728719956595818e-05,
+      "loss": 0.7469,
+      "step": 22440
+    },
+    {
+      "epoch": 0.48223568329252053,
+      "grad_norm": 0.5176688281994023,
+      "learning_rate": 1.0721921401894498e-05,
+      "loss": 0.7501,
+      "step": 22450
+    },
+    {
+      "epoch": 0.4824504876057911,
+      "grad_norm": 0.5137388029359484,
+      "learning_rate": 1.0715122511753897e-05,
+      "loss": 0.7425,
+      "step": 22460
+    },
+    {
+      "epoch": 0.4826652919190617,
+      "grad_norm": 0.5159930397277978,
+      "learning_rate": 1.0708323289333102e-05,
+      "loss": 0.7635,
+      "step": 22470
+    },
+    {
+      "epoch": 0.48288009623233236,
+      "grad_norm": 0.5256807433908982,
+      "learning_rate": 1.0701523737791359e-05,
+      "loss": 0.7319,
+      "step": 22480
+    },
+    {
+      "epoch": 0.48309490054560295,
+      "grad_norm": 0.5210214722370281,
+      "learning_rate": 1.0694723860288063e-05,
+      "loss": 0.7492,
+      "step": 22490
+    },
+    {
+      "epoch": 0.4833097048588736,
+      "grad_norm": 0.5231744475769452,
+      "learning_rate": 1.0687923659982766e-05,
+      "loss": 0.735,
+      "step": 22500
+    },
+    {
+      "epoch": 0.4835245091721442,
+      "grad_norm": 0.5011667972879604,
+      "learning_rate": 1.068112314003517e-05,
+      "loss": 0.7355,
+      "step": 22510
+    },
+    {
+      "epoch": 0.48373931348541477,
+      "grad_norm": 0.5265184060707572,
+      "learning_rate": 1.0674322303605115e-05,
+      "loss": 0.7395,
+      "step": 22520
+    },
+    {
+      "epoch": 0.4839541177986854,
+      "grad_norm": 0.5125129868588316,
+      "learning_rate": 1.0667521153852603e-05,
+      "loss": 0.7268,
+      "step": 22530
+    },
+    {
+      "epoch": 0.484168922111956,
+      "grad_norm": 0.5220283396876785,
+      "learning_rate": 1.0660719693937766e-05,
+      "loss": 0.7373,
+      "step": 22540
+    },
+    {
+      "epoch": 0.48438372642522665,
+      "grad_norm": 0.5091857205706148,
+      "learning_rate": 1.0653917927020894e-05,
+      "loss": 0.7402,
+      "step": 22550
+    },
+    {
+      "epoch": 0.48459853073849724,
+      "grad_norm": 0.5107328961554215,
+      "learning_rate": 1.0647115856262413e-05,
+      "loss": 0.7306,
+      "step": 22560
+    },
+    {
+      "epoch": 0.4848133350517678,
+      "grad_norm": 0.5133105974860892,
+      "learning_rate": 1.0640313484822893e-05,
+      "loss": 0.7324,
+      "step": 22570
+    },
+    {
+      "epoch": 0.48502813936503847,
+      "grad_norm": 0.529832381078322,
+      "learning_rate": 1.0633510815863036e-05,
+      "loss": 0.744,
+      "step": 22580
+    },
+    {
+      "epoch": 0.48524294367830906,
+      "grad_norm": 0.5223357765212729,
+      "learning_rate": 1.0626707852543695e-05,
+      "loss": 0.7363,
+      "step": 22590
+    },
+    {
+      "epoch": 0.48545774799157965,
+      "grad_norm": 0.5300543063286008,
+      "learning_rate": 1.0619904598025846e-05,
+      "loss": 0.7604,
+      "step": 22600
+    },
+    {
+      "epoch": 0.4856725523048503,
+      "grad_norm": 0.5379619059578166,
+      "learning_rate": 1.0613101055470612e-05,
+      "loss": 0.7536,
+      "step": 22610
+    },
+    {
+      "epoch": 0.4858873566181209,
+      "grad_norm": 0.5133418467154645,
+      "learning_rate": 1.0606297228039244e-05,
+      "loss": 0.7463,
+      "step": 22620
+    },
+    {
+      "epoch": 0.4861021609313915,
+      "grad_norm": 0.506231778810294,
+      "learning_rate": 1.0599493118893122e-05,
+      "loss": 0.7364,
+      "step": 22630
+    },
+    {
+      "epoch": 0.4863169652446621,
+      "grad_norm": 0.5019381915489841,
+      "learning_rate": 1.0592688731193768e-05,
+      "loss": 0.7333,
+      "step": 22640
+    },
+    {
+      "epoch": 0.4865317695579327,
+      "grad_norm": 0.5133937862155273,
+      "learning_rate": 1.0585884068102824e-05,
+      "loss": 0.7392,
+      "step": 22650
+    },
+    {
+      "epoch": 0.48674657387120335,
+      "grad_norm": 0.5225027361367924,
+      "learning_rate": 1.0579079132782061e-05,
+      "loss": 0.7376,
+      "step": 22660
+    },
+    {
+      "epoch": 0.48696137818447394,
+      "grad_norm": 0.5266639720499878,
+      "learning_rate": 1.0572273928393379e-05,
+      "loss": 0.7254,
+      "step": 22670
+    },
+    {
+      "epoch": 0.4871761824977446,
+      "grad_norm": 0.5509818361008318,
+      "learning_rate": 1.0565468458098806e-05,
+      "loss": 0.7393,
+      "step": 22680
+    },
+    {
+      "epoch": 0.48739098681101517,
+      "grad_norm": 0.5237573163610023,
+      "learning_rate": 1.0558662725060483e-05,
+      "loss": 0.7359,
+      "step": 22690
+    },
+    {
+      "epoch": 0.48760579112428576,
+      "grad_norm": 0.5107223127670096,
+      "learning_rate": 1.0551856732440681e-05,
+      "loss": 0.7585,
+      "step": 22700
+    },
+    {
+      "epoch": 0.4878205954375564,
+      "grad_norm": 0.5059159746679568,
+      "learning_rate": 1.0545050483401793e-05,
+      "loss": 0.7401,
+      "step": 22710
+    },
+    {
+      "epoch": 0.488035399750827,
+      "grad_norm": 0.5200289327198926,
+      "learning_rate": 1.0538243981106331e-05,
+      "loss": 0.7498,
+      "step": 22720
+    },
+    {
+      "epoch": 0.4882502040640976,
+      "grad_norm": 0.5189637119559256,
+      "learning_rate": 1.0531437228716914e-05,
+      "loss": 0.7482,
+      "step": 22730
+    },
+    {
+      "epoch": 0.4884650083773682,
+      "grad_norm": 0.49679955732243036,
+      "learning_rate": 1.0524630229396295e-05,
+      "loss": 0.7446,
+      "step": 22740
+    },
+    {
+      "epoch": 0.4886798126906388,
+      "grad_norm": 0.5207941387326519,
+      "learning_rate": 1.0517822986307325e-05,
+      "loss": 0.7624,
+      "step": 22750
+    },
+    {
+      "epoch": 0.48889461700390946,
+      "grad_norm": 0.5286403629591689,
+      "learning_rate": 1.0511015502612975e-05,
+      "loss": 0.7414,
+      "step": 22760
+    },
+    {
+      "epoch": 0.48910942131718005,
+      "grad_norm": 0.5120125001043839,
+      "learning_rate": 1.0504207781476334e-05,
+      "loss": 0.7366,
+      "step": 22770
+    },
+    {
+      "epoch": 0.48932422563045064,
+      "grad_norm": 0.5179656319122231,
+      "learning_rate": 1.0497399826060596e-05,
+      "loss": 0.7418,
+      "step": 22780
+    },
+    {
+      "epoch": 0.4895390299437213,
+      "grad_norm": 0.5366945858487391,
+      "learning_rate": 1.0490591639529055e-05,
+      "loss": 0.7357,
+      "step": 22790
+    },
+    {
+      "epoch": 0.48975383425699187,
+      "grad_norm": 0.5154119510735722,
+      "learning_rate": 1.0483783225045126e-05,
+      "loss": 0.7219,
+      "step": 22800
+    },
+    {
+      "epoch": 0.4899686385702625,
+      "grad_norm": 0.518597726407149,
+      "learning_rate": 1.0476974585772323e-05,
+      "loss": 0.741,
+      "step": 22810
+    },
+    {
+      "epoch": 0.4901834428835331,
+      "grad_norm": 0.5140931629957429,
+      "learning_rate": 1.047016572487427e-05,
+      "loss": 0.7578,
+      "step": 22820
+    },
+    {
+      "epoch": 0.4903982471968037,
+      "grad_norm": 0.5104752969998968,
+      "learning_rate": 1.0463356645514687e-05,
+      "loss": 0.7416,
+      "step": 22830
+    },
+    {
+      "epoch": 0.49061305151007434,
+      "grad_norm": 0.5313548540682782,
+      "learning_rate": 1.0456547350857397e-05,
+      "loss": 0.7472,
+      "step": 22840
+    },
+    {
+      "epoch": 0.4908278558233449,
+      "grad_norm": 0.5104717883570301,
+      "learning_rate": 1.0449737844066332e-05,
+      "loss": 0.7415,
+      "step": 22850
+    },
+    {
+      "epoch": 0.4910426601366155,
+      "grad_norm": 0.5321738566400499,
+      "learning_rate": 1.0442928128305504e-05,
+      "loss": 0.7307,
+      "step": 22860
+    },
+    {
+      "epoch": 0.49125746444988616,
+      "grad_norm": 0.5180000313964818,
+      "learning_rate": 1.0436118206739044e-05,
+      "loss": 0.7319,
+      "step": 22870
+    },
+    {
+      "epoch": 0.49147226876315675,
+      "grad_norm": 0.5269259655740001,
+      "learning_rate": 1.0429308082531157e-05,
+      "loss": 0.7297,
+      "step": 22880
+    },
+    {
+      "epoch": 0.4916870730764274,
+      "grad_norm": 0.5162542621179522,
+      "learning_rate": 1.0422497758846166e-05,
+      "loss": 0.7303,
+      "step": 22890
+    },
+    {
+      "epoch": 0.491901877389698,
+      "grad_norm": 0.9082353493403971,
+      "learning_rate": 1.0415687238848465e-05,
+      "loss": 0.7393,
+      "step": 22900
+    },
+    {
+      "epoch": 0.4921166817029686,
+      "grad_norm": 0.5249523721707448,
+      "learning_rate": 1.040887652570255e-05,
+      "loss": 0.7359,
+      "step": 22910
+    },
+    {
+      "epoch": 0.4923314860162392,
+      "grad_norm": 0.5310314571240732,
+      "learning_rate": 1.0402065622573003e-05,
+      "loss": 0.7473,
+      "step": 22920
+    },
+    {
+      "epoch": 0.4925462903295098,
+      "grad_norm": 0.5152700502053869,
+      "learning_rate": 1.03952545326245e-05,
+      "loss": 0.7386,
+      "step": 22930
+    },
+    {
+      "epoch": 0.49276109464278045,
+      "grad_norm": 0.5137486693952514,
+      "learning_rate": 1.0388443259021794e-05,
+      "loss": 0.7416,
+      "step": 22940
+    },
+    {
+      "epoch": 0.49297589895605104,
+      "grad_norm": 0.5147003203872301,
+      "learning_rate": 1.0381631804929737e-05,
+      "loss": 0.7558,
+      "step": 22950
+    },
+    {
+      "epoch": 0.49319070326932163,
+      "grad_norm": 0.5150451603396607,
+      "learning_rate": 1.0374820173513252e-05,
+      "loss": 0.7323,
+      "step": 22960
+    },
+    {
+      "epoch": 0.4934055075825923,
+      "grad_norm": 0.4998638447851373,
+      "learning_rate": 1.0368008367937348e-05,
+      "loss": 0.7394,
+      "step": 22970
+    },
+    {
+      "epoch": 0.49362031189586286,
+      "grad_norm": 0.5065272447150604,
+      "learning_rate": 1.036119639136712e-05,
+      "loss": 0.7547,
+      "step": 22980
+    },
+    {
+      "epoch": 0.4938351162091335,
+      "grad_norm": 0.5106286499648831,
+      "learning_rate": 1.035438424696774e-05,
+      "loss": 0.7497,
+      "step": 22990
+    },
+    {
+      "epoch": 0.4940499205224041,
+      "grad_norm": 0.5101651671094659,
+      "learning_rate": 1.0347571937904452e-05,
+      "loss": 0.7363,
+      "step": 23000
+    },
+    {
+      "epoch": 0.4942647248356747,
+      "grad_norm": 0.5243897522248346,
+      "learning_rate": 1.0340759467342582e-05,
+      "loss": 0.7444,
+      "step": 23010
+    },
+    {
+      "epoch": 0.49447952914894533,
+      "grad_norm": 0.5402515262236681,
+      "learning_rate": 1.0333946838447533e-05,
+      "loss": 0.7315,
+      "step": 23020
+    },
+    {
+      "epoch": 0.4946943334622159,
+      "grad_norm": 0.5051561251334392,
+      "learning_rate": 1.0327134054384778e-05,
+      "loss": 0.7332,
+      "step": 23030
+    },
+    {
+      "epoch": 0.4949091377754865,
+      "grad_norm": 0.5685316863277247,
+      "learning_rate": 1.0320321118319865e-05,
+      "loss": 0.7501,
+      "step": 23040
+    },
+    {
+      "epoch": 0.49512394208875715,
+      "grad_norm": 0.5057771401661005,
+      "learning_rate": 1.0313508033418405e-05,
+      "loss": 0.7315,
+      "step": 23050
+    },
+    {
+      "epoch": 0.49533874640202774,
+      "grad_norm": 0.5152345284853551,
+      "learning_rate": 1.0306694802846089e-05,
+      "loss": 0.7327,
+      "step": 23060
+    },
+    {
+      "epoch": 0.4955535507152984,
+      "grad_norm": 0.5021280830392914,
+      "learning_rate": 1.0299881429768669e-05,
+      "loss": 0.7376,
+      "step": 23070
+    },
+    {
+      "epoch": 0.495768355028569,
+      "grad_norm": 0.5249378964815303,
+      "learning_rate": 1.029306791735196e-05,
+      "loss": 0.7422,
+      "step": 23080
+    },
+    {
+      "epoch": 0.49598315934183956,
+      "grad_norm": 0.5032927828015555,
+      "learning_rate": 1.0286254268761853e-05,
+      "loss": 0.7412,
+      "step": 23090
+    },
+    {
+      "epoch": 0.4961979636551102,
+      "grad_norm": 0.5120900752083057,
+      "learning_rate": 1.0279440487164296e-05,
+      "loss": 0.7411,
+      "step": 23100
+    },
+    {
+      "epoch": 0.4964127679683808,
+      "grad_norm": 0.5194032778917269,
+      "learning_rate": 1.0272626575725296e-05,
+      "loss": 0.7422,
+      "step": 23110
+    },
+    {
+      "epoch": 0.49662757228165144,
+      "grad_norm": 0.5051279599131995,
+      "learning_rate": 1.0265812537610918e-05,
+      "loss": 0.734,
+      "step": 23120
+    },
+    {
+      "epoch": 0.49684237659492203,
+      "grad_norm": 0.524066136680626,
+      "learning_rate": 1.0258998375987297e-05,
+      "loss": 0.7471,
+      "step": 23130
+    },
+    {
+      "epoch": 0.4970571809081926,
+      "grad_norm": 0.5212407578128498,
+      "learning_rate": 1.025218409402062e-05,
+      "loss": 0.7234,
+      "step": 23140
+    },
+    {
+      "epoch": 0.49727198522146326,
+      "grad_norm": 0.5294461859743091,
+      "learning_rate": 1.0245369694877121e-05,
+      "loss": 0.7226,
+      "step": 23150
+    },
+    {
+      "epoch": 0.49748678953473385,
+      "grad_norm": 0.5223634925273634,
+      "learning_rate": 1.0238555181723108e-05,
+      "loss": 0.7453,
+      "step": 23160
+    },
+    {
+      "epoch": 0.49770159384800444,
+      "grad_norm": 0.5233047547217023,
+      "learning_rate": 1.0231740557724922e-05,
+      "loss": 0.7343,
+      "step": 23170
+    },
+    {
+      "epoch": 0.4979163981612751,
+      "grad_norm": 0.507862147559393,
+      "learning_rate": 1.0224925826048966e-05,
+      "loss": 0.7274,
+      "step": 23180
+    },
+    {
+      "epoch": 0.4981312024745457,
+      "grad_norm": 0.5155495724786578,
+      "learning_rate": 1.0218110989861691e-05,
+      "loss": 0.7443,
+      "step": 23190
+    },
+    {
+      "epoch": 0.4983460067878163,
+      "grad_norm": 0.5206048961869586,
+      "learning_rate": 1.0211296052329596e-05,
+      "loss": 0.7242,
+      "step": 23200
+    },
+    {
+      "epoch": 0.4985608111010869,
+      "grad_norm": 0.5119701323767093,
+      "learning_rate": 1.020448101661923e-05,
+      "loss": 0.7229,
+      "step": 23210
+    },
+    {
+      "epoch": 0.4987756154143575,
+      "grad_norm": 0.5253915545306624,
+      "learning_rate": 1.0197665885897184e-05,
+      "loss": 0.7243,
+      "step": 23220
+    },
+    {
+      "epoch": 0.49899041972762814,
+      "grad_norm": 0.4909170913393957,
+      "learning_rate": 1.0190850663330093e-05,
+      "loss": 0.7365,
+      "step": 23230
+    },
+    {
+      "epoch": 0.49920522404089873,
+      "grad_norm": 0.513781488832539,
+      "learning_rate": 1.0184035352084635e-05,
+      "loss": 0.7399,
+      "step": 23240
+    },
+    {
+      "epoch": 0.4994200283541694,
+      "grad_norm": 0.5186220905358521,
+      "learning_rate": 1.0177219955327533e-05,
+      "loss": 0.7416,
+      "step": 23250
+    },
+    {
+      "epoch": 0.49963483266743997,
+      "grad_norm": 0.5273590434779399,
+      "learning_rate": 1.0170404476225546e-05,
+      "loss": 0.7388,
+      "step": 23260
+    },
+    {
+      "epoch": 0.49984963698071055,
+      "grad_norm": 0.5270725741076642,
+      "learning_rate": 1.0163588917945472e-05,
+      "loss": 0.7465,
+      "step": 23270
+    },
+    {
+      "epoch": 0.5000644412939812,
+      "grad_norm": 0.5115666946192222,
+      "learning_rate": 1.0156773283654146e-05,
+      "loss": 0.7514,
+      "step": 23280
+    },
+    {
+      "epoch": 0.5002792456072518,
+      "grad_norm": 0.5140916715485198,
+      "learning_rate": 1.0149957576518444e-05,
+      "loss": 0.743,
+      "step": 23290
+    },
+    {
+      "epoch": 0.5004940499205224,
+      "grad_norm": 0.5049649532600137,
+      "learning_rate": 1.0143141799705259e-05,
+      "loss": 0.7492,
+      "step": 23300
+    },
+    {
+      "epoch": 0.500708854233793,
+      "grad_norm": 0.5148345236098256,
+      "learning_rate": 1.0136325956381535e-05,
+      "loss": 0.754,
+      "step": 23310
+    },
+    {
+      "epoch": 0.5009236585470637,
+      "grad_norm": 0.5081943498814092,
+      "learning_rate": 1.0129510049714238e-05,
+      "loss": 0.7464,
+      "step": 23320
+    },
+    {
+      "epoch": 0.5011384628603343,
+      "grad_norm": 0.5263481032653132,
+      "learning_rate": 1.0122694082870365e-05,
+      "loss": 0.7269,
+      "step": 23330
+    },
+    {
+      "epoch": 0.5013532671736048,
+      "grad_norm": 0.49404517879695686,
+      "learning_rate": 1.0115878059016942e-05,
+      "loss": 0.7431,
+      "step": 23340
+    },
+    {
+      "epoch": 0.5015680714868754,
+      "grad_norm": 0.5058326845907175,
+      "learning_rate": 1.0109061981321015e-05,
+      "loss": 0.7396,
+      "step": 23350
+    },
+    {
+      "epoch": 0.501782875800146,
+      "grad_norm": 0.5264138844155276,
+      "learning_rate": 1.0102245852949668e-05,
+      "loss": 0.7262,
+      "step": 23360
+    },
+    {
+      "epoch": 0.5019976801134167,
+      "grad_norm": 0.528096691604748,
+      "learning_rate": 1.0095429677069997e-05,
+      "loss": 0.7305,
+      "step": 23370
+    },
+    {
+      "epoch": 0.5022124844266873,
+      "grad_norm": 0.5116467023890083,
+      "learning_rate": 1.0088613456849125e-05,
+      "loss": 0.7339,
+      "step": 23380
+    },
+    {
+      "epoch": 0.5024272887399579,
+      "grad_norm": 0.4984116595301803,
+      "learning_rate": 1.0081797195454193e-05,
+      "loss": 0.734,
+      "step": 23390
+    },
+    {
+      "epoch": 0.5026420930532285,
+      "grad_norm": 0.5172883195495444,
+      "learning_rate": 1.0074980896052361e-05,
+      "loss": 0.7202,
+      "step": 23400
+    },
+    {
+      "epoch": 0.5028568973664991,
+      "grad_norm": 0.5069474447864731,
+      "learning_rate": 1.0068164561810814e-05,
+      "loss": 0.7261,
+      "step": 23410
+    },
+    {
+      "epoch": 0.5030717016797698,
+      "grad_norm": 0.5339797128974847,
+      "learning_rate": 1.0061348195896745e-05,
+      "loss": 0.756,
+      "step": 23420
+    },
+    {
+      "epoch": 0.5032865059930404,
+      "grad_norm": 0.5143777722194492,
+      "learning_rate": 1.0054531801477364e-05,
+      "loss": 0.7355,
+      "step": 23430
+    },
+    {
+      "epoch": 0.503501310306311,
+      "grad_norm": 0.5374619747017549,
+      "learning_rate": 1.0047715381719893e-05,
+      "loss": 0.7168,
+      "step": 23440
+    },
+    {
+      "epoch": 0.5037161146195815,
+      "grad_norm": 0.5141935837309727,
+      "learning_rate": 1.004089893979157e-05,
+      "loss": 0.7423,
+      "step": 23450
+    },
+    {
+      "epoch": 0.5039309189328521,
+      "grad_norm": 0.5097667863863926,
+      "learning_rate": 1.003408247885964e-05,
+      "loss": 0.7273,
+      "step": 23460
+    },
+    {
+      "epoch": 0.5041457232461228,
+      "grad_norm": 0.7070241417378597,
+      "learning_rate": 1.0027266002091353e-05,
+      "loss": 0.751,
+      "step": 23470
+    },
+    {
+      "epoch": 0.5043605275593934,
+      "grad_norm": 0.5200900402474737,
+      "learning_rate": 1.0020449512653978e-05,
+      "loss": 0.7479,
+      "step": 23480
+    },
+    {
+      "epoch": 0.504575331872664,
+      "grad_norm": 0.502245321750959,
+      "learning_rate": 1.001363301371478e-05,
+      "loss": 0.7344,
+      "step": 23490
+    },
+    {
+      "epoch": 0.5047901361859346,
+      "grad_norm": 0.5102853378974559,
+      "learning_rate": 1.0006816508441028e-05,
+      "loss": 0.7261,
+      "step": 23500
+    },
+    {
+      "epoch": 0.5050049404992052,
+      "grad_norm": 0.5139914043526608,
+      "learning_rate": 1e-05,
+      "loss": 0.7387,
+      "step": 23510
+    },
+    {
+      "epoch": 0.5052197448124758,
+      "grad_norm": 0.5138651494123287,
+      "learning_rate": 9.993183491558975e-06,
+      "loss": 0.7448,
+      "step": 23520
+    },
+    {
+      "epoch": 0.5054345491257465,
+      "grad_norm": 0.5138929240424788,
+      "learning_rate": 9.986366986285222e-06,
+      "loss": 0.7249,
+      "step": 23530
+    },
+    {
+      "epoch": 0.5056493534390171,
+      "grad_norm": 0.5184056795842351,
+      "learning_rate": 9.979550487346024e-06,
+      "loss": 0.74,
+      "step": 23540
+    },
+    {
+      "epoch": 0.5058641577522877,
+      "grad_norm": 0.5485760882777143,
+      "learning_rate": 9.972733997908648e-06,
+      "loss": 0.7288,
+      "step": 23550
+    },
+    {
+      "epoch": 0.5060789620655582,
+      "grad_norm": 0.5106653968189862,
+      "learning_rate": 9.965917521140365e-06,
+      "loss": 0.7364,
+      "step": 23560
+    },
+    {
+      "epoch": 0.5062937663788288,
+      "grad_norm": 0.49612689415307143,
+      "learning_rate": 9.95910106020843e-06,
+      "loss": 0.7324,
+      "step": 23570
+    },
+    {
+      "epoch": 0.5065085706920995,
+      "grad_norm": 0.5297725980627347,
+      "learning_rate": 9.952284618280108e-06,
+      "loss": 0.7332,
+      "step": 23580
+    },
+    {
+      "epoch": 0.5067233750053701,
+      "grad_norm": 0.5170669007337249,
+      "learning_rate": 9.94546819852264e-06,
+      "loss": 0.7483,
+      "step": 23590
+    },
+    {
+      "epoch": 0.5069381793186407,
+      "grad_norm": 0.5038468226015915,
+      "learning_rate": 9.938651804103257e-06,
+      "loss": 0.7398,
+      "step": 23600
+    },
+    {
+      "epoch": 0.5071529836319113,
+      "grad_norm": 0.5087783532039669,
+      "learning_rate": 9.93183543818919e-06,
+      "loss": 0.7325,
+      "step": 23610
+    },
+    {
+      "epoch": 0.5073677879451819,
+      "grad_norm": 0.5160070069541578,
+      "learning_rate": 9.925019103947639e-06,
+      "loss": 0.7325,
+      "step": 23620
+    },
+    {
+      "epoch": 0.5075825922584526,
+      "grad_norm": 0.5075365995460227,
+      "learning_rate": 9.91820280454581e-06,
+      "loss": 0.7553,
+      "step": 23630
+    },
+    {
+      "epoch": 0.5077973965717232,
+      "grad_norm": 0.5267021157619796,
+      "learning_rate": 9.91138654315088e-06,
+      "loss": 0.7298,
+      "step": 23640
+    },
+    {
+      "epoch": 0.5080122008849938,
+      "grad_norm": 0.5186830564197705,
+      "learning_rate": 9.904570322930006e-06,
+      "loss": 0.7556,
+      "step": 23650
+    },
+    {
+      "epoch": 0.5082270051982644,
+      "grad_norm": 0.5210004208221275,
+      "learning_rate": 9.897754147050335e-06,
+      "loss": 0.7376,
+      "step": 23660
+    },
+    {
+      "epoch": 0.508441809511535,
+      "grad_norm": 0.48685588400494934,
+      "learning_rate": 9.890938018678985e-06,
+      "loss": 0.7347,
+      "step": 23670
+    },
+    {
+      "epoch": 0.5086566138248056,
+      "grad_norm": 0.5539745463393094,
+      "learning_rate": 9.884121940983062e-06,
+      "loss": 0.7515,
+      "step": 23680
+    },
+    {
+      "epoch": 0.5088714181380762,
+      "grad_norm": 0.514283104621466,
+      "learning_rate": 9.877305917129636e-06,
+      "loss": 0.7456,
+      "step": 23690
+    },
+    {
+      "epoch": 0.5090862224513468,
+      "grad_norm": 0.5050582130270673,
+      "learning_rate": 9.870489950285765e-06,
+      "loss": 0.7217,
+      "step": 23700
+    },
+    {
+      "epoch": 0.5093010267646174,
+      "grad_norm": 0.5153338859894899,
+      "learning_rate": 9.86367404361847e-06,
+      "loss": 0.7345,
+      "step": 23710
+    },
+    {
+      "epoch": 0.509515831077888,
+      "grad_norm": 0.5292429611522316,
+      "learning_rate": 9.856858200294742e-06,
+      "loss": 0.7455,
+      "step": 23720
+    },
+    {
+      "epoch": 0.5097306353911587,
+      "grad_norm": 0.535174907505527,
+      "learning_rate": 9.850042423481561e-06,
+      "loss": 0.7321,
+      "step": 23730
+    },
+    {
+      "epoch": 0.5099454397044293,
+      "grad_norm": 0.5137588499191782,
+      "learning_rate": 9.843226716345852e-06,
+      "loss": 0.7283,
+      "step": 23740
+    },
+    {
+      "epoch": 0.5101602440176999,
+      "grad_norm": 0.5261263905458244,
+      "learning_rate": 9.83641108205453e-06,
+      "loss": 0.7356,
+      "step": 23750
+    },
+    {
+      "epoch": 0.5103750483309705,
+      "grad_norm": 0.5057280317999486,
+      "learning_rate": 9.829595523774456e-06,
+      "loss": 0.7346,
+      "step": 23760
+    },
+    {
+      "epoch": 0.5105898526442411,
+      "grad_norm": 0.5366500418456487,
+      "learning_rate": 9.82278004467247e-06,
+      "loss": 0.7353,
+      "step": 23770
+    },
+    {
+      "epoch": 0.5108046569575118,
+      "grad_norm": 0.5064327805142302,
+      "learning_rate": 9.81596464791537e-06,
+      "loss": 0.7224,
+      "step": 23780
+    },
+    {
+      "epoch": 0.5110194612707823,
+      "grad_norm": 0.48473823245015357,
+      "learning_rate": 9.80914933666991e-06,
+      "loss": 0.7302,
+      "step": 23790
+    },
+    {
+      "epoch": 0.5112342655840529,
+      "grad_norm": 0.530367936435279,
+      "learning_rate": 9.802334114102821e-06,
+      "loss": 0.723,
+      "step": 23800
+    },
+    {
+      "epoch": 0.5114490698973235,
+      "grad_norm": 0.53191867394961,
+      "learning_rate": 9.795518983380771e-06,
+      "loss": 0.7322,
+      "step": 23810
+    },
+    {
+      "epoch": 0.5116638742105941,
+      "grad_norm": 0.5145552336460878,
+      "learning_rate": 9.788703947670407e-06,
+      "loss": 0.7174,
+      "step": 23820
+    },
+    {
+      "epoch": 0.5118786785238647,
+      "grad_norm": 0.5160752379900866,
+      "learning_rate": 9.781889010138315e-06,
+      "loss": 0.7371,
+      "step": 23830
+    },
+    {
+      "epoch": 0.5120934828371354,
+      "grad_norm": 0.5215561353383581,
+      "learning_rate": 9.775074173951038e-06,
+      "loss": 0.7386,
+      "step": 23840
+    },
+    {
+      "epoch": 0.512308287150406,
+      "grad_norm": 0.546987216601412,
+      "learning_rate": 9.768259442275083e-06,
+      "loss": 0.7252,
+      "step": 23850
+    },
+    {
+      "epoch": 0.5125230914636766,
+      "grad_norm": 0.5250432835086648,
+      "learning_rate": 9.761444818276895e-06,
+      "loss": 0.7326,
+      "step": 23860
+    },
+    {
+      "epoch": 0.5127378957769472,
+      "grad_norm": 0.5201485746022573,
+      "learning_rate": 9.75463030512288e-06,
+      "loss": 0.7361,
+      "step": 23870
+    },
+    {
+      "epoch": 0.5129527000902178,
+      "grad_norm": 0.5161112017032182,
+      "learning_rate": 9.747815905979382e-06,
+      "loss": 0.7175,
+      "step": 23880
+    },
+    {
+      "epoch": 0.5131675044034885,
+      "grad_norm": 0.5194134872899566,
+      "learning_rate": 9.741001624012706e-06,
+      "loss": 0.7306,
+      "step": 23890
+    },
+    {
+      "epoch": 0.513382308716759,
+      "grad_norm": 0.501372178188598,
+      "learning_rate": 9.734187462389086e-06,
+      "loss": 0.7359,
+      "step": 23900
+    },
+    {
+      "epoch": 0.5135971130300296,
+      "grad_norm": 0.5021971430887777,
+      "learning_rate": 9.72737342427471e-06,
+      "loss": 0.7423,
+      "step": 23910
+    },
+    {
+      "epoch": 0.5138119173433002,
+      "grad_norm": 0.501620537070542,
+      "learning_rate": 9.720559512835708e-06,
+      "loss": 0.7262,
+      "step": 23920
+    },
+    {
+      "epoch": 0.5140267216565708,
+      "grad_norm": 0.5022999836926363,
+      "learning_rate": 9.713745731238147e-06,
+      "loss": 0.7331,
+      "step": 23930
+    },
+    {
+      "epoch": 0.5142415259698415,
+      "grad_norm": 0.48851644845547765,
+      "learning_rate": 9.706932082648043e-06,
+      "loss": 0.7392,
+      "step": 23940
+    },
+    {
+      "epoch": 0.5144563302831121,
+      "grad_norm": 0.5066291795790192,
+      "learning_rate": 9.700118570231333e-06,
+      "loss": 0.7338,
+      "step": 23950
+    },
+    {
+      "epoch": 0.5146711345963827,
+      "grad_norm": 0.5589181767659236,
+      "learning_rate": 9.693305197153914e-06,
+      "loss": 0.7272,
+      "step": 23960
+    },
+    {
+      "epoch": 0.5148859389096533,
+      "grad_norm": 0.5241863149299513,
+      "learning_rate": 9.686491966581598e-06,
+      "loss": 0.7308,
+      "step": 23970
+    },
+    {
+      "epoch": 0.5151007432229239,
+      "grad_norm": 0.540530773434582,
+      "learning_rate": 9.679678881680138e-06,
+      "loss": 0.7351,
+      "step": 23980
+    },
+    {
+      "epoch": 0.5153155475361946,
+      "grad_norm": 0.5390573047706596,
+      "learning_rate": 9.672865945615225e-06,
+      "loss": 0.7376,
+      "step": 23990
+    },
+    {
+      "epoch": 0.5155303518494652,
+      "grad_norm": 0.5375969313392924,
+      "learning_rate": 9.666053161552467e-06,
+      "loss": 0.7415,
+      "step": 24000
+    },
+    {
+      "epoch": 0.5157451561627358,
+      "grad_norm": 0.5327773082890134,
+      "learning_rate": 9.65924053265742e-06,
+      "loss": 0.7533,
+      "step": 24010
+    },
+    {
+      "epoch": 0.5159599604760063,
+      "grad_norm": 0.5150496428189917,
+      "learning_rate": 9.652428062095553e-06,
+      "loss": 0.7342,
+      "step": 24020
+    },
+    {
+      "epoch": 0.5161747647892769,
+      "grad_norm": 0.5334047289102489,
+      "learning_rate": 9.645615753032264e-06,
+      "loss": 0.7336,
+      "step": 24030
+    },
+    {
+      "epoch": 0.5163895691025476,
+      "grad_norm": 0.4961872565305141,
+      "learning_rate": 9.638803608632883e-06,
+      "loss": 0.7264,
+      "step": 24040
+    },
+    {
+      "epoch": 0.5166043734158182,
+      "grad_norm": 0.4911939322989355,
+      "learning_rate": 9.631991632062652e-06,
+      "loss": 0.7257,
+      "step": 24050
+    },
+    {
+      "epoch": 0.5168191777290888,
+      "grad_norm": 0.5200979968551847,
+      "learning_rate": 9.625179826486752e-06,
+      "loss": 0.7361,
+      "step": 24060
+    },
+    {
+      "epoch": 0.5170339820423594,
+      "grad_norm": 0.5127022024427715,
+      "learning_rate": 9.618368195070265e-06,
+      "loss": 0.7463,
+      "step": 24070
+    },
+    {
+      "epoch": 0.51724878635563,
+      "grad_norm": 0.5185873242375817,
+      "learning_rate": 9.611556740978208e-06,
+      "loss": 0.741,
+      "step": 24080
+    },
+    {
+      "epoch": 0.5174635906689007,
+      "grad_norm": 0.5115026188640582,
+      "learning_rate": 9.604745467375507e-06,
+      "loss": 0.727,
+      "step": 24090
+    },
+    {
+      "epoch": 0.5176783949821713,
+      "grad_norm": 0.5114651615343405,
+      "learning_rate": 9.597934377427e-06,
+      "loss": 0.7401,
+      "step": 24100
+    },
+    {
+      "epoch": 0.5178931992954419,
+      "grad_norm": 0.5155760878770381,
+      "learning_rate": 9.591123474297456e-06,
+      "loss": 0.7316,
+      "step": 24110
+    },
+    {
+      "epoch": 0.5181080036087125,
+      "grad_norm": 0.5207602050012875,
+      "learning_rate": 9.584312761151537e-06,
+      "loss": 0.7303,
+      "step": 24120
+    },
+    {
+      "epoch": 0.518322807921983,
+      "grad_norm": 0.5101795182054067,
+      "learning_rate": 9.577502241153836e-06,
+      "loss": 0.7541,
+      "step": 24130
+    },
+    {
+      "epoch": 0.5185376122352536,
+      "grad_norm": 0.5031850419195654,
+      "learning_rate": 9.570691917468841e-06,
+      "loss": 0.7178,
+      "step": 24140
+    },
+    {
+      "epoch": 0.5187524165485243,
+      "grad_norm": 0.5059459890339639,
+      "learning_rate": 9.563881793260961e-06,
+      "loss": 0.74,
+      "step": 24150
+    },
+    {
+      "epoch": 0.5189672208617949,
+      "grad_norm": 0.5252032527071196,
+      "learning_rate": 9.5570718716945e-06,
+      "loss": 0.724,
+      "step": 24160
+    },
+    {
+      "epoch": 0.5191820251750655,
+      "grad_norm": 0.49701627155963785,
+      "learning_rate": 9.55026215593367e-06,
+      "loss": 0.7352,
+      "step": 24170
+    },
+    {
+      "epoch": 0.5193968294883361,
+      "grad_norm": 0.5011050975997907,
+      "learning_rate": 9.543452649142605e-06,
+      "loss": 0.7114,
+      "step": 24180
+    },
+    {
+      "epoch": 0.5196116338016067,
+      "grad_norm": 0.5166863959564101,
+      "learning_rate": 9.536643354485315e-06,
+      "loss": 0.7287,
+      "step": 24190
+    },
+    {
+      "epoch": 0.5198264381148774,
+      "grad_norm": 0.5320535495138456,
+      "learning_rate": 9.529834275125733e-06,
+      "loss": 0.7318,
+      "step": 24200
+    },
+    {
+      "epoch": 0.520041242428148,
+      "grad_norm": 0.5427362518912576,
+      "learning_rate": 9.52302541422768e-06,
+      "loss": 0.7383,
+      "step": 24210
+    },
+    {
+      "epoch": 0.5202560467414186,
+      "grad_norm": 0.518569825254274,
+      "learning_rate": 9.516216774954876e-06,
+      "loss": 0.7255,
+      "step": 24220
+    },
+    {
+      "epoch": 0.5204708510546892,
+      "grad_norm": 0.5023215208325698,
+      "learning_rate": 9.50940836047095e-06,
+      "loss": 0.7217,
+      "step": 24230
+    },
+    {
+      "epoch": 0.5206856553679597,
+      "grad_norm": 0.5280315511540243,
+      "learning_rate": 9.50260017393941e-06,
+      "loss": 0.7396,
+      "step": 24240
+    },
+    {
+      "epoch": 0.5209004596812304,
+      "grad_norm": 0.5262868684024251,
+      "learning_rate": 9.495792218523668e-06,
+      "loss": 0.7404,
+      "step": 24250
+    },
+    {
+      "epoch": 0.521115263994501,
+      "grad_norm": 0.5399291263932274,
+      "learning_rate": 9.488984497387023e-06,
+      "loss": 0.7301,
+      "step": 24260
+    },
+    {
+      "epoch": 0.5213300683077716,
+      "grad_norm": 0.5090315167154729,
+      "learning_rate": 9.482177013692678e-06,
+      "loss": 0.7246,
+      "step": 24270
+    },
+    {
+      "epoch": 0.5215448726210422,
+      "grad_norm": 0.5056887738360717,
+      "learning_rate": 9.47536977060371e-06,
+      "loss": 0.7354,
+      "step": 24280
+    },
+    {
+      "epoch": 0.5217596769343128,
+      "grad_norm": 0.5576273098938848,
+      "learning_rate": 9.468562771283088e-06,
+      "loss": 0.7494,
+      "step": 24290
+    },
+    {
+      "epoch": 0.5219744812475835,
+      "grad_norm": 0.5192891146116362,
+      "learning_rate": 9.461756018893674e-06,
+      "loss": 0.7231,
+      "step": 24300
+    },
+    {
+      "epoch": 0.5221892855608541,
+      "grad_norm": 0.5176441563361027,
+      "learning_rate": 9.454949516598207e-06,
+      "loss": 0.7508,
+      "step": 24310
+    },
+    {
+      "epoch": 0.5224040898741247,
+      "grad_norm": 0.5440570683474872,
+      "learning_rate": 9.448143267559322e-06,
+      "loss": 0.7358,
+      "step": 24320
+    },
+    {
+      "epoch": 0.5226188941873953,
+      "grad_norm": 0.5095332879054673,
+      "learning_rate": 9.441337274939519e-06,
+      "loss": 0.7388,
+      "step": 24330
+    },
+    {
+      "epoch": 0.5228336985006659,
+      "grad_norm": 0.5097526753015743,
+      "learning_rate": 9.434531541901197e-06,
+      "loss": 0.7306,
+      "step": 24340
+    },
+    {
+      "epoch": 0.5230485028139366,
+      "grad_norm": 0.5232789277714874,
+      "learning_rate": 9.427726071606623e-06,
+      "loss": 0.7242,
+      "step": 24350
+    },
+    {
+      "epoch": 0.5232633071272071,
+      "grad_norm": 0.5054634510771875,
+      "learning_rate": 9.42092086721794e-06,
+      "loss": 0.7254,
+      "step": 24360
+    },
+    {
+      "epoch": 0.5234781114404777,
+      "grad_norm": 0.532869087530748,
+      "learning_rate": 9.41411593189718e-06,
+      "loss": 0.7344,
+      "step": 24370
+    },
+    {
+      "epoch": 0.5236929157537483,
+      "grad_norm": 0.5096100205368526,
+      "learning_rate": 9.407311268806232e-06,
+      "loss": 0.7313,
+      "step": 24380
+    },
+    {
+      "epoch": 0.5239077200670189,
+      "grad_norm": 0.5296911724090648,
+      "learning_rate": 9.40050688110688e-06,
+      "loss": 0.7265,
+      "step": 24390
+    },
+    {
+      "epoch": 0.5241225243802896,
+      "grad_norm": 0.5213949606620313,
+      "learning_rate": 9.393702771960763e-06,
+      "loss": 0.7306,
+      "step": 24400
+    },
+    {
+      "epoch": 0.5243373286935602,
+      "grad_norm": 0.518401803230462,
+      "learning_rate": 9.386898944529392e-06,
+      "loss": 0.7404,
+      "step": 24410
+    },
+    {
+      "epoch": 0.5245521330068308,
+      "grad_norm": 0.519834518279471,
+      "learning_rate": 9.380095401974159e-06,
+      "loss": 0.7409,
+      "step": 24420
+    },
+    {
+      "epoch": 0.5247669373201014,
+      "grad_norm": 0.5243292050356617,
+      "learning_rate": 9.373292147456309e-06,
+      "loss": 0.7264,
+      "step": 24430
+    },
+    {
+      "epoch": 0.524981741633372,
+      "grad_norm": 0.5058736452889017,
+      "learning_rate": 9.366489184136966e-06,
+      "loss": 0.7207,
+      "step": 24440
+    },
+    {
+      "epoch": 0.5251965459466426,
+      "grad_norm": 0.5213210419850763,
+      "learning_rate": 9.35968651517711e-06,
+      "loss": 0.746,
+      "step": 24450
+    },
+    {
+      "epoch": 0.5254113502599133,
+      "grad_norm": 0.4984379629009062,
+      "learning_rate": 9.35288414373759e-06,
+      "loss": 0.7372,
+      "step": 24460
+    },
+    {
+      "epoch": 0.5256261545731838,
+      "grad_norm": 0.524707210819013,
+      "learning_rate": 9.346082072979111e-06,
+      "loss": 0.7375,
+      "step": 24470
+    },
+    {
+      "epoch": 0.5258409588864544,
+      "grad_norm": 0.5200185641061693,
+      "learning_rate": 9.339280306062237e-06,
+      "loss": 0.7366,
+      "step": 24480
+    },
+    {
+      "epoch": 0.526055763199725,
+      "grad_norm": 0.5277055511224572,
+      "learning_rate": 9.332478846147404e-06,
+      "loss": 0.7346,
+      "step": 24490
+    },
+    {
+      "epoch": 0.5262705675129956,
+      "grad_norm": 0.5580648886270715,
+      "learning_rate": 9.325677696394887e-06,
+      "loss": 0.7513,
+      "step": 24500
+    },
+    {
+      "epoch": 0.5264853718262663,
+      "grad_norm": 0.4991201127455899,
+      "learning_rate": 9.318876859964832e-06,
+      "loss": 0.738,
+      "step": 24510
+    },
+    {
+      "epoch": 0.5267001761395369,
+      "grad_norm": 0.5098501461011234,
+      "learning_rate": 9.312076340017232e-06,
+      "loss": 0.7287,
+      "step": 24520
+    },
+    {
+      "epoch": 0.5269149804528075,
+      "grad_norm": 0.6021201034831747,
+      "learning_rate": 9.30527613971194e-06,
+      "loss": 0.7499,
+      "step": 24530
+    },
+    {
+      "epoch": 0.5271297847660781,
+      "grad_norm": 0.5252128416172374,
+      "learning_rate": 9.298476262208646e-06,
+      "loss": 0.7326,
+      "step": 24540
+    },
+    {
+      "epoch": 0.5273445890793487,
+      "grad_norm": 0.49998760334312775,
+      "learning_rate": 9.2916767106669e-06,
+      "loss": 0.7307,
+      "step": 24550
+    },
+    {
+      "epoch": 0.5275593933926194,
+      "grad_norm": 0.5125373956927092,
+      "learning_rate": 9.284877488246105e-06,
+      "loss": 0.7298,
+      "step": 24560
+    },
+    {
+      "epoch": 0.52777419770589,
+      "grad_norm": 0.5080343316486924,
+      "learning_rate": 9.278078598105502e-06,
+      "loss": 0.7345,
+      "step": 24570
+    },
+    {
+      "epoch": 0.5279890020191605,
+      "grad_norm": 0.5174244580842269,
+      "learning_rate": 9.271280043404185e-06,
+      "loss": 0.7332,
+      "step": 24580
+    },
+    {
+      "epoch": 0.5282038063324311,
+      "grad_norm": 0.5058297456693376,
+      "learning_rate": 9.264481827301083e-06,
+      "loss": 0.734,
+      "step": 24590
+    },
+    {
+      "epoch": 0.5284186106457017,
+      "grad_norm": 0.5292500345044558,
+      "learning_rate": 9.257683952954973e-06,
+      "loss": 0.7409,
+      "step": 24600
+    },
+    {
+      "epoch": 0.5286334149589724,
+      "grad_norm": 0.5225245379527742,
+      "learning_rate": 9.25088642352448e-06,
+      "loss": 0.7321,
+      "step": 24610
+    },
+    {
+      "epoch": 0.528848219272243,
+      "grad_norm": 0.5224232644015462,
+      "learning_rate": 9.244089242168055e-06,
+      "loss": 0.7172,
+      "step": 24620
+    },
+    {
+      "epoch": 0.5290630235855136,
+      "grad_norm": 0.5062021193853431,
+      "learning_rate": 9.237292412043997e-06,
+      "loss": 0.736,
+      "step": 24630
+    },
+    {
+      "epoch": 0.5292778278987842,
+      "grad_norm": 0.5295018228622798,
+      "learning_rate": 9.230495936310436e-06,
+      "loss": 0.7307,
+      "step": 24640
+    },
+    {
+      "epoch": 0.5294926322120548,
+      "grad_norm": 0.5267263552657305,
+      "learning_rate": 9.223699818125348e-06,
+      "loss": 0.7345,
+      "step": 24650
+    },
+    {
+      "epoch": 0.5297074365253255,
+      "grad_norm": 0.5311126333449129,
+      "learning_rate": 9.21690406064653e-06,
+      "loss": 0.7405,
+      "step": 24660
+    },
+    {
+      "epoch": 0.5299222408385961,
+      "grad_norm": 0.5153915101075538,
+      "learning_rate": 9.210108667031616e-06,
+      "loss": 0.7391,
+      "step": 24670
+    },
+    {
+      "epoch": 0.5301370451518667,
+      "grad_norm": 0.5151283148764161,
+      "learning_rate": 9.203313640438074e-06,
+      "loss": 0.7148,
+      "step": 24680
+    },
+    {
+      "epoch": 0.5303518494651372,
+      "grad_norm": 0.5089214087459585,
+      "learning_rate": 9.196518984023191e-06,
+      "loss": 0.7539,
+      "step": 24690
+    },
+    {
+      "epoch": 0.5305666537784078,
+      "grad_norm": 0.5127764385700229,
+      "learning_rate": 9.189724700944104e-06,
+      "loss": 0.734,
+      "step": 24700
+    },
+    {
+      "epoch": 0.5307814580916784,
+      "grad_norm": 0.4988269906652733,
+      "learning_rate": 9.182930794357749e-06,
+      "loss": 0.7217,
+      "step": 24710
+    },
+    {
+      "epoch": 0.5309962624049491,
+      "grad_norm": 0.5289065899803795,
+      "learning_rate": 9.176137267420913e-06,
+      "loss": 0.7228,
+      "step": 24720
+    },
+    {
+      "epoch": 0.5312110667182197,
+      "grad_norm": 0.520505526330622,
+      "learning_rate": 9.169344123290186e-06,
+      "loss": 0.7126,
+      "step": 24730
+    },
+    {
+      "epoch": 0.5314258710314903,
+      "grad_norm": 0.4998804860251492,
+      "learning_rate": 9.16255136512199e-06,
+      "loss": 0.7288,
+      "step": 24740
+    },
+    {
+      "epoch": 0.5316406753447609,
+      "grad_norm": 0.505080866954583,
+      "learning_rate": 9.155758996072568e-06,
+      "loss": 0.7338,
+      "step": 24750
+    },
+    {
+      "epoch": 0.5318554796580315,
+      "grad_norm": 0.4987180793184345,
+      "learning_rate": 9.148967019297973e-06,
+      "loss": 0.7137,
+      "step": 24760
+    },
+    {
+      "epoch": 0.5320702839713022,
+      "grad_norm": 0.5197695735697442,
+      "learning_rate": 9.142175437954095e-06,
+      "loss": 0.7395,
+      "step": 24770
+    },
+    {
+      "epoch": 0.5322850882845728,
+      "grad_norm": 0.5291148961767198,
+      "learning_rate": 9.13538425519662e-06,
+      "loss": 0.7339,
+      "step": 24780
+    },
+    {
+      "epoch": 0.5324998925978434,
+      "grad_norm": 0.5129546756881223,
+      "learning_rate": 9.128593474181058e-06,
+      "loss": 0.7253,
+      "step": 24790
+    },
+    {
+      "epoch": 0.532714696911114,
+      "grad_norm": 0.5032558888699209,
+      "learning_rate": 9.121803098062732e-06,
+      "loss": 0.728,
+      "step": 24800
+    },
+    {
+      "epoch": 0.5329295012243845,
+      "grad_norm": 0.5159297096786016,
+      "learning_rate": 9.115013129996774e-06,
+      "loss": 0.7179,
+      "step": 24810
+    },
+    {
+      "epoch": 0.5331443055376552,
+      "grad_norm": 0.5167687353856899,
+      "learning_rate": 9.108223573138133e-06,
+      "loss": 0.7252,
+      "step": 24820
+    },
+    {
+      "epoch": 0.5333591098509258,
+      "grad_norm": 0.5259491525566676,
+      "learning_rate": 9.101434430641561e-06,
+      "loss": 0.7201,
+      "step": 24830
+    },
+    {
+      "epoch": 0.5335739141641964,
+      "grad_norm": 0.5236412827950238,
+      "learning_rate": 9.09464570566162e-06,
+      "loss": 0.7237,
+      "step": 24840
+    },
+    {
+      "epoch": 0.533788718477467,
+      "grad_norm": 0.5181715919819297,
+      "learning_rate": 9.087857401352673e-06,
+      "loss": 0.7275,
+      "step": 24850
+    },
+    {
+      "epoch": 0.5340035227907376,
+      "grad_norm": 0.5281987861982105,
+      "learning_rate": 9.081069520868891e-06,
+      "loss": 0.7338,
+      "step": 24860
+    },
+    {
+      "epoch": 0.5342183271040083,
+      "grad_norm": 0.5075187850648184,
+      "learning_rate": 9.074282067364254e-06,
+      "loss": 0.7287,
+      "step": 24870
+    },
+    {
+      "epoch": 0.5344331314172789,
+      "grad_norm": 0.505210251407526,
+      "learning_rate": 9.067495043992532e-06,
+      "loss": 0.7448,
+      "step": 24880
+    },
+    {
+      "epoch": 0.5346479357305495,
+      "grad_norm": 0.5045253045306024,
+      "learning_rate": 9.060708453907312e-06,
+      "loss": 0.7124,
+      "step": 24890
+    },
+    {
+      "epoch": 0.5348627400438201,
+      "grad_norm": 0.5332241555538618,
+      "learning_rate": 9.053922300261957e-06,
+      "loss": 0.7295,
+      "step": 24900
+    },
+    {
+      "epoch": 0.5350775443570907,
+      "grad_norm": 0.5242064871975776,
+      "learning_rate": 9.047136586209646e-06,
+      "loss": 0.7305,
+      "step": 24910
+    },
+    {
+      "epoch": 0.5352923486703614,
+      "grad_norm": 0.5213752485486699,
+      "learning_rate": 9.040351314903346e-06,
+      "loss": 0.7344,
+      "step": 24920
+    },
+    {
+      "epoch": 0.5355071529836319,
+      "grad_norm": 0.5279860659235984,
+      "learning_rate": 9.033566489495815e-06,
+      "loss": 0.7149,
+      "step": 24930
+    },
+    {
+      "epoch": 0.5357219572969025,
+      "grad_norm": 0.5153059474379569,
+      "learning_rate": 9.026782113139614e-06,
+      "loss": 0.7272,
+      "step": 24940
+    },
+    {
+      "epoch": 0.5359367616101731,
+      "grad_norm": 0.502633555256086,
+      "learning_rate": 9.019998188987087e-06,
+      "loss": 0.7358,
+      "step": 24950
+    },
+    {
+      "epoch": 0.5361515659234437,
+      "grad_norm": 0.5227824018715104,
+      "learning_rate": 9.01321472019037e-06,
+      "loss": 0.7322,
+      "step": 24960
+    },
+    {
+      "epoch": 0.5363663702367144,
+      "grad_norm": 0.5096066611505108,
+      "learning_rate": 9.006431709901385e-06,
+      "loss": 0.7291,
+      "step": 24970
+    },
+    {
+      "epoch": 0.536581174549985,
+      "grad_norm": 0.5050380085409679,
+      "learning_rate": 8.999649161271851e-06,
+      "loss": 0.7383,
+      "step": 24980
+    },
+    {
+      "epoch": 0.5367959788632556,
+      "grad_norm": 0.5098544768515563,
+      "learning_rate": 8.99286707745326e-06,
+      "loss": 0.7296,
+      "step": 24990
+    },
+    {
+      "epoch": 0.5370107831765262,
+      "grad_norm": 0.49596489778805924,
+      "learning_rate": 8.986085461596892e-06,
+      "loss": 0.7448,
+      "step": 25000
+    },
+    {
+      "epoch": 0.5372255874897968,
+      "grad_norm": 0.524177361015968,
+      "learning_rate": 8.979304316853816e-06,
+      "loss": 0.7223,
+      "step": 25010
+    },
+    {
+      "epoch": 0.5374403918030674,
+      "grad_norm": 0.5275256334675215,
+      "learning_rate": 8.972523646374868e-06,
+      "loss": 0.7155,
+      "step": 25020
+    },
+    {
+      "epoch": 0.537655196116338,
+      "grad_norm": 0.5230415273183412,
+      "learning_rate": 8.965743453310682e-06,
+      "loss": 0.7244,
+      "step": 25030
+    },
+    {
+      "epoch": 0.5378700004296086,
+      "grad_norm": 0.5056528573977891,
+      "learning_rate": 8.958963740811657e-06,
+      "loss": 0.7373,
+      "step": 25040
+    },
+    {
+      "epoch": 0.5380848047428792,
+      "grad_norm": 0.5129477293931691,
+      "learning_rate": 8.952184512027971e-06,
+      "loss": 0.7409,
+      "step": 25050
+    },
+    {
+      "epoch": 0.5382996090561498,
+      "grad_norm": 0.5254219239381808,
+      "learning_rate": 8.94540577010958e-06,
+      "loss": 0.7289,
+      "step": 25060
+    },
+    {
+      "epoch": 0.5385144133694204,
+      "grad_norm": 0.5137037649048417,
+      "learning_rate": 8.938627518206207e-06,
+      "loss": 0.7346,
+      "step": 25070
+    },
+    {
+      "epoch": 0.5387292176826911,
+      "grad_norm": 0.501166734367294,
+      "learning_rate": 8.931849759467364e-06,
+      "loss": 0.7376,
+      "step": 25080
+    },
+    {
+      "epoch": 0.5389440219959617,
+      "grad_norm": 0.5158686563553051,
+      "learning_rate": 8.925072497042312e-06,
+      "loss": 0.7246,
+      "step": 25090
+    },
+    {
+      "epoch": 0.5391588263092323,
+      "grad_norm": 0.5064258891872856,
+      "learning_rate": 8.9182957340801e-06,
+      "loss": 0.7224,
+      "step": 25100
+    },
+    {
+      "epoch": 0.5393736306225029,
+      "grad_norm": 0.5109514152684812,
+      "learning_rate": 8.911519473729533e-06,
+      "loss": 0.7258,
+      "step": 25110
+    },
+    {
+      "epoch": 0.5395884349357735,
+      "grad_norm": 0.4965294744044547,
+      "learning_rate": 8.904743719139184e-06,
+      "loss": 0.7231,
+      "step": 25120
+    },
+    {
+      "epoch": 0.5398032392490442,
+      "grad_norm": 0.5163978388155885,
+      "learning_rate": 8.897968473457397e-06,
+      "loss": 0.7213,
+      "step": 25130
+    },
+    {
+      "epoch": 0.5400180435623148,
+      "grad_norm": 0.5179202460764463,
+      "learning_rate": 8.89119373983227e-06,
+      "loss": 0.7333,
+      "step": 25140
+    },
+    {
+      "epoch": 0.5402328478755853,
+      "grad_norm": 0.5267218520730645,
+      "learning_rate": 8.884419521411681e-06,
+      "loss": 0.7271,
+      "step": 25150
+    },
+    {
+      "epoch": 0.5404476521888559,
+      "grad_norm": 0.49970760568396944,
+      "learning_rate": 8.877645821343245e-06,
+      "loss": 0.7134,
+      "step": 25160
+    },
+    {
+      "epoch": 0.5406624565021265,
+      "grad_norm": 0.5110857110324059,
+      "learning_rate": 8.870872642774354e-06,
+      "loss": 0.7371,
+      "step": 25170
+    },
+    {
+      "epoch": 0.5408772608153972,
+      "grad_norm": 0.524857212027166,
+      "learning_rate": 8.864099988852149e-06,
+      "loss": 0.7303,
+      "step": 25180
+    },
+    {
+      "epoch": 0.5410920651286678,
+      "grad_norm": 0.5056157960738001,
+      "learning_rate": 8.857327862723524e-06,
+      "loss": 0.7163,
+      "step": 25190
+    },
+    {
+      "epoch": 0.5413068694419384,
+      "grad_norm": 0.4958979488976298,
+      "learning_rate": 8.850556267535146e-06,
+      "loss": 0.7364,
+      "step": 25200
+    },
+    {
+      "epoch": 0.541521673755209,
+      "grad_norm": 0.5132188143017179,
+      "learning_rate": 8.843785206433412e-06,
+      "loss": 0.7335,
+      "step": 25210
+    },
+    {
+      "epoch": 0.5417364780684796,
+      "grad_norm": 0.4994335318852988,
+      "learning_rate": 8.837014682564487e-06,
+      "loss": 0.7257,
+      "step": 25220
+    },
+    {
+      "epoch": 0.5419512823817503,
+      "grad_norm": 0.5384327760658743,
+      "learning_rate": 8.83024469907428e-06,
+      "loss": 0.7351,
+      "step": 25230
+    },
+    {
+      "epoch": 0.5421660866950209,
+      "grad_norm": 0.5261027642763866,
+      "learning_rate": 8.823475259108445e-06,
+      "loss": 0.7356,
+      "step": 25240
+    },
+    {
+      "epoch": 0.5423808910082915,
+      "grad_norm": 0.5060595175869693,
+      "learning_rate": 8.816706365812396e-06,
+      "loss": 0.7136,
+      "step": 25250
+    },
+    {
+      "epoch": 0.542595695321562,
+      "grad_norm": 0.5089315623590392,
+      "learning_rate": 8.809938022331279e-06,
+      "loss": 0.7582,
+      "step": 25260
+    },
+    {
+      "epoch": 0.5428104996348326,
+      "grad_norm": 0.5517010005843618,
+      "learning_rate": 8.803170231809997e-06,
+      "loss": 0.7264,
+      "step": 25270
+    },
+    {
+      "epoch": 0.5430253039481033,
+      "grad_norm": 0.5192946990016879,
+      "learning_rate": 8.796402997393179e-06,
+      "loss": 0.7338,
+      "step": 25280
+    },
+    {
+      "epoch": 0.5432401082613739,
+      "grad_norm": 0.5141174201291905,
+      "learning_rate": 8.789636322225221e-06,
+      "loss": 0.7301,
+      "step": 25290
+    },
+    {
+      "epoch": 0.5434549125746445,
+      "grad_norm": 0.5156454974533952,
+      "learning_rate": 8.782870209450234e-06,
+      "loss": 0.7386,
+      "step": 25300
+    },
+    {
+      "epoch": 0.5436697168879151,
+      "grad_norm": 0.5399434222135454,
+      "learning_rate": 8.776104662212077e-06,
+      "loss": 0.7297,
+      "step": 25310
+    },
+    {
+      "epoch": 0.5438845212011857,
+      "grad_norm": 0.5080337339754446,
+      "learning_rate": 8.769339683654358e-06,
+      "loss": 0.7268,
+      "step": 25320
+    },
+    {
+      "epoch": 0.5440993255144563,
+      "grad_norm": 0.5211965166488158,
+      "learning_rate": 8.762575276920403e-06,
+      "loss": 0.72,
+      "step": 25330
+    },
+    {
+      "epoch": 0.544314129827727,
+      "grad_norm": 0.511040069798665,
+      "learning_rate": 8.755811445153282e-06,
+      "loss": 0.7245,
+      "step": 25340
+    },
+    {
+      "epoch": 0.5445289341409976,
+      "grad_norm": 0.5496977731743303,
+      "learning_rate": 8.749048191495787e-06,
+      "loss": 0.7362,
+      "step": 25350
+    },
+    {
+      "epoch": 0.5447437384542682,
+      "grad_norm": 0.497836078774839,
+      "learning_rate": 8.742285519090465e-06,
+      "loss": 0.7369,
+      "step": 25360
+    },
+    {
+      "epoch": 0.5449585427675387,
+      "grad_norm": 0.4958506383207596,
+      "learning_rate": 8.735523431079567e-06,
+      "loss": 0.6986,
+      "step": 25370
+    },
+    {
+      "epoch": 0.5451733470808093,
+      "grad_norm": 0.5023433708525002,
+      "learning_rate": 8.728761930605086e-06,
+      "loss": 0.7225,
+      "step": 25380
+    },
+    {
+      "epoch": 0.54538815139408,
+      "grad_norm": 0.5115175514759246,
+      "learning_rate": 8.72200102080874e-06,
+      "loss": 0.7311,
+      "step": 25390
+    },
+    {
+      "epoch": 0.5456029557073506,
+      "grad_norm": 0.49400314000603524,
+      "learning_rate": 8.715240704831965e-06,
+      "loss": 0.711,
+      "step": 25400
+    },
+    {
+      "epoch": 0.5458177600206212,
+      "grad_norm": 0.5281302488810649,
+      "learning_rate": 8.70848098581594e-06,
+      "loss": 0.7307,
+      "step": 25410
+    },
+    {
+      "epoch": 0.5460325643338918,
+      "grad_norm": 0.5253108140644606,
+      "learning_rate": 8.701721866901548e-06,
+      "loss": 0.7542,
+      "step": 25420
+    },
+    {
+      "epoch": 0.5462473686471624,
+      "grad_norm": 0.49677506907339725,
+      "learning_rate": 8.694963351229395e-06,
+      "loss": 0.7346,
+      "step": 25430
+    },
+    {
+      "epoch": 0.5464621729604331,
+      "grad_norm": 0.49156308079799516,
+      "learning_rate": 8.68820544193982e-06,
+      "loss": 0.7301,
+      "step": 25440
+    },
+    {
+      "epoch": 0.5466769772737037,
+      "grad_norm": 0.5054408047803833,
+      "learning_rate": 8.681448142172862e-06,
+      "loss": 0.717,
+      "step": 25450
+    },
+    {
+      "epoch": 0.5468917815869743,
+      "grad_norm": 0.5250496149552367,
+      "learning_rate": 8.674691455068296e-06,
+      "loss": 0.7208,
+      "step": 25460
+    },
+    {
+      "epoch": 0.5471065859002449,
+      "grad_norm": 3.4136357360926253,
+      "learning_rate": 8.667935383765595e-06,
+      "loss": 0.735,
+      "step": 25470
+    },
+    {
+      "epoch": 0.5473213902135154,
+      "grad_norm": 0.4871497116063486,
+      "learning_rate": 8.66117993140396e-06,
+      "loss": 0.7342,
+      "step": 25480
+    },
+    {
+      "epoch": 0.5475361945267861,
+      "grad_norm": 0.5209915171759771,
+      "learning_rate": 8.654425101122296e-06,
+      "loss": 0.7336,
+      "step": 25490
+    },
+    {
+      "epoch": 0.5477509988400567,
+      "grad_norm": 0.5156635252595837,
+      "learning_rate": 8.647670896059216e-06,
+      "loss": 0.7293,
+      "step": 25500
+    },
+    {
+      "epoch": 0.5479658031533273,
+      "grad_norm": 0.5152409120879327,
+      "learning_rate": 8.640917319353055e-06,
+      "loss": 0.7273,
+      "step": 25510
+    },
+    {
+      "epoch": 0.5481806074665979,
+      "grad_norm": 0.5416207574425208,
+      "learning_rate": 8.634164374141838e-06,
+      "loss": 0.7258,
+      "step": 25520
+    },
+    {
+      "epoch": 0.5483954117798685,
+      "grad_norm": 0.515769700298451,
+      "learning_rate": 8.62741206356332e-06,
+      "loss": 0.7196,
+      "step": 25530
+    },
+    {
+      "epoch": 0.5486102160931392,
+      "grad_norm": 0.5161072492397152,
+      "learning_rate": 8.62066039075494e-06,
+      "loss": 0.7146,
+      "step": 25540
+    },
+    {
+      "epoch": 0.5488250204064098,
+      "grad_norm": 0.5188545967485997,
+      "learning_rate": 8.61390935885385e-06,
+      "loss": 0.7343,
+      "step": 25550
+    },
+    {
+      "epoch": 0.5490398247196804,
+      "grad_norm": 0.5220576998409723,
+      "learning_rate": 8.607158970996905e-06,
+      "loss": 0.7225,
+      "step": 25560
+    },
+    {
+      "epoch": 0.549254629032951,
+      "grad_norm": 0.4933476797650889,
+      "learning_rate": 8.600409230320652e-06,
+      "loss": 0.727,
+      "step": 25570
+    },
+    {
+      "epoch": 0.5494694333462216,
+      "grad_norm": 0.5360731957654817,
+      "learning_rate": 8.593660139961354e-06,
+      "loss": 0.7304,
+      "step": 25580
+    },
+    {
+      "epoch": 0.5496842376594921,
+      "grad_norm": 0.4967871791391136,
+      "learning_rate": 8.586911703054953e-06,
+      "loss": 0.718,
+      "step": 25590
+    },
+    {
+      "epoch": 0.5498990419727628,
+      "grad_norm": 0.5094344264185746,
+      "learning_rate": 8.580163922737107e-06,
+      "loss": 0.7351,
+      "step": 25600
+    },
+    {
+      "epoch": 0.5501138462860334,
+      "grad_norm": 0.4998333921079445,
+      "learning_rate": 8.573416802143147e-06,
+      "loss": 0.7362,
+      "step": 25610
+    },
+    {
+      "epoch": 0.550328650599304,
+      "grad_norm": 0.49839738637776776,
+      "learning_rate": 8.56667034440811e-06,
+      "loss": 0.7343,
+      "step": 25620
+    },
+    {
+      "epoch": 0.5505434549125746,
+      "grad_norm": 0.5122812659780278,
+      "learning_rate": 8.559924552666731e-06,
+      "loss": 0.7299,
+      "step": 25630
+    },
+    {
+      "epoch": 0.5507582592258452,
+      "grad_norm": 0.510657449522192,
+      "learning_rate": 8.55317943005342e-06,
+      "loss": 0.7323,
+      "step": 25640
+    },
+    {
+      "epoch": 0.5509730635391159,
+      "grad_norm": 0.5029028813355643,
+      "learning_rate": 8.546434979702293e-06,
+      "loss": 0.7287,
+      "step": 25650
+    },
+    {
+      "epoch": 0.5511878678523865,
+      "grad_norm": 0.5228476287215734,
+      "learning_rate": 8.539691204747134e-06,
+      "loss": 0.7321,
+      "step": 25660
+    },
+    {
+      "epoch": 0.5514026721656571,
+      "grad_norm": 0.5359812041522654,
+      "learning_rate": 8.532948108321433e-06,
+      "loss": 0.742,
+      "step": 25670
+    },
+    {
+      "epoch": 0.5516174764789277,
+      "grad_norm": 0.5130595618709932,
+      "learning_rate": 8.526205693558353e-06,
+      "loss": 0.7319,
+      "step": 25680
+    },
+    {
+      "epoch": 0.5518322807921983,
+      "grad_norm": 0.4996255988319911,
+      "learning_rate": 8.51946396359074e-06,
+      "loss": 0.7407,
+      "step": 25690
+    },
+    {
+      "epoch": 0.552047085105469,
+      "grad_norm": 0.49407356190346213,
+      "learning_rate": 8.51272292155113e-06,
+      "loss": 0.717,
+      "step": 25700
+    },
+    {
+      "epoch": 0.5522618894187395,
+      "grad_norm": 0.4898737936647392,
+      "learning_rate": 8.50598257057173e-06,
+      "loss": 0.747,
+      "step": 25710
+    },
+    {
+      "epoch": 0.5524766937320101,
+      "grad_norm": 0.5212613411297291,
+      "learning_rate": 8.499242913784436e-06,
+      "loss": 0.733,
+      "step": 25720
+    },
+    {
+      "epoch": 0.5526914980452807,
+      "grad_norm": 0.5428952625628243,
+      "learning_rate": 8.492503954320804e-06,
+      "loss": 0.7235,
+      "step": 25730
+    },
+    {
+      "epoch": 0.5529063023585513,
+      "grad_norm": 0.5179866204385108,
+      "learning_rate": 8.485765695312091e-06,
+      "loss": 0.7353,
+      "step": 25740
+    },
+    {
+      "epoch": 0.553121106671822,
+      "grad_norm": 0.5143508651285503,
+      "learning_rate": 8.479028139889209e-06,
+      "loss": 0.7092,
+      "step": 25750
+    },
+    {
+      "epoch": 0.5533359109850926,
+      "grad_norm": 0.4899586729640197,
+      "learning_rate": 8.472291291182747e-06,
+      "loss": 0.7214,
+      "step": 25760
+    },
+    {
+      "epoch": 0.5535507152983632,
+      "grad_norm": 0.5104765609615125,
+      "learning_rate": 8.465555152322971e-06,
+      "loss": 0.7343,
+      "step": 25770
+    },
+    {
+      "epoch": 0.5537655196116338,
+      "grad_norm": 0.4936010581482795,
+      "learning_rate": 8.45881972643981e-06,
+      "loss": 0.7201,
+      "step": 25780
+    },
+    {
+      "epoch": 0.5539803239249044,
+      "grad_norm": 0.5103205277286559,
+      "learning_rate": 8.452085016662873e-06,
+      "loss": 0.7249,
+      "step": 25790
+    },
+    {
+      "epoch": 0.5541951282381751,
+      "grad_norm": 0.5337997235376728,
+      "learning_rate": 8.445351026121425e-06,
+      "loss": 0.723,
+      "step": 25800
+    },
+    {
+      "epoch": 0.5544099325514457,
+      "grad_norm": 0.5330888068677769,
+      "learning_rate": 8.438617757944396e-06,
+      "loss": 0.7368,
+      "step": 25810
+    },
+    {
+      "epoch": 0.5546247368647163,
+      "grad_norm": 0.507422198237814,
+      "learning_rate": 8.431885215260393e-06,
+      "loss": 0.7277,
+      "step": 25820
+    },
+    {
+      "epoch": 0.5548395411779868,
+      "grad_norm": 0.5092852644195597,
+      "learning_rate": 8.42515340119767e-06,
+      "loss": 0.7141,
+      "step": 25830
+    },
+    {
+      "epoch": 0.5550543454912574,
+      "grad_norm": 0.4983369620376665,
+      "learning_rate": 8.418422318884158e-06,
+      "loss": 0.7213,
+      "step": 25840
+    },
+    {
+      "epoch": 0.5552691498045281,
+      "grad_norm": 0.5057768818854864,
+      "learning_rate": 8.411691971447437e-06,
+      "loss": 0.7204,
+      "step": 25850
+    },
+    {
+      "epoch": 0.5554839541177987,
+      "grad_norm": 0.4957062726709344,
+      "learning_rate": 8.40496236201475e-06,
+      "loss": 0.7239,
+      "step": 25860
+    },
+    {
+      "epoch": 0.5556987584310693,
+      "grad_norm": 0.5098362148369251,
+      "learning_rate": 8.398233493712997e-06,
+      "loss": 0.7229,
+      "step": 25870
+    },
+    {
+      "epoch": 0.5559135627443399,
+      "grad_norm": 0.540586305639984,
+      "learning_rate": 8.391505369668725e-06,
+      "loss": 0.7223,
+      "step": 25880
+    },
+    {
+      "epoch": 0.5561283670576105,
+      "grad_norm": 0.5164869278284776,
+      "learning_rate": 8.384777993008154e-06,
+      "loss": 0.7241,
+      "step": 25890
+    },
+    {
+      "epoch": 0.5563431713708811,
+      "grad_norm": 0.5178900728279536,
+      "learning_rate": 8.378051366857137e-06,
+      "loss": 0.7324,
+      "step": 25900
+    },
+    {
+      "epoch": 0.5565579756841518,
+      "grad_norm": 0.5093380002448458,
+      "learning_rate": 8.371325494341193e-06,
+      "loss": 0.7337,
+      "step": 25910
+    },
+    {
+      "epoch": 0.5567727799974224,
+      "grad_norm": 0.5228044379168946,
+      "learning_rate": 8.364600378585482e-06,
+      "loss": 0.7399,
+      "step": 25920
+    },
+    {
+      "epoch": 0.556987584310693,
+      "grad_norm": 0.5144671939481396,
+      "learning_rate": 8.357876022714816e-06,
+      "loss": 0.7254,
+      "step": 25930
+    },
+    {
+      "epoch": 0.5572023886239635,
+      "grad_norm": 0.4986034297019067,
+      "learning_rate": 8.351152429853653e-06,
+      "loss": 0.7106,
+      "step": 25940
+    },
+    {
+      "epoch": 0.5574171929372341,
+      "grad_norm": 0.5023603101794233,
+      "learning_rate": 8.34442960312609e-06,
+      "loss": 0.7226,
+      "step": 25950
+    },
+    {
+      "epoch": 0.5576319972505048,
+      "grad_norm": 0.5209318452676072,
+      "learning_rate": 8.337707545655886e-06,
+      "loss": 0.7238,
+      "step": 25960
+    },
+    {
+      "epoch": 0.5578468015637754,
+      "grad_norm": 0.4926074970141638,
+      "learning_rate": 8.330986260566424e-06,
+      "loss": 0.7276,
+      "step": 25970
+    },
+    {
+      "epoch": 0.558061605877046,
+      "grad_norm": 0.4958510123483159,
+      "learning_rate": 8.324265750980738e-06,
+      "loss": 0.7148,
+      "step": 25980
+    },
+    {
+      "epoch": 0.5582764101903166,
+      "grad_norm": 0.5075764876809195,
+      "learning_rate": 8.317546020021498e-06,
+      "loss": 0.7155,
+      "step": 25990
+    },
+    {
+      "epoch": 0.5584912145035872,
+      "grad_norm": 0.5141742208103084,
+      "learning_rate": 8.310827070811008e-06,
+      "loss": 0.7385,
+      "step": 26000
+    },
+    {
+      "epoch": 0.5587060188168579,
+      "grad_norm": 0.5084465568638146,
+      "learning_rate": 8.30410890647122e-06,
+      "loss": 0.7202,
+      "step": 26010
+    },
+    {
+      "epoch": 0.5589208231301285,
+      "grad_norm": 0.502206471808609,
+      "learning_rate": 8.297391530123713e-06,
+      "loss": 0.7237,
+      "step": 26020
+    },
+    {
+      "epoch": 0.5591356274433991,
+      "grad_norm": 0.5086015295612804,
+      "learning_rate": 8.290674944889705e-06,
+      "loss": 0.7387,
+      "step": 26030
+    },
+    {
+      "epoch": 0.5593504317566697,
+      "grad_norm": 0.5252472355926682,
+      "learning_rate": 8.283959153890037e-06,
+      "loss": 0.728,
+      "step": 26040
+    },
+    {
+      "epoch": 0.5595652360699402,
+      "grad_norm": 0.4975594240387986,
+      "learning_rate": 8.277244160245196e-06,
+      "loss": 0.7213,
+      "step": 26050
+    },
+    {
+      "epoch": 0.5597800403832109,
+      "grad_norm": 0.4978146880109962,
+      "learning_rate": 8.270529967075284e-06,
+      "loss": 0.7173,
+      "step": 26060
+    },
+    {
+      "epoch": 0.5599948446964815,
+      "grad_norm": 0.5268438508793595,
+      "learning_rate": 8.263816577500034e-06,
+      "loss": 0.7371,
+      "step": 26070
+    },
+    {
+      "epoch": 0.5602096490097521,
+      "grad_norm": 0.5137983659248461,
+      "learning_rate": 8.257103994638817e-06,
+      "loss": 0.7197,
+      "step": 26080
+    },
+    {
+      "epoch": 0.5604244533230227,
+      "grad_norm": 0.5194520105890317,
+      "learning_rate": 8.250392221610612e-06,
+      "loss": 0.7297,
+      "step": 26090
+    },
+    {
+      "epoch": 0.5606392576362933,
+      "grad_norm": 0.5133858765119592,
+      "learning_rate": 8.24368126153404e-06,
+      "loss": 0.732,
+      "step": 26100
+    },
+    {
+      "epoch": 0.560854061949564,
+      "grad_norm": 0.5029432924404077,
+      "learning_rate": 8.236971117527324e-06,
+      "loss": 0.728,
+      "step": 26110
+    },
+    {
+      "epoch": 0.5610688662628346,
+      "grad_norm": 0.5081831178394076,
+      "learning_rate": 8.230261792708328e-06,
+      "loss": 0.7324,
+      "step": 26120
+    },
+    {
+      "epoch": 0.5612836705761052,
+      "grad_norm": 0.5046477194727244,
+      "learning_rate": 8.223553290194521e-06,
+      "loss": 0.7363,
+      "step": 26130
+    },
+    {
+      "epoch": 0.5614984748893758,
+      "grad_norm": 0.4956098306676567,
+      "learning_rate": 8.216845613102995e-06,
+      "loss": 0.7245,
+      "step": 26140
+    },
+    {
+      "epoch": 0.5617132792026464,
+      "grad_norm": 0.5157536272581615,
+      "learning_rate": 8.21013876455046e-06,
+      "loss": 0.7117,
+      "step": 26150
+    },
+    {
+      "epoch": 0.561928083515917,
+      "grad_norm": 0.5168564344065179,
+      "learning_rate": 8.203432747653234e-06,
+      "loss": 0.7245,
+      "step": 26160
+    },
+    {
+      "epoch": 0.5621428878291876,
+      "grad_norm": 0.505045763218236,
+      "learning_rate": 8.19672756552726e-06,
+      "loss": 0.7176,
+      "step": 26170
+    },
+    {
+      "epoch": 0.5623576921424582,
+      "grad_norm": 0.5239596568078339,
+      "learning_rate": 8.190023221288088e-06,
+      "loss": 0.7326,
+      "step": 26180
+    },
+    {
+      "epoch": 0.5625724964557288,
+      "grad_norm": 0.5302666154295604,
+      "learning_rate": 8.183319718050873e-06,
+      "loss": 0.7422,
+      "step": 26190
+    },
+    {
+      "epoch": 0.5627873007689994,
+      "grad_norm": 0.5003281299438962,
+      "learning_rate": 8.176617058930385e-06,
+      "loss": 0.7284,
+      "step": 26200
+    },
+    {
+      "epoch": 0.56300210508227,
+      "grad_norm": 0.49200878286488725,
+      "learning_rate": 8.169915247040998e-06,
+      "loss": 0.7313,
+      "step": 26210
+    },
+    {
+      "epoch": 0.5632169093955407,
+      "grad_norm": 0.49815572715612766,
+      "learning_rate": 8.163214285496704e-06,
+      "loss": 0.7216,
+      "step": 26220
+    },
+    {
+      "epoch": 0.5634317137088113,
+      "grad_norm": 0.5142148330083207,
+      "learning_rate": 8.15651417741108e-06,
+      "loss": 0.7239,
+      "step": 26230
+    },
+    {
+      "epoch": 0.5636465180220819,
+      "grad_norm": 0.49075756148020155,
+      "learning_rate": 8.149814925897327e-06,
+      "loss": 0.7295,
+      "step": 26240
+    },
+    {
+      "epoch": 0.5638613223353525,
+      "grad_norm": 0.5246558400473214,
+      "learning_rate": 8.143116534068231e-06,
+      "loss": 0.726,
+      "step": 26250
+    },
+    {
+      "epoch": 0.564076126648623,
+      "grad_norm": 0.5246290515230206,
+      "learning_rate": 8.136419005036186e-06,
+      "loss": 0.7182,
+      "step": 26260
+    },
+    {
+      "epoch": 0.5642909309618938,
+      "grad_norm": 0.5036297463223443,
+      "learning_rate": 8.129722341913192e-06,
+      "loss": 0.7384,
+      "step": 26270
+    },
+    {
+      "epoch": 0.5645057352751643,
+      "grad_norm": 0.527968106908949,
+      "learning_rate": 8.12302654781083e-06,
+      "loss": 0.7215,
+      "step": 26280
+    },
+    {
+      "epoch": 0.5647205395884349,
+      "grad_norm": 0.5225876564331369,
+      "learning_rate": 8.116331625840297e-06,
+      "loss": 0.7244,
+      "step": 26290
+    },
+    {
+      "epoch": 0.5649353439017055,
+      "grad_norm": 0.5436618520471259,
+      "learning_rate": 8.109637579112368e-06,
+      "loss": 0.7398,
+      "step": 26300
+    },
+    {
+      "epoch": 0.5651501482149761,
+      "grad_norm": 0.5148458404115451,
+      "learning_rate": 8.102944410737422e-06,
+      "loss": 0.7273,
+      "step": 26310
+    },
+    {
+      "epoch": 0.5653649525282468,
+      "grad_norm": 0.5015994626598883,
+      "learning_rate": 8.096252123825422e-06,
+      "loss": 0.7157,
+      "step": 26320
+    },
+    {
+      "epoch": 0.5655797568415174,
+      "grad_norm": 0.505696959132355,
+      "learning_rate": 8.089560721485922e-06,
+      "loss": 0.7234,
+      "step": 26330
+    },
+    {
+      "epoch": 0.565794561154788,
+      "grad_norm": 0.5359793879908359,
+      "learning_rate": 8.082870206828078e-06,
+      "loss": 0.72,
+      "step": 26340
+    },
+    {
+      "epoch": 0.5660093654680586,
+      "grad_norm": 0.5110305690812829,
+      "learning_rate": 8.076180582960618e-06,
+      "loss": 0.7234,
+      "step": 26350
+    },
+    {
+      "epoch": 0.5662241697813292,
+      "grad_norm": 0.498824135773946,
+      "learning_rate": 8.069491852991861e-06,
+      "loss": 0.7137,
+      "step": 26360
+    },
+    {
+      "epoch": 0.5664389740945999,
+      "grad_norm": 0.5071066304511312,
+      "learning_rate": 8.062804020029716e-06,
+      "loss": 0.7257,
+      "step": 26370
+    },
+    {
+      "epoch": 0.5666537784078705,
+      "grad_norm": 0.5382565877388203,
+      "learning_rate": 8.056117087181663e-06,
+      "loss": 0.7233,
+      "step": 26380
+    },
+    {
+      "epoch": 0.566868582721141,
+      "grad_norm": 0.507293946375138,
+      "learning_rate": 8.04943105755478e-06,
+      "loss": 0.7283,
+      "step": 26390
+    },
+    {
+      "epoch": 0.5670833870344116,
+      "grad_norm": 0.5280130888890053,
+      "learning_rate": 8.04274593425571e-06,
+      "loss": 0.7317,
+      "step": 26400
+    },
+    {
+      "epoch": 0.5672981913476822,
+      "grad_norm": 0.518867511724086,
+      "learning_rate": 8.03606172039069e-06,
+      "loss": 0.7167,
+      "step": 26410
+    },
+    {
+      "epoch": 0.5675129956609529,
+      "grad_norm": 0.5060431523298652,
+      "learning_rate": 8.029378419065515e-06,
+      "loss": 0.717,
+      "step": 26420
+    },
+    {
+      "epoch": 0.5677277999742235,
+      "grad_norm": 0.5063699893804817,
+      "learning_rate": 8.022696033385576e-06,
+      "loss": 0.7272,
+      "step": 26430
+    },
+    {
+      "epoch": 0.5679426042874941,
+      "grad_norm": 0.5240786574435802,
+      "learning_rate": 8.016014566455827e-06,
+      "loss": 0.7104,
+      "step": 26440
+    },
+    {
+      "epoch": 0.5681574086007647,
+      "grad_norm": 0.5023416444895722,
+      "learning_rate": 8.009334021380797e-06,
+      "loss": 0.7271,
+      "step": 26450
+    },
+    {
+      "epoch": 0.5683722129140353,
+      "grad_norm": 0.5364383380095676,
+      "learning_rate": 8.002654401264587e-06,
+      "loss": 0.7263,
+      "step": 26460
+    },
+    {
+      "epoch": 0.568587017227306,
+      "grad_norm": 0.5281110306818434,
+      "learning_rate": 7.995975709210869e-06,
+      "loss": 0.7145,
+      "step": 26470
+    },
+    {
+      "epoch": 0.5688018215405766,
+      "grad_norm": 0.5172649393256681,
+      "learning_rate": 7.989297948322885e-06,
+      "loss": 0.7279,
+      "step": 26480
+    },
+    {
+      "epoch": 0.5690166258538472,
+      "grad_norm": 0.4984810683384183,
+      "learning_rate": 7.982621121703438e-06,
+      "loss": 0.7017,
+      "step": 26490
+    },
+    {
+      "epoch": 0.5692314301671177,
+      "grad_norm": 0.5013994336883097,
+      "learning_rate": 7.97594523245491e-06,
+      "loss": 0.7052,
+      "step": 26500
+    },
+    {
+      "epoch": 0.5694462344803883,
+      "grad_norm": 0.4962092438638248,
+      "learning_rate": 7.969270283679233e-06,
+      "loss": 0.7145,
+      "step": 26510
+    },
+    {
+      "epoch": 0.5696610387936589,
+      "grad_norm": 0.5113556248884656,
+      "learning_rate": 7.96259627847791e-06,
+      "loss": 0.7182,
+      "step": 26520
+    },
+    {
+      "epoch": 0.5698758431069296,
+      "grad_norm": 0.5049922401075675,
+      "learning_rate": 7.955923219952002e-06,
+      "loss": 0.7093,
+      "step": 26530
+    },
+    {
+      "epoch": 0.5700906474202002,
+      "grad_norm": 0.5033957581062982,
+      "learning_rate": 7.949251111202132e-06,
+      "loss": 0.7274,
+      "step": 26540
+    },
+    {
+      "epoch": 0.5703054517334708,
+      "grad_norm": 0.5356557064198388,
+      "learning_rate": 7.942579955328485e-06,
+      "loss": 0.7241,
+      "step": 26550
+    },
+    {
+      "epoch": 0.5705202560467414,
+      "grad_norm": 0.5345233979565142,
+      "learning_rate": 7.935909755430797e-06,
+      "loss": 0.7334,
+      "step": 26560
+    },
+    {
+      "epoch": 0.570735060360012,
+      "grad_norm": 0.49378966951960895,
+      "learning_rate": 7.92924051460836e-06,
+      "loss": 0.7388,
+      "step": 26570
+    },
+    {
+      "epoch": 0.5709498646732827,
+      "grad_norm": 0.5077813116094394,
+      "learning_rate": 7.92257223596003e-06,
+      "loss": 0.7226,
+      "step": 26580
+    },
+    {
+      "epoch": 0.5711646689865533,
+      "grad_norm": 0.5242154347144848,
+      "learning_rate": 7.9159049225842e-06,
+      "loss": 0.7293,
+      "step": 26590
+    },
+    {
+      "epoch": 0.5713794732998239,
+      "grad_norm": 0.4939088384409381,
+      "learning_rate": 7.90923857757883e-06,
+      "loss": 0.7351,
+      "step": 26600
+    },
+    {
+      "epoch": 0.5715942776130944,
+      "grad_norm": 0.49854560906783285,
+      "learning_rate": 7.902573204041422e-06,
+      "loss": 0.7112,
+      "step": 26610
+    },
+    {
+      "epoch": 0.571809081926365,
+      "grad_norm": 0.489407868294081,
+      "learning_rate": 7.895908805069026e-06,
+      "loss": 0.731,
+      "step": 26620
+    },
+    {
+      "epoch": 0.5720238862396357,
+      "grad_norm": 0.5063998256679312,
+      "learning_rate": 7.889245383758247e-06,
+      "loss": 0.733,
+      "step": 26630
+    },
+    {
+      "epoch": 0.5722386905529063,
+      "grad_norm": 0.5118782447758873,
+      "learning_rate": 7.882582943205218e-06,
+      "loss": 0.7172,
+      "step": 26640
+    },
+    {
+      "epoch": 0.5724534948661769,
+      "grad_norm": 0.4961115360495314,
+      "learning_rate": 7.87592148650564e-06,
+      "loss": 0.7183,
+      "step": 26650
+    },
+    {
+      "epoch": 0.5726682991794475,
+      "grad_norm": 0.5280800091056106,
+      "learning_rate": 7.869261016754736e-06,
+      "loss": 0.7295,
+      "step": 26660
+    },
+    {
+      "epoch": 0.5728831034927181,
+      "grad_norm": 0.5289805883203285,
+      "learning_rate": 7.862601537047289e-06,
+      "loss": 0.7265,
+      "step": 26670
+    },
+    {
+      "epoch": 0.5730979078059888,
+      "grad_norm": 0.5193534190312329,
+      "learning_rate": 7.855943050477605e-06,
+      "loss": 0.7333,
+      "step": 26680
+    },
+    {
+      "epoch": 0.5733127121192594,
+      "grad_norm": 0.5290783598213893,
+      "learning_rate": 7.84928556013954e-06,
+      "loss": 0.7221,
+      "step": 26690
+    },
+    {
+      "epoch": 0.57352751643253,
+      "grad_norm": 0.5127746177226163,
+      "learning_rate": 7.842629069126484e-06,
+      "loss": 0.7388,
+      "step": 26700
+    },
+    {
+      "epoch": 0.5737423207458006,
+      "grad_norm": 0.512414892050149,
+      "learning_rate": 7.835973580531353e-06,
+      "loss": 0.7284,
+      "step": 26710
+    },
+    {
+      "epoch": 0.5739571250590711,
+      "grad_norm": 0.5029147333178203,
+      "learning_rate": 7.829319097446617e-06,
+      "loss": 0.716,
+      "step": 26720
+    },
+    {
+      "epoch": 0.5741719293723418,
+      "grad_norm": 0.520040331471592,
+      "learning_rate": 7.82266562296426e-06,
+      "loss": 0.7159,
+      "step": 26730
+    },
+    {
+      "epoch": 0.5743867336856124,
+      "grad_norm": 0.4978396177758656,
+      "learning_rate": 7.816013160175812e-06,
+      "loss": 0.7208,
+      "step": 26740
+    },
+    {
+      "epoch": 0.574601537998883,
+      "grad_norm": 0.538410253835665,
+      "learning_rate": 7.809361712172322e-06,
+      "loss": 0.7146,
+      "step": 26750
+    },
+    {
+      "epoch": 0.5748163423121536,
+      "grad_norm": 0.5071511006184085,
+      "learning_rate": 7.802711282044366e-06,
+      "loss": 0.7314,
+      "step": 26760
+    },
+    {
+      "epoch": 0.5750311466254242,
+      "grad_norm": 0.5026752106850176,
+      "learning_rate": 7.796061872882065e-06,
+      "loss": 0.7353,
+      "step": 26770
+    },
+    {
+      "epoch": 0.5752459509386948,
+      "grad_norm": 0.4928490982584089,
+      "learning_rate": 7.789413487775043e-06,
+      "loss": 0.7161,
+      "step": 26780
+    },
+    {
+      "epoch": 0.5754607552519655,
+      "grad_norm": 0.5054902551078443,
+      "learning_rate": 7.782766129812463e-06,
+      "loss": 0.7227,
+      "step": 26790
+    },
+    {
+      "epoch": 0.5756755595652361,
+      "grad_norm": 0.5106529312162003,
+      "learning_rate": 7.776119802083001e-06,
+      "loss": 0.7173,
+      "step": 26800
+    },
+    {
+      "epoch": 0.5758903638785067,
+      "grad_norm": 0.5157371052564345,
+      "learning_rate": 7.76947450767487e-06,
+      "loss": 0.7122,
+      "step": 26810
+    },
+    {
+      "epoch": 0.5761051681917773,
+      "grad_norm": 0.5284843076164706,
+      "learning_rate": 7.762830249675785e-06,
+      "loss": 0.7256,
+      "step": 26820
+    },
+    {
+      "epoch": 0.5763199725050479,
+      "grad_norm": 0.5456146407952996,
+      "learning_rate": 7.756187031172984e-06,
+      "loss": 0.7391,
+      "step": 26830
+    },
+    {
+      "epoch": 0.5765347768183186,
+      "grad_norm": 0.4917659110447439,
+      "learning_rate": 7.74954485525323e-06,
+      "loss": 0.7185,
+      "step": 26840
+    },
+    {
+      "epoch": 0.5767495811315891,
+      "grad_norm": 0.506243431890311,
+      "learning_rate": 7.74290372500279e-06,
+      "loss": 0.7367,
+      "step": 26850
+    },
+    {
+      "epoch": 0.5769643854448597,
+      "grad_norm": 0.506566764718463,
+      "learning_rate": 7.736263643507462e-06,
+      "loss": 0.7336,
+      "step": 26860
+    },
+    {
+      "epoch": 0.5771791897581303,
+      "grad_norm": 0.5188308701341517,
+      "learning_rate": 7.729624613852533e-06,
+      "loss": 0.7165,
+      "step": 26870
+    },
+    {
+      "epoch": 0.5773939940714009,
+      "grad_norm": 0.5149438496598627,
+      "learning_rate": 7.722986639122827e-06,
+      "loss": 0.7287,
+      "step": 26880
+    },
+    {
+      "epoch": 0.5776087983846716,
+      "grad_norm": 0.5235992368572671,
+      "learning_rate": 7.71634972240266e-06,
+      "loss": 0.7174,
+      "step": 26890
+    },
+    {
+      "epoch": 0.5778236026979422,
+      "grad_norm": 0.5113815635765164,
+      "learning_rate": 7.70971386677586e-06,
+      "loss": 0.7109,
+      "step": 26900
+    },
+    {
+      "epoch": 0.5780384070112128,
+      "grad_norm": 0.49355622034783786,
+      "learning_rate": 7.703079075325764e-06,
+      "loss": 0.7103,
+      "step": 26910
+    },
+    {
+      "epoch": 0.5782532113244834,
+      "grad_norm": 0.519299512412156,
+      "learning_rate": 7.69644535113521e-06,
+      "loss": 0.7369,
+      "step": 26920
+    },
+    {
+      "epoch": 0.578468015637754,
+      "grad_norm": 0.5310388569132752,
+      "learning_rate": 7.689812697286555e-06,
+      "loss": 0.7188,
+      "step": 26930
+    },
+    {
+      "epoch": 0.5786828199510247,
+      "grad_norm": 0.5168701357967611,
+      "learning_rate": 7.683181116861642e-06,
+      "loss": 0.73,
+      "step": 26940
+    },
+    {
+      "epoch": 0.5788976242642953,
+      "grad_norm": 0.5042983692202927,
+      "learning_rate": 7.676550612941816e-06,
+      "loss": 0.7135,
+      "step": 26950
+    },
+    {
+      "epoch": 0.5791124285775658,
+      "grad_norm": 0.5177162051099963,
+      "learning_rate": 7.669921188607935e-06,
+      "loss": 0.7394,
+      "step": 26960
+    },
+    {
+      "epoch": 0.5793272328908364,
+      "grad_norm": 0.5119619648123758,
+      "learning_rate": 7.663292846940343e-06,
+      "loss": 0.7207,
+      "step": 26970
+    },
+    {
+      "epoch": 0.579542037204107,
+      "grad_norm": 0.5362452673286001,
+      "learning_rate": 7.656665591018887e-06,
+      "loss": 0.7143,
+      "step": 26980
+    },
+    {
+      "epoch": 0.5797568415173777,
+      "grad_norm": 0.5129247448122536,
+      "learning_rate": 7.650039423922905e-06,
+      "loss": 0.7224,
+      "step": 26990
+    },
+    {
+      "epoch": 0.5799716458306483,
+      "grad_norm": 0.5088798828391344,
+      "learning_rate": 7.643414348731237e-06,
+      "loss": 0.7148,
+      "step": 27000
+    },
+    {
+      "epoch": 0.5801864501439189,
+      "grad_norm": 0.5027910917195445,
+      "learning_rate": 7.636790368522208e-06,
+      "loss": 0.7219,
+      "step": 27010
+    },
+    {
+      "epoch": 0.5804012544571895,
+      "grad_norm": 0.5185537348120938,
+      "learning_rate": 7.630167486373632e-06,
+      "loss": 0.7221,
+      "step": 27020
+    },
+    {
+      "epoch": 0.5806160587704601,
+      "grad_norm": 0.5272726326291755,
+      "learning_rate": 7.623545705362822e-06,
+      "loss": 0.7175,
+      "step": 27030
+    },
+    {
+      "epoch": 0.5808308630837308,
+      "grad_norm": 0.5166485329784066,
+      "learning_rate": 7.616925028566575e-06,
+      "loss": 0.7341,
+      "step": 27040
+    },
+    {
+      "epoch": 0.5810456673970014,
+      "grad_norm": 0.5315842783462765,
+      "learning_rate": 7.6103054590611755e-06,
+      "loss": 0.7391,
+      "step": 27050
+    },
+    {
+      "epoch": 0.581260471710272,
+      "grad_norm": 0.5015083854889525,
+      "learning_rate": 7.603686999922386e-06,
+      "loss": 0.7144,
+      "step": 27060
+    },
+    {
+      "epoch": 0.5814752760235425,
+      "grad_norm": 0.5184207030120949,
+      "learning_rate": 7.597069654225471e-06,
+      "loss": 0.7248,
+      "step": 27070
+    },
+    {
+      "epoch": 0.5816900803368131,
+      "grad_norm": 0.5316867610900236,
+      "learning_rate": 7.590453425045159e-06,
+      "loss": 0.733,
+      "step": 27080
+    },
+    {
+      "epoch": 0.5819048846500837,
+      "grad_norm": 0.4965452979386377,
+      "learning_rate": 7.583838315455665e-06,
+      "loss": 0.7204,
+      "step": 27090
+    },
+    {
+      "epoch": 0.5821196889633544,
+      "grad_norm": 0.5297374352101749,
+      "learning_rate": 7.577224328530694e-06,
+      "loss": 0.7227,
+      "step": 27100
+    },
+    {
+      "epoch": 0.582334493276625,
+      "grad_norm": 0.5359023033266158,
+      "learning_rate": 7.570611467343414e-06,
+      "loss": 0.7341,
+      "step": 27110
+    },
+    {
+      "epoch": 0.5825492975898956,
+      "grad_norm": 0.5131959097511573,
+      "learning_rate": 7.563999734966483e-06,
+      "loss": 0.7137,
+      "step": 27120
+    },
+    {
+      "epoch": 0.5827641019031662,
+      "grad_norm": 0.5332592734188711,
+      "learning_rate": 7.557389134472021e-06,
+      "loss": 0.7339,
+      "step": 27130
+    },
+    {
+      "epoch": 0.5829789062164368,
+      "grad_norm": 0.4950524149977223,
+      "learning_rate": 7.550779668931641e-06,
+      "loss": 0.7278,
+      "step": 27140
+    },
+    {
+      "epoch": 0.5831937105297075,
+      "grad_norm": 0.4979574277254489,
+      "learning_rate": 7.544171341416409e-06,
+      "loss": 0.7116,
+      "step": 27150
+    },
+    {
+      "epoch": 0.5834085148429781,
+      "grad_norm": 0.5184623277698175,
+      "learning_rate": 7.537564154996871e-06,
+      "loss": 0.7171,
+      "step": 27160
+    },
+    {
+      "epoch": 0.5836233191562487,
+      "grad_norm": 0.5028351179542662,
+      "learning_rate": 7.530958112743048e-06,
+      "loss": 0.7074,
+      "step": 27170
+    },
+    {
+      "epoch": 0.5838381234695192,
+      "grad_norm": 0.5140269218645958,
+      "learning_rate": 7.524353217724414e-06,
+      "loss": 0.7183,
+      "step": 27180
+    },
+    {
+      "epoch": 0.5840529277827898,
+      "grad_norm": 0.5116611764499623,
+      "learning_rate": 7.517749473009931e-06,
+      "loss": 0.731,
+      "step": 27190
+    },
+    {
+      "epoch": 0.5842677320960605,
+      "grad_norm": 0.5034040525475355,
+      "learning_rate": 7.51114688166801e-06,
+      "loss": 0.7026,
+      "step": 27200
+    },
+    {
+      "epoch": 0.5844825364093311,
+      "grad_norm": 0.5120240441103905,
+      "learning_rate": 7.50454544676653e-06,
+      "loss": 0.7298,
+      "step": 27210
+    },
+    {
+      "epoch": 0.5846973407226017,
+      "grad_norm": 0.5141277473304001,
+      "learning_rate": 7.497945171372838e-06,
+      "loss": 0.7169,
+      "step": 27220
+    },
+    {
+      "epoch": 0.5849121450358723,
+      "grad_norm": 0.5092099955268595,
+      "learning_rate": 7.4913460585537314e-06,
+      "loss": 0.7227,
+      "step": 27230
+    },
+    {
+      "epoch": 0.5851269493491429,
+      "grad_norm": 0.5394087392182072,
+      "learning_rate": 7.484748111375482e-06,
+      "loss": 0.7368,
+      "step": 27240
+    },
+    {
+      "epoch": 0.5853417536624136,
+      "grad_norm": 0.5296527582320034,
+      "learning_rate": 7.478151332903807e-06,
+      "loss": 0.7249,
+      "step": 27250
+    },
+    {
+      "epoch": 0.5855565579756842,
+      "grad_norm": 0.527906994069362,
+      "learning_rate": 7.4715557262038904e-06,
+      "loss": 0.702,
+      "step": 27260
+    },
+    {
+      "epoch": 0.5857713622889548,
+      "grad_norm": 0.49003860514458714,
+      "learning_rate": 7.464961294340366e-06,
+      "loss": 0.7216,
+      "step": 27270
+    },
+    {
+      "epoch": 0.5859861666022254,
+      "grad_norm": 0.5158001608175528,
+      "learning_rate": 7.45836804037732e-06,
+      "loss": 0.7076,
+      "step": 27280
+    },
+    {
+      "epoch": 0.586200970915496,
+      "grad_norm": 0.5189638533176454,
+      "learning_rate": 7.451775967378296e-06,
+      "loss": 0.7302,
+      "step": 27290
+    },
+    {
+      "epoch": 0.5864157752287666,
+      "grad_norm": 0.5114303128229795,
+      "learning_rate": 7.4451850784062825e-06,
+      "loss": 0.7195,
+      "step": 27300
+    },
+    {
+      "epoch": 0.5866305795420372,
+      "grad_norm": 0.5141598815151798,
+      "learning_rate": 7.438595376523734e-06,
+      "loss": 0.7259,
+      "step": 27310
+    },
+    {
+      "epoch": 0.5868453838553078,
+      "grad_norm": 0.5115551026689363,
+      "learning_rate": 7.432006864792529e-06,
+      "loss": 0.7363,
+      "step": 27320
+    },
+    {
+      "epoch": 0.5870601881685784,
+      "grad_norm": 0.5067452647479966,
+      "learning_rate": 7.4254195462740165e-06,
+      "loss": 0.7088,
+      "step": 27330
+    },
+    {
+      "epoch": 0.587274992481849,
+      "grad_norm": 0.5201913281991899,
+      "learning_rate": 7.418833424028974e-06,
+      "loss": 0.7268,
+      "step": 27340
+    },
+    {
+      "epoch": 0.5874897967951197,
+      "grad_norm": 0.5178789741191531,
+      "learning_rate": 7.412248501117627e-06,
+      "loss": 0.701,
+      "step": 27350
+    },
+    {
+      "epoch": 0.5877046011083903,
+      "grad_norm": 0.5059570164144281,
+      "learning_rate": 7.4056647805996526e-06,
+      "loss": 0.7159,
+      "step": 27360
+    },
+    {
+      "epoch": 0.5879194054216609,
+      "grad_norm": 0.5130596201558596,
+      "learning_rate": 7.399082265534161e-06,
+      "loss": 0.7208,
+      "step": 27370
+    },
+    {
+      "epoch": 0.5881342097349315,
+      "grad_norm": 0.5129021388577076,
+      "learning_rate": 7.392500958979705e-06,
+      "loss": 0.7268,
+      "step": 27380
+    },
+    {
+      "epoch": 0.5883490140482021,
+      "grad_norm": 0.49928892977313877,
+      "learning_rate": 7.385920863994273e-06,
+      "loss": 0.7291,
+      "step": 27390
+    },
+    {
+      "epoch": 0.5885638183614726,
+      "grad_norm": 0.5013211052439496,
+      "learning_rate": 7.3793419836352884e-06,
+      "loss": 0.7319,
+      "step": 27400
+    },
+    {
+      "epoch": 0.5887786226747433,
+      "grad_norm": 0.5333496221119072,
+      "learning_rate": 7.372764320959624e-06,
+      "loss": 0.7174,
+      "step": 27410
+    },
+    {
+      "epoch": 0.5889934269880139,
+      "grad_norm": 0.516097492110587,
+      "learning_rate": 7.366187879023572e-06,
+      "loss": 0.7144,
+      "step": 27420
+    },
+    {
+      "epoch": 0.5892082313012845,
+      "grad_norm": 0.5135516714260109,
+      "learning_rate": 7.359612660882862e-06,
+      "loss": 0.7107,
+      "step": 27430
+    },
+    {
+      "epoch": 0.5894230356145551,
+      "grad_norm": 0.5339129088339684,
+      "learning_rate": 7.353038669592654e-06,
+      "loss": 0.7238,
+      "step": 27440
+    },
+    {
+      "epoch": 0.5896378399278257,
+      "grad_norm": 0.4964474003475041,
+      "learning_rate": 7.346465908207545e-06,
+      "loss": 0.7176,
+      "step": 27450
+    },
+    {
+      "epoch": 0.5898526442410964,
+      "grad_norm": 0.501452498338148,
+      "learning_rate": 7.339894379781551e-06,
+      "loss": 0.7131,
+      "step": 27460
+    },
+    {
+      "epoch": 0.590067448554367,
+      "grad_norm": 0.5133436088736439,
+      "learning_rate": 7.333324087368117e-06,
+      "loss": 0.7215,
+      "step": 27470
+    },
+    {
+      "epoch": 0.5902822528676376,
+      "grad_norm": 0.5192593089145867,
+      "learning_rate": 7.326755034020122e-06,
+      "loss": 0.7162,
+      "step": 27480
+    },
+    {
+      "epoch": 0.5904970571809082,
+      "grad_norm": 0.5477084008394134,
+      "learning_rate": 7.320187222789856e-06,
+      "loss": 0.7042,
+      "step": 27490
+    },
+    {
+      "epoch": 0.5907118614941788,
+      "grad_norm": 0.5180536818105407,
+      "learning_rate": 7.3136206567290465e-06,
+      "loss": 0.7305,
+      "step": 27500
+    },
+    {
+      "epoch": 0.5909266658074495,
+      "grad_norm": 0.502178932682032,
+      "learning_rate": 7.307055338888826e-06,
+      "loss": 0.7139,
+      "step": 27510
+    },
+    {
+      "epoch": 0.59114147012072,
+      "grad_norm": 0.5157309880089097,
+      "learning_rate": 7.300491272319764e-06,
+      "loss": 0.7163,
+      "step": 27520
+    },
+    {
+      "epoch": 0.5913562744339906,
+      "grad_norm": 0.5148147550554781,
+      "learning_rate": 7.293928460071838e-06,
+      "loss": 0.7175,
+      "step": 27530
+    },
+    {
+      "epoch": 0.5915710787472612,
+      "grad_norm": 0.5082539143510494,
+      "learning_rate": 7.287366905194439e-06,
+      "loss": 0.7155,
+      "step": 27540
+    },
+    {
+      "epoch": 0.5917858830605318,
+      "grad_norm": 0.5141335882021415,
+      "learning_rate": 7.280806610736391e-06,
+      "loss": 0.7287,
+      "step": 27550
+    },
+    {
+      "epoch": 0.5920006873738025,
+      "grad_norm": 0.48870302456568654,
+      "learning_rate": 7.274247579745908e-06,
+      "loss": 0.717,
+      "step": 27560
+    },
+    {
+      "epoch": 0.5922154916870731,
+      "grad_norm": 0.514022070926401,
+      "learning_rate": 7.267689815270642e-06,
+      "loss": 0.7339,
+      "step": 27570
+    },
+    {
+      "epoch": 0.5924302960003437,
+      "grad_norm": 0.5135139016976253,
+      "learning_rate": 7.261133320357641e-06,
+      "loss": 0.7198,
+      "step": 27580
+    },
+    {
+      "epoch": 0.5926451003136143,
+      "grad_norm": 0.53429824873582,
+      "learning_rate": 7.254578098053362e-06,
+      "loss": 0.7349,
+      "step": 27590
+    },
+    {
+      "epoch": 0.5928599046268849,
+      "grad_norm": 0.5058250349786495,
+      "learning_rate": 7.248024151403682e-06,
+      "loss": 0.724,
+      "step": 27600
+    },
+    {
+      "epoch": 0.5930747089401556,
+      "grad_norm": 0.5206560440876598,
+      "learning_rate": 7.2414714834538725e-06,
+      "loss": 0.7337,
+      "step": 27610
+    },
+    {
+      "epoch": 0.5932895132534262,
+      "grad_norm": 0.48980099237908875,
+      "learning_rate": 7.234920097248623e-06,
+      "loss": 0.7181,
+      "step": 27620
+    },
+    {
+      "epoch": 0.5935043175666967,
+      "grad_norm": 0.5115417079620694,
+      "learning_rate": 7.228369995832015e-06,
+      "loss": 0.7297,
+      "step": 27630
+    },
+    {
+      "epoch": 0.5937191218799673,
+      "grad_norm": 0.5016557907859813,
+      "learning_rate": 7.221821182247548e-06,
+      "loss": 0.732,
+      "step": 27640
+    },
+    {
+      "epoch": 0.5939339261932379,
+      "grad_norm": 0.49835395618153433,
+      "learning_rate": 7.215273659538114e-06,
+      "loss": 0.7145,
+      "step": 27650
+    },
+    {
+      "epoch": 0.5941487305065085,
+      "grad_norm": 0.5169175369018176,
+      "learning_rate": 7.2087274307459945e-06,
+      "loss": 0.7091,
+      "step": 27660
+    },
+    {
+      "epoch": 0.5943635348197792,
+      "grad_norm": 0.5068225266138379,
+      "learning_rate": 7.2021824989128915e-06,
+      "loss": 0.7318,
+      "step": 27670
+    },
+    {
+      "epoch": 0.5945783391330498,
+      "grad_norm": 0.5091346931786659,
+      "learning_rate": 7.195638867079889e-06,
+      "loss": 0.7203,
+      "step": 27680
+    },
+    {
+      "epoch": 0.5947931434463204,
+      "grad_norm": 0.5271726430775541,
+      "learning_rate": 7.1890965382874765e-06,
+      "loss": 0.7318,
+      "step": 27690
+    },
+    {
+      "epoch": 0.595007947759591,
+      "grad_norm": 0.5205082559060062,
+      "learning_rate": 7.182555515575531e-06,
+      "loss": 0.7221,
+      "step": 27700
+    },
+    {
+      "epoch": 0.5952227520728616,
+      "grad_norm": 0.5108228067074766,
+      "learning_rate": 7.176015801983326e-06,
+      "loss": 0.7186,
+      "step": 27710
+    },
+    {
+      "epoch": 0.5954375563861323,
+      "grad_norm": 0.5032605086696729,
+      "learning_rate": 7.169477400549525e-06,
+      "loss": 0.7159,
+      "step": 27720
+    },
+    {
+      "epoch": 0.5956523606994029,
+      "grad_norm": 0.5143877348374966,
+      "learning_rate": 7.16294031431218e-06,
+      "loss": 0.7161,
+      "step": 27730
+    },
+    {
+      "epoch": 0.5958671650126734,
+      "grad_norm": 0.5175753229468208,
+      "learning_rate": 7.156404546308741e-06,
+      "loss": 0.7241,
+      "step": 27740
+    },
+    {
+      "epoch": 0.596081969325944,
+      "grad_norm": 0.5160103076524589,
+      "learning_rate": 7.149870099576033e-06,
+      "loss": 0.722,
+      "step": 27750
+    },
+    {
+      "epoch": 0.5962967736392146,
+      "grad_norm": 0.5276236091636177,
+      "learning_rate": 7.143336977150278e-06,
+      "loss": 0.7271,
+      "step": 27760
+    },
+    {
+      "epoch": 0.5965115779524853,
+      "grad_norm": 0.5115582800303922,
+      "learning_rate": 7.136805182067074e-06,
+      "loss": 0.7276,
+      "step": 27770
+    },
+    {
+      "epoch": 0.5967263822657559,
+      "grad_norm": 0.5007489074574742,
+      "learning_rate": 7.130274717361405e-06,
+      "loss": 0.7101,
+      "step": 27780
+    },
+    {
+      "epoch": 0.5969411865790265,
+      "grad_norm": 0.5247645617356222,
+      "learning_rate": 7.123745586067645e-06,
+      "loss": 0.7214,
+      "step": 27790
+    },
+    {
+      "epoch": 0.5971559908922971,
+      "grad_norm": 0.509054996286859,
+      "learning_rate": 7.117217791219533e-06,
+      "loss": 0.7062,
+      "step": 27800
+    },
+    {
+      "epoch": 0.5973707952055677,
+      "grad_norm": 0.5146383947427272,
+      "learning_rate": 7.110691335850202e-06,
+      "loss": 0.727,
+      "step": 27810
+    },
+    {
+      "epoch": 0.5975855995188384,
+      "grad_norm": 0.5107471000601168,
+      "learning_rate": 7.1041662229921485e-06,
+      "loss": 0.7078,
+      "step": 27820
+    },
+    {
+      "epoch": 0.597800403832109,
+      "grad_norm": 0.48699390234221435,
+      "learning_rate": 7.097642455677261e-06,
+      "loss": 0.7127,
+      "step": 27830
+    },
+    {
+      "epoch": 0.5980152081453796,
+      "grad_norm": 0.48779003203616056,
+      "learning_rate": 7.091120036936791e-06,
+      "loss": 0.713,
+      "step": 27840
+    },
+    {
+      "epoch": 0.5982300124586502,
+      "grad_norm": 0.5073544067387802,
+      "learning_rate": 7.084598969801362e-06,
+      "loss": 0.7194,
+      "step": 27850
+    },
+    {
+      "epoch": 0.5984448167719207,
+      "grad_norm": 0.5143979161778222,
+      "learning_rate": 7.0780792573009835e-06,
+      "loss": 0.7171,
+      "step": 27860
+    },
+    {
+      "epoch": 0.5986596210851914,
+      "grad_norm": 0.5100260027061173,
+      "learning_rate": 7.07156090246502e-06,
+      "loss": 0.7219,
+      "step": 27870
+    },
+    {
+      "epoch": 0.598874425398462,
+      "grad_norm": 0.5080855464944665,
+      "learning_rate": 7.065043908322214e-06,
+      "loss": 0.7288,
+      "step": 27880
+    },
+    {
+      "epoch": 0.5990892297117326,
+      "grad_norm": 0.503233048103245,
+      "learning_rate": 7.058528277900669e-06,
+      "loss": 0.708,
+      "step": 27890
+    },
+    {
+      "epoch": 0.5993040340250032,
+      "grad_norm": 0.5262280681492298,
+      "learning_rate": 7.052014014227866e-06,
+      "loss": 0.7167,
+      "step": 27900
+    },
+    {
+      "epoch": 0.5995188383382738,
+      "grad_norm": 0.522581486548443,
+      "learning_rate": 7.045501120330642e-06,
+      "loss": 0.7272,
+      "step": 27910
+    },
+    {
+      "epoch": 0.5997336426515445,
+      "grad_norm": 0.5104789267173563,
+      "learning_rate": 7.0389895992351956e-06,
+      "loss": 0.7087,
+      "step": 27920
+    },
+    {
+      "epoch": 0.5999484469648151,
+      "grad_norm": 0.5146754901769843,
+      "learning_rate": 7.032479453967097e-06,
+      "loss": 0.728,
+      "step": 27930
+    },
+    {
+      "epoch": 0.6001632512780857,
+      "grad_norm": 0.5116496747974748,
+      "learning_rate": 7.0259706875512645e-06,
+      "loss": 0.7125,
+      "step": 27940
+    },
+    {
+      "epoch": 0.6003780555913563,
+      "grad_norm": 0.521148762194517,
+      "learning_rate": 7.019463303011993e-06,
+      "loss": 0.7174,
+      "step": 27950
+    },
+    {
+      "epoch": 0.6005928599046269,
+      "grad_norm": 0.5311474419836788,
+      "learning_rate": 7.012957303372918e-06,
+      "loss": 0.7239,
+      "step": 27960
+    },
+    {
+      "epoch": 0.6008076642178974,
+      "grad_norm": 0.5235689594460002,
+      "learning_rate": 7.006452691657039e-06,
+      "loss": 0.7212,
+      "step": 27970
+    },
+    {
+      "epoch": 0.6010224685311681,
+      "grad_norm": 0.5092035810607163,
+      "learning_rate": 6.999949470886715e-06,
+      "loss": 0.7092,
+      "step": 27980
+    },
+    {
+      "epoch": 0.6012372728444387,
+      "grad_norm": 0.4938851390556326,
+      "learning_rate": 6.9934476440836465e-06,
+      "loss": 0.7244,
+      "step": 27990
+    },
+    {
+      "epoch": 0.6014520771577093,
+      "grad_norm": 0.5309729653806932,
+      "learning_rate": 6.986947214268902e-06,
+      "loss": 0.7263,
+      "step": 28000
+    },
+    {
+      "epoch": 0.6016668814709799,
+      "grad_norm": 0.5203026110014208,
+      "learning_rate": 6.980448184462887e-06,
+      "loss": 0.707,
+      "step": 28010
+    },
+    {
+      "epoch": 0.6018816857842505,
+      "grad_norm": 0.508186230193396,
+      "learning_rate": 6.973950557685366e-06,
+      "loss": 0.7188,
+      "step": 28020
+    },
+    {
+      "epoch": 0.6020964900975212,
+      "grad_norm": 0.5101290046607262,
+      "learning_rate": 6.967454336955447e-06,
+      "loss": 0.7261,
+      "step": 28030
+    },
+    {
+      "epoch": 0.6023112944107918,
+      "grad_norm": 0.5004521118510085,
+      "learning_rate": 6.96095952529158e-06,
+      "loss": 0.7097,
+      "step": 28040
+    },
+    {
+      "epoch": 0.6025260987240624,
+      "grad_norm": 0.49710168230451124,
+      "learning_rate": 6.9544661257115745e-06,
+      "loss": 0.7275,
+      "step": 28050
+    },
+    {
+      "epoch": 0.602740903037333,
+      "grad_norm": 0.5324092271912054,
+      "learning_rate": 6.947974141232568e-06,
+      "loss": 0.7176,
+      "step": 28060
+    },
+    {
+      "epoch": 0.6029557073506036,
+      "grad_norm": 0.5254153398251872,
+      "learning_rate": 6.9414835748710525e-06,
+      "loss": 0.7329,
+      "step": 28070
+    },
+    {
+      "epoch": 0.6031705116638743,
+      "grad_norm": 0.5018438803534653,
+      "learning_rate": 6.934994429642854e-06,
+      "loss": 0.7114,
+      "step": 28080
+    },
+    {
+      "epoch": 0.6033853159771448,
+      "grad_norm": 0.5110608431528573,
+      "learning_rate": 6.928506708563142e-06,
+      "loss": 0.7175,
+      "step": 28090
+    },
+    {
+      "epoch": 0.6036001202904154,
+      "grad_norm": 0.5110626161063511,
+      "learning_rate": 6.922020414646422e-06,
+      "loss": 0.7228,
+      "step": 28100
+    },
+    {
+      "epoch": 0.603814924603686,
+      "grad_norm": 0.5165958575854644,
+      "learning_rate": 6.915535550906532e-06,
+      "loss": 0.7282,
+      "step": 28110
+    },
+    {
+      "epoch": 0.6040297289169566,
+      "grad_norm": 0.5252321669011311,
+      "learning_rate": 6.909052120356659e-06,
+      "loss": 0.7218,
+      "step": 28120
+    },
+    {
+      "epoch": 0.6042445332302273,
+      "grad_norm": 0.5145699979700377,
+      "learning_rate": 6.902570126009309e-06,
+      "loss": 0.721,
+      "step": 28130
+    },
+    {
+      "epoch": 0.6044593375434979,
+      "grad_norm": 0.5257878272393476,
+      "learning_rate": 6.8960895708763335e-06,
+      "loss": 0.7236,
+      "step": 28140
+    },
+    {
+      "epoch": 0.6046741418567685,
+      "grad_norm": 0.5107902953084463,
+      "learning_rate": 6.8896104579689026e-06,
+      "loss": 0.7139,
+      "step": 28150
+    },
+    {
+      "epoch": 0.6048889461700391,
+      "grad_norm": 0.5187512853597482,
+      "learning_rate": 6.8831327902975245e-06,
+      "loss": 0.7211,
+      "step": 28160
+    },
+    {
+      "epoch": 0.6051037504833097,
+      "grad_norm": 0.5179836553200129,
+      "learning_rate": 6.876656570872036e-06,
+      "loss": 0.7157,
+      "step": 28170
+    },
+    {
+      "epoch": 0.6053185547965804,
+      "grad_norm": 0.4961138882817056,
+      "learning_rate": 6.870181802701596e-06,
+      "loss": 0.7144,
+      "step": 28180
+    },
+    {
+      "epoch": 0.605533359109851,
+      "grad_norm": 0.4994657394772925,
+      "learning_rate": 6.8637084887946985e-06,
+      "loss": 0.7154,
+      "step": 28190
+    },
+    {
+      "epoch": 0.6057481634231215,
+      "grad_norm": 0.5360515600802547,
+      "learning_rate": 6.857236632159146e-06,
+      "loss": 0.6994,
+      "step": 28200
+    },
+    {
+      "epoch": 0.6059629677363921,
+      "grad_norm": 0.49991257688614876,
+      "learning_rate": 6.85076623580208e-06,
+      "loss": 0.7249,
+      "step": 28210
+    },
+    {
+      "epoch": 0.6061777720496627,
+      "grad_norm": 0.5108166128653483,
+      "learning_rate": 6.8442973027299565e-06,
+      "loss": 0.7229,
+      "step": 28220
+    },
+    {
+      "epoch": 0.6063925763629334,
+      "grad_norm": 0.49773101558129873,
+      "learning_rate": 6.837829835948546e-06,
+      "loss": 0.7091,
+      "step": 28230
+    },
+    {
+      "epoch": 0.606607380676204,
+      "grad_norm": 0.505655201231906,
+      "learning_rate": 6.8313638384629525e-06,
+      "loss": 0.7206,
+      "step": 28240
+    },
+    {
+      "epoch": 0.6068221849894746,
+      "grad_norm": 0.5131719935934191,
+      "learning_rate": 6.824899313277582e-06,
+      "loss": 0.7252,
+      "step": 28250
+    },
+    {
+      "epoch": 0.6070369893027452,
+      "grad_norm": 0.503882580363197,
+      "learning_rate": 6.8184362633961655e-06,
+      "loss": 0.7305,
+      "step": 28260
+    },
+    {
+      "epoch": 0.6072517936160158,
+      "grad_norm": 0.5001026171381061,
+      "learning_rate": 6.811974691821741e-06,
+      "loss": 0.7131,
+      "step": 28270
+    },
+    {
+      "epoch": 0.6074665979292864,
+      "grad_norm": 0.5018092976471628,
+      "learning_rate": 6.805514601556671e-06,
+      "loss": 0.7179,
+      "step": 28280
+    },
+    {
+      "epoch": 0.6076814022425571,
+      "grad_norm": 0.507419770302612,
+      "learning_rate": 6.799055995602621e-06,
+      "loss": 0.7056,
+      "step": 28290
+    },
+    {
+      "epoch": 0.6078962065558277,
+      "grad_norm": 0.5171706639932914,
+      "learning_rate": 6.7925988769605656e-06,
+      "loss": 0.7265,
+      "step": 28300
+    },
+    {
+      "epoch": 0.6081110108690982,
+      "grad_norm": 0.5057274040421198,
+      "learning_rate": 6.786143248630795e-06,
+      "loss": 0.7055,
+      "step": 28310
+    },
+    {
+      "epoch": 0.6083258151823688,
+      "grad_norm": 0.5132979835569406,
+      "learning_rate": 6.7796891136129e-06,
+      "loss": 0.7117,
+      "step": 28320
+    },
+    {
+      "epoch": 0.6085406194956394,
+      "grad_norm": 0.5100482558341333,
+      "learning_rate": 6.773236474905786e-06,
+      "loss": 0.7099,
+      "step": 28330
+    },
+    {
+      "epoch": 0.6087554238089101,
+      "grad_norm": 0.5068203945434347,
+      "learning_rate": 6.7667853355076565e-06,
+      "loss": 0.7152,
+      "step": 28340
+    },
+    {
+      "epoch": 0.6089702281221807,
+      "grad_norm": 0.4922194906275137,
+      "learning_rate": 6.760335698416016e-06,
+      "loss": 0.7248,
+      "step": 28350
+    },
+    {
+      "epoch": 0.6091850324354513,
+      "grad_norm": 0.4940640802222499,
+      "learning_rate": 6.753887566627682e-06,
+      "loss": 0.7088,
+      "step": 28360
+    },
+    {
+      "epoch": 0.6093998367487219,
+      "grad_norm": 0.4917970774712095,
+      "learning_rate": 6.747440943138757e-06,
+      "loss": 0.7041,
+      "step": 28370
+    },
+    {
+      "epoch": 0.6096146410619925,
+      "grad_norm": 0.5157876999086878,
+      "learning_rate": 6.740995830944658e-06,
+      "loss": 0.7242,
+      "step": 28380
+    },
+    {
+      "epoch": 0.6098294453752632,
+      "grad_norm": 0.638677062933045,
+      "learning_rate": 6.7345522330400915e-06,
+      "loss": 0.7107,
+      "step": 28390
+    },
+    {
+      "epoch": 0.6100442496885338,
+      "grad_norm": 0.5291704161918205,
+      "learning_rate": 6.728110152419063e-06,
+      "loss": 0.7259,
+      "step": 28400
+    },
+    {
+      "epoch": 0.6102590540018044,
+      "grad_norm": 0.5435155538816592,
+      "learning_rate": 6.7216695920748694e-06,
+      "loss": 0.7292,
+      "step": 28410
+    },
+    {
+      "epoch": 0.610473858315075,
+      "grad_norm": 0.5249535872174969,
+      "learning_rate": 6.715230555000102e-06,
+      "loss": 0.7124,
+      "step": 28420
+    },
+    {
+      "epoch": 0.6106886626283455,
+      "grad_norm": 0.5168556551996696,
+      "learning_rate": 6.708793044186652e-06,
+      "loss": 0.7135,
+      "step": 28430
+    },
+    {
+      "epoch": 0.6109034669416162,
+      "grad_norm": 0.506513826242894,
+      "learning_rate": 6.702357062625689e-06,
+      "loss": 0.7171,
+      "step": 28440
+    },
+    {
+      "epoch": 0.6111182712548868,
+      "grad_norm": 0.5122231459063203,
+      "learning_rate": 6.695922613307684e-06,
+      "loss": 0.7366,
+      "step": 28450
+    },
+    {
+      "epoch": 0.6113330755681574,
+      "grad_norm": 0.5266734325137272,
+      "learning_rate": 6.689489699222387e-06,
+      "loss": 0.7177,
+      "step": 28460
+    },
+    {
+      "epoch": 0.611547879881428,
+      "grad_norm": 0.5249480955219177,
+      "learning_rate": 6.6830583233588406e-06,
+      "loss": 0.7177,
+      "step": 28470
+    },
+    {
+      "epoch": 0.6117626841946986,
+      "grad_norm": 0.536864147141546,
+      "learning_rate": 6.67662848870537e-06,
+      "loss": 0.7334,
+      "step": 28480
+    },
+    {
+      "epoch": 0.6119774885079693,
+      "grad_norm": 0.5017540321010265,
+      "learning_rate": 6.670200198249579e-06,
+      "loss": 0.7141,
+      "step": 28490
+    },
+    {
+      "epoch": 0.6121922928212399,
+      "grad_norm": 0.5022749037709945,
+      "learning_rate": 6.663773454978368e-06,
+      "loss": 0.7087,
+      "step": 28500
+    },
+    {
+      "epoch": 0.6124070971345105,
+      "grad_norm": 0.5079751963910859,
+      "learning_rate": 6.657348261877905e-06,
+      "loss": 0.7411,
+      "step": 28510
+    },
+    {
+      "epoch": 0.6126219014477811,
+      "grad_norm": 0.5184599582736278,
+      "learning_rate": 6.650924621933645e-06,
+      "loss": 0.7171,
+      "step": 28520
+    },
+    {
+      "epoch": 0.6128367057610516,
+      "grad_norm": 0.50121952483461,
+      "learning_rate": 6.644502538130317e-06,
+      "loss": 0.7209,
+      "step": 28530
+    },
+    {
+      "epoch": 0.6130515100743223,
+      "grad_norm": 0.515280894738279,
+      "learning_rate": 6.638082013451925e-06,
+      "loss": 0.7198,
+      "step": 28540
+    },
+    {
+      "epoch": 0.6132663143875929,
+      "grad_norm": 0.5136328125428188,
+      "learning_rate": 6.631663050881763e-06,
+      "loss": 0.7099,
+      "step": 28550
+    },
+    {
+      "epoch": 0.6134811187008635,
+      "grad_norm": 0.5174719027017979,
+      "learning_rate": 6.625245653402379e-06,
+      "loss": 0.725,
+      "step": 28560
+    },
+    {
+      "epoch": 0.6136959230141341,
+      "grad_norm": 0.5250930188904968,
+      "learning_rate": 6.6188298239956105e-06,
+      "loss": 0.7072,
+      "step": 28570
+    },
+    {
+      "epoch": 0.6139107273274047,
+      "grad_norm": 0.5369921932935243,
+      "learning_rate": 6.612415565642552e-06,
+      "loss": 0.7229,
+      "step": 28580
+    },
+    {
+      "epoch": 0.6141255316406753,
+      "grad_norm": 0.5023867945450698,
+      "learning_rate": 6.606002881323581e-06,
+      "loss": 0.7225,
+      "step": 28590
+    },
+    {
+      "epoch": 0.614340335953946,
+      "grad_norm": 0.5107950386038079,
+      "learning_rate": 6.599591774018338e-06,
+      "loss": 0.7155,
+      "step": 28600
+    },
+    {
+      "epoch": 0.6145551402672166,
+      "grad_norm": 0.5003163460337205,
+      "learning_rate": 6.5931822467057275e-06,
+      "loss": 0.7168,
+      "step": 28610
+    },
+    {
+      "epoch": 0.6147699445804872,
+      "grad_norm": 0.5273078172630057,
+      "learning_rate": 6.5867743023639255e-06,
+      "loss": 0.7251,
+      "step": 28620
+    },
+    {
+      "epoch": 0.6149847488937578,
+      "grad_norm": 0.5126901132037992,
+      "learning_rate": 6.580367943970365e-06,
+      "loss": 0.7083,
+      "step": 28630
+    },
+    {
+      "epoch": 0.6151995532070283,
+      "grad_norm": 0.5155415180923669,
+      "learning_rate": 6.573963174501755e-06,
+      "loss": 0.7117,
+      "step": 28640
+    },
+    {
+      "epoch": 0.615414357520299,
+      "grad_norm": 0.514330215394583,
+      "learning_rate": 6.567559996934052e-06,
+      "loss": 0.7147,
+      "step": 28650
+    },
+    {
+      "epoch": 0.6156291618335696,
+      "grad_norm": 0.5245843260086881,
+      "learning_rate": 6.5611584142424845e-06,
+      "loss": 0.7109,
+      "step": 28660
+    },
+    {
+      "epoch": 0.6158439661468402,
+      "grad_norm": 0.5167536487547844,
+      "learning_rate": 6.554758429401532e-06,
+      "loss": 0.7216,
+      "step": 28670
+    },
+    {
+      "epoch": 0.6160587704601108,
+      "grad_norm": 0.5109351895706468,
+      "learning_rate": 6.548360045384933e-06,
+      "loss": 0.7056,
+      "step": 28680
+    },
+    {
+      "epoch": 0.6162735747733814,
+      "grad_norm": 0.5190795799887076,
+      "learning_rate": 6.541963265165686e-06,
+      "loss": 0.7242,
+      "step": 28690
+    },
+    {
+      "epoch": 0.6164883790866521,
+      "grad_norm": 0.5106655693293612,
+      "learning_rate": 6.535568091716036e-06,
+      "loss": 0.7168,
+      "step": 28700
+    },
+    {
+      "epoch": 0.6167031833999227,
+      "grad_norm": 0.5047058090402041,
+      "learning_rate": 6.529174528007496e-06,
+      "loss": 0.7412,
+      "step": 28710
+    },
+    {
+      "epoch": 0.6169179877131933,
+      "grad_norm": 0.5197711629668383,
+      "learning_rate": 6.522782577010815e-06,
+      "loss": 0.7214,
+      "step": 28720
+    },
+    {
+      "epoch": 0.6171327920264639,
+      "grad_norm": 0.4994718882123556,
+      "learning_rate": 6.516392241696004e-06,
+      "loss": 0.7331,
+      "step": 28730
+    },
+    {
+      "epoch": 0.6173475963397345,
+      "grad_norm": 0.5072482739785458,
+      "learning_rate": 6.5100035250323155e-06,
+      "loss": 0.7052,
+      "step": 28740
+    },
+    {
+      "epoch": 0.6175624006530052,
+      "grad_norm": 0.4899692225682852,
+      "learning_rate": 6.503616429988253e-06,
+      "loss": 0.6999,
+      "step": 28750
+    },
+    {
+      "epoch": 0.6177772049662758,
+      "grad_norm": 0.5047645583660254,
+      "learning_rate": 6.497230959531573e-06,
+      "loss": 0.7141,
+      "step": 28760
+    },
+    {
+      "epoch": 0.6179920092795463,
+      "grad_norm": 0.5144621325420531,
+      "learning_rate": 6.490847116629267e-06,
+      "loss": 0.721,
+      "step": 28770
+    },
+    {
+      "epoch": 0.6182068135928169,
+      "grad_norm": 0.5073195245109307,
+      "learning_rate": 6.484464904247573e-06,
+      "loss": 0.7095,
+      "step": 28780
+    },
+    {
+      "epoch": 0.6184216179060875,
+      "grad_norm": 0.5122140072591916,
+      "learning_rate": 6.4780843253519766e-06,
+      "loss": 0.7085,
+      "step": 28790
+    },
+    {
+      "epoch": 0.6186364222193582,
+      "grad_norm": 0.5103737664691493,
+      "learning_rate": 6.471705382907194e-06,
+      "loss": 0.7023,
+      "step": 28800
+    },
+    {
+      "epoch": 0.6188512265326288,
+      "grad_norm": 0.49611210872837896,
+      "learning_rate": 6.465328079877196e-06,
+      "loss": 0.7027,
+      "step": 28810
+    },
+    {
+      "epoch": 0.6190660308458994,
+      "grad_norm": 0.5196971341780113,
+      "learning_rate": 6.458952419225175e-06,
+      "loss": 0.7147,
+      "step": 28820
+    },
+    {
+      "epoch": 0.61928083515917,
+      "grad_norm": 0.5200694337234366,
+      "learning_rate": 6.452578403913577e-06,
+      "loss": 0.7129,
+      "step": 28830
+    },
+    {
+      "epoch": 0.6194956394724406,
+      "grad_norm": 0.49974616265603755,
+      "learning_rate": 6.446206036904068e-06,
+      "loss": 0.7025,
+      "step": 28840
+    },
+    {
+      "epoch": 0.6197104437857112,
+      "grad_norm": 0.5036120598369263,
+      "learning_rate": 6.439835321157561e-06,
+      "loss": 0.7117,
+      "step": 28850
+    },
+    {
+      "epoch": 0.6199252480989819,
+      "grad_norm": 0.5312724368454428,
+      "learning_rate": 6.433466259634191e-06,
+      "loss": 0.7174,
+      "step": 28860
+    },
+    {
+      "epoch": 0.6201400524122525,
+      "grad_norm": 0.5424611207118475,
+      "learning_rate": 6.427098855293328e-06,
+      "loss": 0.7106,
+      "step": 28870
+    },
+    {
+      "epoch": 0.620354856725523,
+      "grad_norm": 0.5100284315892076,
+      "learning_rate": 6.42073311109358e-06,
+      "loss": 0.7226,
+      "step": 28880
+    },
+    {
+      "epoch": 0.6205696610387936,
+      "grad_norm": 0.5173898778049251,
+      "learning_rate": 6.414369029992771e-06,
+      "loss": 0.7311,
+      "step": 28890
+    },
+    {
+      "epoch": 0.6207844653520642,
+      "grad_norm": 0.5125989979068201,
+      "learning_rate": 6.408006614947963e-06,
+      "loss": 0.7136,
+      "step": 28900
+    },
+    {
+      "epoch": 0.6209992696653349,
+      "grad_norm": 0.5241560351725514,
+      "learning_rate": 6.401645868915434e-06,
+      "loss": 0.7309,
+      "step": 28910
+    },
+    {
+      "epoch": 0.6212140739786055,
+      "grad_norm": 0.5326577223760485,
+      "learning_rate": 6.39528679485069e-06,
+      "loss": 0.7222,
+      "step": 28920
+    },
+    {
+      "epoch": 0.6214288782918761,
+      "grad_norm": 0.5412010757518007,
+      "learning_rate": 6.388929395708469e-06,
+      "loss": 0.7275,
+      "step": 28930
+    },
+    {
+      "epoch": 0.6216436826051467,
+      "grad_norm": 0.501280757518304,
+      "learning_rate": 6.382573674442719e-06,
+      "loss": 0.7043,
+      "step": 28940
+    },
+    {
+      "epoch": 0.6218584869184173,
+      "grad_norm": 0.5035153061061383,
+      "learning_rate": 6.376219634006614e-06,
+      "loss": 0.72,
+      "step": 28950
+    },
+    {
+      "epoch": 0.622073291231688,
+      "grad_norm": 0.5006644086233506,
+      "learning_rate": 6.3698672773525414e-06,
+      "loss": 0.7375,
+      "step": 28960
+    },
+    {
+      "epoch": 0.6222880955449586,
+      "grad_norm": 0.5148002471276852,
+      "learning_rate": 6.363516607432116e-06,
+      "loss": 0.7029,
+      "step": 28970
+    },
+    {
+      "epoch": 0.6225028998582292,
+      "grad_norm": 0.5117306851220371,
+      "learning_rate": 6.357167627196164e-06,
+      "loss": 0.7268,
+      "step": 28980
+    },
+    {
+      "epoch": 0.6227177041714997,
+      "grad_norm": 0.5040793669298755,
+      "learning_rate": 6.35082033959472e-06,
+      "loss": 0.7366,
+      "step": 28990
+    },
+    {
+      "epoch": 0.6229325084847703,
+      "grad_norm": 0.511200471454493,
+      "learning_rate": 6.344474747577043e-06,
+      "loss": 0.7241,
+      "step": 29000
+    },
+    {
+      "epoch": 0.623147312798041,
+      "grad_norm": 0.5050786822698712,
+      "learning_rate": 6.338130854091595e-06,
+      "loss": 0.7216,
+      "step": 29010
+    },
+    {
+      "epoch": 0.6233621171113116,
+      "grad_norm": 0.5058176815929498,
+      "learning_rate": 6.331788662086058e-06,
+      "loss": 0.7058,
+      "step": 29020
+    },
+    {
+      "epoch": 0.6235769214245822,
+      "grad_norm": 0.4981789798742353,
+      "learning_rate": 6.325448174507312e-06,
+      "loss": 0.7211,
+      "step": 29030
+    },
+    {
+      "epoch": 0.6237917257378528,
+      "grad_norm": 0.496475062930444,
+      "learning_rate": 6.319109394301459e-06,
+      "loss": 0.7246,
+      "step": 29040
+    },
+    {
+      "epoch": 0.6240065300511234,
+      "grad_norm": 0.5206633397961193,
+      "learning_rate": 6.312772324413798e-06,
+      "loss": 0.719,
+      "step": 29050
+    },
+    {
+      "epoch": 0.6242213343643941,
+      "grad_norm": 0.5147683511194359,
+      "learning_rate": 6.30643696778883e-06,
+      "loss": 0.7319,
+      "step": 29060
+    },
+    {
+      "epoch": 0.6244361386776647,
+      "grad_norm": 0.5013409604266713,
+      "learning_rate": 6.300103327370272e-06,
+      "loss": 0.7138,
+      "step": 29070
+    },
+    {
+      "epoch": 0.6246509429909353,
+      "grad_norm": 0.5448909474545818,
+      "learning_rate": 6.29377140610103e-06,
+      "loss": 0.7219,
+      "step": 29080
+    },
+    {
+      "epoch": 0.6248657473042059,
+      "grad_norm": 0.4980788653525116,
+      "learning_rate": 6.287441206923225e-06,
+      "loss": 0.7054,
+      "step": 29090
+    },
+    {
+      "epoch": 0.6250805516174764,
+      "grad_norm": 0.5063361287454743,
+      "learning_rate": 6.28111273277817e-06,
+      "loss": 0.7291,
+      "step": 29100
+    },
+    {
+      "epoch": 0.6252953559307471,
+      "grad_norm": 0.5079410093076961,
+      "learning_rate": 6.274785986606371e-06,
+      "loss": 0.7176,
+      "step": 29110
+    },
+    {
+      "epoch": 0.6255101602440177,
+      "grad_norm": 0.5169067311424639,
+      "learning_rate": 6.2684609713475454e-06,
+      "loss": 0.7153,
+      "step": 29120
+    },
+    {
+      "epoch": 0.6257249645572883,
+      "grad_norm": 0.5053854343865869,
+      "learning_rate": 6.26213768994059e-06,
+      "loss": 0.7257,
+      "step": 29130
+    },
+    {
+      "epoch": 0.6259397688705589,
+      "grad_norm": 0.5019853299444441,
+      "learning_rate": 6.255816145323613e-06,
+      "loss": 0.7243,
+      "step": 29140
+    },
+    {
+      "epoch": 0.6261545731838295,
+      "grad_norm": 0.5136425354507431,
+      "learning_rate": 6.249496340433903e-06,
+      "loss": 0.7133,
+      "step": 29150
+    },
+    {
+      "epoch": 0.6263693774971001,
+      "grad_norm": 0.4961378621057317,
+      "learning_rate": 6.243178278207944e-06,
+      "loss": 0.7201,
+      "step": 29160
+    },
+    {
+      "epoch": 0.6265841818103708,
+      "grad_norm": 0.500774286297422,
+      "learning_rate": 6.236861961581413e-06,
+      "loss": 0.7183,
+      "step": 29170
+    },
+    {
+      "epoch": 0.6267989861236414,
+      "grad_norm": 0.5360545274720105,
+      "learning_rate": 6.230547393489166e-06,
+      "loss": 0.7197,
+      "step": 29180
+    },
+    {
+      "epoch": 0.627013790436912,
+      "grad_norm": 0.5298466892431664,
+      "learning_rate": 6.224234576865264e-06,
+      "loss": 0.7137,
+      "step": 29190
+    },
+    {
+      "epoch": 0.6272285947501826,
+      "grad_norm": 0.5397675431496055,
+      "learning_rate": 6.217923514642938e-06,
+      "loss": 0.7302,
+      "step": 29200
+    },
+    {
+      "epoch": 0.6274433990634531,
+      "grad_norm": 0.5102301973810862,
+      "learning_rate": 6.211614209754615e-06,
+      "loss": 0.7001,
+      "step": 29210
+    },
+    {
+      "epoch": 0.6276582033767238,
+      "grad_norm": 0.49821596276059255,
+      "learning_rate": 6.205306665131892e-06,
+      "loss": 0.7234,
+      "step": 29220
+    },
+    {
+      "epoch": 0.6278730076899944,
+      "grad_norm": 0.5107970378794817,
+      "learning_rate": 6.199000883705563e-06,
+      "loss": 0.7074,
+      "step": 29230
+    },
+    {
+      "epoch": 0.628087812003265,
+      "grad_norm": 0.5262658842700451,
+      "learning_rate": 6.192696868405598e-06,
+      "loss": 0.7082,
+      "step": 29240
+    },
+    {
+      "epoch": 0.6283026163165356,
+      "grad_norm": 0.49640493413761044,
+      "learning_rate": 6.186394622161136e-06,
+      "loss": 0.7117,
+      "step": 29250
+    },
+    {
+      "epoch": 0.6285174206298062,
+      "grad_norm": 0.520113591142662,
+      "learning_rate": 6.1800941479005125e-06,
+      "loss": 0.7089,
+      "step": 29260
+    },
+    {
+      "epoch": 0.6287322249430769,
+      "grad_norm": 0.5152262365486174,
+      "learning_rate": 6.173795448551223e-06,
+      "loss": 0.7041,
+      "step": 29270
+    },
+    {
+      "epoch": 0.6289470292563475,
+      "grad_norm": 0.5044955000995295,
+      "learning_rate": 6.16749852703995e-06,
+      "loss": 0.7066,
+      "step": 29280
+    },
+    {
+      "epoch": 0.6291618335696181,
+      "grad_norm": 0.5039566573378436,
+      "learning_rate": 6.161203386292539e-06,
+      "loss": 0.7168,
+      "step": 29290
+    },
+    {
+      "epoch": 0.6293766378828887,
+      "grad_norm": 0.5132657302646363,
+      "learning_rate": 6.154910029234022e-06,
+      "loss": 0.7282,
+      "step": 29300
+    },
+    {
+      "epoch": 0.6295914421961593,
+      "grad_norm": 0.501215277645178,
+      "learning_rate": 6.148618458788589e-06,
+      "loss": 0.7125,
+      "step": 29310
+    },
+    {
+      "epoch": 0.62980624650943,
+      "grad_norm": 0.5171520975134748,
+      "learning_rate": 6.142328677879604e-06,
+      "loss": 0.7201,
+      "step": 29320
+    },
+    {
+      "epoch": 0.6300210508227005,
+      "grad_norm": 0.5093541985407058,
+      "learning_rate": 6.136040689429606e-06,
+      "loss": 0.7249,
+      "step": 29330
+    },
+    {
+      "epoch": 0.6302358551359711,
+      "grad_norm": 0.5038211596059813,
+      "learning_rate": 6.1297544963602874e-06,
+      "loss": 0.7116,
+      "step": 29340
+    },
+    {
+      "epoch": 0.6304506594492417,
+      "grad_norm": 0.499683261524144,
+      "learning_rate": 6.123470101592524e-06,
+      "loss": 0.7168,
+      "step": 29350
+    },
+    {
+      "epoch": 0.6306654637625123,
+      "grad_norm": 0.5524946516291463,
+      "learning_rate": 6.1171875080463425e-06,
+      "loss": 0.7087,
+      "step": 29360
+    },
+    {
+      "epoch": 0.630880268075783,
+      "grad_norm": 0.5018295390971191,
+      "learning_rate": 6.1109067186409365e-06,
+      "loss": 0.7219,
+      "step": 29370
+    },
+    {
+      "epoch": 0.6310950723890536,
+      "grad_norm": 0.53849226672192,
+      "learning_rate": 6.104627736294664e-06,
+      "loss": 0.7008,
+      "step": 29380
+    },
+    {
+      "epoch": 0.6313098767023242,
+      "grad_norm": 0.5134656847802564,
+      "learning_rate": 6.098350563925035e-06,
+      "loss": 0.7165,
+      "step": 29390
+    },
+    {
+      "epoch": 0.6315246810155948,
+      "grad_norm": 0.5395193132941271,
+      "learning_rate": 6.092075204448734e-06,
+      "loss": 0.7263,
+      "step": 29400
+    },
+    {
+      "epoch": 0.6317394853288654,
+      "grad_norm": 0.5291663702673787,
+      "learning_rate": 6.085801660781585e-06,
+      "loss": 0.7226,
+      "step": 29410
+    },
+    {
+      "epoch": 0.6319542896421361,
+      "grad_norm": 0.4920471964797305,
+      "learning_rate": 6.079529935838584e-06,
+      "loss": 0.7253,
+      "step": 29420
+    },
+    {
+      "epoch": 0.6321690939554067,
+      "grad_norm": 0.5091308822653934,
+      "learning_rate": 6.0732600325338745e-06,
+      "loss": 0.7084,
+      "step": 29430
+    },
+    {
+      "epoch": 0.6323838982686772,
+      "grad_norm": 0.5083144836724668,
+      "learning_rate": 6.066991953780748e-06,
+      "loss": 0.7096,
+      "step": 29440
+    },
+    {
+      "epoch": 0.6325987025819478,
+      "grad_norm": 0.5553675193548707,
+      "learning_rate": 6.06072570249166e-06,
+      "loss": 0.7189,
+      "step": 29450
+    },
+    {
+      "epoch": 0.6328135068952184,
+      "grad_norm": 0.5042084082023858,
+      "learning_rate": 6.054461281578206e-06,
+      "loss": 0.7076,
+      "step": 29460
+    },
+    {
+      "epoch": 0.633028311208489,
+      "grad_norm": 0.4889129315159301,
+      "learning_rate": 6.048198693951142e-06,
+      "loss": 0.7034,
+      "step": 29470
+    },
+    {
+      "epoch": 0.6332431155217597,
+      "grad_norm": 0.49120737847508095,
+      "learning_rate": 6.041937942520363e-06,
+      "loss": 0.7162,
+      "step": 29480
+    },
+    {
+      "epoch": 0.6334579198350303,
+      "grad_norm": 0.5106224481761457,
+      "learning_rate": 6.035679030194917e-06,
+      "loss": 0.7108,
+      "step": 29490
+    },
+    {
+      "epoch": 0.6336727241483009,
+      "grad_norm": 0.5142715172785339,
+      "learning_rate": 6.02942195988299e-06,
+      "loss": 0.7182,
+      "step": 29500
+    },
+    {
+      "epoch": 0.6338875284615715,
+      "grad_norm": 0.5245286567422575,
+      "learning_rate": 6.023166734491919e-06,
+      "loss": 0.7069,
+      "step": 29510
+    },
+    {
+      "epoch": 0.6341023327748421,
+      "grad_norm": 0.49896046500592833,
+      "learning_rate": 6.0169133569281835e-06,
+      "loss": 0.7217,
+      "step": 29520
+    },
+    {
+      "epoch": 0.6343171370881128,
+      "grad_norm": 0.5076374052572614,
+      "learning_rate": 6.010661830097399e-06,
+      "loss": 0.7244,
+      "step": 29530
+    },
+    {
+      "epoch": 0.6345319414013834,
+      "grad_norm": 0.5036960875426778,
+      "learning_rate": 6.004412156904329e-06,
+      "loss": 0.7246,
+      "step": 29540
+    },
+    {
+      "epoch": 0.634746745714654,
+      "grad_norm": 0.5190186034362325,
+      "learning_rate": 5.9981643402528675e-06,
+      "loss": 0.7181,
+      "step": 29550
+    },
+    {
+      "epoch": 0.6349615500279245,
+      "grad_norm": 0.5185275006081458,
+      "learning_rate": 5.991918383046047e-06,
+      "loss": 0.7171,
+      "step": 29560
+    },
+    {
+      "epoch": 0.6351763543411951,
+      "grad_norm": 0.5092203362213314,
+      "learning_rate": 5.985674288186045e-06,
+      "loss": 0.7282,
+      "step": 29570
+    },
+    {
+      "epoch": 0.6353911586544658,
+      "grad_norm": 0.4997149128548273,
+      "learning_rate": 5.979432058574164e-06,
+      "loss": 0.7174,
+      "step": 29580
+    },
+    {
+      "epoch": 0.6356059629677364,
+      "grad_norm": 0.5068981922104722,
+      "learning_rate": 5.973191697110845e-06,
+      "loss": 0.7141,
+      "step": 29590
+    },
+    {
+      "epoch": 0.635820767281007,
+      "grad_norm": 0.5110915167748981,
+      "learning_rate": 5.966953206695653e-06,
+      "loss": 0.7102,
+      "step": 29600
+    },
+    {
+      "epoch": 0.6360355715942776,
+      "grad_norm": 0.511793837071635,
+      "learning_rate": 5.960716590227298e-06,
+      "loss": 0.7029,
+      "step": 29610
+    },
+    {
+      "epoch": 0.6362503759075482,
+      "grad_norm": 0.5010481752806605,
+      "learning_rate": 5.954481850603606e-06,
+      "loss": 0.7052,
+      "step": 29620
+    },
+    {
+      "epoch": 0.6364651802208189,
+      "grad_norm": 0.5184653587723411,
+      "learning_rate": 5.9482489907215324e-06,
+      "loss": 0.709,
+      "step": 29630
+    },
+    {
+      "epoch": 0.6366799845340895,
+      "grad_norm": 0.5186353614978536,
+      "learning_rate": 5.9420180134771735e-06,
+      "loss": 0.7256,
+      "step": 29640
+    },
+    {
+      "epoch": 0.6368947888473601,
+      "grad_norm": 0.5090791214108952,
+      "learning_rate": 5.935788921765728e-06,
+      "loss": 0.7125,
+      "step": 29650
+    },
+    {
+      "epoch": 0.6371095931606306,
+      "grad_norm": 0.5182097600656689,
+      "learning_rate": 5.92956171848154e-06,
+      "loss": 0.7095,
+      "step": 29660
+    },
+    {
+      "epoch": 0.6373243974739012,
+      "grad_norm": 0.49906621293560977,
+      "learning_rate": 5.923336406518059e-06,
+      "loss": 0.7179,
+      "step": 29670
+    },
+    {
+      "epoch": 0.6375392017871719,
+      "grad_norm": 0.49207042909075416,
+      "learning_rate": 5.91711298876787e-06,
+      "loss": 0.7141,
+      "step": 29680
+    },
+    {
+      "epoch": 0.6377540061004425,
+      "grad_norm": 0.491835738604546,
+      "learning_rate": 5.910891468122668e-06,
+      "loss": 0.7078,
+      "step": 29690
+    },
+    {
+      "epoch": 0.6379688104137131,
+      "grad_norm": 0.5078817311304347,
+      "learning_rate": 5.9046718474732666e-06,
+      "loss": 0.7072,
+      "step": 29700
+    },
+    {
+      "epoch": 0.6381836147269837,
+      "grad_norm": 0.5124871601736757,
+      "learning_rate": 5.898454129709606e-06,
+      "loss": 0.7202,
+      "step": 29710
+    },
+    {
+      "epoch": 0.6383984190402543,
+      "grad_norm": 0.5134905108899009,
+      "learning_rate": 5.8922383177207286e-06,
+      "loss": 0.7282,
+      "step": 29720
+    },
+    {
+      "epoch": 0.6386132233535249,
+      "grad_norm": 0.5155981047848882,
+      "learning_rate": 5.886024414394806e-06,
+      "loss": 0.7113,
+      "step": 29730
+    },
+    {
+      "epoch": 0.6388280276667956,
+      "grad_norm": 0.5060176620986123,
+      "learning_rate": 5.879812422619111e-06,
+      "loss": 0.7152,
+      "step": 29740
+    },
+    {
+      "epoch": 0.6390428319800662,
+      "grad_norm": 0.5031393620149552,
+      "learning_rate": 5.873602345280033e-06,
+      "loss": 0.7171,
+      "step": 29750
+    },
+    {
+      "epoch": 0.6392576362933368,
+      "grad_norm": 0.5091217506585529,
+      "learning_rate": 5.8673941852630735e-06,
+      "loss": 0.7072,
+      "step": 29760
+    },
+    {
+      "epoch": 0.6394724406066074,
+      "grad_norm": 0.5124995237491698,
+      "learning_rate": 5.861187945452836e-06,
+      "loss": 0.7179,
+      "step": 29770
+    },
+    {
+      "epoch": 0.6396872449198779,
+      "grad_norm": 0.5111197620823217,
+      "learning_rate": 5.854983628733046e-06,
+      "loss": 0.7265,
+      "step": 29780
+    },
+    {
+      "epoch": 0.6399020492331486,
+      "grad_norm": 0.5121226878357348,
+      "learning_rate": 5.848781237986516e-06,
+      "loss": 0.7058,
+      "step": 29790
+    },
+    {
+      "epoch": 0.6401168535464192,
+      "grad_norm": 0.5029940756710116,
+      "learning_rate": 5.842580776095186e-06,
+      "loss": 0.7242,
+      "step": 29800
+    },
+    {
+      "epoch": 0.6403316578596898,
+      "grad_norm": 0.518317509648807,
+      "learning_rate": 5.8363822459400766e-06,
+      "loss": 0.7122,
+      "step": 29810
+    },
+    {
+      "epoch": 0.6405464621729604,
+      "grad_norm": 0.513796209625438,
+      "learning_rate": 5.830185650401327e-06,
+      "loss": 0.7109,
+      "step": 29820
+    },
+    {
+      "epoch": 0.640761266486231,
+      "grad_norm": 0.5065915184714491,
+      "learning_rate": 5.82399099235817e-06,
+      "loss": 0.7185,
+      "step": 29830
+    },
+    {
+      "epoch": 0.6409760707995017,
+      "grad_norm": 0.5383003131895742,
+      "learning_rate": 5.817798274688945e-06,
+      "loss": 0.7149,
+      "step": 29840
+    },
+    {
+      "epoch": 0.6411908751127723,
+      "grad_norm": 0.5013183139637445,
+      "learning_rate": 5.811607500271086e-06,
+      "loss": 0.7061,
+      "step": 29850
+    },
+    {
+      "epoch": 0.6414056794260429,
+      "grad_norm": 0.5084891027020223,
+      "learning_rate": 5.805418671981112e-06,
+      "loss": 0.7078,
+      "step": 29860
+    },
+    {
+      "epoch": 0.6416204837393135,
+      "grad_norm": 0.5358856173892373,
+      "learning_rate": 5.7992317926946666e-06,
+      "loss": 0.7059,
+      "step": 29870
+    },
+    {
+      "epoch": 0.641835288052584,
+      "grad_norm": 0.5118937595826556,
+      "learning_rate": 5.793046865286457e-06,
+      "loss": 0.7084,
+      "step": 29880
+    },
+    {
+      "epoch": 0.6420500923658548,
+      "grad_norm": 0.5289502239664226,
+      "learning_rate": 5.786863892630301e-06,
+      "loss": 0.7186,
+      "step": 29890
+    },
+    {
+      "epoch": 0.6422648966791253,
+      "grad_norm": 0.5030440160930806,
+      "learning_rate": 5.78068287759911e-06,
+      "loss": 0.7093,
+      "step": 29900
+    },
+    {
+      "epoch": 0.6424797009923959,
+      "grad_norm": 0.5013233963022401,
+      "learning_rate": 5.774503823064865e-06,
+      "loss": 0.7046,
+      "step": 29910
+    },
+    {
+      "epoch": 0.6426945053056665,
+      "grad_norm": 0.5218815308682272,
+      "learning_rate": 5.768326731898668e-06,
+      "loss": 0.7093,
+      "step": 29920
+    },
+    {
+      "epoch": 0.6429093096189371,
+      "grad_norm": 0.4911776039589,
+      "learning_rate": 5.762151606970681e-06,
+      "loss": 0.6985,
+      "step": 29930
+    },
+    {
+      "epoch": 0.6431241139322078,
+      "grad_norm": 0.4990229286965237,
+      "learning_rate": 5.755978451150165e-06,
+      "loss": 0.7039,
+      "step": 29940
+    },
+    {
+      "epoch": 0.6433389182454784,
+      "grad_norm": 0.5017215398112681,
+      "learning_rate": 5.749807267305469e-06,
+      "loss": 0.719,
+      "step": 29950
+    },
+    {
+      "epoch": 0.643553722558749,
+      "grad_norm": 0.49447190334501206,
+      "learning_rate": 5.743638058304009e-06,
+      "loss": 0.7096,
+      "step": 29960
+    },
+    {
+      "epoch": 0.6437685268720196,
+      "grad_norm": 0.4932132883026573,
+      "learning_rate": 5.737470827012309e-06,
+      "loss": 0.719,
+      "step": 29970
+    },
+    {
+      "epoch": 0.6439833311852902,
+      "grad_norm": 0.4990600666743509,
+      "learning_rate": 5.731305576295951e-06,
+      "loss": 0.7119,
+      "step": 29980
+    },
+    {
+      "epoch": 0.6441981354985609,
+      "grad_norm": 0.49829438730309705,
+      "learning_rate": 5.72514230901961e-06,
+      "loss": 0.7111,
+      "step": 29990
+    },
+    {
+      "epoch": 0.6444129398118315,
+      "grad_norm": 0.5142379504720469,
+      "learning_rate": 5.7189810280470374e-06,
+      "loss": 0.7182,
+      "step": 30000
+    },
+    {
+      "epoch": 0.644627744125102,
+      "grad_norm": 0.5004279772636778,
+      "learning_rate": 5.71282173624105e-06,
+      "loss": 0.7123,
+      "step": 30010
+    },
+    {
+      "epoch": 0.6448425484383726,
+      "grad_norm": 0.4997893834727797,
+      "learning_rate": 5.706664436463563e-06,
+      "loss": 0.7069,
+      "step": 30020
+    },
+    {
+      "epoch": 0.6450573527516432,
+      "grad_norm": 0.5224573953092382,
+      "learning_rate": 5.700509131575538e-06,
+      "loss": 0.7315,
+      "step": 30030
+    },
+    {
+      "epoch": 0.6452721570649138,
+      "grad_norm": 0.5157124232819824,
+      "learning_rate": 5.6943558244370435e-06,
+      "loss": 0.7085,
+      "step": 30040
+    },
+    {
+      "epoch": 0.6454869613781845,
+      "grad_norm": 0.5098311956322868,
+      "learning_rate": 5.688204517907184e-06,
+      "loss": 0.698,
+      "step": 30050
+    },
+    {
+      "epoch": 0.6457017656914551,
+      "grad_norm": 0.5585768535083954,
+      "learning_rate": 5.6820552148441595e-06,
+      "loss": 0.721,
+      "step": 30060
+    },
+    {
+      "epoch": 0.6459165700047257,
+      "grad_norm": 0.5191401680283296,
+      "learning_rate": 5.675907918105233e-06,
+      "loss": 0.718,
+      "step": 30070
+    },
+    {
+      "epoch": 0.6461313743179963,
+      "grad_norm": 0.5154719384443771,
+      "learning_rate": 5.669762630546722e-06,
+      "loss": 0.7135,
+      "step": 30080
+    },
+    {
+      "epoch": 0.6463461786312669,
+      "grad_norm": 0.5161802110337572,
+      "learning_rate": 5.663619355024037e-06,
+      "loss": 0.7299,
+      "step": 30090
+    },
+    {
+      "epoch": 0.6465609829445376,
+      "grad_norm": 0.5251836486068403,
+      "learning_rate": 5.6574780943916265e-06,
+      "loss": 0.7139,
+      "step": 30100
+    },
+    {
+      "epoch": 0.6467757872578082,
+      "grad_norm": 0.5253745077383208,
+      "learning_rate": 5.651338851503017e-06,
+      "loss": 0.7081,
+      "step": 30110
+    },
+    {
+      "epoch": 0.6469905915710787,
+      "grad_norm": 0.499068951345218,
+      "learning_rate": 5.645201629210802e-06,
+      "loss": 0.7118,
+      "step": 30120
+    },
+    {
+      "epoch": 0.6472053958843493,
+      "grad_norm": 0.5212681373915045,
+      "learning_rate": 5.639066430366616e-06,
+      "loss": 0.7058,
+      "step": 30130
+    },
+    {
+      "epoch": 0.6474202001976199,
+      "grad_norm": 0.5192551695189365,
+      "learning_rate": 5.63293325782118e-06,
+      "loss": 0.7209,
+      "step": 30140
+    },
+    {
+      "epoch": 0.6476350045108906,
+      "grad_norm": 0.5009778176216481,
+      "learning_rate": 5.626802114424252e-06,
+      "loss": 0.7147,
+      "step": 30150
+    },
+    {
+      "epoch": 0.6478498088241612,
+      "grad_norm": 0.511657473010408,
+      "learning_rate": 5.620673003024654e-06,
+      "loss": 0.7111,
+      "step": 30160
+    },
+    {
+      "epoch": 0.6480646131374318,
+      "grad_norm": 0.4946400904637439,
+      "learning_rate": 5.614545926470272e-06,
+      "loss": 0.7039,
+      "step": 30170
+    },
+    {
+      "epoch": 0.6482794174507024,
+      "grad_norm": 0.5452566834347506,
+      "learning_rate": 5.608420887608033e-06,
+      "loss": 0.7124,
+      "step": 30180
+    },
+    {
+      "epoch": 0.648494221763973,
+      "grad_norm": 0.5187590204840195,
+      "learning_rate": 5.60229788928393e-06,
+      "loss": 0.7138,
+      "step": 30190
+    },
+    {
+      "epoch": 0.6487090260772437,
+      "grad_norm": 0.4961448077999401,
+      "learning_rate": 5.5961769343429885e-06,
+      "loss": 0.7088,
+      "step": 30200
+    },
+    {
+      "epoch": 0.6489238303905143,
+      "grad_norm": 0.49664867664736395,
+      "learning_rate": 5.590058025629315e-06,
+      "loss": 0.7043,
+      "step": 30210
+    },
+    {
+      "epoch": 0.6491386347037849,
+      "grad_norm": 0.5089161076297286,
+      "learning_rate": 5.5839411659860355e-06,
+      "loss": 0.696,
+      "step": 30220
+    },
+    {
+      "epoch": 0.6493534390170554,
+      "grad_norm": 0.5007244262694044,
+      "learning_rate": 5.577826358255339e-06,
+      "loss": 0.7104,
+      "step": 30230
+    },
+    {
+      "epoch": 0.649568243330326,
+      "grad_norm": 0.5116267642012309,
+      "learning_rate": 5.57171360527846e-06,
+      "loss": 0.7176,
+      "step": 30240
+    },
+    {
+      "epoch": 0.6497830476435967,
+      "grad_norm": 0.5110539680425629,
+      "learning_rate": 5.5656029098956755e-06,
+      "loss": 0.7176,
+      "step": 30250
+    },
+    {
+      "epoch": 0.6499978519568673,
+      "grad_norm": 0.5087432238272799,
+      "learning_rate": 5.559494274946311e-06,
+      "loss": 0.7045,
+      "step": 30260
+    },
+    {
+      "epoch": 0.6502126562701379,
+      "grad_norm": 0.5049786677025178,
+      "learning_rate": 5.553387703268725e-06,
+      "loss": 0.7105,
+      "step": 30270
+    },
+    {
+      "epoch": 0.6504274605834085,
+      "grad_norm": 0.5099641743518565,
+      "learning_rate": 5.547283197700324e-06,
+      "loss": 0.7056,
+      "step": 30280
+    },
+    {
+      "epoch": 0.6506422648966791,
+      "grad_norm": 0.5143331023536819,
+      "learning_rate": 5.541180761077556e-06,
+      "loss": 0.6937,
+      "step": 30290
+    },
+    {
+      "epoch": 0.6508570692099498,
+      "grad_norm": 0.49683778878230767,
+      "learning_rate": 5.535080396235906e-06,
+      "loss": 0.7087,
+      "step": 30300
+    },
+    {
+      "epoch": 0.6510718735232204,
+      "grad_norm": 0.533544500826429,
+      "learning_rate": 5.528982106009899e-06,
+      "loss": 0.7144,
+      "step": 30310
+    },
+    {
+      "epoch": 0.651286677836491,
+      "grad_norm": 0.5151062048248995,
+      "learning_rate": 5.5228858932330845e-06,
+      "loss": 0.7009,
+      "step": 30320
+    },
+    {
+      "epoch": 0.6515014821497616,
+      "grad_norm": 0.5122586500692712,
+      "learning_rate": 5.5167917607380605e-06,
+      "loss": 0.7031,
+      "step": 30330
+    },
+    {
+      "epoch": 0.6517162864630321,
+      "grad_norm": 0.5212961303785209,
+      "learning_rate": 5.510699711356451e-06,
+      "loss": 0.6934,
+      "step": 30340
+    },
+    {
+      "epoch": 0.6519310907763027,
+      "grad_norm": 0.5100624104875204,
+      "learning_rate": 5.504609747918916e-06,
+      "loss": 0.7259,
+      "step": 30350
+    },
+    {
+      "epoch": 0.6521458950895734,
+      "grad_norm": 0.5072539914616855,
+      "learning_rate": 5.498521873255145e-06,
+      "loss": 0.7156,
+      "step": 30360
+    },
+    {
+      "epoch": 0.652360699402844,
+      "grad_norm": 0.5025326619318836,
+      "learning_rate": 5.4924360901938574e-06,
+      "loss": 0.6881,
+      "step": 30370
+    },
+    {
+      "epoch": 0.6525755037161146,
+      "grad_norm": 0.5194768705340874,
+      "learning_rate": 5.486352401562796e-06,
+      "loss": 0.6997,
+      "step": 30380
+    },
+    {
+      "epoch": 0.6527903080293852,
+      "grad_norm": 0.4926980584558694,
+      "learning_rate": 5.480270810188732e-06,
+      "loss": 0.7079,
+      "step": 30390
+    },
+    {
+      "epoch": 0.6530051123426558,
+      "grad_norm": 0.5301706061164216,
+      "learning_rate": 5.4741913188974705e-06,
+      "loss": 0.7148,
+      "step": 30400
+    },
+    {
+      "epoch": 0.6532199166559265,
+      "grad_norm": 0.5077273663914399,
+      "learning_rate": 5.468113930513831e-06,
+      "loss": 0.702,
+      "step": 30410
+    },
+    {
+      "epoch": 0.6534347209691971,
+      "grad_norm": 0.5113783715432562,
+      "learning_rate": 5.462038647861659e-06,
+      "loss": 0.7237,
+      "step": 30420
+    },
+    {
+      "epoch": 0.6536495252824677,
+      "grad_norm": 0.5350113300965483,
+      "learning_rate": 5.455965473763824e-06,
+      "loss": 0.7077,
+      "step": 30430
+    },
+    {
+      "epoch": 0.6538643295957383,
+      "grad_norm": 0.534802331833065,
+      "learning_rate": 5.449894411042218e-06,
+      "loss": 0.7098,
+      "step": 30440
+    },
+    {
+      "epoch": 0.6540791339090088,
+      "grad_norm": 0.505768970566755,
+      "learning_rate": 5.4438254625177376e-06,
+      "loss": 0.7037,
+      "step": 30450
+    },
+    {
+      "epoch": 0.6542939382222795,
+      "grad_norm": 0.5243742881101912,
+      "learning_rate": 5.437758631010313e-06,
+      "loss": 0.7184,
+      "step": 30460
+    },
+    {
+      "epoch": 0.6545087425355501,
+      "grad_norm": 0.5174893952785108,
+      "learning_rate": 5.431693919338883e-06,
+      "loss": 0.7074,
+      "step": 30470
+    },
+    {
+      "epoch": 0.6547235468488207,
+      "grad_norm": 0.5018291601456127,
+      "learning_rate": 5.425631330321403e-06,
+      "loss": 0.7181,
+      "step": 30480
+    },
+    {
+      "epoch": 0.6549383511620913,
+      "grad_norm": 0.509935560750042,
+      "learning_rate": 5.41957086677485e-06,
+      "loss": 0.7272,
+      "step": 30490
+    },
+    {
+      "epoch": 0.6551531554753619,
+      "grad_norm": 0.5106281701236529,
+      "learning_rate": 5.413512531515195e-06,
+      "loss": 0.7181,
+      "step": 30500
+    },
+    {
+      "epoch": 0.6553679597886326,
+      "grad_norm": 0.5021707092734177,
+      "learning_rate": 5.407456327357437e-06,
+      "loss": 0.7122,
+      "step": 30510
+    },
+    {
+      "epoch": 0.6555827641019032,
+      "grad_norm": 0.5211648146663314,
+      "learning_rate": 5.401402257115576e-06,
+      "loss": 0.7042,
+      "step": 30520
+    },
+    {
+      "epoch": 0.6557975684151738,
+      "grad_norm": 0.5108248508895996,
+      "learning_rate": 5.395350323602624e-06,
+      "loss": 0.7027,
+      "step": 30530
+    },
+    {
+      "epoch": 0.6560123727284444,
+      "grad_norm": 0.504994079751055,
+      "learning_rate": 5.389300529630603e-06,
+      "loss": 0.7124,
+      "step": 30540
+    },
+    {
+      "epoch": 0.656227177041715,
+      "grad_norm": 0.5035319701169906,
+      "learning_rate": 5.383252878010528e-06,
+      "loss": 0.7117,
+      "step": 30550
+    },
+    {
+      "epoch": 0.6564419813549857,
+      "grad_norm": 0.5140239243917751,
+      "learning_rate": 5.3772073715524405e-06,
+      "loss": 0.7143,
+      "step": 30560
+    },
+    {
+      "epoch": 0.6566567856682562,
+      "grad_norm": 0.5014878071394002,
+      "learning_rate": 5.371164013065362e-06,
+      "loss": 0.7013,
+      "step": 30570
+    },
+    {
+      "epoch": 0.6568715899815268,
+      "grad_norm": 0.530273663858455,
+      "learning_rate": 5.365122805357331e-06,
+      "loss": 0.7137,
+      "step": 30580
+    },
+    {
+      "epoch": 0.6570863942947974,
+      "grad_norm": 0.5249382641477562,
+      "learning_rate": 5.3590837512353855e-06,
+      "loss": 0.7093,
+      "step": 30590
+    },
+    {
+      "epoch": 0.657301198608068,
+      "grad_norm": 0.5048796782346253,
+      "learning_rate": 5.353046853505548e-06,
+      "loss": 0.7091,
+      "step": 30600
+    },
+    {
+      "epoch": 0.6575160029213387,
+      "grad_norm": 0.5091027814069775,
+      "learning_rate": 5.347012114972865e-06,
+      "loss": 0.7028,
+      "step": 30610
+    },
+    {
+      "epoch": 0.6577308072346093,
+      "grad_norm": 0.5113428171167113,
+      "learning_rate": 5.34097953844135e-06,
+      "loss": 0.7142,
+      "step": 30620
+    },
+    {
+      "epoch": 0.6579456115478799,
+      "grad_norm": 0.4914977814112626,
+      "learning_rate": 5.334949126714044e-06,
+      "loss": 0.7137,
+      "step": 30630
+    },
+    {
+      "epoch": 0.6581604158611505,
+      "grad_norm": 0.4964243055566761,
+      "learning_rate": 5.328920882592953e-06,
+      "loss": 0.7087,
+      "step": 30640
+    },
+    {
+      "epoch": 0.6583752201744211,
+      "grad_norm": 0.5207847747861839,
+      "learning_rate": 5.322894808879091e-06,
+      "loss": 0.7134,
+      "step": 30650
+    },
+    {
+      "epoch": 0.6585900244876917,
+      "grad_norm": 0.5043669757937382,
+      "learning_rate": 5.3168709083724664e-06,
+      "loss": 0.7004,
+      "step": 30660
+    },
+    {
+      "epoch": 0.6588048288009624,
+      "grad_norm": 0.511841340255158,
+      "learning_rate": 5.310849183872059e-06,
+      "loss": 0.7081,
+      "step": 30670
+    },
+    {
+      "epoch": 0.659019633114233,
+      "grad_norm": 0.5084802396159405,
+      "learning_rate": 5.304829638175866e-06,
+      "loss": 0.7027,
+      "step": 30680
+    },
+    {
+      "epoch": 0.6592344374275035,
+      "grad_norm": 0.5068077969636335,
+      "learning_rate": 5.298812274080847e-06,
+      "loss": 0.7068,
+      "step": 30690
+    },
+    {
+      "epoch": 0.6594492417407741,
+      "grad_norm": 0.5156077038006672,
+      "learning_rate": 5.29279709438296e-06,
+      "loss": 0.7155,
+      "step": 30700
+    },
+    {
+      "epoch": 0.6596640460540447,
+      "grad_norm": 0.5004198196593579,
+      "learning_rate": 5.286784101877149e-06,
+      "loss": 0.7049,
+      "step": 30710
+    },
+    {
+      "epoch": 0.6598788503673154,
+      "grad_norm": 0.5306581949971477,
+      "learning_rate": 5.2807732993573295e-06,
+      "loss": 0.7078,
+      "step": 30720
+    },
+    {
+      "epoch": 0.660093654680586,
+      "grad_norm": 0.5036574433980202,
+      "learning_rate": 5.2747646896164215e-06,
+      "loss": 0.6997,
+      "step": 30730
+    },
+    {
+      "epoch": 0.6603084589938566,
+      "grad_norm": 0.52149079735347,
+      "learning_rate": 5.268758275446303e-06,
+      "loss": 0.7228,
+      "step": 30740
+    },
+    {
+      "epoch": 0.6605232633071272,
+      "grad_norm": 0.5079017436216969,
+      "learning_rate": 5.262754059637845e-06,
+      "loss": 0.7057,
+      "step": 30750
+    },
+    {
+      "epoch": 0.6607380676203978,
+      "grad_norm": 0.5013233677412027,
+      "learning_rate": 5.2567520449808975e-06,
+      "loss": 0.7175,
+      "step": 30760
+    },
+    {
+      "epoch": 0.6609528719336685,
+      "grad_norm": 0.5172472025182714,
+      "learning_rate": 5.2507522342642725e-06,
+      "loss": 0.7074,
+      "step": 30770
+    },
+    {
+      "epoch": 0.6611676762469391,
+      "grad_norm": 0.5302279379741301,
+      "learning_rate": 5.244754630275786e-06,
+      "loss": 0.7046,
+      "step": 30780
+    },
+    {
+      "epoch": 0.6613824805602097,
+      "grad_norm": 0.5103884640220524,
+      "learning_rate": 5.2387592358021954e-06,
+      "loss": 0.7087,
+      "step": 30790
+    },
+    {
+      "epoch": 0.6615972848734802,
+      "grad_norm": 0.5266498719563615,
+      "learning_rate": 5.2327660536292655e-06,
+      "loss": 0.7208,
+      "step": 30800
+    },
+    {
+      "epoch": 0.6618120891867508,
+      "grad_norm": 0.547112136398287,
+      "learning_rate": 5.226775086541705e-06,
+      "loss": 0.7246,
+      "step": 30810
+    },
+    {
+      "epoch": 0.6620268935000215,
+      "grad_norm": 0.5134807747932422,
+      "learning_rate": 5.220786337323205e-06,
+      "loss": 0.7038,
+      "step": 30820
+    },
+    {
+      "epoch": 0.6622416978132921,
+      "grad_norm": 0.5069417485434651,
+      "learning_rate": 5.214799808756432e-06,
+      "loss": 0.7267,
+      "step": 30830
+    },
+    {
+      "epoch": 0.6624565021265627,
+      "grad_norm": 0.5112475490765994,
+      "learning_rate": 5.208815503623001e-06,
+      "loss": 0.7049,
+      "step": 30840
+    },
+    {
+      "epoch": 0.6626713064398333,
+      "grad_norm": 0.49798758139115207,
+      "learning_rate": 5.202833424703524e-06,
+      "loss": 0.7049,
+      "step": 30850
+    },
+    {
+      "epoch": 0.6628861107531039,
+      "grad_norm": 0.5208558474736136,
+      "learning_rate": 5.196853574777547e-06,
+      "loss": 0.7054,
+      "step": 30860
+    },
+    {
+      "epoch": 0.6631009150663746,
+      "grad_norm": 0.5143732263552888,
+      "learning_rate": 5.190875956623602e-06,
+      "loss": 0.7161,
+      "step": 30870
+    },
+    {
+      "epoch": 0.6633157193796452,
+      "grad_norm": 0.5072200840559442,
+      "learning_rate": 5.184900573019179e-06,
+      "loss": 0.7143,
+      "step": 30880
+    },
+    {
+      "epoch": 0.6635305236929158,
+      "grad_norm": 0.5148338042683931,
+      "learning_rate": 5.1789274267407174e-06,
+      "loss": 0.7197,
+      "step": 30890
+    },
+    {
+      "epoch": 0.6637453280061864,
+      "grad_norm": 0.5355543429844896,
+      "learning_rate": 5.172956520563641e-06,
+      "loss": 0.6953,
+      "step": 30900
+    },
+    {
+      "epoch": 0.6639601323194569,
+      "grad_norm": 0.5111517548781118,
+      "learning_rate": 5.166987857262309e-06,
+      "loss": 0.704,
+      "step": 30910
+    },
+    {
+      "epoch": 0.6641749366327275,
+      "grad_norm": 0.5176946751784909,
+      "learning_rate": 5.16102143961005e-06,
+      "loss": 0.7041,
+      "step": 30920
+    },
+    {
+      "epoch": 0.6643897409459982,
+      "grad_norm": 0.5096883894063213,
+      "learning_rate": 5.155057270379149e-06,
+      "loss": 0.7065,
+      "step": 30930
+    },
+    {
+      "epoch": 0.6646045452592688,
+      "grad_norm": 0.5077111477162392,
+      "learning_rate": 5.149095352340847e-06,
+      "loss": 0.6996,
+      "step": 30940
+    },
+    {
+      "epoch": 0.6648193495725394,
+      "grad_norm": 0.49554593063940544,
+      "learning_rate": 5.143135688265337e-06,
+      "loss": 0.7097,
+      "step": 30950
+    },
+    {
+      "epoch": 0.66503415388581,
+      "grad_norm": 0.499479767542394,
+      "learning_rate": 5.137178280921759e-06,
+      "loss": 0.7176,
+      "step": 30960
+    },
+    {
+      "epoch": 0.6652489581990806,
+      "grad_norm": 0.49131352882866786,
+      "learning_rate": 5.131223133078213e-06,
+      "loss": 0.7166,
+      "step": 30970
+    },
+    {
+      "epoch": 0.6654637625123513,
+      "grad_norm": 0.5522986513731968,
+      "learning_rate": 5.125270247501746e-06,
+      "loss": 0.7076,
+      "step": 30980
+    },
+    {
+      "epoch": 0.6656785668256219,
+      "grad_norm": 0.5241166130212557,
+      "learning_rate": 5.119319626958354e-06,
+      "loss": 0.7071,
+      "step": 30990
+    },
+    {
+      "epoch": 0.6658933711388925,
+      "grad_norm": 0.5161672474322907,
+      "learning_rate": 5.113371274212984e-06,
+      "loss": 0.6892,
+      "step": 31000
+    },
+    {
+      "epoch": 0.666108175452163,
+      "grad_norm": 0.5019103502653095,
+      "learning_rate": 5.1074251920295194e-06,
+      "loss": 0.6991,
+      "step": 31010
+    },
+    {
+      "epoch": 0.6663229797654336,
+      "grad_norm": 0.5032297589781682,
+      "learning_rate": 5.1014813831708035e-06,
+      "loss": 0.7206,
+      "step": 31020
+    },
+    {
+      "epoch": 0.6665377840787043,
+      "grad_norm": 0.5002814476280841,
+      "learning_rate": 5.095539850398605e-06,
+      "loss": 0.7048,
+      "step": 31030
+    },
+    {
+      "epoch": 0.6667525883919749,
+      "grad_norm": 0.5156688113615284,
+      "learning_rate": 5.089600596473649e-06,
+      "loss": 0.7189,
+      "step": 31040
+    },
+    {
+      "epoch": 0.6669673927052455,
+      "grad_norm": 0.5188971013023798,
+      "learning_rate": 5.083663624155598e-06,
+      "loss": 0.7107,
+      "step": 31050
+    },
+    {
+      "epoch": 0.6671821970185161,
+      "grad_norm": 0.5203051062665321,
+      "learning_rate": 5.077728936203055e-06,
+      "loss": 0.7045,
+      "step": 31060
+    },
+    {
+      "epoch": 0.6673970013317867,
+      "grad_norm": 0.5041312268909838,
+      "learning_rate": 5.071796535373561e-06,
+      "loss": 0.7156,
+      "step": 31070
+    },
+    {
+      "epoch": 0.6676118056450574,
+      "grad_norm": 0.5159105151923982,
+      "learning_rate": 5.065866424423589e-06,
+      "loss": 0.6996,
+      "step": 31080
+    },
+    {
+      "epoch": 0.667826609958328,
+      "grad_norm": 0.5186673038935381,
+      "learning_rate": 5.059938606108554e-06,
+      "loss": 0.6947,
+      "step": 31090
+    },
+    {
+      "epoch": 0.6680414142715986,
+      "grad_norm": 0.5233119049957315,
+      "learning_rate": 5.054013083182808e-06,
+      "loss": 0.7172,
+      "step": 31100
+    },
+    {
+      "epoch": 0.6682562185848692,
+      "grad_norm": 0.5067736527978581,
+      "learning_rate": 5.048089858399632e-06,
+      "loss": 0.7021,
+      "step": 31110
+    },
+    {
+      "epoch": 0.6684710228981398,
+      "grad_norm": 0.5039632454745936,
+      "learning_rate": 5.042168934511237e-06,
+      "loss": 0.7059,
+      "step": 31120
+    },
+    {
+      "epoch": 0.6686858272114105,
+      "grad_norm": 0.5504075401326516,
+      "learning_rate": 5.036250314268775e-06,
+      "loss": 0.7125,
+      "step": 31130
+    },
+    {
+      "epoch": 0.668900631524681,
+      "grad_norm": 0.5140327017691517,
+      "learning_rate": 5.030334000422311e-06,
+      "loss": 0.716,
+      "step": 31140
+    },
+    {
+      "epoch": 0.6691154358379516,
+      "grad_norm": 0.5174324057304772,
+      "learning_rate": 5.024419995720854e-06,
+      "loss": 0.7091,
+      "step": 31150
+    },
+    {
+      "epoch": 0.6693302401512222,
+      "grad_norm": 0.5105348539745065,
+      "learning_rate": 5.0185083029123326e-06,
+      "loss": 0.7188,
+      "step": 31160
+    },
+    {
+      "epoch": 0.6695450444644928,
+      "grad_norm": 0.5104216621284471,
+      "learning_rate": 5.012598924743603e-06,
+      "loss": 0.6973,
+      "step": 31170
+    },
+    {
+      "epoch": 0.6697598487777635,
+      "grad_norm": 0.5015920565585724,
+      "learning_rate": 5.006691863960448e-06,
+      "loss": 0.7,
+      "step": 31180
+    },
+    {
+      "epoch": 0.6699746530910341,
+      "grad_norm": 0.5246820583424576,
+      "learning_rate": 5.000787123307562e-06,
+      "loss": 0.7043,
+      "step": 31190
+    },
+    {
+      "epoch": 0.6701894574043047,
+      "grad_norm": 0.5157821738943842,
+      "learning_rate": 4.994884705528583e-06,
+      "loss": 0.7145,
+      "step": 31200
+    },
+    {
+      "epoch": 0.6704042617175753,
+      "grad_norm": 0.49631469998188016,
+      "learning_rate": 4.9889846133660495e-06,
+      "loss": 0.7218,
+      "step": 31210
+    },
+    {
+      "epoch": 0.6706190660308459,
+      "grad_norm": 0.49915868212535197,
+      "learning_rate": 4.983086849561427e-06,
+      "loss": 0.7047,
+      "step": 31220
+    },
+    {
+      "epoch": 0.6708338703441165,
+      "grad_norm": 0.5307081421883926,
+      "learning_rate": 4.977191416855102e-06,
+      "loss": 0.7166,
+      "step": 31230
+    },
+    {
+      "epoch": 0.6710486746573872,
+      "grad_norm": 0.5169884060557652,
+      "learning_rate": 4.971298317986374e-06,
+      "loss": 0.7257,
+      "step": 31240
+    },
+    {
+      "epoch": 0.6712634789706577,
+      "grad_norm": 0.5009977824029985,
+      "learning_rate": 4.965407555693464e-06,
+      "loss": 0.7004,
+      "step": 31250
+    },
+    {
+      "epoch": 0.6714782832839283,
+      "grad_norm": 0.5155270739896842,
+      "learning_rate": 4.9595191327134915e-06,
+      "loss": 0.7153,
+      "step": 31260
+    },
+    {
+      "epoch": 0.6716930875971989,
+      "grad_norm": 0.5044556594755597,
+      "learning_rate": 4.9536330517825085e-06,
+      "loss": 0.7048,
+      "step": 31270
+    },
+    {
+      "epoch": 0.6719078919104695,
+      "grad_norm": 0.5050373362709072,
+      "learning_rate": 4.947749315635467e-06,
+      "loss": 0.7176,
+      "step": 31280
+    },
+    {
+      "epoch": 0.6721226962237402,
+      "grad_norm": 0.4952372981882212,
+      "learning_rate": 4.9418679270062345e-06,
+      "loss": 0.7315,
+      "step": 31290
+    },
+    {
+      "epoch": 0.6723375005370108,
+      "grad_norm": 0.5075005573307158,
+      "learning_rate": 4.935988888627589e-06,
+      "loss": 0.6999,
+      "step": 31300
+    },
+    {
+      "epoch": 0.6725523048502814,
+      "grad_norm": 0.5237607258041448,
+      "learning_rate": 4.930112203231202e-06,
+      "loss": 0.7199,
+      "step": 31310
+    },
+    {
+      "epoch": 0.672767109163552,
+      "grad_norm": 0.5140569272272469,
+      "learning_rate": 4.924237873547678e-06,
+      "loss": 0.7062,
+      "step": 31320
+    },
+    {
+      "epoch": 0.6729819134768226,
+      "grad_norm": 0.5064496524470513,
+      "learning_rate": 4.9183659023065035e-06,
+      "loss": 0.7098,
+      "step": 31330
+    },
+    {
+      "epoch": 0.6731967177900933,
+      "grad_norm": 0.49850427047279267,
+      "learning_rate": 4.912496292236078e-06,
+      "loss": 0.7127,
+      "step": 31340
+    },
+    {
+      "epoch": 0.6734115221033639,
+      "grad_norm": 0.5129869200842382,
+      "learning_rate": 4.906629046063709e-06,
+      "loss": 0.6957,
+      "step": 31350
+    },
+    {
+      "epoch": 0.6736263264166344,
+      "grad_norm": 0.5037255148710768,
+      "learning_rate": 4.900764166515589e-06,
+      "loss": 0.7053,
+      "step": 31360
+    },
+    {
+      "epoch": 0.673841130729905,
+      "grad_norm": 0.5218708185856743,
+      "learning_rate": 4.894901656316837e-06,
+      "loss": 0.6969,
+      "step": 31370
+    },
+    {
+      "epoch": 0.6740559350431756,
+      "grad_norm": 0.5106451275587679,
+      "learning_rate": 4.889041518191442e-06,
+      "loss": 0.7029,
+      "step": 31380
+    },
+    {
+      "epoch": 0.6742707393564463,
+      "grad_norm": 0.504620024778387,
+      "learning_rate": 4.883183754862319e-06,
+      "loss": 0.7108,
+      "step": 31390
+    },
+    {
+      "epoch": 0.6744855436697169,
+      "grad_norm": 0.500245116066161,
+      "learning_rate": 4.877328369051254e-06,
+      "loss": 0.7023,
+      "step": 31400
+    },
+    {
+      "epoch": 0.6747003479829875,
+      "grad_norm": 0.4984112969568294,
+      "learning_rate": 4.871475363478945e-06,
+      "loss": 0.6984,
+      "step": 31410
+    },
+    {
+      "epoch": 0.6749151522962581,
+      "grad_norm": 0.490340528966453,
+      "learning_rate": 4.865624740864981e-06,
+      "loss": 0.7171,
+      "step": 31420
+    },
+    {
+      "epoch": 0.6751299566095287,
+      "grad_norm": 0.5154565088297944,
+      "learning_rate": 4.859776503927831e-06,
+      "loss": 0.7254,
+      "step": 31430
+    },
+    {
+      "epoch": 0.6753447609227994,
+      "grad_norm": 0.5162711441600745,
+      "learning_rate": 4.8539306553848835e-06,
+      "loss": 0.7,
+      "step": 31440
+    },
+    {
+      "epoch": 0.67555956523607,
+      "grad_norm": 0.5296439320774667,
+      "learning_rate": 4.848087197952385e-06,
+      "loss": 0.7201,
+      "step": 31450
+    },
+    {
+      "epoch": 0.6757743695493406,
+      "grad_norm": 0.5211601804495921,
+      "learning_rate": 4.842246134345492e-06,
+      "loss": 0.7077,
+      "step": 31460
+    },
+    {
+      "epoch": 0.6759891738626111,
+      "grad_norm": 0.5204672709973052,
+      "learning_rate": 4.8364074672782445e-06,
+      "loss": 0.7025,
+      "step": 31470
+    },
+    {
+      "epoch": 0.6762039781758817,
+      "grad_norm": 0.5443914115894181,
+      "learning_rate": 4.8305711994635585e-06,
+      "loss": 0.7161,
+      "step": 31480
+    },
+    {
+      "epoch": 0.6764187824891524,
+      "grad_norm": 0.5398665273534337,
+      "learning_rate": 4.8247373336132565e-06,
+      "loss": 0.7152,
+      "step": 31490
+    },
+    {
+      "epoch": 0.676633586802423,
+      "grad_norm": 0.5205394432033161,
+      "learning_rate": 4.818905872438021e-06,
+      "loss": 0.7045,
+      "step": 31500
+    },
+    {
+      "epoch": 0.6768483911156936,
+      "grad_norm": 0.505861773994322,
+      "learning_rate": 4.813076818647434e-06,
+      "loss": 0.7176,
+      "step": 31510
+    },
+    {
+      "epoch": 0.6770631954289642,
+      "grad_norm": 0.5138045231439949,
+      "learning_rate": 4.807250174949955e-06,
+      "loss": 0.7092,
+      "step": 31520
+    },
+    {
+      "epoch": 0.6772779997422348,
+      "grad_norm": 0.5198137824432044,
+      "learning_rate": 4.801425944052911e-06,
+      "loss": 0.7115,
+      "step": 31530
+    },
+    {
+      "epoch": 0.6774928040555054,
+      "grad_norm": 0.5121939716704446,
+      "learning_rate": 4.795604128662536e-06,
+      "loss": 0.7001,
+      "step": 31540
+    },
+    {
+      "epoch": 0.6777076083687761,
+      "grad_norm": 0.5036727571961231,
+      "learning_rate": 4.78978473148391e-06,
+      "loss": 0.7005,
+      "step": 31550
+    },
+    {
+      "epoch": 0.6779224126820467,
+      "grad_norm": 0.4963339926656166,
+      "learning_rate": 4.783967755221008e-06,
+      "loss": 0.6981,
+      "step": 31560
+    },
+    {
+      "epoch": 0.6781372169953173,
+      "grad_norm": 0.5026469805657872,
+      "learning_rate": 4.778153202576678e-06,
+      "loss": 0.7061,
+      "step": 31570
+    },
+    {
+      "epoch": 0.6783520213085878,
+      "grad_norm": 0.4888239012176708,
+      "learning_rate": 4.772341076252639e-06,
+      "loss": 0.6928,
+      "step": 31580
+    },
+    {
+      "epoch": 0.6785668256218584,
+      "grad_norm": 0.5033386278198433,
+      "learning_rate": 4.766531378949487e-06,
+      "loss": 0.7066,
+      "step": 31590
+    },
+    {
+      "epoch": 0.6787816299351291,
+      "grad_norm": 0.5076861079010834,
+      "learning_rate": 4.7607241133666735e-06,
+      "loss": 0.7045,
+      "step": 31600
+    },
+    {
+      "epoch": 0.6789964342483997,
+      "grad_norm": 0.5368479007734174,
+      "learning_rate": 4.75491928220255e-06,
+      "loss": 0.7136,
+      "step": 31610
+    },
+    {
+      "epoch": 0.6792112385616703,
+      "grad_norm": 0.5082100349533373,
+      "learning_rate": 4.749116888154306e-06,
+      "loss": 0.7046,
+      "step": 31620
+    },
+    {
+      "epoch": 0.6794260428749409,
+      "grad_norm": 0.5133784115796547,
+      "learning_rate": 4.743316933918016e-06,
+      "loss": 0.7035,
+      "step": 31630
+    },
+    {
+      "epoch": 0.6796408471882115,
+      "grad_norm": 0.5274702887841751,
+      "learning_rate": 4.737519422188617e-06,
+      "loss": 0.707,
+      "step": 31640
+    },
+    {
+      "epoch": 0.6798556515014822,
+      "grad_norm": 0.5053552682276936,
+      "learning_rate": 4.73172435565991e-06,
+      "loss": 0.7126,
+      "step": 31650
+    },
+    {
+      "epoch": 0.6800704558147528,
+      "grad_norm": 0.529992238236458,
+      "learning_rate": 4.725931737024565e-06,
+      "loss": 0.7124,
+      "step": 31660
+    },
+    {
+      "epoch": 0.6802852601280234,
+      "grad_norm": 0.5316204503933539,
+      "learning_rate": 4.720141568974104e-06,
+      "loss": 0.7188,
+      "step": 31670
+    },
+    {
+      "epoch": 0.680500064441294,
+      "grad_norm": 0.5054012212316678,
+      "learning_rate": 4.714353854198918e-06,
+      "loss": 0.7118,
+      "step": 31680
+    },
+    {
+      "epoch": 0.6807148687545646,
+      "grad_norm": 0.5143920728251262,
+      "learning_rate": 4.708568595388258e-06,
+      "loss": 0.7087,
+      "step": 31690
+    },
+    {
+      "epoch": 0.6809296730678353,
+      "grad_norm": 0.5322332430035215,
+      "learning_rate": 4.7027857952302315e-06,
+      "loss": 0.7123,
+      "step": 31700
+    },
+    {
+      "epoch": 0.6811444773811058,
+      "grad_norm": 0.520262064220361,
+      "learning_rate": 4.697005456411811e-06,
+      "loss": 0.7156,
+      "step": 31710
+    },
+    {
+      "epoch": 0.6813592816943764,
+      "grad_norm": 0.5095960914177793,
+      "learning_rate": 4.691227581618808e-06,
+      "loss": 0.7102,
+      "step": 31720
+    },
+    {
+      "epoch": 0.681574086007647,
+      "grad_norm": 0.509847182822789,
+      "learning_rate": 4.685452173535906e-06,
+      "loss": 0.7073,
+      "step": 31730
+    },
+    {
+      "epoch": 0.6817888903209176,
+      "grad_norm": 0.4928338256132167,
+      "learning_rate": 4.679679234846636e-06,
+      "loss": 0.7094,
+      "step": 31740
+    },
+    {
+      "epoch": 0.6820036946341883,
+      "grad_norm": 0.5281801273797183,
+      "learning_rate": 4.67390876823338e-06,
+      "loss": 0.7103,
+      "step": 31750
+    },
+    {
+      "epoch": 0.6822184989474589,
+      "grad_norm": 0.5147073369745232,
+      "learning_rate": 4.668140776377378e-06,
+      "loss": 0.7138,
+      "step": 31760
+    },
+    {
+      "epoch": 0.6824333032607295,
+      "grad_norm": 0.518041164686382,
+      "learning_rate": 4.66237526195871e-06,
+      "loss": 0.7056,
+      "step": 31770
+    },
+    {
+      "epoch": 0.6826481075740001,
+      "grad_norm": 0.5032736884933615,
+      "learning_rate": 4.656612227656318e-06,
+      "loss": 0.7005,
+      "step": 31780
+    },
+    {
+      "epoch": 0.6828629118872707,
+      "grad_norm": 0.5164912240008823,
+      "learning_rate": 4.650851676147976e-06,
+      "loss": 0.7112,
+      "step": 31790
+    },
+    {
+      "epoch": 0.6830777162005413,
+      "grad_norm": 0.5190164964456715,
+      "learning_rate": 4.645093610110314e-06,
+      "loss": 0.7031,
+      "step": 31800
+    },
+    {
+      "epoch": 0.683292520513812,
+      "grad_norm": 0.499193870377874,
+      "learning_rate": 4.639338032218806e-06,
+      "loss": 0.6953,
+      "step": 31810
+    },
+    {
+      "epoch": 0.6835073248270825,
+      "grad_norm": 0.5107643101481627,
+      "learning_rate": 4.633584945147771e-06,
+      "loss": 0.7058,
+      "step": 31820
+    },
+    {
+      "epoch": 0.6837221291403531,
+      "grad_norm": 0.4788958047068638,
+      "learning_rate": 4.627834351570368e-06,
+      "loss": 0.6999,
+      "step": 31830
+    },
+    {
+      "epoch": 0.6839369334536237,
+      "grad_norm": 0.5243752262151895,
+      "learning_rate": 4.622086254158602e-06,
+      "loss": 0.7093,
+      "step": 31840
+    },
+    {
+      "epoch": 0.6841517377668943,
+      "grad_norm": 0.5117247308515849,
+      "learning_rate": 4.616340655583307e-06,
+      "loss": 0.7127,
+      "step": 31850
+    },
+    {
+      "epoch": 0.684366542080165,
+      "grad_norm": 0.5039057854230863,
+      "learning_rate": 4.610597558514167e-06,
+      "loss": 0.7155,
+      "step": 31860
+    },
+    {
+      "epoch": 0.6845813463934356,
+      "grad_norm": 0.5110748098732393,
+      "learning_rate": 4.6048569656197005e-06,
+      "loss": 0.7182,
+      "step": 31870
+    },
+    {
+      "epoch": 0.6847961507067062,
+      "grad_norm": 0.5033241610487543,
+      "learning_rate": 4.599118879567262e-06,
+      "loss": 0.7178,
+      "step": 31880
+    },
+    {
+      "epoch": 0.6850109550199768,
+      "grad_norm": 0.5330181219085199,
+      "learning_rate": 4.593383303023047e-06,
+      "loss": 0.6988,
+      "step": 31890
+    },
+    {
+      "epoch": 0.6852257593332474,
+      "grad_norm": 0.5146643640112155,
+      "learning_rate": 4.587650238652068e-06,
+      "loss": 0.7071,
+      "step": 31900
+    },
+    {
+      "epoch": 0.6854405636465181,
+      "grad_norm": 0.5558498206129705,
+      "learning_rate": 4.581919689118187e-06,
+      "loss": 0.7089,
+      "step": 31910
+    },
+    {
+      "epoch": 0.6856553679597887,
+      "grad_norm": 0.5051408069436808,
+      "learning_rate": 4.576191657084093e-06,
+      "loss": 0.7171,
+      "step": 31920
+    },
+    {
+      "epoch": 0.6858701722730592,
+      "grad_norm": 0.4947858066673929,
+      "learning_rate": 4.570466145211303e-06,
+      "loss": 0.705,
+      "step": 31930
+    },
+    {
+      "epoch": 0.6860849765863298,
+      "grad_norm": 0.49215658445409516,
+      "learning_rate": 4.564743156160167e-06,
+      "loss": 0.7004,
+      "step": 31940
+    },
+    {
+      "epoch": 0.6862997808996004,
+      "grad_norm": 0.5052799951057044,
+      "learning_rate": 4.559022692589851e-06,
+      "loss": 0.7048,
+      "step": 31950
+    },
+    {
+      "epoch": 0.6865145852128711,
+      "grad_norm": 0.49573841106926664,
+      "learning_rate": 4.553304757158372e-06,
+      "loss": 0.7074,
+      "step": 31960
+    },
+    {
+      "epoch": 0.6867293895261417,
+      "grad_norm": 0.5143793031370387,
+      "learning_rate": 4.547589352522542e-06,
+      "loss": 0.7039,
+      "step": 31970
+    },
+    {
+      "epoch": 0.6869441938394123,
+      "grad_norm": 0.504158519855662,
+      "learning_rate": 4.541876481338019e-06,
+      "loss": 0.7006,
+      "step": 31980
+    },
+    {
+      "epoch": 0.6871589981526829,
+      "grad_norm": 0.503671838515148,
+      "learning_rate": 4.536166146259276e-06,
+      "loss": 0.7147,
+      "step": 31990
+    },
+    {
+      "epoch": 0.6873738024659535,
+      "grad_norm": 0.5287913454757696,
+      "learning_rate": 4.530458349939606e-06,
+      "loss": 0.7243,
+      "step": 32000
+    },
+    {
+      "epoch": 0.6875886067792242,
+      "grad_norm": 0.5016494023049036,
+      "learning_rate": 4.524753095031133e-06,
+      "loss": 0.7028,
+      "step": 32010
+    },
+    {
+      "epoch": 0.6878034110924948,
+      "grad_norm": 0.5166519761915471,
+      "learning_rate": 4.519050384184775e-06,
+      "loss": 0.7129,
+      "step": 32020
+    },
+    {
+      "epoch": 0.6880182154057654,
+      "grad_norm": 0.5158977598310385,
+      "learning_rate": 4.513350220050305e-06,
+      "loss": 0.7071,
+      "step": 32030
+    },
+    {
+      "epoch": 0.6882330197190359,
+      "grad_norm": 0.5016269659965205,
+      "learning_rate": 4.507652605276278e-06,
+      "loss": 0.7047,
+      "step": 32040
+    },
+    {
+      "epoch": 0.6884478240323065,
+      "grad_norm": 0.5163690901069261,
+      "learning_rate": 4.501957542510085e-06,
+      "loss": 0.7117,
+      "step": 32050
+    },
+    {
+      "epoch": 0.6886626283455772,
+      "grad_norm": 0.5108399393202574,
+      "learning_rate": 4.4962650343979255e-06,
+      "loss": 0.7183,
+      "step": 32060
+    },
+    {
+      "epoch": 0.6888774326588478,
+      "grad_norm": 0.5084486965060432,
+      "learning_rate": 4.490575083584804e-06,
+      "loss": 0.7138,
+      "step": 32070
+    },
+    {
+      "epoch": 0.6890922369721184,
+      "grad_norm": 0.5218646619009035,
+      "learning_rate": 4.484887692714557e-06,
+      "loss": 0.7106,
+      "step": 32080
+    },
+    {
+      "epoch": 0.689307041285389,
+      "grad_norm": 0.5201258534596843,
+      "learning_rate": 4.479202864429808e-06,
+      "loss": 0.6951,
+      "step": 32090
+    },
+    {
+      "epoch": 0.6895218455986596,
+      "grad_norm": 0.516396272076133,
+      "learning_rate": 4.473520601372006e-06,
+      "loss": 0.7136,
+      "step": 32100
+    },
+    {
+      "epoch": 0.6897366499119302,
+      "grad_norm": 0.5032875799939698,
+      "learning_rate": 4.467840906181403e-06,
+      "loss": 0.6944,
+      "step": 32110
+    },
+    {
+      "epoch": 0.6899514542252009,
+      "grad_norm": 0.49750240137602164,
+      "learning_rate": 4.462163781497051e-06,
+      "loss": 0.7029,
+      "step": 32120
+    },
+    {
+      "epoch": 0.6901662585384715,
+      "grad_norm": 0.5033581986528483,
+      "learning_rate": 4.456489229956826e-06,
+      "loss": 0.7043,
+      "step": 32130
+    },
+    {
+      "epoch": 0.6903810628517421,
+      "grad_norm": 0.5166768407944105,
+      "learning_rate": 4.450817254197386e-06,
+      "loss": 0.6929,
+      "step": 32140
+    },
+    {
+      "epoch": 0.6905958671650126,
+      "grad_norm": 0.4918195538890641,
+      "learning_rate": 4.4451478568542064e-06,
+      "loss": 0.7109,
+      "step": 32150
+    },
+    {
+      "epoch": 0.6908106714782832,
+      "grad_norm": 0.5145980685669729,
+      "learning_rate": 4.439481040561565e-06,
+      "loss": 0.7008,
+      "step": 32160
+    },
+    {
+      "epoch": 0.6910254757915539,
+      "grad_norm": 0.5170120361869788,
+      "learning_rate": 4.433816807952525e-06,
+      "loss": 0.7102,
+      "step": 32170
+    },
+    {
+      "epoch": 0.6912402801048245,
+      "grad_norm": 0.5390912485506946,
+      "learning_rate": 4.428155161658976e-06,
+      "loss": 0.6978,
+      "step": 32180
+    },
+    {
+      "epoch": 0.6914550844180951,
+      "grad_norm": 0.513671096300247,
+      "learning_rate": 4.422496104311574e-06,
+      "loss": 0.704,
+      "step": 32190
+    },
+    {
+      "epoch": 0.6916698887313657,
+      "grad_norm": 0.5512396747122822,
+      "learning_rate": 4.416839638539804e-06,
+      "loss": 0.7071,
+      "step": 32200
+    },
+    {
+      "epoch": 0.6918846930446363,
+      "grad_norm": 0.5039999780605813,
+      "learning_rate": 4.411185766971919e-06,
+      "loss": 0.7098,
+      "step": 32210
+    },
+    {
+      "epoch": 0.692099497357907,
+      "grad_norm": 0.5056673300275484,
+      "learning_rate": 4.4055344922349845e-06,
+      "loss": 0.7002,
+      "step": 32220
+    },
+    {
+      "epoch": 0.6923143016711776,
+      "grad_norm": 0.5228713724860102,
+      "learning_rate": 4.399885816954855e-06,
+      "loss": 0.7069,
+      "step": 32230
+    },
+    {
+      "epoch": 0.6925291059844482,
+      "grad_norm": 0.5154170548308558,
+      "learning_rate": 4.394239743756166e-06,
+      "loss": 0.7173,
+      "step": 32240
+    },
+    {
+      "epoch": 0.6927439102977188,
+      "grad_norm": 0.4932044034667852,
+      "learning_rate": 4.3885962752623675e-06,
+      "loss": 0.7,
+      "step": 32250
+    },
+    {
+      "epoch": 0.6929587146109893,
+      "grad_norm": 0.5291577432915663,
+      "learning_rate": 4.382955414095675e-06,
+      "loss": 0.707,
+      "step": 32260
+    },
+    {
+      "epoch": 0.69317351892426,
+      "grad_norm": 0.5062081138297674,
+      "learning_rate": 4.3773171628771075e-06,
+      "loss": 0.7053,
+      "step": 32270
+    },
+    {
+      "epoch": 0.6933883232375306,
+      "grad_norm": 0.5235760210430898,
+      "learning_rate": 4.371681524226469e-06,
+      "loss": 0.7203,
+      "step": 32280
+    },
+    {
+      "epoch": 0.6936031275508012,
+      "grad_norm": 0.5061927155930297,
+      "learning_rate": 4.366048500762335e-06,
+      "loss": 0.7112,
+      "step": 32290
+    },
+    {
+      "epoch": 0.6938179318640718,
+      "grad_norm": 0.5131390727212936,
+      "learning_rate": 4.360418095102097e-06,
+      "loss": 0.7095,
+      "step": 32300
+    },
+    {
+      "epoch": 0.6940327361773424,
+      "grad_norm": 0.5280050051289648,
+      "learning_rate": 4.354790309861897e-06,
+      "loss": 0.7088,
+      "step": 32310
+    },
+    {
+      "epoch": 0.6942475404906131,
+      "grad_norm": 0.5170240200200222,
+      "learning_rate": 4.349165147656679e-06,
+      "loss": 0.6946,
+      "step": 32320
+    },
+    {
+      "epoch": 0.6944623448038837,
+      "grad_norm": 0.49194844383818315,
+      "learning_rate": 4.343542611100161e-06,
+      "loss": 0.7109,
+      "step": 32330
+    },
+    {
+      "epoch": 0.6946771491171543,
+      "grad_norm": 0.5021844766055354,
+      "learning_rate": 4.337922702804846e-06,
+      "loss": 0.6949,
+      "step": 32340
+    },
+    {
+      "epoch": 0.6948919534304249,
+      "grad_norm": 0.5099714451400136,
+      "learning_rate": 4.332305425382013e-06,
+      "loss": 0.707,
+      "step": 32350
+    },
+    {
+      "epoch": 0.6951067577436955,
+      "grad_norm": 0.5212733236473239,
+      "learning_rate": 4.326690781441711e-06,
+      "loss": 0.6997,
+      "step": 32360
+    },
+    {
+      "epoch": 0.6953215620569662,
+      "grad_norm": 0.4903110024798184,
+      "learning_rate": 4.3210787735927824e-06,
+      "loss": 0.7021,
+      "step": 32370
+    },
+    {
+      "epoch": 0.6955363663702367,
+      "grad_norm": 0.505886652482517,
+      "learning_rate": 4.315469404442829e-06,
+      "loss": 0.7,
+      "step": 32380
+    },
+    {
+      "epoch": 0.6957511706835073,
+      "grad_norm": 0.5348702621789376,
+      "learning_rate": 4.309862676598233e-06,
+      "loss": 0.6989,
+      "step": 32390
+    },
+    {
+      "epoch": 0.6959659749967779,
+      "grad_norm": 0.5053002509205807,
+      "learning_rate": 4.304258592664151e-06,
+      "loss": 0.705,
+      "step": 32400
+    },
+    {
+      "epoch": 0.6961807793100485,
+      "grad_norm": 0.49313377903105116,
+      "learning_rate": 4.298657155244508e-06,
+      "loss": 0.7126,
+      "step": 32410
+    },
+    {
+      "epoch": 0.6963955836233191,
+      "grad_norm": 0.49836073223251326,
+      "learning_rate": 4.293058366942004e-06,
+      "loss": 0.6942,
+      "step": 32420
+    },
+    {
+      "epoch": 0.6966103879365898,
+      "grad_norm": 0.5033826955151206,
+      "learning_rate": 4.287462230358097e-06,
+      "loss": 0.69,
+      "step": 32430
+    },
+    {
+      "epoch": 0.6968251922498604,
+      "grad_norm": 0.5124591714204908,
+      "learning_rate": 4.281868748093023e-06,
+      "loss": 0.7088,
+      "step": 32440
+    },
+    {
+      "epoch": 0.697039996563131,
+      "grad_norm": 0.5152063252453369,
+      "learning_rate": 4.276277922745784e-06,
+      "loss": 0.7235,
+      "step": 32450
+    },
+    {
+      "epoch": 0.6972548008764016,
+      "grad_norm": 0.5068795899169701,
+      "learning_rate": 4.2706897569141435e-06,
+      "loss": 0.6995,
+      "step": 32460
+    },
+    {
+      "epoch": 0.6974696051896722,
+      "grad_norm": 0.5194180550407479,
+      "learning_rate": 4.2651042531946364e-06,
+      "loss": 0.7132,
+      "step": 32470
+    },
+    {
+      "epoch": 0.6976844095029429,
+      "grad_norm": 0.5093344990238854,
+      "learning_rate": 4.259521414182547e-06,
+      "loss": 0.6991,
+      "step": 32480
+    },
+    {
+      "epoch": 0.6978992138162134,
+      "grad_norm": 0.5226727716813793,
+      "learning_rate": 4.253941242471934e-06,
+      "loss": 0.7052,
+      "step": 32490
+    },
+    {
+      "epoch": 0.698114018129484,
+      "grad_norm": 0.5101918234709942,
+      "learning_rate": 4.248363740655612e-06,
+      "loss": 0.709,
+      "step": 32500
+    },
+    {
+      "epoch": 0.6983288224427546,
+      "grad_norm": 0.5101101585718761,
+      "learning_rate": 4.242788911325156e-06,
+      "loss": 0.7108,
+      "step": 32510
+    },
+    {
+      "epoch": 0.6985436267560252,
+      "grad_norm": 0.4867970175259229,
+      "learning_rate": 4.2372167570709e-06,
+      "loss": 0.7042,
+      "step": 32520
+    },
+    {
+      "epoch": 0.6987584310692959,
+      "grad_norm": 0.5014423746270559,
+      "learning_rate": 4.231647280481936e-06,
+      "loss": 0.694,
+      "step": 32530
+    },
+    {
+      "epoch": 0.6989732353825665,
+      "grad_norm": 0.5146828954090973,
+      "learning_rate": 4.226080484146103e-06,
+      "loss": 0.7207,
+      "step": 32540
+    },
+    {
+      "epoch": 0.6991880396958371,
+      "grad_norm": 0.4901928041129853,
+      "learning_rate": 4.220516370650007e-06,
+      "loss": 0.6995,
+      "step": 32550
+    },
+    {
+      "epoch": 0.6994028440091077,
+      "grad_norm": 0.5066794982082476,
+      "learning_rate": 4.214954942578997e-06,
+      "loss": 0.7031,
+      "step": 32560
+    },
+    {
+      "epoch": 0.6996176483223783,
+      "grad_norm": 0.5117861279536446,
+      "learning_rate": 4.209396202517183e-06,
+      "loss": 0.6918,
+      "step": 32570
+    },
+    {
+      "epoch": 0.699832452635649,
+      "grad_norm": 0.5064003948776196,
+      "learning_rate": 4.203840153047422e-06,
+      "loss": 0.706,
+      "step": 32580
+    },
+    {
+      "epoch": 0.7000472569489196,
+      "grad_norm": 0.5292023184958814,
+      "learning_rate": 4.198286796751318e-06,
+      "loss": 0.7005,
+      "step": 32590
+    },
+    {
+      "epoch": 0.7002620612621901,
+      "grad_norm": 0.5050282381222843,
+      "learning_rate": 4.1927361362092336e-06,
+      "loss": 0.699,
+      "step": 32600
+    },
+    {
+      "epoch": 0.7004768655754607,
+      "grad_norm": 0.5072730993957972,
+      "learning_rate": 4.187188174000262e-06,
+      "loss": 0.6993,
+      "step": 32610
+    },
+    {
+      "epoch": 0.7006916698887313,
+      "grad_norm": 0.5081741609904642,
+      "learning_rate": 4.181642912702256e-06,
+      "loss": 0.7046,
+      "step": 32620
+    },
+    {
+      "epoch": 0.700906474202002,
+      "grad_norm": 0.5132090809761299,
+      "learning_rate": 4.176100354891812e-06,
+      "loss": 0.6907,
+      "step": 32630
+    },
+    {
+      "epoch": 0.7011212785152726,
+      "grad_norm": 0.5114069928059646,
+      "learning_rate": 4.170560503144266e-06,
+      "loss": 0.7162,
+      "step": 32640
+    },
+    {
+      "epoch": 0.7013360828285432,
+      "grad_norm": 0.4963333550516421,
+      "learning_rate": 4.165023360033703e-06,
+      "loss": 0.706,
+      "step": 32650
+    },
+    {
+      "epoch": 0.7015508871418138,
+      "grad_norm": 0.5148865519289305,
+      "learning_rate": 4.159488928132938e-06,
+      "loss": 0.7147,
+      "step": 32660
+    },
+    {
+      "epoch": 0.7017656914550844,
+      "grad_norm": 0.5264557896427997,
+      "learning_rate": 4.153957210013535e-06,
+      "loss": 0.7022,
+      "step": 32670
+    },
+    {
+      "epoch": 0.7019804957683551,
+      "grad_norm": 0.5085093647122813,
+      "learning_rate": 4.148428208245799e-06,
+      "loss": 0.714,
+      "step": 32680
+    },
+    {
+      "epoch": 0.7021953000816257,
+      "grad_norm": 0.5323744427421241,
+      "learning_rate": 4.142901925398766e-06,
+      "loss": 0.6977,
+      "step": 32690
+    },
+    {
+      "epoch": 0.7024101043948963,
+      "grad_norm": 0.5166468220520238,
+      "learning_rate": 4.137378364040216e-06,
+      "loss": 0.6951,
+      "step": 32700
+    },
+    {
+      "epoch": 0.7026249087081669,
+      "grad_norm": 0.5004414110795155,
+      "learning_rate": 4.1318575267366515e-06,
+      "loss": 0.718,
+      "step": 32710
+    },
+    {
+      "epoch": 0.7028397130214374,
+      "grad_norm": 0.5144502686296928,
+      "learning_rate": 4.12633941605333e-06,
+      "loss": 0.6955,
+      "step": 32720
+    },
+    {
+      "epoch": 0.703054517334708,
+      "grad_norm": 0.503363662007366,
+      "learning_rate": 4.120824034554221e-06,
+      "loss": 0.709,
+      "step": 32730
+    },
+    {
+      "epoch": 0.7032693216479787,
+      "grad_norm": 0.532794569128917,
+      "learning_rate": 4.115311384802038e-06,
+      "loss": 0.7011,
+      "step": 32740
+    },
+    {
+      "epoch": 0.7034841259612493,
+      "grad_norm": 0.5310969014155835,
+      "learning_rate": 4.1098014693582265e-06,
+      "loss": 0.7008,
+      "step": 32750
+    },
+    {
+      "epoch": 0.7036989302745199,
+      "grad_norm": 0.4961846506249242,
+      "learning_rate": 4.104294290782946e-06,
+      "loss": 0.6958,
+      "step": 32760
+    },
+    {
+      "epoch": 0.7039137345877905,
+      "grad_norm": 0.5131064931445611,
+      "learning_rate": 4.09878985163511e-06,
+      "loss": 0.7031,
+      "step": 32770
+    },
+    {
+      "epoch": 0.7041285389010611,
+      "grad_norm": 0.501337109243204,
+      "learning_rate": 4.09328815447233e-06,
+      "loss": 0.7002,
+      "step": 32780
+    },
+    {
+      "epoch": 0.7043433432143318,
+      "grad_norm": 0.5010141549810861,
+      "learning_rate": 4.0877892018509735e-06,
+      "loss": 0.6982,
+      "step": 32790
+    },
+    {
+      "epoch": 0.7045581475276024,
+      "grad_norm": 0.5190222646312463,
+      "learning_rate": 4.082292996326107e-06,
+      "loss": 0.6942,
+      "step": 32800
+    },
+    {
+      "epoch": 0.704772951840873,
+      "grad_norm": 0.5021762528824651,
+      "learning_rate": 4.076799540451532e-06,
+      "loss": 0.7036,
+      "step": 32810
+    },
+    {
+      "epoch": 0.7049877561541436,
+      "grad_norm": 0.5216162795537221,
+      "learning_rate": 4.071308836779778e-06,
+      "loss": 0.705,
+      "step": 32820
+    },
+    {
+      "epoch": 0.7052025604674141,
+      "grad_norm": 0.5081741693301844,
+      "learning_rate": 4.065820887862077e-06,
+      "loss": 0.7118,
+      "step": 32830
+    },
+    {
+      "epoch": 0.7054173647806848,
+      "grad_norm": 0.4991610018130501,
+      "learning_rate": 4.0603356962484075e-06,
+      "loss": 0.7004,
+      "step": 32840
+    },
+    {
+      "epoch": 0.7056321690939554,
+      "grad_norm": 0.5315742680425537,
+      "learning_rate": 4.054853264487442e-06,
+      "loss": 0.6897,
+      "step": 32850
+    },
+    {
+      "epoch": 0.705846973407226,
+      "grad_norm": 0.5012628613471524,
+      "learning_rate": 4.049373595126584e-06,
+      "loss": 0.6952,
+      "step": 32860
+    },
+    {
+      "epoch": 0.7060617777204966,
+      "grad_norm": 0.5112201352671399,
+      "learning_rate": 4.043896690711954e-06,
+      "loss": 0.7163,
+      "step": 32870
+    },
+    {
+      "epoch": 0.7062765820337672,
+      "grad_norm": 0.4955072270942802,
+      "learning_rate": 4.0384225537883735e-06,
+      "loss": 0.7037,
+      "step": 32880
+    },
+    {
+      "epoch": 0.7064913863470379,
+      "grad_norm": 0.5272007156898026,
+      "learning_rate": 4.032951186899404e-06,
+      "loss": 0.7045,
+      "step": 32890
+    },
+    {
+      "epoch": 0.7067061906603085,
+      "grad_norm": 0.5131148249079197,
+      "learning_rate": 4.027482592587294e-06,
+      "loss": 0.7132,
+      "step": 32900
+    },
+    {
+      "epoch": 0.7069209949735791,
+      "grad_norm": 0.50205423046405,
+      "learning_rate": 4.022016773393017e-06,
+      "loss": 0.7166,
+      "step": 32910
+    },
+    {
+      "epoch": 0.7071357992868497,
+      "grad_norm": 0.5022328702294467,
+      "learning_rate": 4.0165537318562595e-06,
+      "loss": 0.7065,
+      "step": 32920
+    },
+    {
+      "epoch": 0.7073506036001203,
+      "grad_norm": 0.5175065388500295,
+      "learning_rate": 4.011093470515402e-06,
+      "loss": 0.6979,
+      "step": 32930
+    },
+    {
+      "epoch": 0.707565407913391,
+      "grad_norm": 0.5212370076896278,
+      "learning_rate": 4.005635991907556e-06,
+      "loss": 0.7022,
+      "step": 32940
+    },
+    {
+      "epoch": 0.7077802122266615,
+      "grad_norm": 0.4941680464626519,
+      "learning_rate": 4.000181298568514e-06,
+      "loss": 0.6891,
+      "step": 32950
+    },
+    {
+      "epoch": 0.7079950165399321,
+      "grad_norm": 0.5160432814123912,
+      "learning_rate": 3.994729393032803e-06,
+      "loss": 0.6975,
+      "step": 32960
+    },
+    {
+      "epoch": 0.7082098208532027,
+      "grad_norm": 0.5090291581171473,
+      "learning_rate": 3.989280277833629e-06,
+      "loss": 0.7021,
+      "step": 32970
+    },
+    {
+      "epoch": 0.7084246251664733,
+      "grad_norm": 0.5048922334558574,
+      "learning_rate": 3.983833955502915e-06,
+      "loss": 0.6897,
+      "step": 32980
+    },
+    {
+      "epoch": 0.7086394294797439,
+      "grad_norm": 0.5065038482347751,
+      "learning_rate": 3.978390428571286e-06,
+      "loss": 0.6934,
+      "step": 32990
+    },
+    {
+      "epoch": 0.7088542337930146,
+      "grad_norm": 0.535448626077636,
+      "learning_rate": 3.972949699568057e-06,
+      "loss": 0.6893,
+      "step": 33000
+    },
+    {
+      "epoch": 0.7090690381062852,
+      "grad_norm": 0.5169501432061164,
+      "learning_rate": 3.967511771021264e-06,
+      "loss": 0.7029,
+      "step": 33010
+    },
+    {
+      "epoch": 0.7092838424195558,
+      "grad_norm": 0.511359264294529,
+      "learning_rate": 3.96207664545762e-06,
+      "loss": 0.7034,
+      "step": 33020
+    },
+    {
+      "epoch": 0.7094986467328264,
+      "grad_norm": 0.4960055217526054,
+      "learning_rate": 3.956644325402547e-06,
+      "loss": 0.7024,
+      "step": 33030
+    },
+    {
+      "epoch": 0.709713451046097,
+      "grad_norm": 0.5274217499816306,
+      "learning_rate": 3.951214813380164e-06,
+      "loss": 0.7051,
+      "step": 33040
+    },
+    {
+      "epoch": 0.7099282553593677,
+      "grad_norm": 0.4966382776596969,
+      "learning_rate": 3.9457881119132745e-06,
+      "loss": 0.6994,
+      "step": 33050
+    },
+    {
+      "epoch": 0.7101430596726382,
+      "grad_norm": 0.5360848716987743,
+      "learning_rate": 3.940364223523398e-06,
+      "loss": 0.7146,
+      "step": 33060
+    },
+    {
+      "epoch": 0.7103578639859088,
+      "grad_norm": 0.5765957098247436,
+      "learning_rate": 3.934943150730719e-06,
+      "loss": 0.6882,
+      "step": 33070
+    },
+    {
+      "epoch": 0.7105726682991794,
+      "grad_norm": 0.5219156969848493,
+      "learning_rate": 3.9295248960541355e-06,
+      "loss": 0.7012,
+      "step": 33080
+    },
+    {
+      "epoch": 0.71078747261245,
+      "grad_norm": 0.5177328505359572,
+      "learning_rate": 3.924109462011225e-06,
+      "loss": 0.6934,
+      "step": 33090
+    },
+    {
+      "epoch": 0.7110022769257207,
+      "grad_norm": 0.5083853674745945,
+      "learning_rate": 3.91869685111826e-06,
+      "loss": 0.699,
+      "step": 33100
+    },
+    {
+      "epoch": 0.7112170812389913,
+      "grad_norm": 0.5161556089894507,
+      "learning_rate": 3.913287065890201e-06,
+      "loss": 0.6894,
+      "step": 33110
+    },
+    {
+      "epoch": 0.7114318855522619,
+      "grad_norm": 0.5176090611045527,
+      "learning_rate": 3.907880108840688e-06,
+      "loss": 0.7132,
+      "step": 33120
+    },
+    {
+      "epoch": 0.7116466898655325,
+      "grad_norm": 0.5048282675725968,
+      "learning_rate": 3.902475982482055e-06,
+      "loss": 0.7103,
+      "step": 33130
+    },
+    {
+      "epoch": 0.7118614941788031,
+      "grad_norm": 0.5184451927132917,
+      "learning_rate": 3.897074689325316e-06,
+      "loss": 0.6916,
+      "step": 33140
+    },
+    {
+      "epoch": 0.7120762984920738,
+      "grad_norm": 0.5010826412828934,
+      "learning_rate": 3.891676231880175e-06,
+      "loss": 0.6897,
+      "step": 33150
+    },
+    {
+      "epoch": 0.7122911028053444,
+      "grad_norm": 0.524146057682627,
+      "learning_rate": 3.8862806126550105e-06,
+      "loss": 0.7169,
+      "step": 33160
+    },
+    {
+      "epoch": 0.712505907118615,
+      "grad_norm": 0.5038515983066828,
+      "learning_rate": 3.8808878341568875e-06,
+      "loss": 0.7081,
+      "step": 33170
+    },
+    {
+      "epoch": 0.7127207114318855,
+      "grad_norm": 0.5255219224371609,
+      "learning_rate": 3.875497898891552e-06,
+      "loss": 0.7022,
+      "step": 33180
+    },
+    {
+      "epoch": 0.7129355157451561,
+      "grad_norm": 0.4985578239131686,
+      "learning_rate": 3.87011080936342e-06,
+      "loss": 0.7012,
+      "step": 33190
+    },
+    {
+      "epoch": 0.7131503200584268,
+      "grad_norm": 0.4924989128893002,
+      "learning_rate": 3.864726568075595e-06,
+      "loss": 0.6947,
+      "step": 33200
+    },
+    {
+      "epoch": 0.7133651243716974,
+      "grad_norm": 0.5309886497942049,
+      "learning_rate": 3.859345177529853e-06,
+      "loss": 0.7049,
+      "step": 33210
+    },
+    {
+      "epoch": 0.713579928684968,
+      "grad_norm": 0.5206730299572171,
+      "learning_rate": 3.8539666402266465e-06,
+      "loss": 0.6977,
+      "step": 33220
+    },
+    {
+      "epoch": 0.7137947329982386,
+      "grad_norm": 0.5158488974635094,
+      "learning_rate": 3.848590958665104e-06,
+      "loss": 0.7065,
+      "step": 33230
+    },
+    {
+      "epoch": 0.7140095373115092,
+      "grad_norm": 0.48965491649636267,
+      "learning_rate": 3.843218135343019e-06,
+      "loss": 0.7011,
+      "step": 33240
+    },
+    {
+      "epoch": 0.7142243416247799,
+      "grad_norm": 0.564742066719133,
+      "learning_rate": 3.837848172756865e-06,
+      "loss": 0.7058,
+      "step": 33250
+    },
+    {
+      "epoch": 0.7144391459380505,
+      "grad_norm": 0.5137654269820426,
+      "learning_rate": 3.832481073401786e-06,
+      "loss": 0.7052,
+      "step": 33260
+    },
+    {
+      "epoch": 0.7146539502513211,
+      "grad_norm": 0.5180489971377459,
+      "learning_rate": 3.827116839771593e-06,
+      "loss": 0.7148,
+      "step": 33270
+    },
+    {
+      "epoch": 0.7148687545645916,
+      "grad_norm": 0.5058146041046021,
+      "learning_rate": 3.821755474358764e-06,
+      "loss": 0.6951,
+      "step": 33280
+    },
+    {
+      "epoch": 0.7150835588778622,
+      "grad_norm": 0.5048478043474748,
+      "learning_rate": 3.816396979654452e-06,
+      "loss": 0.708,
+      "step": 33290
+    },
+    {
+      "epoch": 0.7152983631911328,
+      "grad_norm": 0.5200290541035588,
+      "learning_rate": 3.8110413581484628e-06,
+      "loss": 0.7076,
+      "step": 33300
+    },
+    {
+      "epoch": 0.7155131675044035,
+      "grad_norm": 0.5041024765974936,
+      "learning_rate": 3.805688612329279e-06,
+      "loss": 0.7087,
+      "step": 33310
+    },
+    {
+      "epoch": 0.7157279718176741,
+      "grad_norm": 0.517113427361278,
+      "learning_rate": 3.800338744684041e-06,
+      "loss": 0.7035,
+      "step": 33320
+    },
+    {
+      "epoch": 0.7159427761309447,
+      "grad_norm": 0.5180046345306407,
+      "learning_rate": 3.794991757698555e-06,
+      "loss": 0.7112,
+      "step": 33330
+    },
+    {
+      "epoch": 0.7161575804442153,
+      "grad_norm": 0.5088817995846314,
+      "learning_rate": 3.7896476538572914e-06,
+      "loss": 0.7107,
+      "step": 33340
+    },
+    {
+      "epoch": 0.7163723847574859,
+      "grad_norm": 0.49205509109092965,
+      "learning_rate": 3.7843064356433656e-06,
+      "loss": 0.6988,
+      "step": 33350
+    },
+    {
+      "epoch": 0.7165871890707566,
+      "grad_norm": 0.5006056060538654,
+      "learning_rate": 3.7789681055385787e-06,
+      "loss": 0.7052,
+      "step": 33360
+    },
+    {
+      "epoch": 0.7168019933840272,
+      "grad_norm": 0.521153940432141,
+      "learning_rate": 3.7736326660233623e-06,
+      "loss": 0.7135,
+      "step": 33370
+    },
+    {
+      "epoch": 0.7170167976972978,
+      "grad_norm": 0.5069579531044478,
+      "learning_rate": 3.768300119576822e-06,
+      "loss": 0.7183,
+      "step": 33380
+    },
+    {
+      "epoch": 0.7172316020105683,
+      "grad_norm": 0.5215580306799487,
+      "learning_rate": 3.7629704686767144e-06,
+      "loss": 0.6942,
+      "step": 33390
+    },
+    {
+      "epoch": 0.7174464063238389,
+      "grad_norm": 0.5122954931151245,
+      "learning_rate": 3.7576437157994506e-06,
+      "loss": 0.6879,
+      "step": 33400
+    },
+    {
+      "epoch": 0.7176612106371096,
+      "grad_norm": 0.5464224896140304,
+      "learning_rate": 3.7523198634200985e-06,
+      "loss": 0.7155,
+      "step": 33410
+    },
+    {
+      "epoch": 0.7178760149503802,
+      "grad_norm": 0.4963871129860938,
+      "learning_rate": 3.7469989140123696e-06,
+      "loss": 0.6932,
+      "step": 33420
+    },
+    {
+      "epoch": 0.7180908192636508,
+      "grad_norm": 0.5220526668691016,
+      "learning_rate": 3.7416808700486328e-06,
+      "loss": 0.7096,
+      "step": 33430
+    },
+    {
+      "epoch": 0.7183056235769214,
+      "grad_norm": 0.5142411024190581,
+      "learning_rate": 3.7363657339999093e-06,
+      "loss": 0.6957,
+      "step": 33440
+    },
+    {
+      "epoch": 0.718520427890192,
+      "grad_norm": 0.5078089733725818,
+      "learning_rate": 3.7310535083358635e-06,
+      "loss": 0.7011,
+      "step": 33450
+    },
+    {
+      "epoch": 0.7187352322034627,
+      "grad_norm": 0.49558634222610376,
+      "learning_rate": 3.7257441955248153e-06,
+      "loss": 0.6895,
+      "step": 33460
+    },
+    {
+      "epoch": 0.7189500365167333,
+      "grad_norm": 0.48750240509688914,
+      "learning_rate": 3.7204377980337137e-06,
+      "loss": 0.718,
+      "step": 33470
+    },
+    {
+      "epoch": 0.7191648408300039,
+      "grad_norm": 0.5033127653562809,
+      "learning_rate": 3.7151343183281808e-06,
+      "loss": 0.7034,
+      "step": 33480
+    },
+    {
+      "epoch": 0.7193796451432745,
+      "grad_norm": 0.4924112060947027,
+      "learning_rate": 3.7098337588724565e-06,
+      "loss": 0.6955,
+      "step": 33490
+    },
+    {
+      "epoch": 0.719594449456545,
+      "grad_norm": 0.5143671743596542,
+      "learning_rate": 3.7045361221294375e-06,
+      "loss": 0.708,
+      "step": 33500
+    },
+    {
+      "epoch": 0.7198092537698157,
+      "grad_norm": 0.5082249361246125,
+      "learning_rate": 3.6992414105606644e-06,
+      "loss": 0.6782,
+      "step": 33510
+    },
+    {
+      "epoch": 0.7200240580830863,
+      "grad_norm": 0.5043808950755216,
+      "learning_rate": 3.693949626626302e-06,
+      "loss": 0.6979,
+      "step": 33520
+    },
+    {
+      "epoch": 0.7202388623963569,
+      "grad_norm": 0.5219484042509999,
+      "learning_rate": 3.688660772785183e-06,
+      "loss": 0.704,
+      "step": 33530
+    },
+    {
+      "epoch": 0.7204536667096275,
+      "grad_norm": 0.5449994389723714,
+      "learning_rate": 3.6833748514947465e-06,
+      "loss": 0.709,
+      "step": 33540
+    },
+    {
+      "epoch": 0.7206684710228981,
+      "grad_norm": 0.5187402619586201,
+      "learning_rate": 3.6780918652110984e-06,
+      "loss": 0.7063,
+      "step": 33550
+    },
+    {
+      "epoch": 0.7208832753361688,
+      "grad_norm": 0.5025397434112188,
+      "learning_rate": 3.672811816388959e-06,
+      "loss": 0.7125,
+      "step": 33560
+    },
+    {
+      "epoch": 0.7210980796494394,
+      "grad_norm": 0.4922234036325025,
+      "learning_rate": 3.6675347074816948e-06,
+      "loss": 0.699,
+      "step": 33570
+    },
+    {
+      "epoch": 0.72131288396271,
+      "grad_norm": 0.5049263733928765,
+      "learning_rate": 3.662260540941306e-06,
+      "loss": 0.7042,
+      "step": 33580
+    },
+    {
+      "epoch": 0.7215276882759806,
+      "grad_norm": 0.5209764065547637,
+      "learning_rate": 3.6569893192184123e-06,
+      "loss": 0.7002,
+      "step": 33590
+    },
+    {
+      "epoch": 0.7217424925892512,
+      "grad_norm": 0.5046889630269408,
+      "learning_rate": 3.6517210447622918e-06,
+      "loss": 0.7017,
+      "step": 33600
+    },
+    {
+      "epoch": 0.7219572969025218,
+      "grad_norm": 0.5059886025352888,
+      "learning_rate": 3.646455720020826e-06,
+      "loss": 0.6994,
+      "step": 33610
+    },
+    {
+      "epoch": 0.7221721012157925,
+      "grad_norm": 0.5098305495450107,
+      "learning_rate": 3.6411933474405402e-06,
+      "loss": 0.7073,
+      "step": 33620
+    },
+    {
+      "epoch": 0.722386905529063,
+      "grad_norm": 0.5002036914879331,
+      "learning_rate": 3.6359339294665897e-06,
+      "loss": 0.6901,
+      "step": 33630
+    },
+    {
+      "epoch": 0.7226017098423336,
+      "grad_norm": 0.523077232704163,
+      "learning_rate": 3.630677468542739e-06,
+      "loss": 0.7114,
+      "step": 33640
+    },
+    {
+      "epoch": 0.7228165141556042,
+      "grad_norm": 0.5040742343196016,
+      "learning_rate": 3.625423967111409e-06,
+      "loss": 0.6915,
+      "step": 33650
+    },
+    {
+      "epoch": 0.7230313184688748,
+      "grad_norm": 0.504689168487558,
+      "learning_rate": 3.6201734276136156e-06,
+      "loss": 0.7027,
+      "step": 33660
+    },
+    {
+      "epoch": 0.7232461227821455,
+      "grad_norm": 0.5360928374223011,
+      "learning_rate": 3.614925852489015e-06,
+      "loss": 0.7044,
+      "step": 33670
+    },
+    {
+      "epoch": 0.7234609270954161,
+      "grad_norm": 0.508286028640259,
+      "learning_rate": 3.6096812441758865e-06,
+      "loss": 0.7102,
+      "step": 33680
+    },
+    {
+      "epoch": 0.7236757314086867,
+      "grad_norm": 0.4950815691274723,
+      "learning_rate": 3.604439605111114e-06,
+      "loss": 0.7042,
+      "step": 33690
+    },
+    {
+      "epoch": 0.7238905357219573,
+      "grad_norm": 0.4990133005177096,
+      "learning_rate": 3.5992009377302306e-06,
+      "loss": 0.7092,
+      "step": 33700
+    },
+    {
+      "epoch": 0.7241053400352279,
+      "grad_norm": 0.5080144595958782,
+      "learning_rate": 3.5939652444673614e-06,
+      "loss": 0.7057,
+      "step": 33710
+    },
+    {
+      "epoch": 0.7243201443484986,
+      "grad_norm": 0.49504096828193644,
+      "learning_rate": 3.588732527755262e-06,
+      "loss": 0.6866,
+      "step": 33720
+    },
+    {
+      "epoch": 0.7245349486617692,
+      "grad_norm": 0.5010995541979635,
+      "learning_rate": 3.583502790025304e-06,
+      "loss": 0.7017,
+      "step": 33730
+    },
+    {
+      "epoch": 0.7247497529750397,
+      "grad_norm": 0.5259173065794601,
+      "learning_rate": 3.578276033707476e-06,
+      "loss": 0.7094,
+      "step": 33740
+    },
+    {
+      "epoch": 0.7249645572883103,
+      "grad_norm": 0.5119905092520929,
+      "learning_rate": 3.5730522612303808e-06,
+      "loss": 0.701,
+      "step": 33750
+    },
+    {
+      "epoch": 0.7251793616015809,
+      "grad_norm": 0.5647220761659497,
+      "learning_rate": 3.5678314750212253e-06,
+      "loss": 0.6946,
+      "step": 33760
+    },
+    {
+      "epoch": 0.7253941659148516,
+      "grad_norm": 0.5084711336782207,
+      "learning_rate": 3.5626136775058484e-06,
+      "loss": 0.7022,
+      "step": 33770
+    },
+    {
+      "epoch": 0.7256089702281222,
+      "grad_norm": 0.5259298296821393,
+      "learning_rate": 3.5573988711086793e-06,
+      "loss": 0.6953,
+      "step": 33780
+    },
+    {
+      "epoch": 0.7258237745413928,
+      "grad_norm": 0.4968324999154312,
+      "learning_rate": 3.552187058252772e-06,
+      "loss": 0.7083,
+      "step": 33790
+    },
+    {
+      "epoch": 0.7260385788546634,
+      "grad_norm": 0.5111234225003652,
+      "learning_rate": 3.546978241359785e-06,
+      "loss": 0.6995,
+      "step": 33800
+    },
+    {
+      "epoch": 0.726253383167934,
+      "grad_norm": 0.5321941233022512,
+      "learning_rate": 3.541772422849977e-06,
+      "loss": 0.6862,
+      "step": 33810
+    },
+    {
+      "epoch": 0.7264681874812047,
+      "grad_norm": 0.5446186196935507,
+      "learning_rate": 3.5365696051422337e-06,
+      "loss": 0.7064,
+      "step": 33820
+    },
+    {
+      "epoch": 0.7266829917944753,
+      "grad_norm": 0.5339112094968238,
+      "learning_rate": 3.531369790654022e-06,
+      "loss": 0.6939,
+      "step": 33830
+    },
+    {
+      "epoch": 0.7268977961077459,
+      "grad_norm": 0.5116370373744857,
+      "learning_rate": 3.526172981801429e-06,
+      "loss": 0.6848,
+      "step": 33840
+    },
+    {
+      "epoch": 0.7271126004210164,
+      "grad_norm": 0.49391991528376755,
+      "learning_rate": 3.5209791809991424e-06,
+      "loss": 0.6974,
+      "step": 33850
+    },
+    {
+      "epoch": 0.727327404734287,
+      "grad_norm": 0.4986794011059733,
+      "learning_rate": 3.5157883906604484e-06,
+      "loss": 0.6986,
+      "step": 33860
+    },
+    {
+      "epoch": 0.7275422090475576,
+      "grad_norm": 0.5365339519034793,
+      "learning_rate": 3.5106006131972425e-06,
+      "loss": 0.7011,
+      "step": 33870
+    },
+    {
+      "epoch": 0.7277570133608283,
+      "grad_norm": 0.505426889604584,
+      "learning_rate": 3.5054158510200077e-06,
+      "loss": 0.7272,
+      "step": 33880
+    },
+    {
+      "epoch": 0.7279718176740989,
+      "grad_norm": 0.534863544759657,
+      "learning_rate": 3.5002341065378352e-06,
+      "loss": 0.6975,
+      "step": 33890
+    },
+    {
+      "epoch": 0.7281866219873695,
+      "grad_norm": 0.5029648260679437,
+      "learning_rate": 3.4950553821584133e-06,
+      "loss": 0.6967,
+      "step": 33900
+    },
+    {
+      "epoch": 0.7284014263006401,
+      "grad_norm": 0.5014206921527354,
+      "learning_rate": 3.4898796802880253e-06,
+      "loss": 0.7089,
+      "step": 33910
+    },
+    {
+      "epoch": 0.7286162306139107,
+      "grad_norm": 0.5234721876994048,
+      "learning_rate": 3.484707003331549e-06,
+      "loss": 0.7085,
+      "step": 33920
+    },
+    {
+      "epoch": 0.7288310349271814,
+      "grad_norm": 0.5239292676381829,
+      "learning_rate": 3.4795373536924627e-06,
+      "loss": 0.6843,
+      "step": 33930
+    },
+    {
+      "epoch": 0.729045839240452,
+      "grad_norm": 0.4980484392742611,
+      "learning_rate": 3.474370733772827e-06,
+      "loss": 0.6965,
+      "step": 33940
+    },
+    {
+      "epoch": 0.7292606435537226,
+      "grad_norm": 0.5364410045950752,
+      "learning_rate": 3.4692071459733024e-06,
+      "loss": 0.6984,
+      "step": 33950
+    },
+    {
+      "epoch": 0.7294754478669931,
+      "grad_norm": 0.5079175468345598,
+      "learning_rate": 3.464046592693142e-06,
+      "loss": 0.6894,
+      "step": 33960
+    },
+    {
+      "epoch": 0.7296902521802637,
+      "grad_norm": 0.5195456559419563,
+      "learning_rate": 3.4588890763301843e-06,
+      "loss": 0.7166,
+      "step": 33970
+    },
+    {
+      "epoch": 0.7299050564935344,
+      "grad_norm": 0.5069432370518213,
+      "learning_rate": 3.453734599280859e-06,
+      "loss": 0.7039,
+      "step": 33980
+    },
+    {
+      "epoch": 0.730119860806805,
+      "grad_norm": 0.5132347077274939,
+      "learning_rate": 3.4485831639401836e-06,
+      "loss": 0.7007,
+      "step": 33990
+    },
+    {
+      "epoch": 0.7303346651200756,
+      "grad_norm": 0.4969882977586769,
+      "learning_rate": 3.4434347727017645e-06,
+      "loss": 0.7043,
+      "step": 34000
+    },
+    {
+      "epoch": 0.7305494694333462,
+      "grad_norm": 0.519434920750127,
+      "learning_rate": 3.438289427957785e-06,
+      "loss": 0.7021,
+      "step": 34010
+    },
+    {
+      "epoch": 0.7307642737466168,
+      "grad_norm": 0.5195248274879436,
+      "learning_rate": 3.4331471320990216e-06,
+      "loss": 0.7106,
+      "step": 34020
+    },
+    {
+      "epoch": 0.7309790780598875,
+      "grad_norm": 0.5452676568255899,
+      "learning_rate": 3.4280078875148317e-06,
+      "loss": 0.689,
+      "step": 34030
+    },
+    {
+      "epoch": 0.7311938823731581,
+      "grad_norm": 0.5006895911589073,
+      "learning_rate": 3.4228716965931553e-06,
+      "loss": 0.7172,
+      "step": 34040
+    },
+    {
+      "epoch": 0.7314086866864287,
+      "grad_norm": 0.543855444481588,
+      "learning_rate": 3.417738561720515e-06,
+      "loss": 0.7082,
+      "step": 34050
+    },
+    {
+      "epoch": 0.7316234909996993,
+      "grad_norm": 0.5225704241808111,
+      "learning_rate": 3.4126084852820064e-06,
+      "loss": 0.7012,
+      "step": 34060
+    },
+    {
+      "epoch": 0.7318382953129698,
+      "grad_norm": 0.5439757755860316,
+      "learning_rate": 3.40748146966131e-06,
+      "loss": 0.6898,
+      "step": 34070
+    },
+    {
+      "epoch": 0.7320530996262405,
+      "grad_norm": 0.521232354304882,
+      "learning_rate": 3.402357517240684e-06,
+      "loss": 0.7246,
+      "step": 34080
+    },
+    {
+      "epoch": 0.7322679039395111,
+      "grad_norm": 0.504585628259099,
+      "learning_rate": 3.397236630400962e-06,
+      "loss": 0.6846,
+      "step": 34090
+    },
+    {
+      "epoch": 0.7324827082527817,
+      "grad_norm": 0.5105232285125065,
+      "learning_rate": 3.3921188115215574e-06,
+      "loss": 0.709,
+      "step": 34100
+    },
+    {
+      "epoch": 0.7326975125660523,
+      "grad_norm": 0.5118542121131522,
+      "learning_rate": 3.3870040629804445e-06,
+      "loss": 0.6957,
+      "step": 34110
+    },
+    {
+      "epoch": 0.7329123168793229,
+      "grad_norm": 0.494913168921892,
+      "learning_rate": 3.3818923871541932e-06,
+      "loss": 0.6846,
+      "step": 34120
+    },
+    {
+      "epoch": 0.7331271211925936,
+      "grad_norm": 0.5369888915842843,
+      "learning_rate": 3.376783786417922e-06,
+      "loss": 0.7195,
+      "step": 34130
+    },
+    {
+      "epoch": 0.7333419255058642,
+      "grad_norm": 0.49649984813451475,
+      "learning_rate": 3.371678263145337e-06,
+      "loss": 0.6934,
+      "step": 34140
+    },
+    {
+      "epoch": 0.7335567298191348,
+      "grad_norm": 0.5374808788911171,
+      "learning_rate": 3.3665758197087084e-06,
+      "loss": 0.7042,
+      "step": 34150
+    },
+    {
+      "epoch": 0.7337715341324054,
+      "grad_norm": 0.508279517829229,
+      "learning_rate": 3.361476458478875e-06,
+      "loss": 0.7006,
+      "step": 34160
+    },
+    {
+      "epoch": 0.733986338445676,
+      "grad_norm": 0.506858247543215,
+      "learning_rate": 3.3563801818252483e-06,
+      "loss": 0.6905,
+      "step": 34170
+    },
+    {
+      "epoch": 0.7342011427589465,
+      "grad_norm": 0.5233572520758975,
+      "learning_rate": 3.351286992115793e-06,
+      "loss": 0.7044,
+      "step": 34180
+    },
+    {
+      "epoch": 0.7344159470722172,
+      "grad_norm": 0.49926488863527185,
+      "learning_rate": 3.346196891717063e-06,
+      "loss": 0.6974,
+      "step": 34190
+    },
+    {
+      "epoch": 0.7346307513854878,
+      "grad_norm": 0.5127254239928138,
+      "learning_rate": 3.3411098829941513e-06,
+      "loss": 0.6949,
+      "step": 34200
+    },
+    {
+      "epoch": 0.7348455556987584,
+      "grad_norm": 0.5157624607732113,
+      "learning_rate": 3.33602596831073e-06,
+      "loss": 0.7038,
+      "step": 34210
+    },
+    {
+      "epoch": 0.735060360012029,
+      "grad_norm": 0.5324610708151536,
+      "learning_rate": 3.3309451500290336e-06,
+      "loss": 0.7072,
+      "step": 34220
+    },
+    {
+      "epoch": 0.7352751643252996,
+      "grad_norm": 0.5185302310161645,
+      "learning_rate": 3.3258674305098435e-06,
+      "loss": 0.6948,
+      "step": 34230
+    },
+    {
+      "epoch": 0.7354899686385703,
+      "grad_norm": 0.4957613188241341,
+      "learning_rate": 3.3207928121125243e-06,
+      "loss": 0.7011,
+      "step": 34240
+    },
+    {
+      "epoch": 0.7357047729518409,
+      "grad_norm": 0.5441591292886901,
+      "learning_rate": 3.315721297194977e-06,
+      "loss": 0.705,
+      "step": 34250
+    },
+    {
+      "epoch": 0.7359195772651115,
+      "grad_norm": 0.5278633009999051,
+      "learning_rate": 3.3106528881136745e-06,
+      "loss": 0.7038,
+      "step": 34260
+    },
+    {
+      "epoch": 0.7361343815783821,
+      "grad_norm": 0.49493002519430035,
+      "learning_rate": 3.305587587223645e-06,
+      "loss": 0.7014,
+      "step": 34270
+    },
+    {
+      "epoch": 0.7363491858916527,
+      "grad_norm": 0.5164985719789421,
+      "learning_rate": 3.300525396878461e-06,
+      "loss": 0.6918,
+      "step": 34280
+    },
+    {
+      "epoch": 0.7365639902049234,
+      "grad_norm": 0.5385527870192864,
+      "learning_rate": 3.2954663194302715e-06,
+      "loss": 0.7056,
+      "step": 34290
+    },
+    {
+      "epoch": 0.736778794518194,
+      "grad_norm": 0.5222972879975742,
+      "learning_rate": 3.2904103572297565e-06,
+      "loss": 0.7012,
+      "step": 34300
+    },
+    {
+      "epoch": 0.7369935988314645,
+      "grad_norm": 0.515795075393324,
+      "learning_rate": 3.2853575126261606e-06,
+      "loss": 0.704,
+      "step": 34310
+    },
+    {
+      "epoch": 0.7372084031447351,
+      "grad_norm": 0.4972752236528448,
+      "learning_rate": 3.280307787967283e-06,
+      "loss": 0.6891,
+      "step": 34320
+    },
+    {
+      "epoch": 0.7374232074580057,
+      "grad_norm": 0.5276200292672564,
+      "learning_rate": 3.2752611855994566e-06,
+      "loss": 0.7069,
+      "step": 34330
+    },
+    {
+      "epoch": 0.7376380117712764,
+      "grad_norm": 0.5527465990583433,
+      "learning_rate": 3.270217707867588e-06,
+      "loss": 0.6867,
+      "step": 34340
+    },
+    {
+      "epoch": 0.737852816084547,
+      "grad_norm": 0.5098818638723858,
+      "learning_rate": 3.265177357115106e-06,
+      "loss": 0.6933,
+      "step": 34350
+    },
+    {
+      "epoch": 0.7380676203978176,
+      "grad_norm": 0.5001020449149964,
+      "learning_rate": 3.2601401356840112e-06,
+      "loss": 0.7017,
+      "step": 34360
+    },
+    {
+      "epoch": 0.7382824247110882,
+      "grad_norm": 0.5090868444518032,
+      "learning_rate": 3.2551060459148297e-06,
+      "loss": 0.6899,
+      "step": 34370
+    },
+    {
+      "epoch": 0.7384972290243588,
+      "grad_norm": 0.5291041455847213,
+      "learning_rate": 3.250075090146644e-06,
+      "loss": 0.6951,
+      "step": 34380
+    },
+    {
+      "epoch": 0.7387120333376295,
+      "grad_norm": 0.49906694116191236,
+      "learning_rate": 3.2450472707170786e-06,
+      "loss": 0.704,
+      "step": 34390
+    },
+    {
+      "epoch": 0.7389268376509001,
+      "grad_norm": 0.49374014306535213,
+      "learning_rate": 3.2400225899622917e-06,
+      "loss": 0.6808,
+      "step": 34400
+    },
+    {
+      "epoch": 0.7391416419641706,
+      "grad_norm": 0.5331437741952022,
+      "learning_rate": 3.235001050217003e-06,
+      "loss": 0.7108,
+      "step": 34410
+    },
+    {
+      "epoch": 0.7393564462774412,
+      "grad_norm": 0.5433540729594903,
+      "learning_rate": 3.2299826538144506e-06,
+      "loss": 0.7037,
+      "step": 34420
+    },
+    {
+      "epoch": 0.7395712505907118,
+      "grad_norm": 0.49990473998285656,
+      "learning_rate": 3.2249674030864254e-06,
+      "loss": 0.7034,
+      "step": 34430
+    },
+    {
+      "epoch": 0.7397860549039825,
+      "grad_norm": 0.5184986266564906,
+      "learning_rate": 3.2199553003632566e-06,
+      "loss": 0.7059,
+      "step": 34440
+    },
+    {
+      "epoch": 0.7400008592172531,
+      "grad_norm": 0.5255092855129455,
+      "learning_rate": 3.2149463479737974e-06,
+      "loss": 0.7078,
+      "step": 34450
+    },
+    {
+      "epoch": 0.7402156635305237,
+      "grad_norm": 0.5078685785943006,
+      "learning_rate": 3.2099405482454613e-06,
+      "loss": 0.6856,
+      "step": 34460
+    },
+    {
+      "epoch": 0.7404304678437943,
+      "grad_norm": 0.5001995241491086,
+      "learning_rate": 3.2049379035041718e-06,
+      "loss": 0.6947,
+      "step": 34470
+    },
+    {
+      "epoch": 0.7406452721570649,
+      "grad_norm": 0.5229199908697051,
+      "learning_rate": 3.199938416074401e-06,
+      "loss": 0.698,
+      "step": 34480
+    },
+    {
+      "epoch": 0.7408600764703355,
+      "grad_norm": 0.506611104007963,
+      "learning_rate": 3.1949420882791493e-06,
+      "loss": 0.7095,
+      "step": 34490
+    },
+    {
+      "epoch": 0.7410748807836062,
+      "grad_norm": 0.5063586727734837,
+      "learning_rate": 3.1899489224399514e-06,
+      "loss": 0.7033,
+      "step": 34500
+    },
+    {
+      "epoch": 0.7412896850968768,
+      "grad_norm": 0.5267235455499195,
+      "learning_rate": 3.184958920876874e-06,
+      "loss": 0.695,
+      "step": 34510
+    },
+    {
+      "epoch": 0.7415044894101473,
+      "grad_norm": 0.501781417708024,
+      "learning_rate": 3.1799720859085025e-06,
+      "loss": 0.6983,
+      "step": 34520
+    },
+    {
+      "epoch": 0.7417192937234179,
+      "grad_norm": 0.5127416336972987,
+      "learning_rate": 3.174988419851971e-06,
+      "loss": 0.6996,
+      "step": 34530
+    },
+    {
+      "epoch": 0.7419340980366885,
+      "grad_norm": 0.5238966491029383,
+      "learning_rate": 3.170007925022921e-06,
+      "loss": 0.7003,
+      "step": 34540
+    },
+    {
+      "epoch": 0.7421489023499592,
+      "grad_norm": 0.5027088254862997,
+      "learning_rate": 3.1650306037355303e-06,
+      "loss": 0.693,
+      "step": 34550
+    },
+    {
+      "epoch": 0.7423637066632298,
+      "grad_norm": 0.5038648172133018,
+      "learning_rate": 3.160056458302504e-06,
+      "loss": 0.6989,
+      "step": 34560
+    },
+    {
+      "epoch": 0.7425785109765004,
+      "grad_norm": 0.5125824985664078,
+      "learning_rate": 3.155085491035066e-06,
+      "loss": 0.7069,
+      "step": 34570
+    },
+    {
+      "epoch": 0.742793315289771,
+      "grad_norm": 0.5375010969807497,
+      "learning_rate": 3.1501177042429697e-06,
+      "loss": 0.7113,
+      "step": 34580
+    },
+    {
+      "epoch": 0.7430081196030416,
+      "grad_norm": 0.507552166036012,
+      "learning_rate": 3.1451531002344804e-06,
+      "loss": 0.7009,
+      "step": 34590
+    },
+    {
+      "epoch": 0.7432229239163123,
+      "grad_norm": 0.5130116795996187,
+      "learning_rate": 3.140191681316396e-06,
+      "loss": 0.7009,
+      "step": 34600
+    },
+    {
+      "epoch": 0.7434377282295829,
+      "grad_norm": 0.513751129751177,
+      "learning_rate": 3.1352334497940262e-06,
+      "loss": 0.7051,
+      "step": 34610
+    },
+    {
+      "epoch": 0.7436525325428535,
+      "grad_norm": 0.5088528949925636,
+      "learning_rate": 3.1302784079712067e-06,
+      "loss": 0.7024,
+      "step": 34620
+    },
+    {
+      "epoch": 0.743867336856124,
+      "grad_norm": 0.519047865808951,
+      "learning_rate": 3.1253265581502877e-06,
+      "loss": 0.6916,
+      "step": 34630
+    },
+    {
+      "epoch": 0.7440821411693946,
+      "grad_norm": 0.5261411126518063,
+      "learning_rate": 3.1203779026321313e-06,
+      "loss": 0.6956,
+      "step": 34640
+    },
+    {
+      "epoch": 0.7442969454826653,
+      "grad_norm": 0.5173606042960484,
+      "learning_rate": 3.115432443716123e-06,
+      "loss": 0.7191,
+      "step": 34650
+    },
+    {
+      "epoch": 0.7445117497959359,
+      "grad_norm": 0.5197079005508889,
+      "learning_rate": 3.110490183700159e-06,
+      "loss": 0.7082,
+      "step": 34660
+    },
+    {
+      "epoch": 0.7447265541092065,
+      "grad_norm": 0.5109626703024929,
+      "learning_rate": 3.1055511248806514e-06,
+      "loss": 0.6868,
+      "step": 34670
+    },
+    {
+      "epoch": 0.7449413584224771,
+      "grad_norm": 0.5194401187437783,
+      "learning_rate": 3.100615269552523e-06,
+      "loss": 0.6972,
+      "step": 34680
+    },
+    {
+      "epoch": 0.7451561627357477,
+      "grad_norm": 0.5061536798223207,
+      "learning_rate": 3.0956826200092114e-06,
+      "loss": 0.6966,
+      "step": 34690
+    },
+    {
+      "epoch": 0.7453709670490184,
+      "grad_norm": 0.5083820365394748,
+      "learning_rate": 3.090753178542657e-06,
+      "loss": 0.6783,
+      "step": 34700
+    },
+    {
+      "epoch": 0.745585771362289,
+      "grad_norm": 0.5077977495235433,
+      "learning_rate": 3.0858269474433165e-06,
+      "loss": 0.7079,
+      "step": 34710
+    },
+    {
+      "epoch": 0.7458005756755596,
+      "grad_norm": 0.512990160380994,
+      "learning_rate": 3.080903929000153e-06,
+      "loss": 0.6993,
+      "step": 34720
+    },
+    {
+      "epoch": 0.7460153799888302,
+      "grad_norm": 0.517593034507449,
+      "learning_rate": 3.0759841255006386e-06,
+      "loss": 0.7033,
+      "step": 34730
+    },
+    {
+      "epoch": 0.7462301843021008,
+      "grad_norm": 0.4897976460660666,
+      "learning_rate": 3.0710675392307477e-06,
+      "loss": 0.6933,
+      "step": 34740
+    },
+    {
+      "epoch": 0.7464449886153715,
+      "grad_norm": 0.4947523248312534,
+      "learning_rate": 3.066154172474962e-06,
+      "loss": 0.6946,
+      "step": 34750
+    },
+    {
+      "epoch": 0.746659792928642,
+      "grad_norm": 0.5091626362163222,
+      "learning_rate": 3.0612440275162727e-06,
+      "loss": 0.6867,
+      "step": 34760
+    },
+    {
+      "epoch": 0.7468745972419126,
+      "grad_norm": 0.5114642942026933,
+      "learning_rate": 3.056337106636159e-06,
+      "loss": 0.6996,
+      "step": 34770
+    },
+    {
+      "epoch": 0.7470894015551832,
+      "grad_norm": 0.5014471572550719,
+      "learning_rate": 3.0514334121146173e-06,
+      "loss": 0.7086,
+      "step": 34780
+    },
+    {
+      "epoch": 0.7473042058684538,
+      "grad_norm": 0.5084778446042698,
+      "learning_rate": 3.046532946230136e-06,
+      "loss": 0.6935,
+      "step": 34790
+    },
+    {
+      "epoch": 0.7475190101817244,
+      "grad_norm": 0.5229812004697073,
+      "learning_rate": 3.0416357112597108e-06,
+      "loss": 0.7039,
+      "step": 34800
+    },
+    {
+      "epoch": 0.7477338144949951,
+      "grad_norm": 0.5019941677521293,
+      "learning_rate": 3.0367417094788308e-06,
+      "loss": 0.7018,
+      "step": 34810
+    },
+    {
+      "epoch": 0.7479486188082657,
+      "grad_norm": 0.5255533916684315,
+      "learning_rate": 3.0318509431614794e-06,
+      "loss": 0.7103,
+      "step": 34820
+    },
+    {
+      "epoch": 0.7481634231215363,
+      "grad_norm": 0.5094742009361463,
+      "learning_rate": 3.026963414580145e-06,
+      "loss": 0.6916,
+      "step": 34830
+    },
+    {
+      "epoch": 0.7483782274348069,
+      "grad_norm": 0.503694503035299,
+      "learning_rate": 3.0220791260058057e-06,
+      "loss": 0.7062,
+      "step": 34840
+    },
+    {
+      "epoch": 0.7485930317480775,
+      "grad_norm": 0.5306897240771807,
+      "learning_rate": 3.017198079707937e-06,
+      "loss": 0.7022,
+      "step": 34850
+    },
+    {
+      "epoch": 0.7488078360613482,
+      "grad_norm": 0.5166101019632601,
+      "learning_rate": 3.01232027795451e-06,
+      "loss": 0.7041,
+      "step": 34860
+    },
+    {
+      "epoch": 0.7490226403746187,
+      "grad_norm": 0.5145609190625092,
+      "learning_rate": 3.0074457230119768e-06,
+      "loss": 0.6888,
+      "step": 34870
+    },
+    {
+      "epoch": 0.7492374446878893,
+      "grad_norm": 0.508976620646315,
+      "learning_rate": 3.0025744171452997e-06,
+      "loss": 0.7115,
+      "step": 34880
+    },
+    {
+      "epoch": 0.7494522490011599,
+      "grad_norm": 0.5253897144615868,
+      "learning_rate": 2.9977063626179128e-06,
+      "loss": 0.6893,
+      "step": 34890
+    },
+    {
+      "epoch": 0.7496670533144305,
+      "grad_norm": 0.5137649765422893,
+      "learning_rate": 2.9928415616917505e-06,
+      "loss": 0.6867,
+      "step": 34900
+    },
+    {
+      "epoch": 0.7498818576277012,
+      "grad_norm": 0.49499574504817595,
+      "learning_rate": 2.9879800166272355e-06,
+      "loss": 0.7002,
+      "step": 34910
+    },
+    {
+      "epoch": 0.7500966619409718,
+      "grad_norm": 0.519084029719955,
+      "learning_rate": 2.983121729683265e-06,
+      "loss": 0.7104,
+      "step": 34920
+    },
+    {
+      "epoch": 0.7503114662542424,
+      "grad_norm": 0.4981518656316401,
+      "learning_rate": 2.9782667031172454e-06,
+      "loss": 0.7028,
+      "step": 34930
+    },
+    {
+      "epoch": 0.750526270567513,
+      "grad_norm": 0.51741442278905,
+      "learning_rate": 2.973414939185041e-06,
+      "loss": 0.6982,
+      "step": 34940
+    },
+    {
+      "epoch": 0.7507410748807836,
+      "grad_norm": 0.5060387675297806,
+      "learning_rate": 2.9685664401410277e-06,
+      "loss": 0.697,
+      "step": 34950
+    },
+    {
+      "epoch": 0.7509558791940543,
+      "grad_norm": 0.5048062525549186,
+      "learning_rate": 2.9637212082380395e-06,
+      "loss": 0.71,
+      "step": 34960
+    },
+    {
+      "epoch": 0.7511706835073249,
+      "grad_norm": 0.5212032772870495,
+      "learning_rate": 2.9588792457274075e-06,
+      "loss": 0.6943,
+      "step": 34970
+    },
+    {
+      "epoch": 0.7513854878205954,
+      "grad_norm": 0.5107973393581554,
+      "learning_rate": 2.9540405548589434e-06,
+      "loss": 0.7015,
+      "step": 34980
+    },
+    {
+      "epoch": 0.751600292133866,
+      "grad_norm": 0.49864860876835415,
+      "learning_rate": 2.949205137880924e-06,
+      "loss": 0.69,
+      "step": 34990
+    },
+    {
+      "epoch": 0.7518150964471366,
+      "grad_norm": 0.5307082939307299,
+      "learning_rate": 2.944372997040129e-06,
+      "loss": 0.7061,
+      "step": 35000
+    },
+    {
+      "epoch": 0.7520299007604073,
+      "grad_norm": 0.5193800817097175,
+      "learning_rate": 2.9395441345817932e-06,
+      "loss": 0.6902,
+      "step": 35010
+    },
+    {
+      "epoch": 0.7522447050736779,
+      "grad_norm": 0.5177658570066975,
+      "learning_rate": 2.9347185527496403e-06,
+      "loss": 0.6956,
+      "step": 35020
+    },
+    {
+      "epoch": 0.7524595093869485,
+      "grad_norm": 0.5015771316930479,
+      "learning_rate": 2.929896253785871e-06,
+      "loss": 0.7147,
+      "step": 35030
+    },
+    {
+      "epoch": 0.7526743137002191,
+      "grad_norm": 0.4998092573649113,
+      "learning_rate": 2.9250772399311457e-06,
+      "loss": 0.7021,
+      "step": 35040
+    },
+    {
+      "epoch": 0.7528891180134897,
+      "grad_norm": 0.5249322187174793,
+      "learning_rate": 2.9202615134246225e-06,
+      "loss": 0.6977,
+      "step": 35050
+    },
+    {
+      "epoch": 0.7531039223267603,
+      "grad_norm": 0.5148512865886042,
+      "learning_rate": 2.91544907650391e-06,
+      "loss": 0.6941,
+      "step": 35060
+    },
+    {
+      "epoch": 0.753318726640031,
+      "grad_norm": 0.5031383431200541,
+      "learning_rate": 2.9106399314050993e-06,
+      "loss": 0.7068,
+      "step": 35070
+    },
+    {
+      "epoch": 0.7535335309533016,
+      "grad_norm": 0.5077882216038482,
+      "learning_rate": 2.905834080362754e-06,
+      "loss": 0.7062,
+      "step": 35080
+    },
+    {
+      "epoch": 0.7537483352665721,
+      "grad_norm": 0.5259578113827944,
+      "learning_rate": 2.901031525609891e-06,
+      "loss": 0.6936,
+      "step": 35090
+    },
+    {
+      "epoch": 0.7539631395798427,
+      "grad_norm": 0.5193893619415478,
+      "learning_rate": 2.896232269378022e-06,
+      "loss": 0.7213,
+      "step": 35100
+    },
+    {
+      "epoch": 0.7541779438931133,
+      "grad_norm": 0.5323126089856631,
+      "learning_rate": 2.8914363138970992e-06,
+      "loss": 0.703,
+      "step": 35110
+    },
+    {
+      "epoch": 0.754392748206384,
+      "grad_norm": 0.5173517751625921,
+      "learning_rate": 2.886643661395564e-06,
+      "loss": 0.7092,
+      "step": 35120
+    },
+    {
+      "epoch": 0.7546075525196546,
+      "grad_norm": 0.6741926928108194,
+      "learning_rate": 2.8818543141003043e-06,
+      "loss": 0.6951,
+      "step": 35130
+    },
+    {
+      "epoch": 0.7548223568329252,
+      "grad_norm": 0.5178832293172879,
+      "learning_rate": 2.877068274236683e-06,
+      "loss": 0.6924,
+      "step": 35140
+    },
+    {
+      "epoch": 0.7550371611461958,
+      "grad_norm": 0.5132574620286683,
+      "learning_rate": 2.872285544028528e-06,
+      "loss": 0.6977,
+      "step": 35150
+    },
+    {
+      "epoch": 0.7552519654594664,
+      "grad_norm": 0.523341159554861,
+      "learning_rate": 2.867506125698114e-06,
+      "loss": 0.7007,
+      "step": 35160
+    },
+    {
+      "epoch": 0.7554667697727371,
+      "grad_norm": 0.559952979776443,
+      "learning_rate": 2.862730021466201e-06,
+      "loss": 0.7094,
+      "step": 35170
+    },
+    {
+      "epoch": 0.7556815740860077,
+      "grad_norm": 0.5149609461834663,
+      "learning_rate": 2.8579572335519866e-06,
+      "loss": 0.6943,
+      "step": 35180
+    },
+    {
+      "epoch": 0.7558963783992783,
+      "grad_norm": 0.507539558550583,
+      "learning_rate": 2.853187764173141e-06,
+      "loss": 0.7008,
+      "step": 35190
+    },
+    {
+      "epoch": 0.7561111827125488,
+      "grad_norm": 0.49479288184079256,
+      "learning_rate": 2.848421615545789e-06,
+      "loss": 0.7042,
+      "step": 35200
+    },
+    {
+      "epoch": 0.7563259870258194,
+      "grad_norm": 0.5189990316887987,
+      "learning_rate": 2.8436587898845035e-06,
+      "loss": 0.6971,
+      "step": 35210
+    },
+    {
+      "epoch": 0.7565407913390901,
+      "grad_norm": 0.5162232202770758,
+      "learning_rate": 2.838899289402335e-06,
+      "loss": 0.7109,
+      "step": 35220
+    },
+    {
+      "epoch": 0.7567555956523607,
+      "grad_norm": 0.5165521047716195,
+      "learning_rate": 2.8341431163107648e-06,
+      "loss": 0.6952,
+      "step": 35230
+    },
+    {
+      "epoch": 0.7569703999656313,
+      "grad_norm": 0.4956540738020749,
+      "learning_rate": 2.829390272819742e-06,
+      "loss": 0.6871,
+      "step": 35240
+    },
+    {
+      "epoch": 0.7571852042789019,
+      "grad_norm": 0.513330560976213,
+      "learning_rate": 2.824640761137667e-06,
+      "loss": 0.6974,
+      "step": 35250
+    },
+    {
+      "epoch": 0.7574000085921725,
+      "grad_norm": 0.5032601979915532,
+      "learning_rate": 2.8198945834713885e-06,
+      "loss": 0.6913,
+      "step": 35260
+    },
+    {
+      "epoch": 0.7576148129054432,
+      "grad_norm": 0.5128246318375348,
+      "learning_rate": 2.815151742026213e-06,
+      "loss": 0.7056,
+      "step": 35270
+    },
+    {
+      "epoch": 0.7578296172187138,
+      "grad_norm": 0.5030995060530129,
+      "learning_rate": 2.810412239005885e-06,
+      "loss": 0.713,
+      "step": 35280
+    },
+    {
+      "epoch": 0.7580444215319844,
+      "grad_norm": 0.5038841598854293,
+      "learning_rate": 2.805676076612608e-06,
+      "loss": 0.6996,
+      "step": 35290
+    },
+    {
+      "epoch": 0.758259225845255,
+      "grad_norm": 0.5122170399356182,
+      "learning_rate": 2.8009432570470296e-06,
+      "loss": 0.7059,
+      "step": 35300
+    },
+    {
+      "epoch": 0.7584740301585255,
+      "grad_norm": 0.5156117155914561,
+      "learning_rate": 2.7962137825082446e-06,
+      "loss": 0.684,
+      "step": 35310
+    },
+    {
+      "epoch": 0.7586888344717962,
+      "grad_norm": 0.5049866933728283,
+      "learning_rate": 2.7914876551937953e-06,
+      "loss": 0.6942,
+      "step": 35320
+    },
+    {
+      "epoch": 0.7589036387850668,
+      "grad_norm": 0.5094420858773214,
+      "learning_rate": 2.7867648772996634e-06,
+      "loss": 0.708,
+      "step": 35330
+    },
+    {
+      "epoch": 0.7591184430983374,
+      "grad_norm": 0.5011895237248974,
+      "learning_rate": 2.7820454510202843e-06,
+      "loss": 0.6918,
+      "step": 35340
+    },
+    {
+      "epoch": 0.759333247411608,
+      "grad_norm": 0.504089241568298,
+      "learning_rate": 2.777329378548522e-06,
+      "loss": 0.7058,
+      "step": 35350
+    },
+    {
+      "epoch": 0.7595480517248786,
+      "grad_norm": 0.5365866928409356,
+      "learning_rate": 2.7726166620756934e-06,
+      "loss": 0.692,
+      "step": 35360
+    },
+    {
+      "epoch": 0.7597628560381492,
+      "grad_norm": 0.4984929620909205,
+      "learning_rate": 2.7679073037915516e-06,
+      "loss": 0.6874,
+      "step": 35370
+    },
+    {
+      "epoch": 0.7599776603514199,
+      "grad_norm": 0.5410453059904045,
+      "learning_rate": 2.763201305884291e-06,
+      "loss": 0.7145,
+      "step": 35380
+    },
+    {
+      "epoch": 0.7601924646646905,
+      "grad_norm": 0.5522929417691606,
+      "learning_rate": 2.758498670540546e-06,
+      "loss": 0.7132,
+      "step": 35390
+    },
+    {
+      "epoch": 0.7604072689779611,
+      "grad_norm": 0.5191149506874112,
+      "learning_rate": 2.7537993999453818e-06,
+      "loss": 0.7094,
+      "step": 35400
+    },
+    {
+      "epoch": 0.7606220732912317,
+      "grad_norm": 0.5364420903269488,
+      "learning_rate": 2.749103496282306e-06,
+      "loss": 0.7105,
+      "step": 35410
+    },
+    {
+      "epoch": 0.7608368776045022,
+      "grad_norm": 0.5307396572245091,
+      "learning_rate": 2.7444109617332614e-06,
+      "loss": 0.7067,
+      "step": 35420
+    },
+    {
+      "epoch": 0.761051681917773,
+      "grad_norm": 0.5189987491520354,
+      "learning_rate": 2.739721798478625e-06,
+      "loss": 0.6967,
+      "step": 35430
+    },
+    {
+      "epoch": 0.7612664862310435,
+      "grad_norm": 0.5522945068168588,
+      "learning_rate": 2.735036008697205e-06,
+      "loss": 0.7211,
+      "step": 35440
+    },
+    {
+      "epoch": 0.7614812905443141,
+      "grad_norm": 0.49783370978462715,
+      "learning_rate": 2.7303535945662485e-06,
+      "loss": 0.6972,
+      "step": 35450
+    },
+    {
+      "epoch": 0.7616960948575847,
+      "grad_norm": 0.5134039941417213,
+      "learning_rate": 2.725674558261423e-06,
+      "loss": 0.6903,
+      "step": 35460
+    },
+    {
+      "epoch": 0.7619108991708553,
+      "grad_norm": 0.5056018475415773,
+      "learning_rate": 2.720998901956836e-06,
+      "loss": 0.6978,
+      "step": 35470
+    },
+    {
+      "epoch": 0.762125703484126,
+      "grad_norm": 0.5016409320759787,
+      "learning_rate": 2.7163266278250222e-06,
+      "loss": 0.7092,
+      "step": 35480
+    },
+    {
+      "epoch": 0.7623405077973966,
+      "grad_norm": 0.5090009069207025,
+      "learning_rate": 2.7116577380369434e-06,
+      "loss": 0.7068,
+      "step": 35490
+    },
+    {
+      "epoch": 0.7625553121106672,
+      "grad_norm": 0.5138423219447262,
+      "learning_rate": 2.7069922347619926e-06,
+      "loss": 0.6999,
+      "step": 35500
+    },
+    {
+      "epoch": 0.7627701164239378,
+      "grad_norm": 0.5079624500816473,
+      "learning_rate": 2.7023301201679763e-06,
+      "loss": 0.7128,
+      "step": 35510
+    },
+    {
+      "epoch": 0.7629849207372084,
+      "grad_norm": 0.519763641884365,
+      "learning_rate": 2.69767139642115e-06,
+      "loss": 0.6984,
+      "step": 35520
+    },
+    {
+      "epoch": 0.7631997250504791,
+      "grad_norm": 0.526599503026077,
+      "learning_rate": 2.6930160656861704e-06,
+      "loss": 0.6996,
+      "step": 35530
+    },
+    {
+      "epoch": 0.7634145293637496,
+      "grad_norm": 0.527777664386869,
+      "learning_rate": 2.688364130126131e-06,
+      "loss": 0.7001,
+      "step": 35540
+    },
+    {
+      "epoch": 0.7636293336770202,
+      "grad_norm": 0.5167948414932604,
+      "learning_rate": 2.6837155919025426e-06,
+      "loss": 0.6886,
+      "step": 35550
+    },
+    {
+      "epoch": 0.7638441379902908,
+      "grad_norm": 0.5005085326477348,
+      "learning_rate": 2.6790704531753385e-06,
+      "loss": 0.6922,
+      "step": 35560
+    },
+    {
+      "epoch": 0.7640589423035614,
+      "grad_norm": 0.5008777298038203,
+      "learning_rate": 2.6744287161028782e-06,
+      "loss": 0.6941,
+      "step": 35570
+    },
+    {
+      "epoch": 0.7642737466168321,
+      "grad_norm": 0.5168378019628393,
+      "learning_rate": 2.669790382841928e-06,
+      "loss": 0.7073,
+      "step": 35580
+    },
+    {
+      "epoch": 0.7644885509301027,
+      "grad_norm": 0.5064269251779776,
+      "learning_rate": 2.665155455547682e-06,
+      "loss": 0.7061,
+      "step": 35590
+    },
+    {
+      "epoch": 0.7647033552433733,
+      "grad_norm": 0.5045676176502482,
+      "learning_rate": 2.66052393637375e-06,
+      "loss": 0.7015,
+      "step": 35600
+    },
+    {
+      "epoch": 0.7649181595566439,
+      "grad_norm": 0.4956210829558816,
+      "learning_rate": 2.655895827472158e-06,
+      "loss": 0.6862,
+      "step": 35610
+    },
+    {
+      "epoch": 0.7651329638699145,
+      "grad_norm": 0.5097661066228676,
+      "learning_rate": 2.6512711309933503e-06,
+      "loss": 0.6848,
+      "step": 35620
+    },
+    {
+      "epoch": 0.7653477681831852,
+      "grad_norm": 0.510005397820754,
+      "learning_rate": 2.6466498490861734e-06,
+      "loss": 0.6905,
+      "step": 35630
+    },
+    {
+      "epoch": 0.7655625724964558,
+      "grad_norm": 0.5253052971997748,
+      "learning_rate": 2.642031983897909e-06,
+      "loss": 0.6904,
+      "step": 35640
+    },
+    {
+      "epoch": 0.7657773768097264,
+      "grad_norm": 0.5027879257266836,
+      "learning_rate": 2.6374175375742285e-06,
+      "loss": 0.685,
+      "step": 35650
+    },
+    {
+      "epoch": 0.7659921811229969,
+      "grad_norm": 0.519001207669886,
+      "learning_rate": 2.6328065122592284e-06,
+      "loss": 0.6946,
+      "step": 35660
+    },
+    {
+      "epoch": 0.7662069854362675,
+      "grad_norm": 0.5156599981554072,
+      "learning_rate": 2.6281989100954155e-06,
+      "loss": 0.6955,
+      "step": 35670
+    },
+    {
+      "epoch": 0.7664217897495381,
+      "grad_norm": 0.5065842964681723,
+      "learning_rate": 2.623594733223692e-06,
+      "loss": 0.6901,
+      "step": 35680
+    },
+    {
+      "epoch": 0.7666365940628088,
+      "grad_norm": 0.5043045850760082,
+      "learning_rate": 2.6189939837833934e-06,
+      "loss": 0.6807,
+      "step": 35690
+    },
+    {
+      "epoch": 0.7668513983760794,
+      "grad_norm": 0.5203139211551309,
+      "learning_rate": 2.6143966639122343e-06,
+      "loss": 0.7069,
+      "step": 35700
+    },
+    {
+      "epoch": 0.76706620268935,
+      "grad_norm": 0.5059570893757329,
+      "learning_rate": 2.609802775746363e-06,
+      "loss": 0.6829,
+      "step": 35710
+    },
+    {
+      "epoch": 0.7672810070026206,
+      "grad_norm": 0.5292183230754148,
+      "learning_rate": 2.6052123214203106e-06,
+      "loss": 0.7116,
+      "step": 35720
+    },
+    {
+      "epoch": 0.7674958113158912,
+      "grad_norm": 0.5050171351436751,
+      "learning_rate": 2.6006253030670246e-06,
+      "loss": 0.6933,
+      "step": 35730
+    },
+    {
+      "epoch": 0.7677106156291619,
+      "grad_norm": 0.5367079725288396,
+      "learning_rate": 2.596041722817859e-06,
+      "loss": 0.6986,
+      "step": 35740
+    },
+    {
+      "epoch": 0.7679254199424325,
+      "grad_norm": 0.5157066385692741,
+      "learning_rate": 2.5914615828025534e-06,
+      "loss": 0.7029,
+      "step": 35750
+    },
+    {
+      "epoch": 0.768140224255703,
+      "grad_norm": 0.5036694678541429,
+      "learning_rate": 2.5868848851492733e-06,
+      "loss": 0.688,
+      "step": 35760
+    },
+    {
+      "epoch": 0.7683550285689736,
+      "grad_norm": 0.5071285342347541,
+      "learning_rate": 2.5823116319845633e-06,
+      "loss": 0.6878,
+      "step": 35770
+    },
+    {
+      "epoch": 0.7685698328822442,
+      "grad_norm": 0.5068127626353475,
+      "learning_rate": 2.5777418254333775e-06,
+      "loss": 0.6938,
+      "step": 35780
+    },
+    {
+      "epoch": 0.7687846371955149,
+      "grad_norm": 0.5198659666241273,
+      "learning_rate": 2.5731754676190725e-06,
+      "loss": 0.6919,
+      "step": 35790
+    },
+    {
+      "epoch": 0.7689994415087855,
+      "grad_norm": 0.5040011486527468,
+      "learning_rate": 2.568612560663385e-06,
+      "loss": 0.6883,
+      "step": 35800
+    },
+    {
+      "epoch": 0.7692142458220561,
+      "grad_norm": 0.5095849632751419,
+      "learning_rate": 2.5640531066864737e-06,
+      "loss": 0.6935,
+      "step": 35810
+    },
+    {
+      "epoch": 0.7694290501353267,
+      "grad_norm": 0.5031387126877207,
+      "learning_rate": 2.559497107806871e-06,
+      "loss": 0.6873,
+      "step": 35820
+    },
+    {
+      "epoch": 0.7696438544485973,
+      "grad_norm": 0.5069738792715828,
+      "learning_rate": 2.554944566141515e-06,
+      "loss": 0.6938,
+      "step": 35830
+    },
+    {
+      "epoch": 0.769858658761868,
+      "grad_norm": 0.5225177154611406,
+      "learning_rate": 2.5503954838057367e-06,
+      "loss": 0.6986,
+      "step": 35840
+    },
+    {
+      "epoch": 0.7700734630751386,
+      "grad_norm": 0.5337253376024659,
+      "learning_rate": 2.54584986291325e-06,
+      "loss": 0.6938,
+      "step": 35850
+    },
+    {
+      "epoch": 0.7702882673884092,
+      "grad_norm": 0.4974683440922778,
+      "learning_rate": 2.54130770557618e-06,
+      "loss": 0.7002,
+      "step": 35860
+    },
+    {
+      "epoch": 0.7705030717016798,
+      "grad_norm": 0.5164372270920562,
+      "learning_rate": 2.536769013905022e-06,
+      "loss": 0.7042,
+      "step": 35870
+    },
+    {
+      "epoch": 0.7707178760149503,
+      "grad_norm": 0.5045862598256815,
+      "learning_rate": 2.532233790008671e-06,
+      "loss": 0.7122,
+      "step": 35880
+    },
+    {
+      "epoch": 0.770932680328221,
+      "grad_norm": 0.5039493617623143,
+      "learning_rate": 2.5277020359944114e-06,
+      "loss": 0.6996,
+      "step": 35890
+    },
+    {
+      "epoch": 0.7711474846414916,
+      "grad_norm": 0.4974176270519638,
+      "learning_rate": 2.5231737539679124e-06,
+      "loss": 0.7015,
+      "step": 35900
+    },
+    {
+      "epoch": 0.7713622889547622,
+      "grad_norm": 0.5183097985832663,
+      "learning_rate": 2.518648946033234e-06,
+      "loss": 0.6908,
+      "step": 35910
+    },
+    {
+      "epoch": 0.7715770932680328,
+      "grad_norm": 0.525976552546943,
+      "learning_rate": 2.5141276142928093e-06,
+      "loss": 0.6969,
+      "step": 35920
+    },
+    {
+      "epoch": 0.7717918975813034,
+      "grad_norm": 0.5332780233463744,
+      "learning_rate": 2.509609760847479e-06,
+      "loss": 0.6916,
+      "step": 35930
+    },
+    {
+      "epoch": 0.772006701894574,
+      "grad_norm": 0.5111293385118552,
+      "learning_rate": 2.5050953877964446e-06,
+      "loss": 0.7115,
+      "step": 35940
+    },
+    {
+      "epoch": 0.7722215062078447,
+      "grad_norm": 0.5304473121055377,
+      "learning_rate": 2.500584497237303e-06,
+      "loss": 0.6997,
+      "step": 35950
+    },
+    {
+      "epoch": 0.7724363105211153,
+      "grad_norm": 0.5202204337822746,
+      "learning_rate": 2.4960770912660324e-06,
+      "loss": 0.6949,
+      "step": 35960
+    },
+    {
+      "epoch": 0.7726511148343859,
+      "grad_norm": 0.5082441180489494,
+      "learning_rate": 2.491573171976982e-06,
+      "loss": 0.6936,
+      "step": 35970
+    },
+    {
+      "epoch": 0.7728659191476565,
+      "grad_norm": 0.5085533382278553,
+      "learning_rate": 2.4870727414629005e-06,
+      "loss": 0.6995,
+      "step": 35980
+    },
+    {
+      "epoch": 0.773080723460927,
+      "grad_norm": 0.5110984394108816,
+      "learning_rate": 2.482575801814894e-06,
+      "loss": 0.6912,
+      "step": 35990
+    },
+    {
+      "epoch": 0.7732955277741977,
+      "grad_norm": 0.5333234352124336,
+      "learning_rate": 2.4780823551224586e-06,
+      "loss": 0.7008,
+      "step": 36000
+    },
+    {
+      "epoch": 0.7735103320874683,
+      "grad_norm": 0.5104268992806394,
+      "learning_rate": 2.473592403473466e-06,
+      "loss": 0.6972,
+      "step": 36010
+    },
+    {
+      "epoch": 0.7737251364007389,
+      "grad_norm": 0.5095548549309175,
+      "learning_rate": 2.4691059489541637e-06,
+      "loss": 0.6833,
+      "step": 36020
+    },
+    {
+      "epoch": 0.7739399407140095,
+      "grad_norm": 0.5379203466934536,
+      "learning_rate": 2.464622993649174e-06,
+      "loss": 0.6885,
+      "step": 36030
+    },
+    {
+      "epoch": 0.7741547450272801,
+      "grad_norm": 0.5103185774726944,
+      "learning_rate": 2.46014353964149e-06,
+      "loss": 0.6972,
+      "step": 36040
+    },
+    {
+      "epoch": 0.7743695493405508,
+      "grad_norm": 0.5108530905870736,
+      "learning_rate": 2.4556675890124803e-06,
+      "loss": 0.6918,
+      "step": 36050
+    },
+    {
+      "epoch": 0.7745843536538214,
+      "grad_norm": 0.5114449569980617,
+      "learning_rate": 2.451195143841889e-06,
+      "loss": 0.6982,
+      "step": 36060
+    },
+    {
+      "epoch": 0.774799157967092,
+      "grad_norm": 0.5117859340036564,
+      "learning_rate": 2.446726206207827e-06,
+      "loss": 0.7139,
+      "step": 36070
+    },
+    {
+      "epoch": 0.7750139622803626,
+      "grad_norm": 0.5482258536620003,
+      "learning_rate": 2.442260778186777e-06,
+      "loss": 0.7027,
+      "step": 36080
+    },
+    {
+      "epoch": 0.7752287665936332,
+      "grad_norm": 0.5072515818925265,
+      "learning_rate": 2.4377988618535943e-06,
+      "loss": 0.6881,
+      "step": 36090
+    },
+    {
+      "epoch": 0.7754435709069039,
+      "grad_norm": 0.4985942219561439,
+      "learning_rate": 2.4333404592814923e-06,
+      "loss": 0.6863,
+      "step": 36100
+    },
+    {
+      "epoch": 0.7756583752201744,
+      "grad_norm": 0.5028179650880126,
+      "learning_rate": 2.4288855725420625e-06,
+      "loss": 0.6791,
+      "step": 36110
+    },
+    {
+      "epoch": 0.775873179533445,
+      "grad_norm": 0.5497148140060542,
+      "learning_rate": 2.424434203705257e-06,
+      "loss": 0.7117,
+      "step": 36120
+    },
+    {
+      "epoch": 0.7760879838467156,
+      "grad_norm": 0.5121974367383552,
+      "learning_rate": 2.4199863548393978e-06,
+      "loss": 0.6947,
+      "step": 36130
+    },
+    {
+      "epoch": 0.7763027881599862,
+      "grad_norm": 0.5024469961030319,
+      "learning_rate": 2.4155420280111673e-06,
+      "loss": 0.7016,
+      "step": 36140
+    },
+    {
+      "epoch": 0.7765175924732569,
+      "grad_norm": 0.49804034625215227,
+      "learning_rate": 2.411101225285616e-06,
+      "loss": 0.6963,
+      "step": 36150
+    },
+    {
+      "epoch": 0.7767323967865275,
+      "grad_norm": 0.5197970242944042,
+      "learning_rate": 2.406663948726147e-06,
+      "loss": 0.6917,
+      "step": 36160
+    },
+    {
+      "epoch": 0.7769472010997981,
+      "grad_norm": 0.5243687759024098,
+      "learning_rate": 2.4022302003945343e-06,
+      "loss": 0.6886,
+      "step": 36170
+    },
+    {
+      "epoch": 0.7771620054130687,
+      "grad_norm": 0.5067864445033509,
+      "learning_rate": 2.3977999823509112e-06,
+      "loss": 0.7043,
+      "step": 36180
+    },
+    {
+      "epoch": 0.7773768097263393,
+      "grad_norm": 0.4969747214927612,
+      "learning_rate": 2.3933732966537683e-06,
+      "loss": 0.7067,
+      "step": 36190
+    },
+    {
+      "epoch": 0.77759161403961,
+      "grad_norm": 0.5173099358575356,
+      "learning_rate": 2.3889501453599575e-06,
+      "loss": 0.7042,
+      "step": 36200
+    },
+    {
+      "epoch": 0.7778064183528806,
+      "grad_norm": 0.5247869692545869,
+      "learning_rate": 2.384530530524688e-06,
+      "loss": 0.6935,
+      "step": 36210
+    },
+    {
+      "epoch": 0.7780212226661511,
+      "grad_norm": 0.5564841459852391,
+      "learning_rate": 2.3801144542015197e-06,
+      "loss": 0.7144,
+      "step": 36220
+    },
+    {
+      "epoch": 0.7782360269794217,
+      "grad_norm": 0.5269720198709235,
+      "learning_rate": 2.3757019184423756e-06,
+      "loss": 0.7053,
+      "step": 36230
+    },
+    {
+      "epoch": 0.7784508312926923,
+      "grad_norm": 0.4911231646939383,
+      "learning_rate": 2.3712929252975327e-06,
+      "loss": 0.705,
+      "step": 36240
+    },
+    {
+      "epoch": 0.7786656356059629,
+      "grad_norm": 0.5168096144036898,
+      "learning_rate": 2.366887476815619e-06,
+      "loss": 0.7133,
+      "step": 36250
+    },
+    {
+      "epoch": 0.7788804399192336,
+      "grad_norm": 0.4905748157218672,
+      "learning_rate": 2.3624855750436206e-06,
+      "loss": 0.7,
+      "step": 36260
+    },
+    {
+      "epoch": 0.7790952442325042,
+      "grad_norm": 0.5059325828111079,
+      "learning_rate": 2.3580872220268623e-06,
+      "loss": 0.6928,
+      "step": 36270
+    },
+    {
+      "epoch": 0.7793100485457748,
+      "grad_norm": 0.5163071515499138,
+      "learning_rate": 2.3536924198090437e-06,
+      "loss": 0.7068,
+      "step": 36280
+    },
+    {
+      "epoch": 0.7795248528590454,
+      "grad_norm": 0.5432830473338806,
+      "learning_rate": 2.34930117043219e-06,
+      "loss": 0.7001,
+      "step": 36290
+    },
+    {
+      "epoch": 0.779739657172316,
+      "grad_norm": 0.5014221779118088,
+      "learning_rate": 2.344913475936689e-06,
+      "loss": 0.7067,
+      "step": 36300
+    },
+    {
+      "epoch": 0.7799544614855867,
+      "grad_norm": 0.5190757630635009,
+      "learning_rate": 2.340529338361275e-06,
+      "loss": 0.7077,
+      "step": 36310
+    },
+    {
+      "epoch": 0.7801692657988573,
+      "grad_norm": 0.5123110569142603,
+      "learning_rate": 2.3361487597430265e-06,
+      "loss": 0.71,
+      "step": 36320
+    },
+    {
+      "epoch": 0.7803840701121278,
+      "grad_norm": 0.5208933584534376,
+      "learning_rate": 2.3317717421173747e-06,
+      "loss": 0.7115,
+      "step": 36330
+    },
+    {
+      "epoch": 0.7805988744253984,
+      "grad_norm": 0.5101156973048868,
+      "learning_rate": 2.3273982875180832e-06,
+      "loss": 0.6866,
+      "step": 36340
+    },
+    {
+      "epoch": 0.780813678738669,
+      "grad_norm": 0.5277994764678194,
+      "learning_rate": 2.3230283979772795e-06,
+      "loss": 0.7037,
+      "step": 36350
+    },
+    {
+      "epoch": 0.7810284830519397,
+      "grad_norm": 0.5609421047646885,
+      "learning_rate": 2.318662075525415e-06,
+      "loss": 0.7071,
+      "step": 36360
+    },
+    {
+      "epoch": 0.7812432873652103,
+      "grad_norm": 0.5162031245256725,
+      "learning_rate": 2.3142993221912968e-06,
+      "loss": 0.6859,
+      "step": 36370
+    },
+    {
+      "epoch": 0.7814580916784809,
+      "grad_norm": 0.526786457219609,
+      "learning_rate": 2.3099401400020693e-06,
+      "loss": 0.7008,
+      "step": 36380
+    },
+    {
+      "epoch": 0.7816728959917515,
+      "grad_norm": 0.5462499765613852,
+      "learning_rate": 2.3055845309832113e-06,
+      "loss": 0.6883,
+      "step": 36390
+    },
+    {
+      "epoch": 0.7818877003050221,
+      "grad_norm": 0.517477588318434,
+      "learning_rate": 2.3012324971585575e-06,
+      "loss": 0.7137,
+      "step": 36400
+    },
+    {
+      "epoch": 0.7821025046182928,
+      "grad_norm": 0.5159739518570036,
+      "learning_rate": 2.2968840405502636e-06,
+      "loss": 0.6996,
+      "step": 36410
+    },
+    {
+      "epoch": 0.7823173089315634,
+      "grad_norm": 0.511918446890818,
+      "learning_rate": 2.292539163178834e-06,
+      "loss": 0.6882,
+      "step": 36420
+    },
+    {
+      "epoch": 0.782532113244834,
+      "grad_norm": 0.5294058439127566,
+      "learning_rate": 2.288197867063109e-06,
+      "loss": 0.6968,
+      "step": 36430
+    },
+    {
+      "epoch": 0.7827469175581045,
+      "grad_norm": 0.5295192397598697,
+      "learning_rate": 2.2838601542202543e-06,
+      "loss": 0.699,
+      "step": 36440
+    },
+    {
+      "epoch": 0.7829617218713751,
+      "grad_norm": 0.5163303982789976,
+      "learning_rate": 2.2795260266657905e-06,
+      "loss": 0.6876,
+      "step": 36450
+    },
+    {
+      "epoch": 0.7831765261846458,
+      "grad_norm": 0.528330319215088,
+      "learning_rate": 2.275195486413554e-06,
+      "loss": 0.6974,
+      "step": 36460
+    },
+    {
+      "epoch": 0.7833913304979164,
+      "grad_norm": 0.5239254843684411,
+      "learning_rate": 2.270868535475722e-06,
+      "loss": 0.6845,
+      "step": 36470
+    },
+    {
+      "epoch": 0.783606134811187,
+      "grad_norm": 0.5129180784875449,
+      "learning_rate": 2.266545175862809e-06,
+      "loss": 0.6919,
+      "step": 36480
+    },
+    {
+      "epoch": 0.7838209391244576,
+      "grad_norm": 0.553503515608531,
+      "learning_rate": 2.262225409583643e-06,
+      "loss": 0.6931,
+      "step": 36490
+    },
+    {
+      "epoch": 0.7840357434377282,
+      "grad_norm": 0.5193440729217803,
+      "learning_rate": 2.257909238645408e-06,
+      "loss": 0.691,
+      "step": 36500
+    },
+    {
+      "epoch": 0.7842505477509989,
+      "grad_norm": 0.5409849700495803,
+      "learning_rate": 2.253596665053592e-06,
+      "loss": 0.6901,
+      "step": 36510
+    },
+    {
+      "epoch": 0.7844653520642695,
+      "grad_norm": 0.50913652185814,
+      "learning_rate": 2.2492876908120355e-06,
+      "loss": 0.6969,
+      "step": 36520
+    },
+    {
+      "epoch": 0.7846801563775401,
+      "grad_norm": 0.5190130256439345,
+      "learning_rate": 2.2449823179228846e-06,
+      "loss": 0.7067,
+      "step": 36530
+    },
+    {
+      "epoch": 0.7848949606908107,
+      "grad_norm": 0.5142508244546575,
+      "learning_rate": 2.240680548386626e-06,
+      "loss": 0.6986,
+      "step": 36540
+    },
+    {
+      "epoch": 0.7851097650040813,
+      "grad_norm": 0.5389748247074397,
+      "learning_rate": 2.2363823842020694e-06,
+      "loss": 0.7,
+      "step": 36550
+    },
+    {
+      "epoch": 0.7853245693173518,
+      "grad_norm": 0.5077039536630492,
+      "learning_rate": 2.2320878273663402e-06,
+      "loss": 0.6961,
+      "step": 36560
+    },
+    {
+      "epoch": 0.7855393736306225,
+      "grad_norm": 0.5387782887838709,
+      "learning_rate": 2.2277968798749074e-06,
+      "loss": 0.7003,
+      "step": 36570
+    },
+    {
+      "epoch": 0.7857541779438931,
+      "grad_norm": 0.5181751565840694,
+      "learning_rate": 2.2235095437215416e-06,
+      "loss": 0.6927,
+      "step": 36580
+    },
+    {
+      "epoch": 0.7859689822571637,
+      "grad_norm": 0.5150089685898623,
+      "learning_rate": 2.2192258208983474e-06,
+      "loss": 0.6922,
+      "step": 36590
+    },
+    {
+      "epoch": 0.7861837865704343,
+      "grad_norm": 0.5203313759269558,
+      "learning_rate": 2.2149457133957498e-06,
+      "loss": 0.6952,
+      "step": 36600
+    },
+    {
+      "epoch": 0.7863985908837049,
+      "grad_norm": 0.513794624005246,
+      "learning_rate": 2.210669223202485e-06,
+      "loss": 0.7039,
+      "step": 36610
+    },
+    {
+      "epoch": 0.7866133951969756,
+      "grad_norm": 0.49987037240309906,
+      "learning_rate": 2.2063963523056265e-06,
+      "loss": 0.6922,
+      "step": 36620
+    },
+    {
+      "epoch": 0.7868281995102462,
+      "grad_norm": 0.5086035486321187,
+      "learning_rate": 2.2021271026905444e-06,
+      "loss": 0.6855,
+      "step": 36630
+    },
+    {
+      "epoch": 0.7870430038235168,
+      "grad_norm": 0.5105940396334009,
+      "learning_rate": 2.1978614763409424e-06,
+      "loss": 0.6969,
+      "step": 36640
+    },
+    {
+      "epoch": 0.7872578081367874,
+      "grad_norm": 0.5037059973128649,
+      "learning_rate": 2.1935994752388323e-06,
+      "loss": 0.696,
+      "step": 36650
+    },
+    {
+      "epoch": 0.787472612450058,
+      "grad_norm": 0.528348362552611,
+      "learning_rate": 2.189341101364546e-06,
+      "loss": 0.6948,
+      "step": 36660
+    },
+    {
+      "epoch": 0.7876874167633287,
+      "grad_norm": 0.5312271902558634,
+      "learning_rate": 2.1850863566967296e-06,
+      "loss": 0.7015,
+      "step": 36670
+    },
+    {
+      "epoch": 0.7879022210765992,
+      "grad_norm": 0.5205776526916766,
+      "learning_rate": 2.1808352432123338e-06,
+      "loss": 0.6935,
+      "step": 36680
+    },
+    {
+      "epoch": 0.7881170253898698,
+      "grad_norm": 0.517986778286259,
+      "learning_rate": 2.1765877628866405e-06,
+      "loss": 0.695,
+      "step": 36690
+    },
+    {
+      "epoch": 0.7883318297031404,
+      "grad_norm": 0.5059031168383098,
+      "learning_rate": 2.172343917693225e-06,
+      "loss": 0.694,
+      "step": 36700
+    },
+    {
+      "epoch": 0.788546634016411,
+      "grad_norm": 0.4977347989860999,
+      "learning_rate": 2.1681037096039826e-06,
+      "loss": 0.6954,
+      "step": 36710
+    },
+    {
+      "epoch": 0.7887614383296817,
+      "grad_norm": 0.5166359451995465,
+      "learning_rate": 2.1638671405891177e-06,
+      "loss": 0.6997,
+      "step": 36720
+    },
+    {
+      "epoch": 0.7889762426429523,
+      "grad_norm": 0.5074952802695203,
+      "learning_rate": 2.159634212617143e-06,
+      "loss": 0.6871,
+      "step": 36730
+    },
+    {
+      "epoch": 0.7891910469562229,
+      "grad_norm": 0.525379462136144,
+      "learning_rate": 2.155404927654884e-06,
+      "loss": 0.6867,
+      "step": 36740
+    },
+    {
+      "epoch": 0.7894058512694935,
+      "grad_norm": 0.5055105654931678,
+      "learning_rate": 2.151179287667463e-06,
+      "loss": 0.685,
+      "step": 36750
+    },
+    {
+      "epoch": 0.7896206555827641,
+      "grad_norm": 0.5392743199170078,
+      "learning_rate": 2.146957294618316e-06,
+      "loss": 0.7023,
+      "step": 36760
+    },
+    {
+      "epoch": 0.7898354598960348,
+      "grad_norm": 0.5245335783053737,
+      "learning_rate": 2.1427389504691854e-06,
+      "loss": 0.687,
+      "step": 36770
+    },
+    {
+      "epoch": 0.7900502642093054,
+      "grad_norm": 0.5382517802199199,
+      "learning_rate": 2.1385242571801145e-06,
+      "loss": 0.702,
+      "step": 36780
+    },
+    {
+      "epoch": 0.7902650685225759,
+      "grad_norm": 0.5268825342443021,
+      "learning_rate": 2.134313216709456e-06,
+      "loss": 0.6948,
+      "step": 36790
+    },
+    {
+      "epoch": 0.7904798728358465,
+      "grad_norm": 0.5169835737072852,
+      "learning_rate": 2.1301058310138556e-06,
+      "loss": 0.676,
+      "step": 36800
+    },
+    {
+      "epoch": 0.7906946771491171,
+      "grad_norm": 0.5044997612963653,
+      "learning_rate": 2.1259021020482674e-06,
+      "loss": 0.7033,
+      "step": 36810
+    },
+    {
+      "epoch": 0.7909094814623878,
+      "grad_norm": 0.514705637218055,
+      "learning_rate": 2.1217020317659463e-06,
+      "loss": 0.6949,
+      "step": 36820
+    },
+    {
+      "epoch": 0.7911242857756584,
+      "grad_norm": 0.5165996618712064,
+      "learning_rate": 2.1175056221184465e-06,
+      "loss": 0.7022,
+      "step": 36830
+    },
+    {
+      "epoch": 0.791339090088929,
+      "grad_norm": 0.5027344033789356,
+      "learning_rate": 2.113312875055621e-06,
+      "loss": 0.6877,
+      "step": 36840
+    },
+    {
+      "epoch": 0.7915538944021996,
+      "grad_norm": 0.5122991945340712,
+      "learning_rate": 2.1091237925256235e-06,
+      "loss": 0.6949,
+      "step": 36850
+    },
+    {
+      "epoch": 0.7917686987154702,
+      "grad_norm": 0.5295857931561139,
+      "learning_rate": 2.1049383764748977e-06,
+      "loss": 0.6942,
+      "step": 36860
+    },
+    {
+      "epoch": 0.7919835030287408,
+      "grad_norm": 0.5120717805001744,
+      "learning_rate": 2.1007566288481905e-06,
+      "loss": 0.6997,
+      "step": 36870
+    },
+    {
+      "epoch": 0.7921983073420115,
+      "grad_norm": 0.519141610866559,
+      "learning_rate": 2.0965785515885416e-06,
+      "loss": 0.6988,
+      "step": 36880
+    },
+    {
+      "epoch": 0.792413111655282,
+      "grad_norm": 0.5333090103926987,
+      "learning_rate": 2.0924041466372878e-06,
+      "loss": 0.7008,
+      "step": 36890
+    },
+    {
+      "epoch": 0.7926279159685526,
+      "grad_norm": 0.5168789441545899,
+      "learning_rate": 2.0882334159340566e-06,
+      "loss": 0.6837,
+      "step": 36900
+    },
+    {
+      "epoch": 0.7928427202818232,
+      "grad_norm": 0.5249764329034577,
+      "learning_rate": 2.0840663614167698e-06,
+      "loss": 0.699,
+      "step": 36910
+    },
+    {
+      "epoch": 0.7930575245950938,
+      "grad_norm": 0.5146899618296968,
+      "learning_rate": 2.0799029850216424e-06,
+      "loss": 0.6942,
+      "step": 36920
+    },
+    {
+      "epoch": 0.7932723289083645,
+      "grad_norm": 0.5166243708864191,
+      "learning_rate": 2.075743288683174e-06,
+      "loss": 0.7027,
+      "step": 36930
+    },
+    {
+      "epoch": 0.7934871332216351,
+      "grad_norm": 0.520103403645907,
+      "learning_rate": 2.0715872743341613e-06,
+      "loss": 0.6913,
+      "step": 36940
+    },
+    {
+      "epoch": 0.7937019375349057,
+      "grad_norm": 0.5661241316056728,
+      "learning_rate": 2.0674349439056884e-06,
+      "loss": 0.6916,
+      "step": 36950
+    },
+    {
+      "epoch": 0.7939167418481763,
+      "grad_norm": 0.5413841506028565,
+      "learning_rate": 2.0632862993271264e-06,
+      "loss": 0.6956,
+      "step": 36960
+    },
+    {
+      "epoch": 0.7941315461614469,
+      "grad_norm": 0.5057462755052492,
+      "learning_rate": 2.0591413425261364e-06,
+      "loss": 0.6967,
+      "step": 36970
+    },
+    {
+      "epoch": 0.7943463504747176,
+      "grad_norm": 0.5036998109773616,
+      "learning_rate": 2.0550000754286603e-06,
+      "loss": 0.6904,
+      "step": 36980
+    },
+    {
+      "epoch": 0.7945611547879882,
+      "grad_norm": 0.5141790017851666,
+      "learning_rate": 2.05086249995893e-06,
+      "loss": 0.6862,
+      "step": 36990
+    },
+    {
+      "epoch": 0.7947759591012588,
+      "grad_norm": 0.504848106422294,
+      "learning_rate": 2.046728618039464e-06,
+      "loss": 0.6813,
+      "step": 37000
+    },
+    {
+      "epoch": 0.7949907634145293,
+      "grad_norm": 0.5069448860207141,
+      "learning_rate": 2.0425984315910597e-06,
+      "loss": 0.6896,
+      "step": 37010
+    },
+    {
+      "epoch": 0.7952055677277999,
+      "grad_norm": 0.5179956584180097,
+      "learning_rate": 2.0384719425328025e-06,
+      "loss": 0.6923,
+      "step": 37020
+    },
+    {
+      "epoch": 0.7954203720410706,
+      "grad_norm": 0.5137679574180213,
+      "learning_rate": 2.0343491527820504e-06,
+      "loss": 0.7031,
+      "step": 37030
+    },
+    {
+      "epoch": 0.7956351763543412,
+      "grad_norm": 0.5076257848053785,
+      "learning_rate": 2.0302300642544583e-06,
+      "loss": 0.6845,
+      "step": 37040
+    },
+    {
+      "epoch": 0.7958499806676118,
+      "grad_norm": 0.5259087825677118,
+      "learning_rate": 2.0261146788639453e-06,
+      "loss": 0.6858,
+      "step": 37050
+    },
+    {
+      "epoch": 0.7960647849808824,
+      "grad_norm": 0.5244573164373156,
+      "learning_rate": 2.0220029985227175e-06,
+      "loss": 0.6967,
+      "step": 37060
+    },
+    {
+      "epoch": 0.796279589294153,
+      "grad_norm": 0.5289311220896381,
+      "learning_rate": 2.017895025141264e-06,
+      "loss": 0.7002,
+      "step": 37070
+    },
+    {
+      "epoch": 0.7964943936074237,
+      "grad_norm": 0.5141566850943272,
+      "learning_rate": 2.013790760628336e-06,
+      "loss": 0.694,
+      "step": 37080
+    },
+    {
+      "epoch": 0.7967091979206943,
+      "grad_norm": 0.5130864502827068,
+      "learning_rate": 2.009690206890984e-06,
+      "loss": 0.7102,
+      "step": 37090
+    },
+    {
+      "epoch": 0.7969240022339649,
+      "grad_norm": 0.5258889044125008,
+      "learning_rate": 2.0055933658345094e-06,
+      "loss": 0.7073,
+      "step": 37100
+    },
+    {
+      "epoch": 0.7971388065472355,
+      "grad_norm": 0.5220405637403223,
+      "learning_rate": 2.0015002393625114e-06,
+      "loss": 0.6862,
+      "step": 37110
+    },
+    {
+      "epoch": 0.797353610860506,
+      "grad_norm": 0.5143925147907012,
+      "learning_rate": 1.997410829376847e-06,
+      "loss": 0.6992,
+      "step": 37120
+    },
+    {
+      "epoch": 0.7975684151737766,
+      "grad_norm": 0.501877470630008,
+      "learning_rate": 1.993325137777652e-06,
+      "loss": 0.6885,
+      "step": 37130
+    },
+    {
+      "epoch": 0.7977832194870473,
+      "grad_norm": 0.5203927062197888,
+      "learning_rate": 1.9892431664633393e-06,
+      "loss": 0.697,
+      "step": 37140
+    },
+    {
+      "epoch": 0.7979980238003179,
+      "grad_norm": 0.5025995904904371,
+      "learning_rate": 1.9851649173305798e-06,
+      "loss": 0.6942,
+      "step": 37150
+    },
+    {
+      "epoch": 0.7982128281135885,
+      "grad_norm": 0.5191545346856352,
+      "learning_rate": 1.981090392274333e-06,
+      "loss": 0.6966,
+      "step": 37160
+    },
+    {
+      "epoch": 0.7984276324268591,
+      "grad_norm": 0.5131619635945721,
+      "learning_rate": 1.9770195931878123e-06,
+      "loss": 0.6826,
+      "step": 37170
+    },
+    {
+      "epoch": 0.7986424367401297,
+      "grad_norm": 0.5218712729260536,
+      "learning_rate": 1.9729525219625077e-06,
+      "loss": 0.6889,
+      "step": 37180
+    },
+    {
+      "epoch": 0.7988572410534004,
+      "grad_norm": 0.5326896100885731,
+      "learning_rate": 1.968889180488178e-06,
+      "loss": 0.6956,
+      "step": 37190
+    },
+    {
+      "epoch": 0.799072045366671,
+      "grad_norm": 0.5179996887681789,
+      "learning_rate": 1.9648295706528385e-06,
+      "loss": 0.691,
+      "step": 37200
+    },
+    {
+      "epoch": 0.7992868496799416,
+      "grad_norm": 0.49638076183346225,
+      "learning_rate": 1.96077369434279e-06,
+      "loss": 0.6881,
+      "step": 37210
+    },
+    {
+      "epoch": 0.7995016539932122,
+      "grad_norm": 0.5142154100208954,
+      "learning_rate": 1.9567215534425777e-06,
+      "loss": 0.6821,
+      "step": 37220
+    },
+    {
+      "epoch": 0.7997164583064827,
+      "grad_norm": 0.5097975763047736,
+      "learning_rate": 1.952673149835025e-06,
+      "loss": 0.6934,
+      "step": 37230
+    },
+    {
+      "epoch": 0.7999312626197534,
+      "grad_norm": 0.5188923865330697,
+      "learning_rate": 1.948628485401215e-06,
+      "loss": 0.7017,
+      "step": 37240
+    },
+    {
+      "epoch": 0.800146066933024,
+      "grad_norm": 0.5393099873888024,
+      "learning_rate": 1.9445875620204846e-06,
+      "loss": 0.6985,
+      "step": 37250
+    },
+    {
+      "epoch": 0.8003608712462946,
+      "grad_norm": 0.5270181645580231,
+      "learning_rate": 1.940550381570453e-06,
+      "loss": 0.7074,
+      "step": 37260
+    },
+    {
+      "epoch": 0.8005756755595652,
+      "grad_norm": 0.49448978860517945,
+      "learning_rate": 1.9365169459269763e-06,
+      "loss": 0.704,
+      "step": 37270
+    },
+    {
+      "epoch": 0.8007904798728358,
+      "grad_norm": 0.5090418918653451,
+      "learning_rate": 1.932487256964191e-06,
+      "loss": 0.6871,
+      "step": 37280
+    },
+    {
+      "epoch": 0.8010052841861065,
+      "grad_norm": 0.5372331103704078,
+      "learning_rate": 1.9284613165544776e-06,
+      "loss": 0.7009,
+      "step": 37290
+    },
+    {
+      "epoch": 0.8012200884993771,
+      "grad_norm": 0.5123892333670363,
+      "learning_rate": 1.9244391265684836e-06,
+      "loss": 0.6939,
+      "step": 37300
+    },
+    {
+      "epoch": 0.8014348928126477,
+      "grad_norm": 0.4978925292624661,
+      "learning_rate": 1.9204206888751133e-06,
+      "loss": 0.6927,
+      "step": 37310
+    },
+    {
+      "epoch": 0.8016496971259183,
+      "grad_norm": 0.4954993561307306,
+      "learning_rate": 1.916406005341517e-06,
+      "loss": 0.6832,
+      "step": 37320
+    },
+    {
+      "epoch": 0.8018645014391889,
+      "grad_norm": 0.5248932509325274,
+      "learning_rate": 1.9123950778331204e-06,
+      "loss": 0.6889,
+      "step": 37330
+    },
+    {
+      "epoch": 0.8020793057524596,
+      "grad_norm": 0.503520388001021,
+      "learning_rate": 1.908387908213585e-06,
+      "loss": 0.6941,
+      "step": 37340
+    },
+    {
+      "epoch": 0.8022941100657301,
+      "grad_norm": 0.5217962284837745,
+      "learning_rate": 1.9043844983448356e-06,
+      "loss": 0.7014,
+      "step": 37350
+    },
+    {
+      "epoch": 0.8025089143790007,
+      "grad_norm": 0.5172482221861939,
+      "learning_rate": 1.9003848500870514e-06,
+      "loss": 0.6816,
+      "step": 37360
+    },
+    {
+      "epoch": 0.8027237186922713,
+      "grad_norm": 0.5171999415635867,
+      "learning_rate": 1.8963889652986533e-06,
+      "loss": 0.6918,
+      "step": 37370
+    },
+    {
+      "epoch": 0.8029385230055419,
+      "grad_norm": 0.531206719078998,
+      "learning_rate": 1.8923968458363307e-06,
+      "loss": 0.6894,
+      "step": 37380
+    },
+    {
+      "epoch": 0.8031533273188126,
+      "grad_norm": 0.509749112200991,
+      "learning_rate": 1.8884084935550063e-06,
+      "loss": 0.6937,
+      "step": 37390
+    },
+    {
+      "epoch": 0.8033681316320832,
+      "grad_norm": 0.5234029461890899,
+      "learning_rate": 1.884423910307861e-06,
+      "loss": 0.7173,
+      "step": 37400
+    },
+    {
+      "epoch": 0.8035829359453538,
+      "grad_norm": 0.520100769338295,
+      "learning_rate": 1.880443097946325e-06,
+      "loss": 0.6965,
+      "step": 37410
+    },
+    {
+      "epoch": 0.8037977402586244,
+      "grad_norm": 0.5198521028897318,
+      "learning_rate": 1.8764660583200733e-06,
+      "loss": 0.6857,
+      "step": 37420
+    },
+    {
+      "epoch": 0.804012544571895,
+      "grad_norm": 0.5170110623698808,
+      "learning_rate": 1.872492793277032e-06,
+      "loss": 0.6751,
+      "step": 37430
+    },
+    {
+      "epoch": 0.8042273488851656,
+      "grad_norm": 0.5058623102726381,
+      "learning_rate": 1.8685233046633655e-06,
+      "loss": 0.6925,
+      "step": 37440
+    },
+    {
+      "epoch": 0.8044421531984363,
+      "grad_norm": 0.5111088811413376,
+      "learning_rate": 1.8645575943234906e-06,
+      "loss": 0.6838,
+      "step": 37450
+    },
+    {
+      "epoch": 0.8046569575117068,
+      "grad_norm": 0.5010466802240249,
+      "learning_rate": 1.8605956641000678e-06,
+      "loss": 0.6894,
+      "step": 37460
+    },
+    {
+      "epoch": 0.8048717618249774,
+      "grad_norm": 0.5003570547858869,
+      "learning_rate": 1.8566375158339977e-06,
+      "loss": 0.6896,
+      "step": 37470
+    },
+    {
+      "epoch": 0.805086566138248,
+      "grad_norm": 0.5067814043345279,
+      "learning_rate": 1.852683151364426e-06,
+      "loss": 0.6804,
+      "step": 37480
+    },
+    {
+      "epoch": 0.8053013704515186,
+      "grad_norm": 0.5060047213750052,
+      "learning_rate": 1.8487325725287419e-06,
+      "loss": 0.7015,
+      "step": 37490
+    },
+    {
+      "epoch": 0.8055161747647893,
+      "grad_norm": 0.5139506510936809,
+      "learning_rate": 1.8447857811625747e-06,
+      "loss": 0.6823,
+      "step": 37500
+    },
+    {
+      "epoch": 0.8057309790780599,
+      "grad_norm": 0.4996768937608163,
+      "learning_rate": 1.8408427790997873e-06,
+      "loss": 0.6838,
+      "step": 37510
+    },
+    {
+      "epoch": 0.8059457833913305,
+      "grad_norm": 0.5125914344469875,
+      "learning_rate": 1.8369035681724912e-06,
+      "loss": 0.6789,
+      "step": 37520
+    },
+    {
+      "epoch": 0.8061605877046011,
+      "grad_norm": 0.5101593507880621,
+      "learning_rate": 1.8329681502110308e-06,
+      "loss": 0.7152,
+      "step": 37530
+    },
+    {
+      "epoch": 0.8063753920178717,
+      "grad_norm": 0.5089159371492329,
+      "learning_rate": 1.8290365270439926e-06,
+      "loss": 0.6919,
+      "step": 37540
+    },
+    {
+      "epoch": 0.8065901963311424,
+      "grad_norm": 0.5046175462761139,
+      "learning_rate": 1.8251087004981972e-06,
+      "loss": 0.6962,
+      "step": 37550
+    },
+    {
+      "epoch": 0.806805000644413,
+      "grad_norm": 0.535765255941607,
+      "learning_rate": 1.821184672398698e-06,
+      "loss": 0.6824,
+      "step": 37560
+    },
+    {
+      "epoch": 0.8070198049576836,
+      "grad_norm": 0.5057658285384766,
+      "learning_rate": 1.817264444568787e-06,
+      "loss": 0.7004,
+      "step": 37570
+    },
+    {
+      "epoch": 0.8072346092709541,
+      "grad_norm": 0.5305219423002547,
+      "learning_rate": 1.8133480188299913e-06,
+      "loss": 0.6881,
+      "step": 37580
+    },
+    {
+      "epoch": 0.8074494135842247,
+      "grad_norm": 0.5093435581220985,
+      "learning_rate": 1.8094353970020705e-06,
+      "loss": 0.7052,
+      "step": 37590
+    },
+    {
+      "epoch": 0.8076642178974954,
+      "grad_norm": 0.5057006548211692,
+      "learning_rate": 1.8055265809030142e-06,
+      "loss": 0.6937,
+      "step": 37600
+    },
+    {
+      "epoch": 0.807879022210766,
+      "grad_norm": 0.5062825370919449,
+      "learning_rate": 1.8016215723490504e-06,
+      "loss": 0.711,
+      "step": 37610
+    },
+    {
+      "epoch": 0.8080938265240366,
+      "grad_norm": 0.5082475799530384,
+      "learning_rate": 1.7977203731546266e-06,
+      "loss": 0.7022,
+      "step": 37620
+    },
+    {
+      "epoch": 0.8083086308373072,
+      "grad_norm": 0.5006515892573048,
+      "learning_rate": 1.7938229851324308e-06,
+      "loss": 0.6874,
+      "step": 37630
+    },
+    {
+      "epoch": 0.8085234351505778,
+      "grad_norm": 0.5148519330223464,
+      "learning_rate": 1.789929410093375e-06,
+      "loss": 0.6914,
+      "step": 37640
+    },
+    {
+      "epoch": 0.8087382394638485,
+      "grad_norm": 0.5026476503083747,
+      "learning_rate": 1.7860396498466004e-06,
+      "loss": 0.6916,
+      "step": 37650
+    },
+    {
+      "epoch": 0.8089530437771191,
+      "grad_norm": 0.5054813405354616,
+      "learning_rate": 1.782153706199481e-06,
+      "loss": 0.6939,
+      "step": 37660
+    },
+    {
+      "epoch": 0.8091678480903897,
+      "grad_norm": 0.5104786331574065,
+      "learning_rate": 1.7782715809576023e-06,
+      "loss": 0.708,
+      "step": 37670
+    },
+    {
+      "epoch": 0.8093826524036603,
+      "grad_norm": 0.5104889216759052,
+      "learning_rate": 1.7743932759247973e-06,
+      "loss": 0.6941,
+      "step": 37680
+    },
+    {
+      "epoch": 0.8095974567169308,
+      "grad_norm": 0.5301500050999222,
+      "learning_rate": 1.7705187929031042e-06,
+      "loss": 0.6935,
+      "step": 37690
+    },
+    {
+      "epoch": 0.8098122610302015,
+      "grad_norm": 0.5008139461716258,
+      "learning_rate": 1.7666481336927965e-06,
+      "loss": 0.6966,
+      "step": 37700
+    },
+    {
+      "epoch": 0.8100270653434721,
+      "grad_norm": 0.5292331018191944,
+      "learning_rate": 1.7627813000923677e-06,
+      "loss": 0.6869,
+      "step": 37710
+    },
+    {
+      "epoch": 0.8102418696567427,
+      "grad_norm": 0.5214981520880394,
+      "learning_rate": 1.7589182938985338e-06,
+      "loss": 0.6881,
+      "step": 37720
+    },
+    {
+      "epoch": 0.8104566739700133,
+      "grad_norm": 0.5162236641163164,
+      "learning_rate": 1.755059116906236e-06,
+      "loss": 0.6898,
+      "step": 37730
+    },
+    {
+      "epoch": 0.8106714782832839,
+      "grad_norm": 0.5157714549239334,
+      "learning_rate": 1.751203770908627e-06,
+      "loss": 0.6875,
+      "step": 37740
+    },
+    {
+      "epoch": 0.8108862825965545,
+      "grad_norm": 0.5107688828062216,
+      "learning_rate": 1.7473522576970881e-06,
+      "loss": 0.7055,
+      "step": 37750
+    },
+    {
+      "epoch": 0.8111010869098252,
+      "grad_norm": 0.5467930765197646,
+      "learning_rate": 1.743504579061216e-06,
+      "loss": 0.6868,
+      "step": 37760
+    },
+    {
+      "epoch": 0.8113158912230958,
+      "grad_norm": 0.5160466173487962,
+      "learning_rate": 1.7396607367888286e-06,
+      "loss": 0.6901,
+      "step": 37770
+    },
+    {
+      "epoch": 0.8115306955363664,
+      "grad_norm": 0.5405614071125214,
+      "learning_rate": 1.7358207326659604e-06,
+      "loss": 0.689,
+      "step": 37780
+    },
+    {
+      "epoch": 0.811745499849637,
+      "grad_norm": 0.5336778524063787,
+      "learning_rate": 1.7319845684768533e-06,
+      "loss": 0.6959,
+      "step": 37790
+    },
+    {
+      "epoch": 0.8119603041629075,
+      "grad_norm": 0.5104195973429427,
+      "learning_rate": 1.7281522460039845e-06,
+      "loss": 0.6881,
+      "step": 37800
+    },
+    {
+      "epoch": 0.8121751084761782,
+      "grad_norm": 0.5172156188555316,
+      "learning_rate": 1.7243237670280267e-06,
+      "loss": 0.6996,
+      "step": 37810
+    },
+    {
+      "epoch": 0.8123899127894488,
+      "grad_norm": 0.5140221172801053,
+      "learning_rate": 1.7204991333278776e-06,
+      "loss": 0.7085,
+      "step": 37820
+    },
+    {
+      "epoch": 0.8126047171027194,
+      "grad_norm": 0.5224661451049724,
+      "learning_rate": 1.7166783466806458e-06,
+      "loss": 0.7031,
+      "step": 37830
+    },
+    {
+      "epoch": 0.81281952141599,
+      "grad_norm": 0.5110409285199197,
+      "learning_rate": 1.712861408861647e-06,
+      "loss": 0.7039,
+      "step": 37840
+    },
+    {
+      "epoch": 0.8130343257292606,
+      "grad_norm": 0.5427771272170174,
+      "learning_rate": 1.709048321644422e-06,
+      "loss": 0.7081,
+      "step": 37850
+    },
+    {
+      "epoch": 0.8132491300425313,
+      "grad_norm": 0.5132044659392205,
+      "learning_rate": 1.705239086800704e-06,
+      "loss": 0.7014,
+      "step": 37860
+    },
+    {
+      "epoch": 0.8134639343558019,
+      "grad_norm": 0.5148999630994164,
+      "learning_rate": 1.701433706100457e-06,
+      "loss": 0.6973,
+      "step": 37870
+    },
+    {
+      "epoch": 0.8136787386690725,
+      "grad_norm": 0.5079680548504395,
+      "learning_rate": 1.6976321813118346e-06,
+      "loss": 0.7022,
+      "step": 37880
+    },
+    {
+      "epoch": 0.8138935429823431,
+      "grad_norm": 0.4960125156821642,
+      "learning_rate": 1.6938345142012102e-06,
+      "loss": 0.6785,
+      "step": 37890
+    },
+    {
+      "epoch": 0.8141083472956137,
+      "grad_norm": 0.5272204610524729,
+      "learning_rate": 1.6900407065331649e-06,
+      "loss": 0.6939,
+      "step": 37900
+    },
+    {
+      "epoch": 0.8143231516088844,
+      "grad_norm": 0.4848903289756121,
+      "learning_rate": 1.6862507600704748e-06,
+      "loss": 0.6998,
+      "step": 37910
+    },
+    {
+      "epoch": 0.8145379559221549,
+      "grad_norm": 0.5052375107473448,
+      "learning_rate": 1.6824646765741426e-06,
+      "loss": 0.6828,
+      "step": 37920
+    },
+    {
+      "epoch": 0.8147527602354255,
+      "grad_norm": 0.5116653542411183,
+      "learning_rate": 1.6786824578033556e-06,
+      "loss": 0.6925,
+      "step": 37930
+    },
+    {
+      "epoch": 0.8149675645486961,
+      "grad_norm": 0.5191668583354877,
+      "learning_rate": 1.6749041055155157e-06,
+      "loss": 0.7069,
+      "step": 37940
+    },
+    {
+      "epoch": 0.8151823688619667,
+      "grad_norm": 0.5216137206032925,
+      "learning_rate": 1.6711296214662308e-06,
+      "loss": 0.7014,
+      "step": 37950
+    },
+    {
+      "epoch": 0.8153971731752374,
+      "grad_norm": 0.5212090406729156,
+      "learning_rate": 1.6673590074092981e-06,
+      "loss": 0.6915,
+      "step": 37960
+    },
+    {
+      "epoch": 0.815611977488508,
+      "grad_norm": 0.5280444927816935,
+      "learning_rate": 1.6635922650967363e-06,
+      "loss": 0.7045,
+      "step": 37970
+    },
+    {
+      "epoch": 0.8158267818017786,
+      "grad_norm": 0.5302776730833281,
+      "learning_rate": 1.659829396278746e-06,
+      "loss": 0.698,
+      "step": 37980
+    },
+    {
+      "epoch": 0.8160415861150492,
+      "grad_norm": 0.5194339368072196,
+      "learning_rate": 1.65607040270374e-06,
+      "loss": 0.6844,
+      "step": 37990
+    },
+    {
+      "epoch": 0.8162563904283198,
+      "grad_norm": 0.5268789032071284,
+      "learning_rate": 1.6523152861183288e-06,
+      "loss": 0.6933,
+      "step": 38000
+    },
+    {
+      "epoch": 0.8164711947415904,
+      "grad_norm": 0.5034790229387067,
+      "learning_rate": 1.6485640482673126e-06,
+      "loss": 0.7029,
+      "step": 38010
+    },
+    {
+      "epoch": 0.8166859990548611,
+      "grad_norm": 0.5238642863385727,
+      "learning_rate": 1.6448166908937046e-06,
+      "loss": 0.6984,
+      "step": 38020
+    },
+    {
+      "epoch": 0.8169008033681316,
+      "grad_norm": 0.5384949344125787,
+      "learning_rate": 1.641073215738702e-06,
+      "loss": 0.6989,
+      "step": 38030
+    },
+    {
+      "epoch": 0.8171156076814022,
+      "grad_norm": 0.515676479439265,
+      "learning_rate": 1.6373336245417025e-06,
+      "loss": 0.6967,
+      "step": 38040
+    },
+    {
+      "epoch": 0.8173304119946728,
+      "grad_norm": 0.5176807526555749,
+      "learning_rate": 1.6335979190403006e-06,
+      "loss": 0.6783,
+      "step": 38050
+    },
+    {
+      "epoch": 0.8175452163079434,
+      "grad_norm": 0.5006394103833677,
+      "learning_rate": 1.6298661009702855e-06,
+      "loss": 0.6995,
+      "step": 38060
+    },
+    {
+      "epoch": 0.8177600206212141,
+      "grad_norm": 0.49950489053942465,
+      "learning_rate": 1.6261381720656378e-06,
+      "loss": 0.6855,
+      "step": 38070
+    },
+    {
+      "epoch": 0.8179748249344847,
+      "grad_norm": 0.5114816231641758,
+      "learning_rate": 1.6224141340585276e-06,
+      "loss": 0.7026,
+      "step": 38080
+    },
+    {
+      "epoch": 0.8181896292477553,
+      "grad_norm": 0.5242198840868686,
+      "learning_rate": 1.6186939886793307e-06,
+      "loss": 0.6848,
+      "step": 38090
+    },
+    {
+      "epoch": 0.8184044335610259,
+      "grad_norm": 0.5396691916322732,
+      "learning_rate": 1.6149777376565967e-06,
+      "loss": 0.689,
+      "step": 38100
+    },
+    {
+      "epoch": 0.8186192378742965,
+      "grad_norm": 0.5157018020298604,
+      "learning_rate": 1.6112653827170754e-06,
+      "loss": 0.6761,
+      "step": 38110
+    },
+    {
+      "epoch": 0.8188340421875672,
+      "grad_norm": 0.5249870157381407,
+      "learning_rate": 1.6075569255857104e-06,
+      "loss": 0.6992,
+      "step": 38120
+    },
+    {
+      "epoch": 0.8190488465008378,
+      "grad_norm": 0.5053189312234057,
+      "learning_rate": 1.6038523679856189e-06,
+      "loss": 0.6875,
+      "step": 38130
+    },
+    {
+      "epoch": 0.8192636508141083,
+      "grad_norm": 0.5293662990085333,
+      "learning_rate": 1.600151711638126e-06,
+      "loss": 0.6911,
+      "step": 38140
+    },
+    {
+      "epoch": 0.8194784551273789,
+      "grad_norm": 0.5048218038817608,
+      "learning_rate": 1.5964549582627276e-06,
+      "loss": 0.6813,
+      "step": 38150
+    },
+    {
+      "epoch": 0.8196932594406495,
+      "grad_norm": 0.5288004984283555,
+      "learning_rate": 1.592762109577114e-06,
+      "loss": 0.6959,
+      "step": 38160
+    },
+    {
+      "epoch": 0.8199080637539202,
+      "grad_norm": 0.5294144163040986,
+      "learning_rate": 1.5890731672971594e-06,
+      "loss": 0.7005,
+      "step": 38170
+    },
+    {
+      "epoch": 0.8201228680671908,
+      "grad_norm": 0.5143583114789181,
+      "learning_rate": 1.5853881331369247e-06,
+      "loss": 0.6998,
+      "step": 38180
+    },
+    {
+      "epoch": 0.8203376723804614,
+      "grad_norm": 0.5009264407402928,
+      "learning_rate": 1.581707008808655e-06,
+      "loss": 0.6915,
+      "step": 38190
+    },
+    {
+      "epoch": 0.820552476693732,
+      "grad_norm": 0.4988425919957875,
+      "learning_rate": 1.5780297960227708e-06,
+      "loss": 0.6996,
+      "step": 38200
+    },
+    {
+      "epoch": 0.8207672810070026,
+      "grad_norm": 0.5001446365022995,
+      "learning_rate": 1.5743564964878866e-06,
+      "loss": 0.7043,
+      "step": 38210
+    },
+    {
+      "epoch": 0.8209820853202733,
+      "grad_norm": 0.5238040005094907,
+      "learning_rate": 1.5706871119107914e-06,
+      "loss": 0.6975,
+      "step": 38220
+    },
+    {
+      "epoch": 0.8211968896335439,
+      "grad_norm": 0.49772899954971367,
+      "learning_rate": 1.567021643996458e-06,
+      "loss": 0.69,
+      "step": 38230
+    },
+    {
+      "epoch": 0.8214116939468145,
+      "grad_norm": 0.5226931397210955,
+      "learning_rate": 1.5633600944480377e-06,
+      "loss": 0.7145,
+      "step": 38240
+    },
+    {
+      "epoch": 0.821626498260085,
+      "grad_norm": 0.5111897161908127,
+      "learning_rate": 1.5597024649668645e-06,
+      "loss": 0.6868,
+      "step": 38250
+    },
+    {
+      "epoch": 0.8218413025733556,
+      "grad_norm": 0.5194930113430709,
+      "learning_rate": 1.5560487572524452e-06,
+      "loss": 0.694,
+      "step": 38260
+    },
+    {
+      "epoch": 0.8220561068866263,
+      "grad_norm": 0.5122991460889587,
+      "learning_rate": 1.552398973002467e-06,
+      "loss": 0.6853,
+      "step": 38270
+    },
+    {
+      "epoch": 0.8222709111998969,
+      "grad_norm": 0.5144915325254265,
+      "learning_rate": 1.5487531139127976e-06,
+      "loss": 0.6981,
+      "step": 38280
+    },
+    {
+      "epoch": 0.8224857155131675,
+      "grad_norm": 0.5099897016133259,
+      "learning_rate": 1.5451111816774756e-06,
+      "loss": 0.7032,
+      "step": 38290
+    },
+    {
+      "epoch": 0.8227005198264381,
+      "grad_norm": 0.5162555967078819,
+      "learning_rate": 1.5414731779887182e-06,
+      "loss": 0.6842,
+      "step": 38300
+    },
+    {
+      "epoch": 0.8229153241397087,
+      "grad_norm": 0.5338368411100364,
+      "learning_rate": 1.537839104536918e-06,
+      "loss": 0.688,
+      "step": 38310
+    },
+    {
+      "epoch": 0.8231301284529793,
+      "grad_norm": 0.5112979366014295,
+      "learning_rate": 1.534208963010636e-06,
+      "loss": 0.6911,
+      "step": 38320
+    },
+    {
+      "epoch": 0.82334493276625,
+      "grad_norm": 0.5087777288883957,
+      "learning_rate": 1.5305827550966113e-06,
+      "loss": 0.6872,
+      "step": 38330
+    },
+    {
+      "epoch": 0.8235597370795206,
+      "grad_norm": 0.49160735113437615,
+      "learning_rate": 1.5269604824797556e-06,
+      "loss": 0.6877,
+      "step": 38340
+    },
+    {
+      "epoch": 0.8237745413927912,
+      "grad_norm": 0.5281039504251113,
+      "learning_rate": 1.5233421468431475e-06,
+      "loss": 0.6916,
+      "step": 38350
+    },
+    {
+      "epoch": 0.8239893457060617,
+      "grad_norm": 0.49685966763451844,
+      "learning_rate": 1.519727749868042e-06,
+      "loss": 0.6868,
+      "step": 38360
+    },
+    {
+      "epoch": 0.8242041500193323,
+      "grad_norm": 0.5292477798818466,
+      "learning_rate": 1.516117293233862e-06,
+      "loss": 0.6813,
+      "step": 38370
+    },
+    {
+      "epoch": 0.824418954332603,
+      "grad_norm": 0.4922303762705952,
+      "learning_rate": 1.5125107786181948e-06,
+      "loss": 0.6939,
+      "step": 38380
+    },
+    {
+      "epoch": 0.8246337586458736,
+      "grad_norm": 0.5098238815457077,
+      "learning_rate": 1.508908207696801e-06,
+      "loss": 0.7038,
+      "step": 38390
+    },
+    {
+      "epoch": 0.8248485629591442,
+      "grad_norm": 0.5204133525606097,
+      "learning_rate": 1.5053095821436092e-06,
+      "loss": 0.6912,
+      "step": 38400
+    },
+    {
+      "epoch": 0.8250633672724148,
+      "grad_norm": 0.5110478312144826,
+      "learning_rate": 1.501714903630712e-06,
+      "loss": 0.6935,
+      "step": 38410
+    },
+    {
+      "epoch": 0.8252781715856854,
+      "grad_norm": 0.4977353499043285,
+      "learning_rate": 1.4981241738283724e-06,
+      "loss": 0.6926,
+      "step": 38420
+    },
+    {
+      "epoch": 0.8254929758989561,
+      "grad_norm": 0.49726992306043954,
+      "learning_rate": 1.494537394405008e-06,
+      "loss": 0.6898,
+      "step": 38430
+    },
+    {
+      "epoch": 0.8257077802122267,
+      "grad_norm": 0.5192905390752384,
+      "learning_rate": 1.4909545670272186e-06,
+      "loss": 0.6868,
+      "step": 38440
+    },
+    {
+      "epoch": 0.8259225845254973,
+      "grad_norm": 0.5050774765754228,
+      "learning_rate": 1.4873756933597505e-06,
+      "loss": 0.6783,
+      "step": 38450
+    },
+    {
+      "epoch": 0.8261373888387679,
+      "grad_norm": 0.5182788128937821,
+      "learning_rate": 1.4838007750655214e-06,
+      "loss": 0.6937,
+      "step": 38460
+    },
+    {
+      "epoch": 0.8263521931520385,
+      "grad_norm": 0.5075379442809032,
+      "learning_rate": 1.4802298138056104e-06,
+      "loss": 0.6824,
+      "step": 38470
+    },
+    {
+      "epoch": 0.8265669974653092,
+      "grad_norm": 0.5149603108095413,
+      "learning_rate": 1.4766628112392566e-06,
+      "loss": 0.7117,
+      "step": 38480
+    },
+    {
+      "epoch": 0.8267818017785797,
+      "grad_norm": 0.5170003208350263,
+      "learning_rate": 1.4730997690238635e-06,
+      "loss": 0.6859,
+      "step": 38490
+    },
+    {
+      "epoch": 0.8269966060918503,
+      "grad_norm": 0.5188232774350414,
+      "learning_rate": 1.4695406888149832e-06,
+      "loss": 0.6959,
+      "step": 38500
+    },
+    {
+      "epoch": 0.8272114104051209,
+      "grad_norm": 0.6124746424283791,
+      "learning_rate": 1.4659855722663453e-06,
+      "loss": 0.6881,
+      "step": 38510
+    },
+    {
+      "epoch": 0.8274262147183915,
+      "grad_norm": 0.5820473251466972,
+      "learning_rate": 1.4624344210298202e-06,
+      "loss": 0.7056,
+      "step": 38520
+    },
+    {
+      "epoch": 0.8276410190316622,
+      "grad_norm": 0.5487848245547172,
+      "learning_rate": 1.4588872367554452e-06,
+      "loss": 0.6813,
+      "step": 38530
+    },
+    {
+      "epoch": 0.8278558233449328,
+      "grad_norm": 0.5052583706726699,
+      "learning_rate": 1.455344021091414e-06,
+      "loss": 0.702,
+      "step": 38540
+    },
+    {
+      "epoch": 0.8280706276582034,
+      "grad_norm": 0.5449975100876051,
+      "learning_rate": 1.451804775684067e-06,
+      "loss": 0.6823,
+      "step": 38550
+    },
+    {
+      "epoch": 0.828285431971474,
+      "grad_norm": 0.5086547663242421,
+      "learning_rate": 1.448269502177918e-06,
+      "loss": 0.6842,
+      "step": 38560
+    },
+    {
+      "epoch": 0.8285002362847446,
+      "grad_norm": 0.5137832027200622,
+      "learning_rate": 1.4447382022156165e-06,
+      "loss": 0.687,
+      "step": 38570
+    },
+    {
+      "epoch": 0.8287150405980153,
+      "grad_norm": 0.49746601253508876,
+      "learning_rate": 1.4412108774379775e-06,
+      "loss": 0.6818,
+      "step": 38580
+    },
+    {
+      "epoch": 0.8289298449112859,
+      "grad_norm": 0.5103152319908182,
+      "learning_rate": 1.4376875294839643e-06,
+      "loss": 0.6914,
+      "step": 38590
+    },
+    {
+      "epoch": 0.8291446492245564,
+      "grad_norm": 0.5275562256088486,
+      "learning_rate": 1.4341681599906897e-06,
+      "loss": 0.6901,
+      "step": 38600
+    },
+    {
+      "epoch": 0.829359453537827,
+      "grad_norm": 0.5062231253926271,
+      "learning_rate": 1.4306527705934293e-06,
+      "loss": 0.6869,
+      "step": 38610
+    },
+    {
+      "epoch": 0.8295742578510976,
+      "grad_norm": 0.5338108634050998,
+      "learning_rate": 1.4271413629255937e-06,
+      "loss": 0.6922,
+      "step": 38620
+    },
+    {
+      "epoch": 0.8297890621643682,
+      "grad_norm": 0.5016771702455557,
+      "learning_rate": 1.4236339386187547e-06,
+      "loss": 0.6893,
+      "step": 38630
+    },
+    {
+      "epoch": 0.8300038664776389,
+      "grad_norm": 0.5191984822566196,
+      "learning_rate": 1.4201304993026321e-06,
+      "loss": 0.6925,
+      "step": 38640
+    },
+    {
+      "epoch": 0.8302186707909095,
+      "grad_norm": 0.5207586260457043,
+      "learning_rate": 1.4166310466050836e-06,
+      "loss": 0.6847,
+      "step": 38650
+    },
+    {
+      "epoch": 0.8304334751041801,
+      "grad_norm": 0.5139953998878591,
+      "learning_rate": 1.413135582152133e-06,
+      "loss": 0.685,
+      "step": 38660
+    },
+    {
+      "epoch": 0.8306482794174507,
+      "grad_norm": 0.5263182428133765,
+      "learning_rate": 1.4096441075679325e-06,
+      "loss": 0.6863,
+      "step": 38670
+    },
+    {
+      "epoch": 0.8308630837307213,
+      "grad_norm": 0.5282654822309718,
+      "learning_rate": 1.4061566244747937e-06,
+      "loss": 0.6957,
+      "step": 38680
+    },
+    {
+      "epoch": 0.831077888043992,
+      "grad_norm": 0.5209258215224781,
+      "learning_rate": 1.4026731344931655e-06,
+      "loss": 0.6956,
+      "step": 38690
+    },
+    {
+      "epoch": 0.8312926923572626,
+      "grad_norm": 0.5106724966574794,
+      "learning_rate": 1.399193639241645e-06,
+      "loss": 0.684,
+      "step": 38700
+    },
+    {
+      "epoch": 0.8315074966705331,
+      "grad_norm": 0.5033855375018464,
+      "learning_rate": 1.3957181403369747e-06,
+      "loss": 0.6867,
+      "step": 38710
+    },
+    {
+      "epoch": 0.8317223009838037,
+      "grad_norm": 0.5085681808640763,
+      "learning_rate": 1.3922466393940315e-06,
+      "loss": 0.6937,
+      "step": 38720
+    },
+    {
+      "epoch": 0.8319371052970743,
+      "grad_norm": 0.5034486647222803,
+      "learning_rate": 1.3887791380258509e-06,
+      "loss": 0.7001,
+      "step": 38730
+    },
+    {
+      "epoch": 0.832151909610345,
+      "grad_norm": 0.5080003477917522,
+      "learning_rate": 1.3853156378435916e-06,
+      "loss": 0.6964,
+      "step": 38740
+    },
+    {
+      "epoch": 0.8323667139236156,
+      "grad_norm": 0.5111977023138986,
+      "learning_rate": 1.3818561404565645e-06,
+      "loss": 0.6881,
+      "step": 38750
+    },
+    {
+      "epoch": 0.8325815182368862,
+      "grad_norm": 0.5061875410904043,
+      "learning_rate": 1.3784006474722212e-06,
+      "loss": 0.6964,
+      "step": 38760
+    },
+    {
+      "epoch": 0.8327963225501568,
+      "grad_norm": 0.5333922720806704,
+      "learning_rate": 1.374949160496143e-06,
+      "loss": 0.6946,
+      "step": 38770
+    },
+    {
+      "epoch": 0.8330111268634274,
+      "grad_norm": 0.5285837150812875,
+      "learning_rate": 1.3715016811320626e-06,
+      "loss": 0.6896,
+      "step": 38780
+    },
+    {
+      "epoch": 0.8332259311766981,
+      "grad_norm": 0.519609938847976,
+      "learning_rate": 1.3680582109818396e-06,
+      "loss": 0.6826,
+      "step": 38790
+    },
+    {
+      "epoch": 0.8334407354899687,
+      "grad_norm": 0.5263908151996978,
+      "learning_rate": 1.3646187516454778e-06,
+      "loss": 0.6948,
+      "step": 38800
+    },
+    {
+      "epoch": 0.8336555398032393,
+      "grad_norm": 0.50222280387675,
+      "learning_rate": 1.3611833047211132e-06,
+      "loss": 0.6938,
+      "step": 38810
+    },
+    {
+      "epoch": 0.8338703441165098,
+      "grad_norm": 0.5295682152493145,
+      "learning_rate": 1.3577518718050199e-06,
+      "loss": 0.6847,
+      "step": 38820
+    },
+    {
+      "epoch": 0.8340851484297804,
+      "grad_norm": 0.526283044923053,
+      "learning_rate": 1.3543244544916078e-06,
+      "loss": 0.6799,
+      "step": 38830
+    },
+    {
+      "epoch": 0.8342999527430511,
+      "grad_norm": 0.521205682856861,
+      "learning_rate": 1.3509010543734158e-06,
+      "loss": 0.7013,
+      "step": 38840
+    },
+    {
+      "epoch": 0.8345147570563217,
+      "grad_norm": 0.499623044668992,
+      "learning_rate": 1.347481673041121e-06,
+      "loss": 0.6916,
+      "step": 38850
+    },
+    {
+      "epoch": 0.8347295613695923,
+      "grad_norm": 0.49936320967671904,
+      "learning_rate": 1.3440663120835329e-06,
+      "loss": 0.6938,
+      "step": 38860
+    },
+    {
+      "epoch": 0.8349443656828629,
+      "grad_norm": 0.5293633851844111,
+      "learning_rate": 1.3406549730875916e-06,
+      "loss": 0.6836,
+      "step": 38870
+    },
+    {
+      "epoch": 0.8351591699961335,
+      "grad_norm": 0.5168353636514088,
+      "learning_rate": 1.3372476576383675e-06,
+      "loss": 0.6904,
+      "step": 38880
+    },
+    {
+      "epoch": 0.8353739743094042,
+      "grad_norm": 0.49422042008773903,
+      "learning_rate": 1.3338443673190637e-06,
+      "loss": 0.695,
+      "step": 38890
+    },
+    {
+      "epoch": 0.8355887786226748,
+      "grad_norm": 0.5056409802616189,
+      "learning_rate": 1.3304451037110132e-06,
+      "loss": 0.6968,
+      "step": 38900
+    },
+    {
+      "epoch": 0.8358035829359454,
+      "grad_norm": 0.517589863459402,
+      "learning_rate": 1.3270498683936738e-06,
+      "loss": 0.69,
+      "step": 38910
+    },
+    {
+      "epoch": 0.836018387249216,
+      "grad_norm": 0.5214346220274612,
+      "learning_rate": 1.3236586629446369e-06,
+      "loss": 0.6914,
+      "step": 38920
+    },
+    {
+      "epoch": 0.8362331915624865,
+      "grad_norm": 0.5087168684935114,
+      "learning_rate": 1.3202714889396172e-06,
+      "loss": 0.6795,
+      "step": 38930
+    },
+    {
+      "epoch": 0.8364479958757571,
+      "grad_norm": 0.5205133962298867,
+      "learning_rate": 1.3168883479524598e-06,
+      "loss": 0.6909,
+      "step": 38940
+    },
+    {
+      "epoch": 0.8366628001890278,
+      "grad_norm": 0.5318929561552359,
+      "learning_rate": 1.3135092415551354e-06,
+      "loss": 0.7047,
+      "step": 38950
+    },
+    {
+      "epoch": 0.8368776045022984,
+      "grad_norm": 0.5028926526152647,
+      "learning_rate": 1.3101341713177351e-06,
+      "loss": 0.6878,
+      "step": 38960
+    },
+    {
+      "epoch": 0.837092408815569,
+      "grad_norm": 0.5277421426222694,
+      "learning_rate": 1.306763138808479e-06,
+      "loss": 0.6932,
+      "step": 38970
+    },
+    {
+      "epoch": 0.8373072131288396,
+      "grad_norm": 0.4805738830452809,
+      "learning_rate": 1.3033961455937127e-06,
+      "loss": 0.6822,
+      "step": 38980
+    },
+    {
+      "epoch": 0.8375220174421102,
+      "grad_norm": 0.5108469134488737,
+      "learning_rate": 1.3000331932379007e-06,
+      "loss": 0.6893,
+      "step": 38990
+    },
+    {
+      "epoch": 0.8377368217553809,
+      "grad_norm": 0.5201560237520049,
+      "learning_rate": 1.2966742833036327e-06,
+      "loss": 0.6857,
+      "step": 39000
+    },
+    {
+      "epoch": 0.8379516260686515,
+      "grad_norm": 0.5188163877062854,
+      "learning_rate": 1.2933194173516218e-06,
+      "loss": 0.7,
+      "step": 39010
+    },
+    {
+      "epoch": 0.8381664303819221,
+      "grad_norm": 0.5093643945181999,
+      "learning_rate": 1.289968596940694e-06,
+      "loss": 0.7081,
+      "step": 39020
+    },
+    {
+      "epoch": 0.8383812346951927,
+      "grad_norm": 0.5203767610471154,
+      "learning_rate": 1.2866218236278038e-06,
+      "loss": 0.6819,
+      "step": 39030
+    },
+    {
+      "epoch": 0.8385960390084632,
+      "grad_norm": 0.5184824000269601,
+      "learning_rate": 1.2832790989680221e-06,
+      "loss": 0.6855,
+      "step": 39040
+    },
+    {
+      "epoch": 0.838810843321734,
+      "grad_norm": 0.5019301811954551,
+      "learning_rate": 1.2799404245145396e-06,
+      "loss": 0.6947,
+      "step": 39050
+    },
+    {
+      "epoch": 0.8390256476350045,
+      "grad_norm": 0.5346206336302403,
+      "learning_rate": 1.2766058018186645e-06,
+      "loss": 0.703,
+      "step": 39060
+    },
+    {
+      "epoch": 0.8392404519482751,
+      "grad_norm": 0.5227834077619942,
+      "learning_rate": 1.2732752324298225e-06,
+      "loss": 0.6945,
+      "step": 39070
+    },
+    {
+      "epoch": 0.8394552562615457,
+      "grad_norm": 0.5216423504327496,
+      "learning_rate": 1.2699487178955573e-06,
+      "loss": 0.6927,
+      "step": 39080
+    },
+    {
+      "epoch": 0.8396700605748163,
+      "grad_norm": 0.5446631129970325,
+      "learning_rate": 1.266626259761523e-06,
+      "loss": 0.6824,
+      "step": 39090
+    },
+    {
+      "epoch": 0.839884864888087,
+      "grad_norm": 0.49940235609166955,
+      "learning_rate": 1.2633078595714943e-06,
+      "loss": 0.7022,
+      "step": 39100
+    },
+    {
+      "epoch": 0.8400996692013576,
+      "grad_norm": 0.5079994069706872,
+      "learning_rate": 1.2599935188673596e-06,
+      "loss": 0.6923,
+      "step": 39110
+    },
+    {
+      "epoch": 0.8403144735146282,
+      "grad_norm": 0.5057473773527276,
+      "learning_rate": 1.2566832391891215e-06,
+      "loss": 0.7097,
+      "step": 39120
+    },
+    {
+      "epoch": 0.8405292778278988,
+      "grad_norm": 0.5100763490334336,
+      "learning_rate": 1.2533770220748942e-06,
+      "loss": 0.679,
+      "step": 39130
+    },
+    {
+      "epoch": 0.8407440821411694,
+      "grad_norm": 0.5003721745532178,
+      "learning_rate": 1.2500748690609033e-06,
+      "loss": 0.699,
+      "step": 39140
+    },
+    {
+      "epoch": 0.8409588864544401,
+      "grad_norm": 0.5137872175833427,
+      "learning_rate": 1.246776781681487e-06,
+      "loss": 0.6928,
+      "step": 39150
+    },
+    {
+      "epoch": 0.8411736907677106,
+      "grad_norm": 0.497587322672259,
+      "learning_rate": 1.2434827614690958e-06,
+      "loss": 0.6988,
+      "step": 39160
+    },
+    {
+      "epoch": 0.8413884950809812,
+      "grad_norm": 0.5202834990746827,
+      "learning_rate": 1.2401928099542892e-06,
+      "loss": 0.6861,
+      "step": 39170
+    },
+    {
+      "epoch": 0.8416032993942518,
+      "grad_norm": 0.5189296596390629,
+      "learning_rate": 1.2369069286657388e-06,
+      "loss": 0.6805,
+      "step": 39180
+    },
+    {
+      "epoch": 0.8418181037075224,
+      "grad_norm": 0.5141868855723348,
+      "learning_rate": 1.2336251191302162e-06,
+      "loss": 0.6897,
+      "step": 39190
+    },
+    {
+      "epoch": 0.842032908020793,
+      "grad_norm": 0.5124494606085882,
+      "learning_rate": 1.2303473828726153e-06,
+      "loss": 0.6957,
+      "step": 39200
+    },
+    {
+      "epoch": 0.8422477123340637,
+      "grad_norm": 0.5214072352573822,
+      "learning_rate": 1.2270737214159245e-06,
+      "loss": 0.6796,
+      "step": 39210
+    },
+    {
+      "epoch": 0.8424625166473343,
+      "grad_norm": 0.528792736816499,
+      "learning_rate": 1.2238041362812435e-06,
+      "loss": 0.6899,
+      "step": 39220
+    },
+    {
+      "epoch": 0.8426773209606049,
+      "grad_norm": 0.5093238296040209,
+      "learning_rate": 1.2205386289877829e-06,
+      "loss": 0.6911,
+      "step": 39230
+    },
+    {
+      "epoch": 0.8428921252738755,
+      "grad_norm": 0.4978808338740638,
+      "learning_rate": 1.2172772010528455e-06,
+      "loss": 0.6933,
+      "step": 39240
+    },
+    {
+      "epoch": 0.8431069295871461,
+      "grad_norm": 0.5325179259250296,
+      "learning_rate": 1.2140198539918568e-06,
+      "loss": 0.6933,
+      "step": 39250
+    },
+    {
+      "epoch": 0.8433217339004168,
+      "grad_norm": 0.5239028106635986,
+      "learning_rate": 1.2107665893183274e-06,
+      "loss": 0.6845,
+      "step": 39260
+    },
+    {
+      "epoch": 0.8435365382136873,
+      "grad_norm": 0.5206218974407494,
+      "learning_rate": 1.2075174085438878e-06,
+      "loss": 0.704,
+      "step": 39270
+    },
+    {
+      "epoch": 0.8437513425269579,
+      "grad_norm": 0.5234721000920526,
+      "learning_rate": 1.2042723131782586e-06,
+      "loss": 0.6886,
+      "step": 39280
+    },
+    {
+      "epoch": 0.8439661468402285,
+      "grad_norm": 0.5166325698154588,
+      "learning_rate": 1.2010313047292677e-06,
+      "loss": 0.6921,
+      "step": 39290
+    },
+    {
+      "epoch": 0.8441809511534991,
+      "grad_norm": 0.5013497589587982,
+      "learning_rate": 1.197794384702844e-06,
+      "loss": 0.6839,
+      "step": 39300
+    },
+    {
+      "epoch": 0.8443957554667698,
+      "grad_norm": 0.5392745947364839,
+      "learning_rate": 1.1945615546030109e-06,
+      "loss": 0.6848,
+      "step": 39310
+    },
+    {
+      "epoch": 0.8446105597800404,
+      "grad_norm": 0.5083861122426512,
+      "learning_rate": 1.1913328159319049e-06,
+      "loss": 0.6914,
+      "step": 39320
+    },
+    {
+      "epoch": 0.844825364093311,
+      "grad_norm": 0.49717604044905356,
+      "learning_rate": 1.1881081701897446e-06,
+      "loss": 0.6882,
+      "step": 39330
+    },
+    {
+      "epoch": 0.8450401684065816,
+      "grad_norm": 0.5273892484603656,
+      "learning_rate": 1.1848876188748583e-06,
+      "loss": 0.7001,
+      "step": 39340
+    },
+    {
+      "epoch": 0.8452549727198522,
+      "grad_norm": 0.5274494392930585,
+      "learning_rate": 1.1816711634836708e-06,
+      "loss": 0.6842,
+      "step": 39350
+    },
+    {
+      "epoch": 0.8454697770331229,
+      "grad_norm": 0.5550754818367603,
+      "learning_rate": 1.1784588055106937e-06,
+      "loss": 0.6957,
+      "step": 39360
+    },
+    {
+      "epoch": 0.8456845813463935,
+      "grad_norm": 0.5093805942815454,
+      "learning_rate": 1.175250546448553e-06,
+      "loss": 0.69,
+      "step": 39370
+    },
+    {
+      "epoch": 0.845899385659664,
+      "grad_norm": 0.5291605321484583,
+      "learning_rate": 1.1720463877879518e-06,
+      "loss": 0.6954,
+      "step": 39380
+    },
+    {
+      "epoch": 0.8461141899729346,
+      "grad_norm": 0.5213523166182829,
+      "learning_rate": 1.1688463310176979e-06,
+      "loss": 0.7063,
+      "step": 39390
+    },
+    {
+      "epoch": 0.8463289942862052,
+      "grad_norm": 0.5043482415928106,
+      "learning_rate": 1.1656503776246942e-06,
+      "loss": 0.6896,
+      "step": 39400
+    },
+    {
+      "epoch": 0.8465437985994759,
+      "grad_norm": 0.5340179003407842,
+      "learning_rate": 1.1624585290939261e-06,
+      "loss": 0.6895,
+      "step": 39410
+    },
+    {
+      "epoch": 0.8467586029127465,
+      "grad_norm": 0.5334819112678492,
+      "learning_rate": 1.1592707869084896e-06,
+      "loss": 0.6925,
+      "step": 39420
+    },
+    {
+      "epoch": 0.8469734072260171,
+      "grad_norm": 0.5196334748388658,
+      "learning_rate": 1.156087152549552e-06,
+      "loss": 0.6972,
+      "step": 39430
+    },
+    {
+      "epoch": 0.8471882115392877,
+      "grad_norm": 0.5212173781531991,
+      "learning_rate": 1.1529076274963925e-06,
+      "loss": 0.6837,
+      "step": 39440
+    },
+    {
+      "epoch": 0.8474030158525583,
+      "grad_norm": 0.5271644084715066,
+      "learning_rate": 1.1497322132263643e-06,
+      "loss": 0.6781,
+      "step": 39450
+    },
+    {
+      "epoch": 0.847617820165829,
+      "grad_norm": 0.5251522649812483,
+      "learning_rate": 1.1465609112149178e-06,
+      "loss": 0.6966,
+      "step": 39460
+    },
+    {
+      "epoch": 0.8478326244790996,
+      "grad_norm": 0.5028358197340315,
+      "learning_rate": 1.1433937229355951e-06,
+      "loss": 0.682,
+      "step": 39470
+    },
+    {
+      "epoch": 0.8480474287923702,
+      "grad_norm": 0.5260187617540428,
+      "learning_rate": 1.140230649860018e-06,
+      "loss": 0.7039,
+      "step": 39480
+    },
+    {
+      "epoch": 0.8482622331056408,
+      "grad_norm": 0.49354427489124325,
+      "learning_rate": 1.1370716934579086e-06,
+      "loss": 0.6806,
+      "step": 39490
+    },
+    {
+      "epoch": 0.8484770374189113,
+      "grad_norm": 0.5353136217093745,
+      "learning_rate": 1.1339168551970647e-06,
+      "loss": 0.7164,
+      "step": 39500
+    },
+    {
+      "epoch": 0.8486918417321819,
+      "grad_norm": 0.5172265496588412,
+      "learning_rate": 1.1307661365433763e-06,
+      "loss": 0.6857,
+      "step": 39510
+    },
+    {
+      "epoch": 0.8489066460454526,
+      "grad_norm": 0.5240872817688168,
+      "learning_rate": 1.1276195389608201e-06,
+      "loss": 0.6842,
+      "step": 39520
+    },
+    {
+      "epoch": 0.8491214503587232,
+      "grad_norm": 0.5120247735049234,
+      "learning_rate": 1.1244770639114522e-06,
+      "loss": 0.7057,
+      "step": 39530
+    },
+    {
+      "epoch": 0.8493362546719938,
+      "grad_norm": 0.49496102620753335,
+      "learning_rate": 1.1213387128554232e-06,
+      "loss": 0.6822,
+      "step": 39540
+    },
+    {
+      "epoch": 0.8495510589852644,
+      "grad_norm": 0.5000858443900138,
+      "learning_rate": 1.1182044872509556e-06,
+      "loss": 0.6881,
+      "step": 39550
+    },
+    {
+      "epoch": 0.849765863298535,
+      "grad_norm": 0.5040809516882646,
+      "learning_rate": 1.1150743885543636e-06,
+      "loss": 0.6947,
+      "step": 39560
+    },
+    {
+      "epoch": 0.8499806676118057,
+      "grad_norm": 0.5420538811255085,
+      "learning_rate": 1.11194841822004e-06,
+      "loss": 0.6772,
+      "step": 39570
+    },
+    {
+      "epoch": 0.8501954719250763,
+      "grad_norm": 0.504747244903184,
+      "learning_rate": 1.1088265777004613e-06,
+      "loss": 0.6809,
+      "step": 39580
+    },
+    {
+      "epoch": 0.8504102762383469,
+      "grad_norm": 0.49931071427553786,
+      "learning_rate": 1.105708868446187e-06,
+      "loss": 0.6829,
+      "step": 39590
+    },
+    {
+      "epoch": 0.8506250805516175,
+      "grad_norm": 0.5090934536047,
+      "learning_rate": 1.102595291905848e-06,
+      "loss": 0.694,
+      "step": 39600
+    },
+    {
+      "epoch": 0.850839884864888,
+      "grad_norm": 0.4994127034268232,
+      "learning_rate": 1.099485849526165e-06,
+      "loss": 0.6967,
+      "step": 39610
+    },
+    {
+      "epoch": 0.8510546891781587,
+      "grad_norm": 0.5202678470784433,
+      "learning_rate": 1.0963805427519335e-06,
+      "loss": 0.6893,
+      "step": 39620
+    },
+    {
+      "epoch": 0.8512694934914293,
+      "grad_norm": 0.5007238242310413,
+      "learning_rate": 1.0932793730260272e-06,
+      "loss": 0.6945,
+      "step": 39630
+    },
+    {
+      "epoch": 0.8514842978046999,
+      "grad_norm": 0.5208651575984488,
+      "learning_rate": 1.0901823417893997e-06,
+      "loss": 0.6903,
+      "step": 39640
+    },
+    {
+      "epoch": 0.8516991021179705,
+      "grad_norm": 0.511764861180484,
+      "learning_rate": 1.087089450481078e-06,
+      "loss": 0.6934,
+      "step": 39650
+    },
+    {
+      "epoch": 0.8519139064312411,
+      "grad_norm": 0.5194391424779614,
+      "learning_rate": 1.0840007005381714e-06,
+      "loss": 0.6772,
+      "step": 39660
+    },
+    {
+      "epoch": 0.8521287107445118,
+      "grad_norm": 0.5000894132679686,
+      "learning_rate": 1.0809160933958574e-06,
+      "loss": 0.6918,
+      "step": 39670
+    },
+    {
+      "epoch": 0.8523435150577824,
+      "grad_norm": 0.5349623963637503,
+      "learning_rate": 1.0778356304873927e-06,
+      "loss": 0.6931,
+      "step": 39680
+    },
+    {
+      "epoch": 0.852558319371053,
+      "grad_norm": 0.5127707494353972,
+      "learning_rate": 1.074759313244108e-06,
+      "loss": 0.6858,
+      "step": 39690
+    },
+    {
+      "epoch": 0.8527731236843236,
+      "grad_norm": 0.5050351323720006,
+      "learning_rate": 1.07168714309541e-06,
+      "loss": 0.6983,
+      "step": 39700
+    },
+    {
+      "epoch": 0.8529879279975942,
+      "grad_norm": 0.5042118701302163,
+      "learning_rate": 1.0686191214687758e-06,
+      "loss": 0.6907,
+      "step": 39710
+    },
+    {
+      "epoch": 0.8532027323108649,
+      "grad_norm": 0.5319983433434204,
+      "learning_rate": 1.0655552497897525e-06,
+      "loss": 0.6797,
+      "step": 39720
+    },
+    {
+      "epoch": 0.8534175366241354,
+      "grad_norm": 0.5126038269348526,
+      "learning_rate": 1.0624955294819628e-06,
+      "loss": 0.6798,
+      "step": 39730
+    },
+    {
+      "epoch": 0.853632340937406,
+      "grad_norm": 0.5005716309497225,
+      "learning_rate": 1.0594399619671013e-06,
+      "loss": 0.6933,
+      "step": 39740
+    },
+    {
+      "epoch": 0.8538471452506766,
+      "grad_norm": 0.5167440374916815,
+      "learning_rate": 1.0563885486649284e-06,
+      "loss": 0.6905,
+      "step": 39750
+    },
+    {
+      "epoch": 0.8540619495639472,
+      "grad_norm": 0.504354990255589,
+      "learning_rate": 1.0533412909932793e-06,
+      "loss": 0.6799,
+      "step": 39760
+    },
+    {
+      "epoch": 0.8542767538772179,
+      "grad_norm": 0.5143195869144642,
+      "learning_rate": 1.0502981903680565e-06,
+      "loss": 0.6833,
+      "step": 39770
+    },
+    {
+      "epoch": 0.8544915581904885,
+      "grad_norm": 0.49947041857768637,
+      "learning_rate": 1.0472592482032272e-06,
+      "loss": 0.6746,
+      "step": 39780
+    },
+    {
+      "epoch": 0.8547063625037591,
+      "grad_norm": 0.5217962095232429,
+      "learning_rate": 1.0442244659108325e-06,
+      "loss": 0.7035,
+      "step": 39790
+    },
+    {
+      "epoch": 0.8549211668170297,
+      "grad_norm": 0.5096704985368335,
+      "learning_rate": 1.0411938449009752e-06,
+      "loss": 0.6825,
+      "step": 39800
+    },
+    {
+      "epoch": 0.8551359711303003,
+      "grad_norm": 0.5135857160178005,
+      "learning_rate": 1.0381673865818288e-06,
+      "loss": 0.6831,
+      "step": 39810
+    },
+    {
+      "epoch": 0.8553507754435709,
+      "grad_norm": 0.522664943240014,
+      "learning_rate": 1.0351450923596317e-06,
+      "loss": 0.6931,
+      "step": 39820
+    },
+    {
+      "epoch": 0.8555655797568416,
+      "grad_norm": 0.49812520742428695,
+      "learning_rate": 1.0321269636386833e-06,
+      "loss": 0.6881,
+      "step": 39830
+    },
+    {
+      "epoch": 0.8557803840701121,
+      "grad_norm": 0.5120365797945067,
+      "learning_rate": 1.0291130018213558e-06,
+      "loss": 0.6995,
+      "step": 39840
+    },
+    {
+      "epoch": 0.8559951883833827,
+      "grad_norm": 0.5392653302148872,
+      "learning_rate": 1.0261032083080746e-06,
+      "loss": 0.6926,
+      "step": 39850
+    },
+    {
+      "epoch": 0.8562099926966533,
+      "grad_norm": 1.451337023254773,
+      "learning_rate": 1.023097584497339e-06,
+      "loss": 0.6877,
+      "step": 39860
+    },
+    {
+      "epoch": 0.8564247970099239,
+      "grad_norm": 0.5476743766148364,
+      "learning_rate": 1.0200961317857027e-06,
+      "loss": 0.6825,
+      "step": 39870
+    },
+    {
+      "epoch": 0.8566396013231946,
+      "grad_norm": 0.5313081010660222,
+      "learning_rate": 1.017098851567785e-06,
+      "loss": 0.7,
+      "step": 39880
+    },
+    {
+      "epoch": 0.8568544056364652,
+      "grad_norm": 0.5362672147803951,
+      "learning_rate": 1.014105745236269e-06,
+      "loss": 0.6909,
+      "step": 39890
+    },
+    {
+      "epoch": 0.8570692099497358,
+      "grad_norm": 0.5217352952005082,
+      "learning_rate": 1.0111168141818916e-06,
+      "loss": 0.6921,
+      "step": 39900
+    },
+    {
+      "epoch": 0.8572840142630064,
+      "grad_norm": 0.5111751841927796,
+      "learning_rate": 1.0081320597934542e-06,
+      "loss": 0.6819,
+      "step": 39910
+    },
+    {
+      "epoch": 0.857498818576277,
+      "grad_norm": 0.4993174113029642,
+      "learning_rate": 1.0051514834578169e-06,
+      "loss": 0.6958,
+      "step": 39920
+    },
+    {
+      "epoch": 0.8577136228895477,
+      "grad_norm": 0.5053368624600403,
+      "learning_rate": 1.002175086559899e-06,
+      "loss": 0.6881,
+      "step": 39930
+    },
+    {
+      "epoch": 0.8579284272028183,
+      "grad_norm": 0.5130032786788024,
+      "learning_rate": 9.992028704826785e-07,
+      "loss": 0.7007,
+      "step": 39940
+    },
+    {
+      "epoch": 0.8581432315160888,
+      "grad_norm": 0.5157365810431223,
+      "learning_rate": 9.96234836607184e-07,
+      "loss": 0.6744,
+      "step": 39950
+    },
+    {
+      "epoch": 0.8583580358293594,
+      "grad_norm": 0.527685117613382,
+      "learning_rate": 9.932709863125133e-07,
+      "loss": 0.6949,
+      "step": 39960
+    },
+    {
+      "epoch": 0.85857284014263,
+      "grad_norm": 0.512839755051422,
+      "learning_rate": 9.903113209758098e-07,
+      "loss": 0.6948,
+      "step": 39970
+    },
+    {
+      "epoch": 0.8587876444559007,
+      "grad_norm": 0.5211080071086338,
+      "learning_rate": 9.87355841972274e-07,
+      "loss": 0.71,
+      "step": 39980
+    },
+    {
+      "epoch": 0.8590024487691713,
+      "grad_norm": 0.5098392488580249,
+      "learning_rate": 9.844045506751687e-07,
+      "loss": 0.6889,
+      "step": 39990
+    },
+    {
+      "epoch": 0.8592172530824419,
+      "grad_norm": 0.5218396209074099,
+      "learning_rate": 9.81457448455797e-07,
+      "loss": 0.6872,
+      "step": 40000
+    },
+    {
+      "epoch": 0.8594320573957125,
+      "grad_norm": 0.5008077047222401,
+      "learning_rate": 9.785145366835325e-07,
+      "loss": 0.6867,
+      "step": 40010
+    },
+    {
+      "epoch": 0.8596468617089831,
+      "grad_norm": 0.5055415929729381,
+      "learning_rate": 9.755758167257857e-07,
+      "loss": 0.7009,
+      "step": 40020
+    },
+    {
+      "epoch": 0.8598616660222538,
+      "grad_norm": 0.5309493067199006,
+      "learning_rate": 9.726412899480342e-07,
+      "loss": 0.689,
+      "step": 40030
+    },
+    {
+      "epoch": 0.8600764703355244,
+      "grad_norm": 0.5187592677496787,
+      "learning_rate": 9.697109577137943e-07,
+      "loss": 0.6768,
+      "step": 40040
+    },
+    {
+      "epoch": 0.860291274648795,
+      "grad_norm": 0.5192073916886506,
+      "learning_rate": 9.667848213846386e-07,
+      "loss": 0.6891,
+      "step": 40050
+    },
+    {
+      "epoch": 0.8605060789620655,
+      "grad_norm": 0.5259783631972406,
+      "learning_rate": 9.638628823201956e-07,
+      "loss": 0.6895,
+      "step": 40060
+    },
+    {
+      "epoch": 0.8607208832753361,
+      "grad_norm": 0.5226106882651611,
+      "learning_rate": 9.609451418781302e-07,
+      "loss": 0.69,
+      "step": 40070
+    },
+    {
+      "epoch": 0.8609356875886067,
+      "grad_norm": 0.5071335472658586,
+      "learning_rate": 9.580316014141722e-07,
+      "loss": 0.6846,
+      "step": 40080
+    },
+    {
+      "epoch": 0.8611504919018774,
+      "grad_norm": 0.5119813837953315,
+      "learning_rate": 9.551222622820866e-07,
+      "loss": 0.695,
+      "step": 40090
+    },
+    {
+      "epoch": 0.861365296215148,
+      "grad_norm": 0.4918570321661607,
+      "learning_rate": 9.522171258336943e-07,
+      "loss": 0.6898,
+      "step": 40100
+    },
+    {
+      "epoch": 0.8615801005284186,
+      "grad_norm": 0.5038278185618428,
+      "learning_rate": 9.493161934188632e-07,
+      "loss": 0.6902,
+      "step": 40110
+    },
+    {
+      "epoch": 0.8617949048416892,
+      "grad_norm": 0.5131068608059737,
+      "learning_rate": 9.464194663854986e-07,
+      "loss": 0.687,
+      "step": 40120
+    },
+    {
+      "epoch": 0.8620097091549598,
+      "grad_norm": 0.5313015107754585,
+      "learning_rate": 9.435269460795671e-07,
+      "loss": 0.6973,
+      "step": 40130
+    },
+    {
+      "epoch": 0.8622245134682305,
+      "grad_norm": 0.509195051088408,
+      "learning_rate": 9.406386338450657e-07,
+      "loss": 0.6946,
+      "step": 40140
+    },
+    {
+      "epoch": 0.8624393177815011,
+      "grad_norm": 0.502073057996939,
+      "learning_rate": 9.377545310240454e-07,
+      "loss": 0.6909,
+      "step": 40150
+    },
+    {
+      "epoch": 0.8626541220947717,
+      "grad_norm": 0.5081753672118122,
+      "learning_rate": 9.34874638956601e-07,
+      "loss": 0.6946,
+      "step": 40160
+    },
+    {
+      "epoch": 0.8628689264080422,
+      "grad_norm": 0.5075208175361569,
+      "learning_rate": 9.31998958980862e-07,
+      "loss": 0.6846,
+      "step": 40170
+    },
+    {
+      "epoch": 0.8630837307213128,
+      "grad_norm": 0.5559833588569257,
+      "learning_rate": 9.291274924330141e-07,
+      "loss": 0.6993,
+      "step": 40180
+    },
+    {
+      "epoch": 0.8632985350345835,
+      "grad_norm": 0.5406798025815065,
+      "learning_rate": 9.262602406472732e-07,
+      "loss": 0.6948,
+      "step": 40190
+    },
+    {
+      "epoch": 0.8635133393478541,
+      "grad_norm": 0.5074440350796556,
+      "learning_rate": 9.233972049559037e-07,
+      "loss": 0.6845,
+      "step": 40200
+    },
+    {
+      "epoch": 0.8637281436611247,
+      "grad_norm": 0.5231140038044005,
+      "learning_rate": 9.205383866892092e-07,
+      "loss": 0.6887,
+      "step": 40210
+    },
+    {
+      "epoch": 0.8639429479743953,
+      "grad_norm": 0.5120556983369559,
+      "learning_rate": 9.176837871755351e-07,
+      "loss": 0.6885,
+      "step": 40220
+    },
+    {
+      "epoch": 0.8641577522876659,
+      "grad_norm": 0.5083353819396028,
+      "learning_rate": 9.148334077412646e-07,
+      "loss": 0.6913,
+      "step": 40230
+    },
+    {
+      "epoch": 0.8643725566009366,
+      "grad_norm": 0.5038742161832033,
+      "learning_rate": 9.119872497108162e-07,
+      "loss": 0.691,
+      "step": 40240
+    },
+    {
+      "epoch": 0.8645873609142072,
+      "grad_norm": 0.5099893292803395,
+      "learning_rate": 9.091453144066587e-07,
+      "loss": 0.6849,
+      "step": 40250
+    },
+    {
+      "epoch": 0.8648021652274778,
+      "grad_norm": 0.5172315105235323,
+      "learning_rate": 9.063076031492857e-07,
+      "loss": 0.6825,
+      "step": 40260
+    },
+    {
+      "epoch": 0.8650169695407484,
+      "grad_norm": 0.5187093289593909,
+      "learning_rate": 9.034741172572359e-07,
+      "loss": 0.6824,
+      "step": 40270
+    },
+    {
+      "epoch": 0.865231773854019,
+      "grad_norm": 0.522303012114229,
+      "learning_rate": 9.006448580470839e-07,
+      "loss": 0.6887,
+      "step": 40280
+    },
+    {
+      "epoch": 0.8654465781672896,
+      "grad_norm": 0.49894862369670223,
+      "learning_rate": 8.978198268334348e-07,
+      "loss": 0.6978,
+      "step": 40290
+    },
+    {
+      "epoch": 0.8656613824805602,
+      "grad_norm": 0.5002768042699983,
+      "learning_rate": 8.949990249289409e-07,
+      "loss": 0.7016,
+      "step": 40300
+    },
+    {
+      "epoch": 0.8658761867938308,
+      "grad_norm": 0.5333278616857066,
+      "learning_rate": 8.921824536442747e-07,
+      "loss": 0.7005,
+      "step": 40310
+    },
+    {
+      "epoch": 0.8660909911071014,
+      "grad_norm": 0.4947738923548671,
+      "learning_rate": 8.893701142881539e-07,
+      "loss": 0.6847,
+      "step": 40320
+    },
+    {
+      "epoch": 0.866305795420372,
+      "grad_norm": 0.5180236450171597,
+      "learning_rate": 8.865620081673243e-07,
+      "loss": 0.6955,
+      "step": 40330
+    },
+    {
+      "epoch": 0.8665205997336427,
+      "grad_norm": 0.5318690818922256,
+      "learning_rate": 8.837581365865688e-07,
+      "loss": 0.6963,
+      "step": 40340
+    },
+    {
+      "epoch": 0.8667354040469133,
+      "grad_norm": 0.5330479773777189,
+      "learning_rate": 8.809585008487009e-07,
+      "loss": 0.6853,
+      "step": 40350
+    },
+    {
+      "epoch": 0.8669502083601839,
+      "grad_norm": 0.516868926152685,
+      "learning_rate": 8.781631022545611e-07,
+      "loss": 0.6887,
+      "step": 40360
+    },
+    {
+      "epoch": 0.8671650126734545,
+      "grad_norm": 0.4970913477817239,
+      "learning_rate": 8.753719421030294e-07,
+      "loss": 0.6974,
+      "step": 40370
+    },
+    {
+      "epoch": 0.8673798169867251,
+      "grad_norm": 0.5426821783948291,
+      "learning_rate": 8.725850216910115e-07,
+      "loss": 0.6916,
+      "step": 40380
+    },
+    {
+      "epoch": 0.8675946212999956,
+      "grad_norm": 0.5320712513650713,
+      "learning_rate": 8.69802342313445e-07,
+      "loss": 0.6862,
+      "step": 40390
+    },
+    {
+      "epoch": 0.8678094256132663,
+      "grad_norm": 0.514150609745708,
+      "learning_rate": 8.670239052632945e-07,
+      "loss": 0.6902,
+      "step": 40400
+    },
+    {
+      "epoch": 0.8680242299265369,
+      "grad_norm": 0.5027008974103963,
+      "learning_rate": 8.642497118315584e-07,
+      "loss": 0.6743,
+      "step": 40410
+    },
+    {
+      "epoch": 0.8682390342398075,
+      "grad_norm": 0.49920507536090114,
+      "learning_rate": 8.614797633072547e-07,
+      "loss": 0.6853,
+      "step": 40420
+    },
+    {
+      "epoch": 0.8684538385530781,
+      "grad_norm": 0.5078214025245414,
+      "learning_rate": 8.58714060977438e-07,
+      "loss": 0.6927,
+      "step": 40430
+    },
+    {
+      "epoch": 0.8686686428663487,
+      "grad_norm": 0.5068985776833694,
+      "learning_rate": 8.55952606127185e-07,
+      "loss": 0.6877,
+      "step": 40440
+    },
+    {
+      "epoch": 0.8688834471796194,
+      "grad_norm": 0.5042210366606157,
+      "learning_rate": 8.531954000395992e-07,
+      "loss": 0.6797,
+      "step": 40450
+    },
+    {
+      "epoch": 0.86909825149289,
+      "grad_norm": 0.5041201672207695,
+      "learning_rate": 8.504424439958125e-07,
+      "loss": 0.6912,
+      "step": 40460
+    },
+    {
+      "epoch": 0.8693130558061606,
+      "grad_norm": 0.49939475577575226,
+      "learning_rate": 8.476937392749817e-07,
+      "loss": 0.689,
+      "step": 40470
+    },
+    {
+      "epoch": 0.8695278601194312,
+      "grad_norm": 0.508423114252119,
+      "learning_rate": 8.449492871542819e-07,
+      "loss": 0.6967,
+      "step": 40480
+    },
+    {
+      "epoch": 0.8697426644327018,
+      "grad_norm": 0.5049557993031742,
+      "learning_rate": 8.422090889089196e-07,
+      "loss": 0.6842,
+      "step": 40490
+    },
+    {
+      "epoch": 0.8699574687459725,
+      "grad_norm": 0.5210648597690205,
+      "learning_rate": 8.394731458121219e-07,
+      "loss": 0.6926,
+      "step": 40500
+    },
+    {
+      "epoch": 0.870172273059243,
+      "grad_norm": 0.5043799660787082,
+      "learning_rate": 8.367414591351408e-07,
+      "loss": 0.69,
+      "step": 40510
+    },
+    {
+      "epoch": 0.8703870773725136,
+      "grad_norm": 0.5317434333391947,
+      "learning_rate": 8.340140301472466e-07,
+      "loss": 0.6889,
+      "step": 40520
+    },
+    {
+      "epoch": 0.8706018816857842,
+      "grad_norm": 0.5400007003253303,
+      "learning_rate": 8.312908601157355e-07,
+      "loss": 0.6937,
+      "step": 40530
+    },
+    {
+      "epoch": 0.8708166859990548,
+      "grad_norm": 0.5127402676218981,
+      "learning_rate": 8.285719503059209e-07,
+      "loss": 0.6916,
+      "step": 40540
+    },
+    {
+      "epoch": 0.8710314903123255,
+      "grad_norm": 0.5151193689571636,
+      "learning_rate": 8.258573019811389e-07,
+      "loss": 0.6951,
+      "step": 40550
+    },
+    {
+      "epoch": 0.8712462946255961,
+      "grad_norm": 0.5239133915563157,
+      "learning_rate": 8.231469164027461e-07,
+      "loss": 0.691,
+      "step": 40560
+    },
+    {
+      "epoch": 0.8714610989388667,
+      "grad_norm": 0.5147336548996377,
+      "learning_rate": 8.204407948301174e-07,
+      "loss": 0.6881,
+      "step": 40570
+    },
+    {
+      "epoch": 0.8716759032521373,
+      "grad_norm": 0.5191148599179711,
+      "learning_rate": 8.177389385206469e-07,
+      "loss": 0.6861,
+      "step": 40580
+    },
+    {
+      "epoch": 0.8718907075654079,
+      "grad_norm": 0.5159462930405265,
+      "learning_rate": 8.150413487297438e-07,
+      "loss": 0.6907,
+      "step": 40590
+    },
+    {
+      "epoch": 0.8721055118786786,
+      "grad_norm": 0.5512478977489063,
+      "learning_rate": 8.12348026710843e-07,
+      "loss": 0.6971,
+      "step": 40600
+    },
+    {
+      "epoch": 0.8723203161919492,
+      "grad_norm": 0.512581984887224,
+      "learning_rate": 8.096589737153859e-07,
+      "loss": 0.6738,
+      "step": 40610
+    },
+    {
+      "epoch": 0.8725351205052198,
+      "grad_norm": 0.5228312882242846,
+      "learning_rate": 8.069741909928374e-07,
+      "loss": 0.688,
+      "step": 40620
+    },
+    {
+      "epoch": 0.8727499248184903,
+      "grad_norm": 0.5006383946158728,
+      "learning_rate": 8.042936797906753e-07,
+      "loss": 0.687,
+      "step": 40630
+    },
+    {
+      "epoch": 0.8729647291317609,
+      "grad_norm": 0.5006156486251359,
+      "learning_rate": 8.016174413543953e-07,
+      "loss": 0.6831,
+      "step": 40640
+    },
+    {
+      "epoch": 0.8731795334450316,
+      "grad_norm": 0.5111940788173959,
+      "learning_rate": 7.989454769275073e-07,
+      "loss": 0.6885,
+      "step": 40650
+    },
+    {
+      "epoch": 0.8733943377583022,
+      "grad_norm": 0.5449360427418215,
+      "learning_rate": 7.962777877515293e-07,
+      "loss": 0.6927,
+      "step": 40660
+    },
+    {
+      "epoch": 0.8736091420715728,
+      "grad_norm": 0.5089460028641523,
+      "learning_rate": 7.936143750660008e-07,
+      "loss": 0.6742,
+      "step": 40670
+    },
+    {
+      "epoch": 0.8738239463848434,
+      "grad_norm": 0.5017956450345394,
+      "learning_rate": 7.909552401084697e-07,
+      "loss": 0.6715,
+      "step": 40680
+    },
+    {
+      "epoch": 0.874038750698114,
+      "grad_norm": 0.49880673793119984,
+      "learning_rate": 7.883003841144976e-07,
+      "loss": 0.679,
+      "step": 40690
+    },
+    {
+      "epoch": 0.8742535550113846,
+      "grad_norm": 0.4907335395967764,
+      "learning_rate": 7.856498083176612e-07,
+      "loss": 0.6785,
+      "step": 40700
+    },
+    {
+      "epoch": 0.8744683593246553,
+      "grad_norm": 0.5329308849961445,
+      "learning_rate": 7.830035139495384e-07,
+      "loss": 0.6933,
+      "step": 40710
+    },
+    {
+      "epoch": 0.8746831636379259,
+      "grad_norm": 0.5130223382623801,
+      "learning_rate": 7.803615022397315e-07,
+      "loss": 0.7,
+      "step": 40720
+    },
+    {
+      "epoch": 0.8748979679511965,
+      "grad_norm": 0.5211784970357225,
+      "learning_rate": 7.777237744158406e-07,
+      "loss": 0.7002,
+      "step": 40730
+    },
+    {
+      "epoch": 0.875112772264467,
+      "grad_norm": 0.5102775700430415,
+      "learning_rate": 7.750903317034831e-07,
+      "loss": 0.692,
+      "step": 40740
+    },
+    {
+      "epoch": 0.8753275765777376,
+      "grad_norm": 0.5042499830623806,
+      "learning_rate": 7.724611753262834e-07,
+      "loss": 0.6793,
+      "step": 40750
+    },
+    {
+      "epoch": 0.8755423808910083,
+      "grad_norm": 0.5160190803282572,
+      "learning_rate": 7.698363065058689e-07,
+      "loss": 0.6889,
+      "step": 40760
+    },
+    {
+      "epoch": 0.8757571852042789,
+      "grad_norm": 0.50370584207469,
+      "learning_rate": 7.672157264618852e-07,
+      "loss": 0.6773,
+      "step": 40770
+    },
+    {
+      "epoch": 0.8759719895175495,
+      "grad_norm": 0.5051104461571446,
+      "learning_rate": 7.645994364119758e-07,
+      "loss": 0.6992,
+      "step": 40780
+    },
+    {
+      "epoch": 0.8761867938308201,
+      "grad_norm": 0.4946566053954112,
+      "learning_rate": 7.61987437571795e-07,
+      "loss": 0.6878,
+      "step": 40790
+    },
+    {
+      "epoch": 0.8764015981440907,
+      "grad_norm": 0.4996685140970515,
+      "learning_rate": 7.593797311550055e-07,
+      "loss": 0.6808,
+      "step": 40800
+    },
+    {
+      "epoch": 0.8766164024573614,
+      "grad_norm": 0.5172882831148481,
+      "learning_rate": 7.567763183732668e-07,
+      "loss": 0.6892,
+      "step": 40810
+    },
+    {
+      "epoch": 0.876831206770632,
+      "grad_norm": 0.5091750814973897,
+      "learning_rate": 7.541772004362557e-07,
+      "loss": 0.693,
+      "step": 40820
+    },
+    {
+      "epoch": 0.8770460110839026,
+      "grad_norm": 0.500703308186769,
+      "learning_rate": 7.515823785516418e-07,
+      "loss": 0.6884,
+      "step": 40830
+    },
+    {
+      "epoch": 0.8772608153971732,
+      "grad_norm": 0.5224680213539507,
+      "learning_rate": 7.489918539251085e-07,
+      "loss": 0.7,
+      "step": 40840
+    },
+    {
+      "epoch": 0.8774756197104437,
+      "grad_norm": 0.5058732772246898,
+      "learning_rate": 7.464056277603326e-07,
+      "loss": 0.6924,
+      "step": 40850
+    },
+    {
+      "epoch": 0.8776904240237144,
+      "grad_norm": 0.5113746801292882,
+      "learning_rate": 7.438237012590033e-07,
+      "loss": 0.6887,
+      "step": 40860
+    },
+    {
+      "epoch": 0.877905228336985,
+      "grad_norm": 0.5305907079168781,
+      "learning_rate": 7.412460756208051e-07,
+      "loss": 0.6798,
+      "step": 40870
+    },
+    {
+      "epoch": 0.8781200326502556,
+      "grad_norm": 0.5126751045545114,
+      "learning_rate": 7.386727520434245e-07,
+      "loss": 0.6885,
+      "step": 40880
+    },
+    {
+      "epoch": 0.8783348369635262,
+      "grad_norm": 0.5022215260758026,
+      "learning_rate": 7.361037317225561e-07,
+      "loss": 0.6854,
+      "step": 40890
+    },
+    {
+      "epoch": 0.8785496412767968,
+      "grad_norm": 0.5192119588923588,
+      "learning_rate": 7.335390158518852e-07,
+      "loss": 0.6853,
+      "step": 40900
+    },
+    {
+      "epoch": 0.8787644455900675,
+      "grad_norm": 0.49672397477009517,
+      "learning_rate": 7.309786056231039e-07,
+      "loss": 0.6874,
+      "step": 40910
+    },
+    {
+      "epoch": 0.8789792499033381,
+      "grad_norm": 0.5145358299532603,
+      "learning_rate": 7.284225022259028e-07,
+      "loss": 0.6935,
+      "step": 40920
+    },
+    {
+      "epoch": 0.8791940542166087,
+      "grad_norm": 0.5226202982810969,
+      "learning_rate": 7.258707068479642e-07,
+      "loss": 0.6899,
+      "step": 40930
+    },
+    {
+      "epoch": 0.8794088585298793,
+      "grad_norm": 0.513542215006283,
+      "learning_rate": 7.23323220674983e-07,
+      "loss": 0.6793,
+      "step": 40940
+    },
+    {
+      "epoch": 0.8796236628431499,
+      "grad_norm": 0.5225094478413966,
+      "learning_rate": 7.207800448906366e-07,
+      "loss": 0.7004,
+      "step": 40950
+    },
+    {
+      "epoch": 0.8798384671564206,
+      "grad_norm": 0.5150496074301979,
+      "learning_rate": 7.182411806766088e-07,
+      "loss": 0.6911,
+      "step": 40960
+    },
+    {
+      "epoch": 0.8800532714696911,
+      "grad_norm": 0.49652687262660034,
+      "learning_rate": 7.157066292125769e-07,
+      "loss": 0.6853,
+      "step": 40970
+    },
+    {
+      "epoch": 0.8802680757829617,
+      "grad_norm": 0.5241260101055815,
+      "learning_rate": 7.131763916762169e-07,
+      "loss": 0.6935,
+      "step": 40980
+    },
+    {
+      "epoch": 0.8804828800962323,
+      "grad_norm": 0.48817871163918836,
+      "learning_rate": 7.106504692431981e-07,
+      "loss": 0.693,
+      "step": 40990
+    },
+    {
+      "epoch": 0.8806976844095029,
+      "grad_norm": 0.529292181472698,
+      "learning_rate": 7.081288630871819e-07,
+      "loss": 0.7034,
+      "step": 41000
+    },
+    {
+      "epoch": 0.8809124887227735,
+      "grad_norm": 0.5059883454495374,
+      "learning_rate": 7.056115743798309e-07,
+      "loss": 0.6962,
+      "step": 41010
+    },
+    {
+      "epoch": 0.8811272930360442,
+      "grad_norm": 0.5470086285335591,
+      "learning_rate": 7.030986042907962e-07,
+      "loss": 0.6866,
+      "step": 41020
+    },
+    {
+      "epoch": 0.8813420973493148,
+      "grad_norm": 0.5236923421331561,
+      "learning_rate": 7.005899539877248e-07,
+      "loss": 0.6817,
+      "step": 41030
+    },
+    {
+      "epoch": 0.8815569016625854,
+      "grad_norm": 0.5089883929134034,
+      "learning_rate": 6.980856246362566e-07,
+      "loss": 0.6948,
+      "step": 41040
+    },
+    {
+      "epoch": 0.881771705975856,
+      "grad_norm": 0.5218195607179691,
+      "learning_rate": 6.955856174000208e-07,
+      "loss": 0.6975,
+      "step": 41050
+    },
+    {
+      "epoch": 0.8819865102891266,
+      "grad_norm": 0.5082317075268683,
+      "learning_rate": 6.93089933440645e-07,
+      "loss": 0.7135,
+      "step": 41060
+    },
+    {
+      "epoch": 0.8822013146023973,
+      "grad_norm": 0.5088831853959854,
+      "learning_rate": 6.905985739177379e-07,
+      "loss": 0.6887,
+      "step": 41070
+    },
+    {
+      "epoch": 0.8824161189156678,
+      "grad_norm": 0.5185177820509219,
+      "learning_rate": 6.881115399889083e-07,
+      "loss": 0.6974,
+      "step": 41080
+    },
+    {
+      "epoch": 0.8826309232289384,
+      "grad_norm": 0.5063149686563211,
+      "learning_rate": 6.856288328097505e-07,
+      "loss": 0.6998,
+      "step": 41090
+    },
+    {
+      "epoch": 0.882845727542209,
+      "grad_norm": 0.5250508854568222,
+      "learning_rate": 6.831504535338485e-07,
+      "loss": 0.6856,
+      "step": 41100
+    },
+    {
+      "epoch": 0.8830605318554796,
+      "grad_norm": 0.5124153836462256,
+      "learning_rate": 6.806764033127778e-07,
+      "loss": 0.6836,
+      "step": 41110
+    },
+    {
+      "epoch": 0.8832753361687503,
+      "grad_norm": 0.5274948224156129,
+      "learning_rate": 6.782066832960987e-07,
+      "loss": 0.6924,
+      "step": 41120
+    },
+    {
+      "epoch": 0.8834901404820209,
+      "grad_norm": 0.6205306011716412,
+      "learning_rate": 6.757412946313613e-07,
+      "loss": 0.6819,
+      "step": 41130
+    },
+    {
+      "epoch": 0.8837049447952915,
+      "grad_norm": 0.5245818667913486,
+      "learning_rate": 6.732802384641057e-07,
+      "loss": 0.6844,
+      "step": 41140
+    },
+    {
+      "epoch": 0.8839197491085621,
+      "grad_norm": 0.5205364438282932,
+      "learning_rate": 6.708235159378551e-07,
+      "loss": 0.6859,
+      "step": 41150
+    },
+    {
+      "epoch": 0.8841345534218327,
+      "grad_norm": 0.5131908898170064,
+      "learning_rate": 6.683711281941196e-07,
+      "loss": 0.6902,
+      "step": 41160
+    },
+    {
+      "epoch": 0.8843493577351034,
+      "grad_norm": 0.5062297624465112,
+      "learning_rate": 6.659230763724001e-07,
+      "loss": 0.6906,
+      "step": 41170
+    },
+    {
+      "epoch": 0.884564162048374,
+      "grad_norm": 0.5026327253540595,
+      "learning_rate": 6.634793616101732e-07,
+      "loss": 0.6722,
+      "step": 41180
+    },
+    {
+      "epoch": 0.8847789663616445,
+      "grad_norm": 0.5009299214524836,
+      "learning_rate": 6.610399850429094e-07,
+      "loss": 0.6872,
+      "step": 41190
+    },
+    {
+      "epoch": 0.8849937706749151,
+      "grad_norm": 0.5238156380409812,
+      "learning_rate": 6.58604947804059e-07,
+      "loss": 0.6917,
+      "step": 41200
+    },
+    {
+      "epoch": 0.8852085749881857,
+      "grad_norm": 0.5083745372288995,
+      "learning_rate": 6.561742510250569e-07,
+      "loss": 0.6809,
+      "step": 41210
+    },
+    {
+      "epoch": 0.8854233793014564,
+      "grad_norm": 0.5383915042584236,
+      "learning_rate": 6.537478958353216e-07,
+      "loss": 0.6979,
+      "step": 41220
+    },
+    {
+      "epoch": 0.885638183614727,
+      "grad_norm": 0.5043226314095659,
+      "learning_rate": 6.513258833622537e-07,
+      "loss": 0.682,
+      "step": 41230
+    },
+    {
+      "epoch": 0.8858529879279976,
+      "grad_norm": 0.5127868644616522,
+      "learning_rate": 6.489082147312387e-07,
+      "loss": 0.6821,
+      "step": 41240
+    },
+    {
+      "epoch": 0.8860677922412682,
+      "grad_norm": 0.4979888752333213,
+      "learning_rate": 6.464948910656377e-07,
+      "loss": 0.685,
+      "step": 41250
+    },
+    {
+      "epoch": 0.8862825965545388,
+      "grad_norm": 0.5128304605983001,
+      "learning_rate": 6.440859134867972e-07,
+      "loss": 0.6962,
+      "step": 41260
+    },
+    {
+      "epoch": 0.8864974008678094,
+      "grad_norm": 0.5051244175392152,
+      "learning_rate": 6.416812831140451e-07,
+      "loss": 0.6909,
+      "step": 41270
+    },
+    {
+      "epoch": 0.8867122051810801,
+      "grad_norm": 0.5101008415865377,
+      "learning_rate": 6.392810010646866e-07,
+      "loss": 0.6855,
+      "step": 41280
+    },
+    {
+      "epoch": 0.8869270094943507,
+      "grad_norm": 0.5202190940464628,
+      "learning_rate": 6.368850684540106e-07,
+      "loss": 0.7058,
+      "step": 41290
+    },
+    {
+      "epoch": 0.8871418138076212,
+      "grad_norm": 0.5186081016150472,
+      "learning_rate": 6.344934863952779e-07,
+      "loss": 0.6781,
+      "step": 41300
+    },
+    {
+      "epoch": 0.8873566181208918,
+      "grad_norm": 0.5329527297577857,
+      "learning_rate": 6.321062559997349e-07,
+      "loss": 0.7019,
+      "step": 41310
+    },
+    {
+      "epoch": 0.8875714224341624,
+      "grad_norm": 0.5217265024678235,
+      "learning_rate": 6.297233783766022e-07,
+      "loss": 0.6985,
+      "step": 41320
+    },
+    {
+      "epoch": 0.8877862267474331,
+      "grad_norm": 0.5194330367298682,
+      "learning_rate": 6.273448546330796e-07,
+      "loss": 0.6839,
+      "step": 41330
+    },
+    {
+      "epoch": 0.8880010310607037,
+      "grad_norm": 0.5097902257790524,
+      "learning_rate": 6.249706858743443e-07,
+      "loss": 0.6973,
+      "step": 41340
+    },
+    {
+      "epoch": 0.8882158353739743,
+      "grad_norm": 0.5142203693554276,
+      "learning_rate": 6.22600873203545e-07,
+      "loss": 0.6985,
+      "step": 41350
+    },
+    {
+      "epoch": 0.8884306396872449,
+      "grad_norm": 0.5090401566472215,
+      "learning_rate": 6.202354177218161e-07,
+      "loss": 0.6795,
+      "step": 41360
+    },
+    {
+      "epoch": 0.8886454440005155,
+      "grad_norm": 0.49677462148757334,
+      "learning_rate": 6.178743205282567e-07,
+      "loss": 0.6861,
+      "step": 41370
+    },
+    {
+      "epoch": 0.8888602483137862,
+      "grad_norm": 0.521382571115747,
+      "learning_rate": 6.155175827199467e-07,
+      "loss": 0.7016,
+      "step": 41380
+    },
+    {
+      "epoch": 0.8890750526270568,
+      "grad_norm": 0.5317933595388324,
+      "learning_rate": 6.131652053919424e-07,
+      "loss": 0.6907,
+      "step": 41390
+    },
+    {
+      "epoch": 0.8892898569403274,
+      "grad_norm": 0.5213449864141357,
+      "learning_rate": 6.10817189637265e-07,
+      "loss": 0.702,
+      "step": 41400
+    },
+    {
+      "epoch": 0.889504661253598,
+      "grad_norm": 0.5514505419123323,
+      "learning_rate": 6.084735365469229e-07,
+      "loss": 0.7022,
+      "step": 41410
+    },
+    {
+      "epoch": 0.8897194655668685,
+      "grad_norm": 0.5138208717231828,
+      "learning_rate": 6.061342472098819e-07,
+      "loss": 0.6742,
+      "step": 41420
+    },
+    {
+      "epoch": 0.8899342698801392,
+      "grad_norm": 0.5071996581331448,
+      "learning_rate": 6.037993227130945e-07,
+      "loss": 0.6922,
+      "step": 41430
+    },
+    {
+      "epoch": 0.8901490741934098,
+      "grad_norm": 0.5102097970519316,
+      "learning_rate": 6.014687641414751e-07,
+      "loss": 0.6929,
+      "step": 41440
+    },
+    {
+      "epoch": 0.8903638785066804,
+      "grad_norm": 0.5143324484042711,
+      "learning_rate": 5.991425725779132e-07,
+      "loss": 0.6975,
+      "step": 41450
+    },
+    {
+      "epoch": 0.890578682819951,
+      "grad_norm": 0.518475417925795,
+      "learning_rate": 5.968207491032707e-07,
+      "loss": 0.6917,
+      "step": 41460
+    },
+    {
+      "epoch": 0.8907934871332216,
+      "grad_norm": 0.4994682556931308,
+      "learning_rate": 5.945032947963736e-07,
+      "loss": 0.6811,
+      "step": 41470
+    },
+    {
+      "epoch": 0.8910082914464923,
+      "grad_norm": 0.5070960760778904,
+      "learning_rate": 5.921902107340282e-07,
+      "loss": 0.7016,
+      "step": 41480
+    },
+    {
+      "epoch": 0.8912230957597629,
+      "grad_norm": 0.5191444966925488,
+      "learning_rate": 5.898814979909995e-07,
+      "loss": 0.7052,
+      "step": 41490
+    },
+    {
+      "epoch": 0.8914379000730335,
+      "grad_norm": 0.5035080497362481,
+      "learning_rate": 5.87577157640028e-07,
+      "loss": 0.6973,
+      "step": 41500
+    },
+    {
+      "epoch": 0.8916527043863041,
+      "grad_norm": 0.5333052134877714,
+      "learning_rate": 5.852771907518228e-07,
+      "loss": 0.7011,
+      "step": 41510
+    },
+    {
+      "epoch": 0.8918675086995747,
+      "grad_norm": 0.5155211109068989,
+      "learning_rate": 5.829815983950526e-07,
+      "loss": 0.693,
+      "step": 41520
+    },
+    {
+      "epoch": 0.8920823130128454,
+      "grad_norm": 0.5023502013771952,
+      "learning_rate": 5.806903816363663e-07,
+      "loss": 0.6906,
+      "step": 41530
+    },
+    {
+      "epoch": 0.8922971173261159,
+      "grad_norm": 0.5214356225380249,
+      "learning_rate": 5.784035415403688e-07,
+      "loss": 0.6883,
+      "step": 41540
+    },
+    {
+      "epoch": 0.8925119216393865,
+      "grad_norm": 0.5159695126204489,
+      "learning_rate": 5.761210791696381e-07,
+      "loss": 0.6897,
+      "step": 41550
+    },
+    {
+      "epoch": 0.8927267259526571,
+      "grad_norm": 0.5209933564528445,
+      "learning_rate": 5.738429955847147e-07,
+      "loss": 0.7004,
+      "step": 41560
+    },
+    {
+      "epoch": 0.8929415302659277,
+      "grad_norm": 0.498039038963185,
+      "learning_rate": 5.715692918441029e-07,
+      "loss": 0.683,
+      "step": 41570
+    },
+    {
+      "epoch": 0.8931563345791983,
+      "grad_norm": 0.5006203623139068,
+      "learning_rate": 5.692999690042799e-07,
+      "loss": 0.6945,
+      "step": 41580
+    },
+    {
+      "epoch": 0.893371138892469,
+      "grad_norm": 0.5233476464091876,
+      "learning_rate": 5.670350281196768e-07,
+      "loss": 0.6987,
+      "step": 41590
+    },
+    {
+      "epoch": 0.8935859432057396,
+      "grad_norm": 0.5136980303605799,
+      "learning_rate": 5.64774470242696e-07,
+      "loss": 0.7007,
+      "step": 41600
+    },
+    {
+      "epoch": 0.8938007475190102,
+      "grad_norm": 0.512143271533227,
+      "learning_rate": 5.625182964236997e-07,
+      "loss": 0.6913,
+      "step": 41610
+    },
+    {
+      "epoch": 0.8940155518322808,
+      "grad_norm": 0.512132113394993,
+      "learning_rate": 5.602665077110169e-07,
+      "loss": 0.6854,
+      "step": 41620
+    },
+    {
+      "epoch": 0.8942303561455514,
+      "grad_norm": 0.51811405156484,
+      "learning_rate": 5.580191051509354e-07,
+      "loss": 0.691,
+      "step": 41630
+    },
+    {
+      "epoch": 0.894445160458822,
+      "grad_norm": 0.5115508553053043,
+      "learning_rate": 5.55776089787703e-07,
+      "loss": 0.6748,
+      "step": 41640
+    },
+    {
+      "epoch": 0.8946599647720926,
+      "grad_norm": 0.5241262214299848,
+      "learning_rate": 5.535374626635381e-07,
+      "loss": 0.6934,
+      "step": 41650
+    },
+    {
+      "epoch": 0.8948747690853632,
+      "grad_norm": 0.5244130502553322,
+      "learning_rate": 5.51303224818609e-07,
+      "loss": 0.6871,
+      "step": 41660
+    },
+    {
+      "epoch": 0.8950895733986338,
+      "grad_norm": 0.4994599674132126,
+      "learning_rate": 5.490733772910517e-07,
+      "loss": 0.6813,
+      "step": 41670
+    },
+    {
+      "epoch": 0.8953043777119044,
+      "grad_norm": 0.4986836428006602,
+      "learning_rate": 5.46847921116962e-07,
+      "loss": 0.6902,
+      "step": 41680
+    },
+    {
+      "epoch": 0.8955191820251751,
+      "grad_norm": 0.5152538290727082,
+      "learning_rate": 5.446268573303881e-07,
+      "loss": 0.6833,
+      "step": 41690
+    },
+    {
+      "epoch": 0.8957339863384457,
+      "grad_norm": 0.502386100555975,
+      "learning_rate": 5.424101869633491e-07,
+      "loss": 0.6927,
+      "step": 41700
+    },
+    {
+      "epoch": 0.8959487906517163,
+      "grad_norm": 0.5074939313677448,
+      "learning_rate": 5.401979110458133e-07,
+      "loss": 0.6758,
+      "step": 41710
+    },
+    {
+      "epoch": 0.8961635949649869,
+      "grad_norm": 0.5058971850323379,
+      "learning_rate": 5.379900306057084e-07,
+      "loss": 0.6797,
+      "step": 41720
+    },
+    {
+      "epoch": 0.8963783992782575,
+      "grad_norm": 0.4926506337654991,
+      "learning_rate": 5.357865466689249e-07,
+      "loss": 0.6822,
+      "step": 41730
+    },
+    {
+      "epoch": 0.8965932035915282,
+      "grad_norm": 0.5202246837788422,
+      "learning_rate": 5.33587460259305e-07,
+      "loss": 0.686,
+      "step": 41740
+    },
+    {
+      "epoch": 0.8968080079047988,
+      "grad_norm": 0.5224471201252612,
+      "learning_rate": 5.313927723986511e-07,
+      "loss": 0.6988,
+      "step": 41750
+    },
+    {
+      "epoch": 0.8970228122180693,
+      "grad_norm": 0.499998013117945,
+      "learning_rate": 5.292024841067179e-07,
+      "loss": 0.6913,
+      "step": 41760
+    },
+    {
+      "epoch": 0.8972376165313399,
+      "grad_norm": 0.5068130297990905,
+      "learning_rate": 5.270165964012198e-07,
+      "loss": 0.6918,
+      "step": 41770
+    },
+    {
+      "epoch": 0.8974524208446105,
+      "grad_norm": 0.5178316720171696,
+      "learning_rate": 5.248351102978255e-07,
+      "loss": 0.7096,
+      "step": 41780
+    },
+    {
+      "epoch": 0.8976672251578812,
+      "grad_norm": 0.5351505572464849,
+      "learning_rate": 5.226580268101566e-07,
+      "loss": 0.6831,
+      "step": 41790
+    },
+    {
+      "epoch": 0.8978820294711518,
+      "grad_norm": 0.5113178684301397,
+      "learning_rate": 5.204853469497917e-07,
+      "loss": 0.6945,
+      "step": 41800
+    },
+    {
+      "epoch": 0.8980968337844224,
+      "grad_norm": 0.5047128008727567,
+      "learning_rate": 5.18317071726262e-07,
+      "loss": 0.688,
+      "step": 41810
+    },
+    {
+      "epoch": 0.898311638097693,
+      "grad_norm": 0.5100213323202718,
+      "learning_rate": 5.161532021470527e-07,
+      "loss": 0.6941,
+      "step": 41820
+    },
+    {
+      "epoch": 0.8985264424109636,
+      "grad_norm": 0.5412542192317169,
+      "learning_rate": 5.139937392175987e-07,
+      "loss": 0.6804,
+      "step": 41830
+    },
+    {
+      "epoch": 0.8987412467242343,
+      "grad_norm": 0.5102282656359679,
+      "learning_rate": 5.118386839412925e-07,
+      "loss": 0.6867,
+      "step": 41840
+    },
+    {
+      "epoch": 0.8989560510375049,
+      "grad_norm": 0.5108191464876962,
+      "learning_rate": 5.096880373194745e-07,
+      "loss": 0.694,
+      "step": 41850
+    },
+    {
+      "epoch": 0.8991708553507755,
+      "grad_norm": 0.5080216890454713,
+      "learning_rate": 5.075418003514388e-07,
+      "loss": 0.6807,
+      "step": 41860
+    },
+    {
+      "epoch": 0.899385659664046,
+      "grad_norm": 0.5088339519727141,
+      "learning_rate": 5.053999740344306e-07,
+      "loss": 0.6868,
+      "step": 41870
+    },
+    {
+      "epoch": 0.8996004639773166,
+      "grad_norm": 0.5152387132782632,
+      "learning_rate": 5.032625593636442e-07,
+      "loss": 0.7018,
+      "step": 41880
+    },
+    {
+      "epoch": 0.8998152682905872,
+      "grad_norm": 0.4901427047062954,
+      "learning_rate": 5.01129557332225e-07,
+      "loss": 0.6845,
+      "step": 41890
+    },
+    {
+      "epoch": 0.9000300726038579,
+      "grad_norm": 0.5129493119966259,
+      "learning_rate": 4.990009689312669e-07,
+      "loss": 0.6826,
+      "step": 41900
+    },
+    {
+      "epoch": 0.9002448769171285,
+      "grad_norm": 0.5223099648979949,
+      "learning_rate": 4.968767951498166e-07,
+      "loss": 0.7005,
+      "step": 41910
+    },
+    {
+      "epoch": 0.9004596812303991,
+      "grad_norm": 0.4891041542553347,
+      "learning_rate": 4.947570369748656e-07,
+      "loss": 0.688,
+      "step": 41920
+    },
+    {
+      "epoch": 0.9006744855436697,
+      "grad_norm": 0.49608949461071944,
+      "learning_rate": 4.92641695391356e-07,
+      "loss": 0.6922,
+      "step": 41930
+    },
+    {
+      "epoch": 0.9008892898569403,
+      "grad_norm": 0.521684682281621,
+      "learning_rate": 4.905307713821761e-07,
+      "loss": 0.6964,
+      "step": 41940
+    },
+    {
+      "epoch": 0.901104094170211,
+      "grad_norm": 0.5052599021707938,
+      "learning_rate": 4.884242659281613e-07,
+      "loss": 0.6806,
+      "step": 41950
+    },
+    {
+      "epoch": 0.9013188984834816,
+      "grad_norm": 0.5154672787022767,
+      "learning_rate": 4.863221800080964e-07,
+      "loss": 0.685,
+      "step": 41960
+    },
+    {
+      "epoch": 0.9015337027967522,
+      "grad_norm": 0.5384871380139898,
+      "learning_rate": 4.842245145987112e-07,
+      "loss": 0.6828,
+      "step": 41970
+    },
+    {
+      "epoch": 0.9017485071100227,
+      "grad_norm": 0.5399455434197028,
+      "learning_rate": 4.821312706746817e-07,
+      "loss": 0.6962,
+      "step": 41980
+    },
+    {
+      "epoch": 0.9019633114232933,
+      "grad_norm": 0.5155738707830432,
+      "learning_rate": 4.800424492086275e-07,
+      "loss": 0.683,
+      "step": 41990
+    },
+    {
+      "epoch": 0.902178115736564,
+      "grad_norm": 0.5046827454219124,
+      "learning_rate": 4.779580511711191e-07,
+      "loss": 0.6852,
+      "step": 42000
+    },
+    {
+      "epoch": 0.9023929200498346,
+      "grad_norm": 0.5236358307924249,
+      "learning_rate": 4.758780775306637e-07,
+      "loss": 0.6973,
+      "step": 42010
+    },
+    {
+      "epoch": 0.9026077243631052,
+      "grad_norm": 0.5308120199608228,
+      "learning_rate": 4.738025292537185e-07,
+      "loss": 0.6911,
+      "step": 42020
+    },
+    {
+      "epoch": 0.9028225286763758,
+      "grad_norm": 0.5135158557502522,
+      "learning_rate": 4.71731407304683e-07,
+      "loss": 0.6889,
+      "step": 42030
+    },
+    {
+      "epoch": 0.9030373329896464,
+      "grad_norm": 0.5192712739754789,
+      "learning_rate": 4.6966471264589865e-07,
+      "loss": 0.6966,
+      "step": 42040
+    },
+    {
+      "epoch": 0.9032521373029171,
+      "grad_norm": 0.5213999046870769,
+      "learning_rate": 4.6760244623765384e-07,
+      "loss": 0.7042,
+      "step": 42050
+    },
+    {
+      "epoch": 0.9034669416161877,
+      "grad_norm": 0.5105007301231993,
+      "learning_rate": 4.65544609038171e-07,
+      "loss": 0.6781,
+      "step": 42060
+    },
+    {
+      "epoch": 0.9036817459294583,
+      "grad_norm": 0.500621312368029,
+      "learning_rate": 4.634912020036242e-07,
+      "loss": 0.6749,
+      "step": 42070
+    },
+    {
+      "epoch": 0.9038965502427289,
+      "grad_norm": 0.505010805688616,
+      "learning_rate": 4.614422260881235e-07,
+      "loss": 0.6732,
+      "step": 42080
+    },
+    {
+      "epoch": 0.9041113545559994,
+      "grad_norm": 0.5342937103145424,
+      "learning_rate": 4.593976822437207e-07,
+      "loss": 0.6844,
+      "step": 42090
+    },
+    {
+      "epoch": 0.9043261588692701,
+      "grad_norm": 0.5183262843971005,
+      "learning_rate": 4.573575714204115e-07,
+      "loss": 0.6984,
+      "step": 42100
+    },
+    {
+      "epoch": 0.9045409631825407,
+      "grad_norm": 0.517561745524737,
+      "learning_rate": 4.5532189456612306e-07,
+      "loss": 0.6889,
+      "step": 42110
+    },
+    {
+      "epoch": 0.9047557674958113,
+      "grad_norm": 0.5026306781586204,
+      "learning_rate": 4.5329065262673666e-07,
+      "loss": 0.6823,
+      "step": 42120
+    },
+    {
+      "epoch": 0.9049705718090819,
+      "grad_norm": 0.5427704129364224,
+      "learning_rate": 4.512638465460584e-07,
+      "loss": 0.6906,
+      "step": 42130
+    },
+    {
+      "epoch": 0.9051853761223525,
+      "grad_norm": 0.5074939200787636,
+      "learning_rate": 4.4924147726584153e-07,
+      "loss": 0.672,
+      "step": 42140
+    },
+    {
+      "epoch": 0.9054001804356231,
+      "grad_norm": 0.504225946197821,
+      "learning_rate": 4.4722354572577785e-07,
+      "loss": 0.6911,
+      "step": 42150
+    },
+    {
+      "epoch": 0.9056149847488938,
+      "grad_norm": 0.5145785005958214,
+      "learning_rate": 4.452100528634906e-07,
+      "loss": 0.694,
+      "step": 42160
+    },
+    {
+      "epoch": 0.9058297890621644,
+      "grad_norm": 0.49881418371078307,
+      "learning_rate": 4.432009996145492e-07,
+      "loss": 0.6794,
+      "step": 42170
+    },
+    {
+      "epoch": 0.906044593375435,
+      "grad_norm": 0.5176800894518867,
+      "learning_rate": 4.411963869124525e-07,
+      "loss": 0.6945,
+      "step": 42180
+    },
+    {
+      "epoch": 0.9062593976887056,
+      "grad_norm": 0.5090959735816222,
+      "learning_rate": 4.3919621568864313e-07,
+      "loss": 0.6875,
+      "step": 42190
+    },
+    {
+      "epoch": 0.9064742020019761,
+      "grad_norm": 0.5075300086152117,
+      "learning_rate": 4.372004868724944e-07,
+      "loss": 0.6857,
+      "step": 42200
+    },
+    {
+      "epoch": 0.9066890063152468,
+      "grad_norm": 0.5005344252080504,
+      "learning_rate": 4.3520920139131784e-07,
+      "loss": 0.6908,
+      "step": 42210
+    },
+    {
+      "epoch": 0.9069038106285174,
+      "grad_norm": 0.5092091177314088,
+      "learning_rate": 4.3322236017036114e-07,
+      "loss": 0.684,
+      "step": 42220
+    },
+    {
+      "epoch": 0.907118614941788,
+      "grad_norm": 0.5128571527964733,
+      "learning_rate": 4.3123996413280356e-07,
+      "loss": 0.6903,
+      "step": 42230
+    },
+    {
+      "epoch": 0.9073334192550586,
+      "grad_norm": 0.5236930918406155,
+      "learning_rate": 4.29262014199765e-07,
+      "loss": 0.6884,
+      "step": 42240
+    },
+    {
+      "epoch": 0.9075482235683292,
+      "grad_norm": 0.522031014128186,
+      "learning_rate": 4.272885112902925e-07,
+      "loss": 0.6862,
+      "step": 42250
+    },
+    {
+      "epoch": 0.9077630278815999,
+      "grad_norm": 0.534657132580671,
+      "learning_rate": 4.253194563213714e-07,
+      "loss": 0.6976,
+      "step": 42260
+    },
+    {
+      "epoch": 0.9079778321948705,
+      "grad_norm": 0.5270962563830953,
+      "learning_rate": 4.233548502079199e-07,
+      "loss": 0.6871,
+      "step": 42270
+    },
+    {
+      "epoch": 0.9081926365081411,
+      "grad_norm": 0.5397710000399429,
+      "learning_rate": 4.2139469386278445e-07,
+      "loss": 0.6787,
+      "step": 42280
+    },
+    {
+      "epoch": 0.9084074408214117,
+      "grad_norm": 0.5157952407298018,
+      "learning_rate": 4.194389881967531e-07,
+      "loss": 0.6929,
+      "step": 42290
+    },
+    {
+      "epoch": 0.9086222451346823,
+      "grad_norm": 0.5155262585551267,
+      "learning_rate": 4.174877341185368e-07,
+      "loss": 0.699,
+      "step": 42300
+    },
+    {
+      "epoch": 0.908837049447953,
+      "grad_norm": 0.6004351700870802,
+      "learning_rate": 4.155409325347826e-07,
+      "loss": 0.68,
+      "step": 42310
+    },
+    {
+      "epoch": 0.9090518537612235,
+      "grad_norm": 0.5179978734902895,
+      "learning_rate": 4.135985843500678e-07,
+      "loss": 0.6855,
+      "step": 42320
+    },
+    {
+      "epoch": 0.9092666580744941,
+      "grad_norm": 0.5172988314550917,
+      "learning_rate": 4.116606904668996e-07,
+      "loss": 0.6903,
+      "step": 42330
+    },
+    {
+      "epoch": 0.9094814623877647,
+      "grad_norm": 0.527441343917074,
+      "learning_rate": 4.097272517857187e-07,
+      "loss": 0.6835,
+      "step": 42340
+    },
+    {
+      "epoch": 0.9096962667010353,
+      "grad_norm": 0.5094088299463104,
+      "learning_rate": 4.07798269204891e-07,
+      "loss": 0.6745,
+      "step": 42350
+    },
+    {
+      "epoch": 0.909911071014306,
+      "grad_norm": 0.49824225660299976,
+      "learning_rate": 4.0587374362071495e-07,
+      "loss": 0.692,
+      "step": 42360
+    },
+    {
+      "epoch": 0.9101258753275766,
+      "grad_norm": 0.4958646778169003,
+      "learning_rate": 4.0395367592741876e-07,
+      "loss": 0.6927,
+      "step": 42370
+    },
+    {
+      "epoch": 0.9103406796408472,
+      "grad_norm": 0.5330942753471896,
+      "learning_rate": 4.020380670171553e-07,
+      "loss": 0.6929,
+      "step": 42380
+    },
+    {
+      "epoch": 0.9105554839541178,
+      "grad_norm": 0.5233184041925335,
+      "learning_rate": 4.001269177800116e-07,
+      "loss": 0.6965,
+      "step": 42390
+    },
+    {
+      "epoch": 0.9107702882673884,
+      "grad_norm": 0.5120728262031152,
+      "learning_rate": 3.9822022910399496e-07,
+      "loss": 0.6741,
+      "step": 42400
+    },
+    {
+      "epoch": 0.9109850925806591,
+      "grad_norm": 0.5064121099027354,
+      "learning_rate": 3.96318001875049e-07,
+      "loss": 0.6784,
+      "step": 42410
+    },
+    {
+      "epoch": 0.9111998968939297,
+      "grad_norm": 0.5602640114772643,
+      "learning_rate": 3.944202369770367e-07,
+      "loss": 0.6961,
+      "step": 42420
+    },
+    {
+      "epoch": 0.9114147012072003,
+      "grad_norm": 0.5432682414153873,
+      "learning_rate": 3.925269352917505e-07,
+      "loss": 0.6895,
+      "step": 42430
+    },
+    {
+      "epoch": 0.9116295055204708,
+      "grad_norm": 0.5159534104410033,
+      "learning_rate": 3.90638097698911e-07,
+      "loss": 0.6967,
+      "step": 42440
+    },
+    {
+      "epoch": 0.9118443098337414,
+      "grad_norm": 0.5421695386729521,
+      "learning_rate": 3.887537250761597e-07,
+      "loss": 0.6906,
+      "step": 42450
+    },
+    {
+      "epoch": 0.912059114147012,
+      "grad_norm": 0.5269412372870173,
+      "learning_rate": 3.8687381829906944e-07,
+      "loss": 0.6921,
+      "step": 42460
+    },
+    {
+      "epoch": 0.9122739184602827,
+      "grad_norm": 0.509562529599299,
+      "learning_rate": 3.849983782411337e-07,
+      "loss": 0.6885,
+      "step": 42470
+    },
+    {
+      "epoch": 0.9124887227735533,
+      "grad_norm": 0.510134688474772,
+      "learning_rate": 3.8312740577377214e-07,
+      "loss": 0.6815,
+      "step": 42480
+    },
+    {
+      "epoch": 0.9127035270868239,
+      "grad_norm": 0.512454202372446,
+      "learning_rate": 3.812609017663271e-07,
+      "loss": 0.7021,
+      "step": 42490
+    },
+    {
+      "epoch": 0.9129183314000945,
+      "grad_norm": 0.5069629345687713,
+      "learning_rate": 3.793988670860671e-07,
+      "loss": 0.6908,
+      "step": 42500
+    },
+    {
+      "epoch": 0.9131331357133651,
+      "grad_norm": 0.5145243709537642,
+      "learning_rate": 3.7754130259818334e-07,
+      "loss": 0.6761,
+      "step": 42510
+    },
+    {
+      "epoch": 0.9133479400266358,
+      "grad_norm": 0.5057114456413828,
+      "learning_rate": 3.7568820916578765e-07,
+      "loss": 0.7133,
+      "step": 42520
+    },
+    {
+      "epoch": 0.9135627443399064,
+      "grad_norm": 0.5027601053480496,
+      "learning_rate": 3.738395876499157e-07,
+      "loss": 0.68,
+      "step": 42530
+    },
+    {
+      "epoch": 0.913777548653177,
+      "grad_norm": 0.500024246376363,
+      "learning_rate": 3.7199543890952817e-07,
+      "loss": 0.6921,
+      "step": 42540
+    },
+    {
+      "epoch": 0.9139923529664475,
+      "grad_norm": 0.5082257613290262,
+      "learning_rate": 3.7015576380150187e-07,
+      "loss": 0.6812,
+      "step": 42550
+    },
+    {
+      "epoch": 0.9142071572797181,
+      "grad_norm": 0.530878498561435,
+      "learning_rate": 3.683205631806408e-07,
+      "loss": 0.6864,
+      "step": 42560
+    },
+    {
+      "epoch": 0.9144219615929888,
+      "grad_norm": 0.5061215824429818,
+      "learning_rate": 3.664898378996662e-07,
+      "loss": 0.6769,
+      "step": 42570
+    },
+    {
+      "epoch": 0.9146367659062594,
+      "grad_norm": 0.4972627962848076,
+      "learning_rate": 3.6466358880921984e-07,
+      "loss": 0.6796,
+      "step": 42580
+    },
+    {
+      "epoch": 0.91485157021953,
+      "grad_norm": 0.514903636534617,
+      "learning_rate": 3.628418167578651e-07,
+      "loss": 0.6843,
+      "step": 42590
+    },
+    {
+      "epoch": 0.9150663745328006,
+      "grad_norm": 0.5047790440228213,
+      "learning_rate": 3.610245225920839e-07,
+      "loss": 0.6826,
+      "step": 42600
+    },
+    {
+      "epoch": 0.9152811788460712,
+      "grad_norm": 0.5083554832988898,
+      "learning_rate": 3.5921170715627953e-07,
+      "loss": 0.6869,
+      "step": 42610
+    },
+    {
+      "epoch": 0.9154959831593419,
+      "grad_norm": 0.5133159115347935,
+      "learning_rate": 3.574033712927727e-07,
+      "loss": 0.6966,
+      "step": 42620
+    },
+    {
+      "epoch": 0.9157107874726125,
+      "grad_norm": 0.5164607440700887,
+      "learning_rate": 3.5559951584180355e-07,
+      "loss": 0.6882,
+      "step": 42630
+    },
+    {
+      "epoch": 0.9159255917858831,
+      "grad_norm": 0.513461482277684,
+      "learning_rate": 3.538001416415282e-07,
+      "loss": 0.6884,
+      "step": 42640
+    },
+    {
+      "epoch": 0.9161403960991537,
+      "grad_norm": 0.5008580969233325,
+      "learning_rate": 3.520052495280213e-07,
+      "loss": 0.6853,
+      "step": 42650
+    },
+    {
+      "epoch": 0.9163552004124242,
+      "grad_norm": 0.5047493200035524,
+      "learning_rate": 3.50214840335279e-07,
+      "loss": 0.6987,
+      "step": 42660
+    },
+    {
+      "epoch": 0.9165700047256949,
+      "grad_norm": 0.5126226903956973,
+      "learning_rate": 3.4842891489520803e-07,
+      "loss": 0.6782,
+      "step": 42670
+    },
+    {
+      "epoch": 0.9167848090389655,
+      "grad_norm": 0.5250140948834445,
+      "learning_rate": 3.466474740376369e-07,
+      "loss": 0.6831,
+      "step": 42680
+    },
+    {
+      "epoch": 0.9169996133522361,
+      "grad_norm": 0.4970994749703241,
+      "learning_rate": 3.448705185903101e-07,
+      "loss": 0.6813,
+      "step": 42690
+    },
+    {
+      "epoch": 0.9172144176655067,
+      "grad_norm": 0.4957874759690821,
+      "learning_rate": 3.4309804937888155e-07,
+      "loss": 0.6812,
+      "step": 42700
+    },
+    {
+      "epoch": 0.9174292219787773,
+      "grad_norm": 0.5126567512309492,
+      "learning_rate": 3.413300672269282e-07,
+      "loss": 0.6904,
+      "step": 42710
+    },
+    {
+      "epoch": 0.917644026292048,
+      "grad_norm": 0.5269391777536964,
+      "learning_rate": 3.3956657295593944e-07,
+      "loss": 0.6799,
+      "step": 42720
+    },
+    {
+      "epoch": 0.9178588306053186,
+      "grad_norm": 0.5037049888031642,
+      "learning_rate": 3.378075673853187e-07,
+      "loss": 0.6747,
+      "step": 42730
+    },
+    {
+      "epoch": 0.9180736349185892,
+      "grad_norm": 0.5180923103541171,
+      "learning_rate": 3.360530513323845e-07,
+      "loss": 0.696,
+      "step": 42740
+    },
+    {
+      "epoch": 0.9182884392318598,
+      "grad_norm": 0.5377551337981332,
+      "learning_rate": 3.3430302561236806e-07,
+      "loss": 0.7026,
+      "step": 42750
+    },
+    {
+      "epoch": 0.9185032435451304,
+      "grad_norm": 0.5086131451212155,
+      "learning_rate": 3.325574910384177e-07,
+      "loss": 0.6867,
+      "step": 42760
+    },
+    {
+      "epoch": 0.9187180478584009,
+      "grad_norm": 0.5360041477265602,
+      "learning_rate": 3.3081644842158924e-07,
+      "loss": 0.6868,
+      "step": 42770
+    },
+    {
+      "epoch": 0.9189328521716716,
+      "grad_norm": 0.5259395774668065,
+      "learning_rate": 3.290798985708576e-07,
+      "loss": 0.6822,
+      "step": 42780
+    },
+    {
+      "epoch": 0.9191476564849422,
+      "grad_norm": 0.511030877333274,
+      "learning_rate": 3.2734784229310425e-07,
+      "loss": 0.6748,
+      "step": 42790
+    },
+    {
+      "epoch": 0.9193624607982128,
+      "grad_norm": 0.507095211823108,
+      "learning_rate": 3.2562028039312746e-07,
+      "loss": 0.6916,
+      "step": 42800
+    },
+    {
+      "epoch": 0.9195772651114834,
+      "grad_norm": 0.5108278970810879,
+      "learning_rate": 3.2389721367363623e-07,
+      "loss": 0.6905,
+      "step": 42810
+    },
+    {
+      "epoch": 0.919792069424754,
+      "grad_norm": 0.5063482123733604,
+      "learning_rate": 3.221786429352458e-07,
+      "loss": 0.6867,
+      "step": 42820
+    },
+    {
+      "epoch": 0.9200068737380247,
+      "grad_norm": 0.5321360439147953,
+      "learning_rate": 3.204645689764907e-07,
+      "loss": 0.6799,
+      "step": 42830
+    },
+    {
+      "epoch": 0.9202216780512953,
+      "grad_norm": 0.5010045520484869,
+      "learning_rate": 3.187549925938094e-07,
+      "loss": 0.6897,
+      "step": 42840
+    },
+    {
+      "epoch": 0.9204364823645659,
+      "grad_norm": 0.5146448011202501,
+      "learning_rate": 3.170499145815542e-07,
+      "loss": 0.6936,
+      "step": 42850
+    },
+    {
+      "epoch": 0.9206512866778365,
+      "grad_norm": 0.522131748562637,
+      "learning_rate": 3.15349335731987e-07,
+      "loss": 0.6758,
+      "step": 42860
+    },
+    {
+      "epoch": 0.9208660909911071,
+      "grad_norm": 0.5084382700595336,
+      "learning_rate": 3.136532568352746e-07,
+      "loss": 0.6872,
+      "step": 42870
+    },
+    {
+      "epoch": 0.9210808953043778,
+      "grad_norm": 0.5063314963517712,
+      "learning_rate": 3.11961678679501e-07,
+      "loss": 0.6763,
+      "step": 42880
+    },
+    {
+      "epoch": 0.9212956996176483,
+      "grad_norm": 0.529160941637022,
+      "learning_rate": 3.1027460205065194e-07,
+      "loss": 0.6898,
+      "step": 42890
+    },
+    {
+      "epoch": 0.9215105039309189,
+      "grad_norm": 0.5245818745748351,
+      "learning_rate": 3.085920277326249e-07,
+      "loss": 0.6639,
+      "step": 42900
+    },
+    {
+      "epoch": 0.9217253082441895,
+      "grad_norm": 0.5217386624549989,
+      "learning_rate": 3.0691395650722434e-07,
+      "loss": 0.6969,
+      "step": 42910
+    },
+    {
+      "epoch": 0.9219401125574601,
+      "grad_norm": 0.5129921994136167,
+      "learning_rate": 3.0524038915416e-07,
+      "loss": 0.6841,
+      "step": 42920
+    },
+    {
+      "epoch": 0.9221549168707308,
+      "grad_norm": 0.5021469780724663,
+      "learning_rate": 3.035713264510565e-07,
+      "loss": 0.6826,
+      "step": 42930
+    },
+    {
+      "epoch": 0.9223697211840014,
+      "grad_norm": 0.5116730541146018,
+      "learning_rate": 3.0190676917343673e-07,
+      "loss": 0.6923,
+      "step": 42940
+    },
+    {
+      "epoch": 0.922584525497272,
+      "grad_norm": 0.5041634183521139,
+      "learning_rate": 3.002467180947355e-07,
+      "loss": 0.6801,
+      "step": 42950
+    },
+    {
+      "epoch": 0.9227993298105426,
+      "grad_norm": 0.5071126517885879,
+      "learning_rate": 2.9859117398629236e-07,
+      "loss": 0.6707,
+      "step": 42960
+    },
+    {
+      "epoch": 0.9230141341238132,
+      "grad_norm": 0.5191938232678536,
+      "learning_rate": 2.969401376173486e-07,
+      "loss": 0.6887,
+      "step": 42970
+    },
+    {
+      "epoch": 0.9232289384370839,
+      "grad_norm": 0.51739812424603,
+      "learning_rate": 2.9529360975506074e-07,
+      "loss": 0.6799,
+      "step": 42980
+    },
+    {
+      "epoch": 0.9234437427503545,
+      "grad_norm": 0.5153545341424074,
+      "learning_rate": 2.9365159116447886e-07,
+      "loss": 0.6835,
+      "step": 42990
+    },
+    {
+      "epoch": 0.923658547063625,
+      "grad_norm": 0.5306207680932173,
+      "learning_rate": 2.920140826085682e-07,
+      "loss": 0.6986,
+      "step": 43000
+    },
+    {
+      "epoch": 0.9238733513768956,
+      "grad_norm": 0.5055030531585294,
+      "learning_rate": 2.9038108484819137e-07,
+      "loss": 0.6895,
+      "step": 43010
+    },
+    {
+      "epoch": 0.9240881556901662,
+      "grad_norm": 0.5063675711910615,
+      "learning_rate": 2.887525986421169e-07,
+      "loss": 0.6995,
+      "step": 43020
+    },
+    {
+      "epoch": 0.9243029600034369,
+      "grad_norm": 0.4996360884485676,
+      "learning_rate": 2.8712862474702063e-07,
+      "loss": 0.6906,
+      "step": 43030
+    },
+    {
+      "epoch": 0.9245177643167075,
+      "grad_norm": 0.5106698528538595,
+      "learning_rate": 2.855091639174734e-07,
+      "loss": 0.6989,
+      "step": 43040
+    },
+    {
+      "epoch": 0.9247325686299781,
+      "grad_norm": 0.5074236659920939,
+      "learning_rate": 2.8389421690595995e-07,
+      "loss": 0.6839,
+      "step": 43050
+    },
+    {
+      "epoch": 0.9249473729432487,
+      "grad_norm": 0.572337745383503,
+      "learning_rate": 2.8228378446285784e-07,
+      "loss": 0.6902,
+      "step": 43060
+    },
+    {
+      "epoch": 0.9251621772565193,
+      "grad_norm": 0.5068410597647415,
+      "learning_rate": 2.8067786733645296e-07,
+      "loss": 0.6867,
+      "step": 43070
+    },
+    {
+      "epoch": 0.9253769815697899,
+      "grad_norm": 0.5186456864886076,
+      "learning_rate": 2.7907646627293284e-07,
+      "loss": 0.6919,
+      "step": 43080
+    },
+    {
+      "epoch": 0.9255917858830606,
+      "grad_norm": 0.5126033061284958,
+      "learning_rate": 2.7747958201638113e-07,
+      "loss": 0.6965,
+      "step": 43090
+    },
+    {
+      "epoch": 0.9258065901963312,
+      "grad_norm": 0.5161766653294025,
+      "learning_rate": 2.75887215308791e-07,
+      "loss": 0.6901,
+      "step": 43100
+    },
+    {
+      "epoch": 0.9260213945096017,
+      "grad_norm": 0.5049885765836929,
+      "learning_rate": 2.7429936689004956e-07,
+      "loss": 0.6788,
+      "step": 43110
+    },
+    {
+      "epoch": 0.9262361988228723,
+      "grad_norm": 0.5106995321298553,
+      "learning_rate": 2.7271603749794763e-07,
+      "loss": 0.6804,
+      "step": 43120
+    },
+    {
+      "epoch": 0.9264510031361429,
+      "grad_norm": 0.5080317309925427,
+      "learning_rate": 2.711372278681779e-07,
+      "loss": 0.6844,
+      "step": 43130
+    },
+    {
+      "epoch": 0.9266658074494136,
+      "grad_norm": 0.5164907399314109,
+      "learning_rate": 2.69562938734329e-07,
+      "loss": 0.6936,
+      "step": 43140
+    },
+    {
+      "epoch": 0.9268806117626842,
+      "grad_norm": 0.5205333655443777,
+      "learning_rate": 2.6799317082789267e-07,
+      "loss": 0.6772,
+      "step": 43150
+    },
+    {
+      "epoch": 0.9270954160759548,
+      "grad_norm": 0.5259790796204346,
+      "learning_rate": 2.664279248782564e-07,
+      "loss": 0.7015,
+      "step": 43160
+    },
+    {
+      "epoch": 0.9273102203892254,
+      "grad_norm": 0.48989809551529545,
+      "learning_rate": 2.648672016127096e-07,
+      "loss": 0.6818,
+      "step": 43170
+    },
+    {
+      "epoch": 0.927525024702496,
+      "grad_norm": 0.5186292029129187,
+      "learning_rate": 2.633110017564389e-07,
+      "loss": 0.6841,
+      "step": 43180
+    },
+    {
+      "epoch": 0.9277398290157667,
+      "grad_norm": 0.494677339162897,
+      "learning_rate": 2.6175932603253017e-07,
+      "loss": 0.6857,
+      "step": 43190
+    },
+    {
+      "epoch": 0.9279546333290373,
+      "grad_norm": 0.513254704122593,
+      "learning_rate": 2.6021217516196464e-07,
+      "loss": 0.6844,
+      "step": 43200
+    },
+    {
+      "epoch": 0.9281694376423079,
+      "grad_norm": 0.5097903856248501,
+      "learning_rate": 2.586695498636238e-07,
+      "loss": 0.6865,
+      "step": 43210
+    },
+    {
+      "epoch": 0.9283842419555784,
+      "grad_norm": 0.5158784848429144,
+      "learning_rate": 2.571314508542866e-07,
+      "loss": 0.6883,
+      "step": 43220
+    },
+    {
+      "epoch": 0.928599046268849,
+      "grad_norm": 0.49451018639662936,
+      "learning_rate": 2.5559787884862465e-07,
+      "loss": 0.6923,
+      "step": 43230
+    },
+    {
+      "epoch": 0.9288138505821197,
+      "grad_norm": 0.5206523113168011,
+      "learning_rate": 2.5406883455921015e-07,
+      "loss": 0.6845,
+      "step": 43240
+    },
+    {
+      "epoch": 0.9290286548953903,
+      "grad_norm": 0.5183662832534691,
+      "learning_rate": 2.5254431869651154e-07,
+      "loss": 0.6888,
+      "step": 43250
+    },
+    {
+      "epoch": 0.9292434592086609,
+      "grad_norm": 0.5012033057724504,
+      "learning_rate": 2.51024331968891e-07,
+      "loss": 0.6838,
+      "step": 43260
+    },
+    {
+      "epoch": 0.9294582635219315,
+      "grad_norm": 0.5271739265196879,
+      "learning_rate": 2.49508875082608e-07,
+      "loss": 0.6925,
+      "step": 43270
+    },
+    {
+      "epoch": 0.9296730678352021,
+      "grad_norm": 0.5196998730244764,
+      "learning_rate": 2.4799794874181496e-07,
+      "loss": 0.6826,
+      "step": 43280
+    },
+    {
+      "epoch": 0.9298878721484728,
+      "grad_norm": 0.5154557177789315,
+      "learning_rate": 2.4649155364856127e-07,
+      "loss": 0.6947,
+      "step": 43290
+    },
+    {
+      "epoch": 0.9301026764617434,
+      "grad_norm": 0.5158926177913861,
+      "learning_rate": 2.449896905027904e-07,
+      "loss": 0.6774,
+      "step": 43300
+    },
+    {
+      "epoch": 0.930317480775014,
+      "grad_norm": 0.5452488869938171,
+      "learning_rate": 2.434923600023409e-07,
+      "loss": 0.6895,
+      "step": 43310
+    },
+    {
+      "epoch": 0.9305322850882846,
+      "grad_norm": 0.5240815118076753,
+      "learning_rate": 2.419995628429428e-07,
+      "loss": 0.6814,
+      "step": 43320
+    },
+    {
+      "epoch": 0.9307470894015551,
+      "grad_norm": 0.5208459297385867,
+      "learning_rate": 2.405112997182224e-07,
+      "loss": 0.6986,
+      "step": 43330
+    },
+    {
+      "epoch": 0.9309618937148257,
+      "grad_norm": 0.5146723954727245,
+      "learning_rate": 2.3902757131969657e-07,
+      "loss": 0.6936,
+      "step": 43340
+    },
+    {
+      "epoch": 0.9311766980280964,
+      "grad_norm": 0.4988199872705836,
+      "learning_rate": 2.3754837833677823e-07,
+      "loss": 0.6921,
+      "step": 43350
+    },
+    {
+      "epoch": 0.931391502341367,
+      "grad_norm": 0.5061737031935741,
+      "learning_rate": 2.3607372145677098e-07,
+      "loss": 0.6819,
+      "step": 43360
+    },
+    {
+      "epoch": 0.9316063066546376,
+      "grad_norm": 0.5118209164152153,
+      "learning_rate": 2.3460360136487004e-07,
+      "loss": 0.6938,
+      "step": 43370
+    },
+    {
+      "epoch": 0.9318211109679082,
+      "grad_norm": 0.5028715193115553,
+      "learning_rate": 2.3313801874416452e-07,
+      "loss": 0.6838,
+      "step": 43380
+    },
+    {
+      "epoch": 0.9320359152811788,
+      "grad_norm": 0.5037421436122391,
+      "learning_rate": 2.3167697427563418e-07,
+      "loss": 0.6695,
+      "step": 43390
+    },
+    {
+      "epoch": 0.9322507195944495,
+      "grad_norm": 0.5056308074418332,
+      "learning_rate": 2.302204686381515e-07,
+      "loss": 0.6915,
+      "step": 43400
+    },
+    {
+      "epoch": 0.9324655239077201,
+      "grad_norm": 0.4973980726367497,
+      "learning_rate": 2.2876850250847626e-07,
+      "loss": 0.6902,
+      "step": 43410
+    },
+    {
+      "epoch": 0.9326803282209907,
+      "grad_norm": 0.49416302287965536,
+      "learning_rate": 2.2732107656126435e-07,
+      "loss": 0.6778,
+      "step": 43420
+    },
+    {
+      "epoch": 0.9328951325342613,
+      "grad_norm": 0.5373794494908628,
+      "learning_rate": 2.2587819146905665e-07,
+      "loss": 0.6905,
+      "step": 43430
+    },
+    {
+      "epoch": 0.9331099368475319,
+      "grad_norm": 0.527952990743072,
+      "learning_rate": 2.2443984790228802e-07,
+      "loss": 0.6798,
+      "step": 43440
+    },
+    {
+      "epoch": 0.9333247411608026,
+      "grad_norm": 0.5510638102933684,
+      "learning_rate": 2.230060465292827e-07,
+      "loss": 0.7014,
+      "step": 43450
+    },
+    {
+      "epoch": 0.9335395454740731,
+      "grad_norm": 0.534593564488736,
+      "learning_rate": 2.2157678801625115e-07,
+      "loss": 0.6812,
+      "step": 43460
+    },
+    {
+      "epoch": 0.9337543497873437,
+      "grad_norm": 0.51290579083354,
+      "learning_rate": 2.201520730272966e-07,
+      "loss": 0.6848,
+      "step": 43470
+    },
+    {
+      "epoch": 0.9339691541006143,
+      "grad_norm": 0.498976254097296,
+      "learning_rate": 2.1873190222441054e-07,
+      "loss": 0.7,
+      "step": 43480
+    },
+    {
+      "epoch": 0.9341839584138849,
+      "grad_norm": 0.5110569428279691,
+      "learning_rate": 2.1731627626747186e-07,
+      "loss": 0.691,
+      "step": 43490
+    },
+    {
+      "epoch": 0.9343987627271556,
+      "grad_norm": 0.5108780076816333,
+      "learning_rate": 2.1590519581424885e-07,
+      "loss": 0.6902,
+      "step": 43500
+    },
+    {
+      "epoch": 0.9346135670404262,
+      "grad_norm": 0.5104285936640807,
+      "learning_rate": 2.144986615203959e-07,
+      "loss": 0.6925,
+      "step": 43510
+    },
+    {
+      "epoch": 0.9348283713536968,
+      "grad_norm": 0.5096080388433524,
+      "learning_rate": 2.13096674039458e-07,
+      "loss": 0.6992,
+      "step": 43520
+    },
+    {
+      "epoch": 0.9350431756669674,
+      "grad_norm": 0.5402406252479437,
+      "learning_rate": 2.1169923402286298e-07,
+      "loss": 0.6953,
+      "step": 43530
+    },
+    {
+      "epoch": 0.935257979980238,
+      "grad_norm": 0.5096415078767643,
+      "learning_rate": 2.1030634211993028e-07,
+      "loss": 0.6897,
+      "step": 43540
+    },
+    {
+      "epoch": 0.9354727842935087,
+      "grad_norm": 0.5172382489410695,
+      "learning_rate": 2.089179989778656e-07,
+      "loss": 0.6967,
+      "step": 43550
+    },
+    {
+      "epoch": 0.9356875886067793,
+      "grad_norm": 0.507182558755288,
+      "learning_rate": 2.0753420524175616e-07,
+      "loss": 0.6952,
+      "step": 43560
+    },
+    {
+      "epoch": 0.9359023929200498,
+      "grad_norm": 0.525056213053443,
+      "learning_rate": 2.0615496155458214e-07,
+      "loss": 0.684,
+      "step": 43570
+    },
+    {
+      "epoch": 0.9361171972333204,
+      "grad_norm": 0.5190206893050133,
+      "learning_rate": 2.0478026855720313e-07,
+      "loss": 0.6954,
+      "step": 43580
+    },
+    {
+      "epoch": 0.936332001546591,
+      "grad_norm": 0.5217080830284128,
+      "learning_rate": 2.0341012688837146e-07,
+      "loss": 0.6891,
+      "step": 43590
+    },
+    {
+      "epoch": 0.9365468058598617,
+      "grad_norm": 0.5373069241110895,
+      "learning_rate": 2.0204453718471795e-07,
+      "loss": 0.6683,
+      "step": 43600
+    },
+    {
+      "epoch": 0.9367616101731323,
+      "grad_norm": 0.512356091417784,
+      "learning_rate": 2.0068350008076055e-07,
+      "loss": 0.6877,
+      "step": 43610
+    },
+    {
+      "epoch": 0.9369764144864029,
+      "grad_norm": 0.5038542302769475,
+      "learning_rate": 1.9932701620890448e-07,
+      "loss": 0.6708,
+      "step": 43620
+    },
+    {
+      "epoch": 0.9371912187996735,
+      "grad_norm": 0.5079741863034887,
+      "learning_rate": 1.979750861994334e-07,
+      "loss": 0.6676,
+      "step": 43630
+    },
+    {
+      "epoch": 0.9374060231129441,
+      "grad_norm": 0.516342583535674,
+      "learning_rate": 1.9662771068052367e-07,
+      "loss": 0.6884,
+      "step": 43640
+    },
+    {
+      "epoch": 0.9376208274262147,
+      "grad_norm": 0.5147837537823089,
+      "learning_rate": 1.9528489027822784e-07,
+      "loss": 0.6785,
+      "step": 43650
+    },
+    {
+      "epoch": 0.9378356317394854,
+      "grad_norm": 0.5238895174683065,
+      "learning_rate": 1.9394662561648458e-07,
+      "loss": 0.6855,
+      "step": 43660
+    },
+    {
+      "epoch": 0.938050436052756,
+      "grad_norm": 0.5038680839138875,
+      "learning_rate": 1.926129173171165e-07,
+      "loss": 0.6833,
+      "step": 43670
+    },
+    {
+      "epoch": 0.9382652403660265,
+      "grad_norm": 0.510424924822099,
+      "learning_rate": 1.912837659998268e-07,
+      "loss": 0.6974,
+      "step": 43680
+    },
+    {
+      "epoch": 0.9384800446792971,
+      "grad_norm": 0.5039561381967255,
+      "learning_rate": 1.8995917228220473e-07,
+      "loss": 0.6938,
+      "step": 43690
+    },
+    {
+      "epoch": 0.9386948489925677,
+      "grad_norm": 0.5135618071345601,
+      "learning_rate": 1.8863913677972023e-07,
+      "loss": 0.6808,
+      "step": 43700
+    },
+    {
+      "epoch": 0.9389096533058384,
+      "grad_norm": 0.517271399611518,
+      "learning_rate": 1.8732366010572268e-07,
+      "loss": 0.673,
+      "step": 43710
+    },
+    {
+      "epoch": 0.939124457619109,
+      "grad_norm": 0.5125943059284018,
+      "learning_rate": 1.8601274287144865e-07,
+      "loss": 0.6988,
+      "step": 43720
+    },
+    {
+      "epoch": 0.9393392619323796,
+      "grad_norm": 0.5196803747985291,
+      "learning_rate": 1.8470638568601096e-07,
+      "loss": 0.6921,
+      "step": 43730
+    },
+    {
+      "epoch": 0.9395540662456502,
+      "grad_norm": 0.49370739380736406,
+      "learning_rate": 1.8340458915640623e-07,
+      "loss": 0.6917,
+      "step": 43740
+    },
+    {
+      "epoch": 0.9397688705589208,
+      "grad_norm": 0.5091442466929231,
+      "learning_rate": 1.8210735388751177e-07,
+      "loss": 0.684,
+      "step": 43750
+    },
+    {
+      "epoch": 0.9399836748721915,
+      "grad_norm": 0.5047518312454556,
+      "learning_rate": 1.8081468048208539e-07,
+      "loss": 0.6877,
+      "step": 43760
+    },
+    {
+      "epoch": 0.9401984791854621,
+      "grad_norm": 0.5206191020901989,
+      "learning_rate": 1.7952656954076443e-07,
+      "loss": 0.7039,
+      "step": 43770
+    },
+    {
+      "epoch": 0.9404132834987327,
+      "grad_norm": 0.5123513333596542,
+      "learning_rate": 1.7824302166206786e-07,
+      "loss": 0.683,
+      "step": 43780
+    },
+    {
+      "epoch": 0.9406280878120032,
+      "grad_norm": 0.5064144872086758,
+      "learning_rate": 1.7696403744239422e-07,
+      "loss": 0.682,
+      "step": 43790
+    },
+    {
+      "epoch": 0.9408428921252738,
+      "grad_norm": 0.5332117626196589,
+      "learning_rate": 1.7568961747601808e-07,
+      "loss": 0.6859,
+      "step": 43800
+    },
+    {
+      "epoch": 0.9410576964385445,
+      "grad_norm": 0.500584413978133,
+      "learning_rate": 1.7441976235509917e-07,
+      "loss": 0.6863,
+      "step": 43810
+    },
+    {
+      "epoch": 0.9412725007518151,
+      "grad_norm": 0.4941849890778416,
+      "learning_rate": 1.731544726696699e-07,
+      "loss": 0.7026,
+      "step": 43820
+    },
+    {
+      "epoch": 0.9414873050650857,
+      "grad_norm": 0.5196772492638178,
+      "learning_rate": 1.7189374900764776e-07,
+      "loss": 0.6894,
+      "step": 43830
+    },
+    {
+      "epoch": 0.9417021093783563,
+      "grad_norm": 0.5164400053424427,
+      "learning_rate": 1.706375919548231e-07,
+      "loss": 0.6822,
+      "step": 43840
+    },
+    {
+      "epoch": 0.9419169136916269,
+      "grad_norm": 0.53777605471878,
+      "learning_rate": 1.6938600209486678e-07,
+      "loss": 0.6878,
+      "step": 43850
+    },
+    {
+      "epoch": 0.9421317180048976,
+      "grad_norm": 0.5217587662345642,
+      "learning_rate": 1.6813898000933025e-07,
+      "loss": 0.6864,
+      "step": 43860
+    },
+    {
+      "epoch": 0.9423465223181682,
+      "grad_norm": 0.5253001291239328,
+      "learning_rate": 1.6689652627763563e-07,
+      "loss": 0.6871,
+      "step": 43870
+    },
+    {
+      "epoch": 0.9425613266314388,
+      "grad_norm": 0.5145763066504357,
+      "learning_rate": 1.6565864147708888e-07,
+      "loss": 0.6927,
+      "step": 43880
+    },
+    {
+      "epoch": 0.9427761309447094,
+      "grad_norm": 0.5166068562999455,
+      "learning_rate": 1.6442532618286987e-07,
+      "loss": 0.6903,
+      "step": 43890
+    },
+    {
+      "epoch": 0.94299093525798,
+      "grad_norm": 0.5020503378401979,
+      "learning_rate": 1.631965809680369e-07,
+      "loss": 0.6952,
+      "step": 43900
+    },
+    {
+      "epoch": 0.9432057395712506,
+      "grad_norm": 0.5118081944837007,
+      "learning_rate": 1.619724064035233e-07,
+      "loss": 0.6976,
+      "step": 43910
+    },
+    {
+      "epoch": 0.9434205438845212,
+      "grad_norm": 0.4971652889934386,
+      "learning_rate": 1.6075280305813845e-07,
+      "loss": 0.6816,
+      "step": 43920
+    },
+    {
+      "epoch": 0.9436353481977918,
+      "grad_norm": 0.5322834743237593,
+      "learning_rate": 1.5953777149857018e-07,
+      "loss": 0.6843,
+      "step": 43930
+    },
+    {
+      "epoch": 0.9438501525110624,
+      "grad_norm": 0.5255977281637059,
+      "learning_rate": 1.5832731228937915e-07,
+      "loss": 0.6849,
+      "step": 43940
+    },
+    {
+      "epoch": 0.944064956824333,
+      "grad_norm": 0.5143738059522022,
+      "learning_rate": 1.5712142599300206e-07,
+      "loss": 0.6919,
+      "step": 43950
+    },
+    {
+      "epoch": 0.9442797611376036,
+      "grad_norm": 0.5106099594288762,
+      "learning_rate": 1.5592011316975297e-07,
+      "loss": 0.6896,
+      "step": 43960
+    },
+    {
+      "epoch": 0.9444945654508743,
+      "grad_norm": 0.4995085706778138,
+      "learning_rate": 1.5472337437781982e-07,
+      "loss": 0.6837,
+      "step": 43970
+    },
+    {
+      "epoch": 0.9447093697641449,
+      "grad_norm": 0.5261391145616446,
+      "learning_rate": 1.5353121017326334e-07,
+      "loss": 0.674,
+      "step": 43980
+    },
+    {
+      "epoch": 0.9449241740774155,
+      "grad_norm": 0.5230909604182425,
+      "learning_rate": 1.5234362111002043e-07,
+      "loss": 0.6711,
+      "step": 43990
+    },
+    {
+      "epoch": 0.9451389783906861,
+      "grad_norm": 0.5062065384872058,
+      "learning_rate": 1.5116060773990304e-07,
+      "loss": 0.6824,
+      "step": 44000
+    },
+    {
+      "epoch": 0.9453537827039566,
+      "grad_norm": 0.533824670779333,
+      "learning_rate": 1.499821706125937e-07,
+      "loss": 0.6932,
+      "step": 44010
+    },
+    {
+      "epoch": 0.9455685870172273,
+      "grad_norm": 0.5288369825844738,
+      "learning_rate": 1.4880831027565213e-07,
+      "loss": 0.6815,
+      "step": 44020
+    },
+    {
+      "epoch": 0.9457833913304979,
+      "grad_norm": 0.5279056326291518,
+      "learning_rate": 1.4763902727451206e-07,
+      "loss": 0.6859,
+      "step": 44030
+    },
+    {
+      "epoch": 0.9459981956437685,
+      "grad_norm": 0.5141616425492499,
+      "learning_rate": 1.4647432215247448e-07,
+      "loss": 0.688,
+      "step": 44040
+    },
+    {
+      "epoch": 0.9462129999570391,
+      "grad_norm": 0.506499141279277,
+      "learning_rate": 1.4531419545071866e-07,
+      "loss": 0.6842,
+      "step": 44050
+    },
+    {
+      "epoch": 0.9464278042703097,
+      "grad_norm": 0.5137357443604906,
+      "learning_rate": 1.441586477082968e-07,
+      "loss": 0.7037,
+      "step": 44060
+    },
+    {
+      "epoch": 0.9466426085835804,
+      "grad_norm": 0.5179057297643055,
+      "learning_rate": 1.4300767946213046e-07,
+      "loss": 0.6836,
+      "step": 44070
+    },
+    {
+      "epoch": 0.946857412896851,
+      "grad_norm": 0.5062004724754988,
+      "learning_rate": 1.4186129124701408e-07,
+      "loss": 0.6934,
+      "step": 44080
+    },
+    {
+      "epoch": 0.9470722172101216,
+      "grad_norm": 0.516759003331539,
+      "learning_rate": 1.4071948359561605e-07,
+      "loss": 0.6876,
+      "step": 44090
+    },
+    {
+      "epoch": 0.9472870215233922,
+      "grad_norm": 0.5121867129971354,
+      "learning_rate": 1.3958225703847305e-07,
+      "loss": 0.6866,
+      "step": 44100
+    },
+    {
+      "epoch": 0.9475018258366628,
+      "grad_norm": 0.5029706915206402,
+      "learning_rate": 1.3844961210399576e-07,
+      "loss": 0.6829,
+      "step": 44110
+    },
+    {
+      "epoch": 0.9477166301499335,
+      "grad_norm": 0.5223591095442727,
+      "learning_rate": 1.3732154931846652e-07,
+      "loss": 0.6899,
+      "step": 44120
+    },
+    {
+      "epoch": 0.947931434463204,
+      "grad_norm": 0.5029921805410988,
+      "learning_rate": 1.3619806920603608e-07,
+      "loss": 0.6945,
+      "step": 44130
+    },
+    {
+      "epoch": 0.9481462387764746,
+      "grad_norm": 0.4862859723168768,
+      "learning_rate": 1.350791722887279e-07,
+      "loss": 0.6706,
+      "step": 44140
+    },
+    {
+      "epoch": 0.9483610430897452,
+      "grad_norm": 0.4934171630865501,
+      "learning_rate": 1.3396485908643398e-07,
+      "loss": 0.6844,
+      "step": 44150
+    },
+    {
+      "epoch": 0.9485758474030158,
+      "grad_norm": 0.4991882715586846,
+      "learning_rate": 1.3285513011691898e-07,
+      "loss": 0.6998,
+      "step": 44160
+    },
+    {
+      "epoch": 0.9487906517162865,
+      "grad_norm": 0.5065887293501903,
+      "learning_rate": 1.317499858958149e-07,
+      "loss": 0.6983,
+      "step": 44170
+    },
+    {
+      "epoch": 0.9490054560295571,
+      "grad_norm": 0.5217077228947482,
+      "learning_rate": 1.306494269366254e-07,
+      "loss": 0.6968,
+      "step": 44180
+    },
+    {
+      "epoch": 0.9492202603428277,
+      "grad_norm": 0.5284986091713506,
+      "learning_rate": 1.295534537507237e-07,
+      "loss": 0.6784,
+      "step": 44190
+    },
+    {
+      "epoch": 0.9494350646560983,
+      "grad_norm": 0.5007427227768886,
+      "learning_rate": 1.284620668473502e-07,
+      "loss": 0.6892,
+      "step": 44200
+    },
+    {
+      "epoch": 0.9496498689693689,
+      "grad_norm": 0.5364368649236175,
+      "learning_rate": 1.273752667336159e-07,
+      "loss": 0.7067,
+      "step": 44210
+    },
+    {
+      "epoch": 0.9498646732826395,
+      "grad_norm": 0.5149978301773179,
+      "learning_rate": 1.262930539145013e-07,
+      "loss": 0.6788,
+      "step": 44220
+    },
+    {
+      "epoch": 0.9500794775959102,
+      "grad_norm": 0.48810830703974617,
+      "learning_rate": 1.25215428892852e-07,
+      "loss": 0.6894,
+      "step": 44230
+    },
+    {
+      "epoch": 0.9502942819091807,
+      "grad_norm": 0.5094043254510362,
+      "learning_rate": 1.2414239216938628e-07,
+      "loss": 0.6734,
+      "step": 44240
+    },
+    {
+      "epoch": 0.9505090862224513,
+      "grad_norm": 0.5180377313222472,
+      "learning_rate": 1.2307394424268758e-07,
+      "loss": 0.6876,
+      "step": 44250
+    },
+    {
+      "epoch": 0.9507238905357219,
+      "grad_norm": 0.5003767010301355,
+      "learning_rate": 1.220100856092088e-07,
+      "loss": 0.6943,
+      "step": 44260
+    },
+    {
+      "epoch": 0.9509386948489925,
+      "grad_norm": 0.5036430920842442,
+      "learning_rate": 1.2095081676326669e-07,
+      "loss": 0.6908,
+      "step": 44270
+    },
+    {
+      "epoch": 0.9511534991622632,
+      "grad_norm": 0.5045971916714835,
+      "learning_rate": 1.1989613819705314e-07,
+      "loss": 0.6761,
+      "step": 44280
+    },
+    {
+      "epoch": 0.9513683034755338,
+      "grad_norm": 0.538898900267551,
+      "learning_rate": 1.1884605040061947e-07,
+      "loss": 0.6907,
+      "step": 44290
+    },
+    {
+      "epoch": 0.9515831077888044,
+      "grad_norm": 0.5035666702349213,
+      "learning_rate": 1.178005538618865e-07,
+      "loss": 0.6878,
+      "step": 44300
+    },
+    {
+      "epoch": 0.951797912102075,
+      "grad_norm": 0.5149697763083855,
+      "learning_rate": 1.167596490666445e-07,
+      "loss": 0.6937,
+      "step": 44310
+    },
+    {
+      "epoch": 0.9520127164153456,
+      "grad_norm": 0.5107879059485395,
+      "learning_rate": 1.1572333649854328e-07,
+      "loss": 0.6783,
+      "step": 44320
+    },
+    {
+      "epoch": 0.9522275207286163,
+      "grad_norm": 0.5232337790057829,
+      "learning_rate": 1.1469161663910877e-07,
+      "loss": 0.6997,
+      "step": 44330
+    },
+    {
+      "epoch": 0.9524423250418869,
+      "grad_norm": 0.5034816506492367,
+      "learning_rate": 1.1366448996772195e-07,
+      "loss": 0.6856,
+      "step": 44340
+    },
+    {
+      "epoch": 0.9526571293551575,
+      "grad_norm": 0.524026297949951,
+      "learning_rate": 1.1264195696163993e-07,
+      "loss": 0.6832,
+      "step": 44350
+    },
+    {
+      "epoch": 0.952871933668428,
+      "grad_norm": 0.5103648732522968,
+      "learning_rate": 1.1162401809597822e-07,
+      "loss": 0.6933,
+      "step": 44360
+    },
+    {
+      "epoch": 0.9530867379816986,
+      "grad_norm": 0.5041210092671914,
+      "learning_rate": 1.106106738437196e-07,
+      "loss": 0.6835,
+      "step": 44370
+    },
+    {
+      "epoch": 0.9533015422949693,
+      "grad_norm": 0.505579894206394,
+      "learning_rate": 1.0960192467571407e-07,
+      "loss": 0.6848,
+      "step": 44380
+    },
+    {
+      "epoch": 0.9535163466082399,
+      "grad_norm": 0.48925078556562024,
+      "learning_rate": 1.0859777106067226e-07,
+      "loss": 0.6867,
+      "step": 44390
+    },
+    {
+      "epoch": 0.9537311509215105,
+      "grad_norm": 0.510694819304209,
+      "learning_rate": 1.0759821346517541e-07,
+      "loss": 0.6942,
+      "step": 44400
+    },
+    {
+      "epoch": 0.9539459552347811,
+      "grad_norm": 0.5020801402247028,
+      "learning_rate": 1.0660325235366309e-07,
+      "loss": 0.6752,
+      "step": 44410
+    },
+    {
+      "epoch": 0.9541607595480517,
+      "grad_norm": 0.514838466266744,
+      "learning_rate": 1.0561288818844217e-07,
+      "loss": 0.6812,
+      "step": 44420
+    },
+    {
+      "epoch": 0.9543755638613224,
+      "grad_norm": 0.5136075576538729,
+      "learning_rate": 1.0462712142968567e-07,
+      "loss": 0.6984,
+      "step": 44430
+    },
+    {
+      "epoch": 0.954590368174593,
+      "grad_norm": 0.5373229072892426,
+      "learning_rate": 1.0364595253542498e-07,
+      "loss": 0.6844,
+      "step": 44440
+    },
+    {
+      "epoch": 0.9548051724878636,
+      "grad_norm": 0.5177203696614198,
+      "learning_rate": 1.0266938196156096e-07,
+      "loss": 0.6842,
+      "step": 44450
+    },
+    {
+      "epoch": 0.9550199768011342,
+      "grad_norm": 0.5304130297606695,
+      "learning_rate": 1.0169741016185286e-07,
+      "loss": 0.698,
+      "step": 44460
+    },
+    {
+      "epoch": 0.9552347811144047,
+      "grad_norm": 0.521535821590921,
+      "learning_rate": 1.0073003758792721e-07,
+      "loss": 0.6791,
+      "step": 44470
+    },
+    {
+      "epoch": 0.9554495854276754,
+      "grad_norm": 0.5216014836223208,
+      "learning_rate": 9.976726468927112e-08,
+      "loss": 0.6875,
+      "step": 44480
+    },
+    {
+      "epoch": 0.955664389740946,
+      "grad_norm": 0.4958526215570128,
+      "learning_rate": 9.88090919132334e-08,
+      "loss": 0.6845,
+      "step": 44490
+    },
+    {
+      "epoch": 0.9558791940542166,
+      "grad_norm": 0.491068148568293,
+      "learning_rate": 9.785551970502904e-08,
+      "loss": 0.6901,
+      "step": 44500
+    },
+    {
+      "epoch": 0.9560939983674872,
+      "grad_norm": 0.5334092820305429,
+      "learning_rate": 9.690654850773251e-08,
+      "loss": 0.6981,
+      "step": 44510
+    },
+    {
+      "epoch": 0.9563088026807578,
+      "grad_norm": 0.5455356568529366,
+      "learning_rate": 9.596217876228109e-08,
+      "loss": 0.6899,
+      "step": 44520
+    },
+    {
+      "epoch": 0.9565236069940284,
+      "grad_norm": 0.5233272097407236,
+      "learning_rate": 9.502241090747488e-08,
+      "loss": 0.6965,
+      "step": 44530
+    },
+    {
+      "epoch": 0.9567384113072991,
+      "grad_norm": 0.5221024613165127,
+      "learning_rate": 9.40872453799746e-08,
+      "loss": 0.6846,
+      "step": 44540
+    },
+    {
+      "epoch": 0.9569532156205697,
+      "grad_norm": 0.5392768910456224,
+      "learning_rate": 9.315668261430378e-08,
+      "loss": 0.6956,
+      "step": 44550
+    },
+    {
+      "epoch": 0.9571680199338403,
+      "grad_norm": 0.5040996409629596,
+      "learning_rate": 9.223072304284542e-08,
+      "loss": 0.6804,
+      "step": 44560
+    },
+    {
+      "epoch": 0.9573828242471109,
+      "grad_norm": 0.5214855537136885,
+      "learning_rate": 9.130936709584537e-08,
+      "loss": 0.7017,
+      "step": 44570
+    },
+    {
+      "epoch": 0.9575976285603814,
+      "grad_norm": 0.5117354205252992,
+      "learning_rate": 9.039261520141008e-08,
+      "loss": 0.6884,
+      "step": 44580
+    },
+    {
+      "epoch": 0.9578124328736521,
+      "grad_norm": 0.5130259202484456,
+      "learning_rate": 8.948046778550546e-08,
+      "loss": 0.6864,
+      "step": 44590
+    },
+    {
+      "epoch": 0.9580272371869227,
+      "grad_norm": 0.5153683100477022,
+      "learning_rate": 8.857292527195916e-08,
+      "loss": 0.6873,
+      "step": 44600
+    },
+    {
+      "epoch": 0.9582420415001933,
+      "grad_norm": 0.52407348167293,
+      "learning_rate": 8.76699880824583e-08,
+      "loss": 0.6873,
+      "step": 44610
+    },
+    {
+      "epoch": 0.9584568458134639,
+      "grad_norm": 0.5284487657831876,
+      "learning_rate": 8.677165663655396e-08,
+      "loss": 0.6678,
+      "step": 44620
+    },
+    {
+      "epoch": 0.9586716501267345,
+      "grad_norm": 0.5136375221731139,
+      "learning_rate": 8.587793135165001e-08,
+      "loss": 0.6918,
+      "step": 44630
+    },
+    {
+      "epoch": 0.9588864544400052,
+      "grad_norm": 0.5376228954573528,
+      "learning_rate": 8.49888126430154e-08,
+      "loss": 0.6918,
+      "step": 44640
+    },
+    {
+      "epoch": 0.9591012587532758,
+      "grad_norm": 0.5102793357805696,
+      "learning_rate": 8.410430092377853e-08,
+      "loss": 0.6878,
+      "step": 44650
+    },
+    {
+      "epoch": 0.9593160630665464,
+      "grad_norm": 0.5191445781226015,
+      "learning_rate": 8.322439660492398e-08,
+      "loss": 0.6882,
+      "step": 44660
+    },
+    {
+      "epoch": 0.959530867379817,
+      "grad_norm": 0.5030540214464223,
+      "learning_rate": 8.234910009529917e-08,
+      "loss": 0.6846,
+      "step": 44670
+    },
+    {
+      "epoch": 0.9597456716930876,
+      "grad_norm": 0.5221867031193459,
+      "learning_rate": 8.147841180160765e-08,
+      "loss": 0.6923,
+      "step": 44680
+    },
+    {
+      "epoch": 0.9599604760063583,
+      "grad_norm": 0.49462846118517095,
+      "learning_rate": 8.061233212841358e-08,
+      "loss": 0.6725,
+      "step": 44690
+    },
+    {
+      "epoch": 0.9601752803196288,
+      "grad_norm": 0.5156449438921737,
+      "learning_rate": 7.975086147813837e-08,
+      "loss": 0.6805,
+      "step": 44700
+    },
+    {
+      "epoch": 0.9603900846328994,
+      "grad_norm": 0.5045341788319067,
+      "learning_rate": 7.889400025106409e-08,
+      "loss": 0.6827,
+      "step": 44710
+    },
+    {
+      "epoch": 0.96060488894617,
+      "grad_norm": 0.5165527445661976,
+      "learning_rate": 7.804174884532778e-08,
+      "loss": 0.6875,
+      "step": 44720
+    },
+    {
+      "epoch": 0.9608196932594406,
+      "grad_norm": 0.5290209182103901,
+      "learning_rate": 7.719410765692825e-08,
+      "loss": 0.6818,
+      "step": 44730
+    },
+    {
+      "epoch": 0.9610344975727113,
+      "grad_norm": 0.5045124524630176,
+      "learning_rate": 7.635107707971712e-08,
+      "loss": 0.683,
+      "step": 44740
+    },
+    {
+      "epoch": 0.9612493018859819,
+      "grad_norm": 0.5110671930296831,
+      "learning_rate": 7.551265750540993e-08,
+      "loss": 0.6716,
+      "step": 44750
+    },
+    {
+      "epoch": 0.9614641061992525,
+      "grad_norm": 0.5184014503506628,
+      "learning_rate": 7.467884932357505e-08,
+      "loss": 0.6953,
+      "step": 44760
+    },
+    {
+      "epoch": 0.9616789105125231,
+      "grad_norm": 0.5025771257461924,
+      "learning_rate": 7.384965292164037e-08,
+      "loss": 0.6875,
+      "step": 44770
+    },
+    {
+      "epoch": 0.9618937148257937,
+      "grad_norm": 0.5106235992412884,
+      "learning_rate": 7.30250686848888e-08,
+      "loss": 0.6924,
+      "step": 44780
+    },
+    {
+      "epoch": 0.9621085191390644,
+      "grad_norm": 0.5139107082255027,
+      "learning_rate": 7.220509699646383e-08,
+      "loss": 0.7049,
+      "step": 44790
+    },
+    {
+      "epoch": 0.962323323452335,
+      "grad_norm": 0.5025433150135805,
+      "learning_rate": 7.138973823736295e-08,
+      "loss": 0.6851,
+      "step": 44800
+    },
+    {
+      "epoch": 0.9625381277656055,
+      "grad_norm": 0.5025163153971417,
+      "learning_rate": 7.057899278643975e-08,
+      "loss": 0.6925,
+      "step": 44810
+    },
+    {
+      "epoch": 0.9627529320788761,
+      "grad_norm": 0.5101640172429046,
+      "learning_rate": 6.977286102040625e-08,
+      "loss": 0.6688,
+      "step": 44820
+    },
+    {
+      "epoch": 0.9629677363921467,
+      "grad_norm": 0.5162577234774317,
+      "learning_rate": 6.89713433138306e-08,
+      "loss": 0.683,
+      "step": 44830
+    },
+    {
+      "epoch": 0.9631825407054173,
+      "grad_norm": 0.5102954616732902,
+      "learning_rate": 6.817444003913487e-08,
+      "loss": 0.6883,
+      "step": 44840
+    },
+    {
+      "epoch": 0.963397345018688,
+      "grad_norm": 0.5156868683615887,
+      "learning_rate": 6.738215156659955e-08,
+      "loss": 0.671,
+      "step": 44850
+    },
+    {
+      "epoch": 0.9636121493319586,
+      "grad_norm": 0.514482173761437,
+      "learning_rate": 6.659447826436017e-08,
+      "loss": 0.6849,
+      "step": 44860
+    },
+    {
+      "epoch": 0.9638269536452292,
+      "grad_norm": 0.5251149204765465,
+      "learning_rate": 6.581142049840616e-08,
+      "loss": 0.6903,
+      "step": 44870
+    },
+    {
+      "epoch": 0.9640417579584998,
+      "grad_norm": 0.4910026180097444,
+      "learning_rate": 6.503297863258429e-08,
+      "loss": 0.6878,
+      "step": 44880
+    },
+    {
+      "epoch": 0.9642565622717704,
+      "grad_norm": 0.5436572691315237,
+      "learning_rate": 6.425915302859631e-08,
+      "loss": 0.6933,
+      "step": 44890
+    },
+    {
+      "epoch": 0.9644713665850411,
+      "grad_norm": 0.4946026585653437,
+      "learning_rate": 6.348994404599907e-08,
+      "loss": 0.6943,
+      "step": 44900
+    },
+    {
+      "epoch": 0.9646861708983117,
+      "grad_norm": 0.5030459362111663,
+      "learning_rate": 6.27253520422022e-08,
+      "loss": 0.6879,
+      "step": 44910
+    },
+    {
+      "epoch": 0.9649009752115822,
+      "grad_norm": 0.5093527545070409,
+      "learning_rate": 6.196537737247488e-08,
+      "loss": 0.6776,
+      "step": 44920
+    },
+    {
+      "epoch": 0.9651157795248528,
+      "grad_norm": 0.534202285869032,
+      "learning_rate": 6.12100203899335e-08,
+      "loss": 0.7175,
+      "step": 44930
+    },
+    {
+      "epoch": 0.9653305838381234,
+      "grad_norm": 0.5112645845888609,
+      "learning_rate": 6.045928144555736e-08,
+      "loss": 0.6764,
+      "step": 44940
+    },
+    {
+      "epoch": 0.9655453881513941,
+      "grad_norm": 0.5085395742615043,
+      "learning_rate": 5.97131608881718e-08,
+      "loss": 0.6778,
+      "step": 44950
+    },
+    {
+      "epoch": 0.9657601924646647,
+      "grad_norm": 0.5335040181019025,
+      "learning_rate": 5.8971659064464006e-08,
+      "loss": 0.6836,
+      "step": 44960
+    },
+    {
+      "epoch": 0.9659749967779353,
+      "grad_norm": 0.5132355793000835,
+      "learning_rate": 5.823477631896945e-08,
+      "loss": 0.6993,
+      "step": 44970
+    },
+    {
+      "epoch": 0.9661898010912059,
+      "grad_norm": 0.5136374270218453,
+      "learning_rate": 5.750251299407761e-08,
+      "loss": 0.6885,
+      "step": 44980
+    },
+    {
+      "epoch": 0.9664046054044765,
+      "grad_norm": 0.5109545044167216,
+      "learning_rate": 5.6774869430036296e-08,
+      "loss": 0.6972,
+      "step": 44990
+    },
+    {
+      "epoch": 0.9666194097177472,
+      "grad_norm": 0.5188694431833389,
+      "learning_rate": 5.605184596494062e-08,
+      "loss": 0.6786,
+      "step": 45000
+    },
+    {
+      "epoch": 0.9668342140310178,
+      "grad_norm": 0.4958762948588162,
+      "learning_rate": 5.5333442934744074e-08,
+      "loss": 0.6846,
+      "step": 45010
+    },
+    {
+      "epoch": 0.9670490183442884,
+      "grad_norm": 0.5273914852077115,
+      "learning_rate": 5.461966067325075e-08,
+      "loss": 0.6889,
+      "step": 45020
+    },
+    {
+      "epoch": 0.967263822657559,
+      "grad_norm": 0.5090438225495914,
+      "learning_rate": 5.3910499512116465e-08,
+      "loss": 0.6946,
+      "step": 45030
+    },
+    {
+      "epoch": 0.9674786269708295,
+      "grad_norm": 0.5283538397180535,
+      "learning_rate": 5.32059597808543e-08,
+      "loss": 0.6821,
+      "step": 45040
+    },
+    {
+      "epoch": 0.9676934312841002,
+      "grad_norm": 0.5129695644458037,
+      "learning_rate": 5.2506041806824614e-08,
+      "loss": 0.6836,
+      "step": 45050
+    },
+    {
+      "epoch": 0.9679082355973708,
+      "grad_norm": 0.5357563822720558,
+      "learning_rate": 5.181074591524393e-08,
+      "loss": 0.6875,
+      "step": 45060
+    },
+    {
+      "epoch": 0.9681230399106414,
+      "grad_norm": 0.9871280435330488,
+      "learning_rate": 5.112007242918049e-08,
+      "loss": 0.6777,
+      "step": 45070
+    },
+    {
+      "epoch": 0.968337844223912,
+      "grad_norm": 0.5259865369064295,
+      "learning_rate": 5.043402166955314e-08,
+      "loss": 0.6878,
+      "step": 45080
+    },
+    {
+      "epoch": 0.9685526485371826,
+      "grad_norm": 0.49858883676118315,
+      "learning_rate": 4.9752593955134655e-08,
+      "loss": 0.6797,
+      "step": 45090
+    },
+    {
+      "epoch": 0.9687674528504533,
+      "grad_norm": 0.5177420266411118,
+      "learning_rate": 4.907578960254955e-08,
+      "loss": 0.696,
+      "step": 45100
+    },
+    {
+      "epoch": 0.9689822571637239,
+      "grad_norm": 0.5112656951608426,
+      "learning_rate": 4.840360892627183e-08,
+      "loss": 0.6812,
+      "step": 45110
+    },
+    {
+      "epoch": 0.9691970614769945,
+      "grad_norm": 0.5048791811705977,
+      "learning_rate": 4.773605223863054e-08,
+      "loss": 0.679,
+      "step": 45120
+    },
+    {
+      "epoch": 0.9694118657902651,
+      "grad_norm": 0.5184010359011654,
+      "learning_rate": 4.707311984980423e-08,
+      "loss": 0.694,
+      "step": 45130
+    },
+    {
+      "epoch": 0.9696266701035356,
+      "grad_norm": 0.526339150759906,
+      "learning_rate": 4.6414812067822056e-08,
+      "loss": 0.6889,
+      "step": 45140
+    },
+    {
+      "epoch": 0.9698414744168062,
+      "grad_norm": 0.5192950033137796,
+      "learning_rate": 4.576112919856601e-08,
+      "loss": 0.6996,
+      "step": 45150
+    },
+    {
+      "epoch": 0.9700562787300769,
+      "grad_norm": 0.5052304888836145,
+      "learning_rate": 4.51120715457698e-08,
+      "loss": 0.6903,
+      "step": 45160
+    },
+    {
+      "epoch": 0.9702710830433475,
+      "grad_norm": 0.5084386454674246,
+      "learning_rate": 4.44676394110144e-08,
+      "loss": 0.693,
+      "step": 45170
+    },
+    {
+      "epoch": 0.9704858873566181,
+      "grad_norm": 0.5136496837941966,
+      "learning_rate": 4.382783309373473e-08,
+      "loss": 0.6925,
+      "step": 45180
+    },
+    {
+      "epoch": 0.9707006916698887,
+      "grad_norm": 0.49863491668395693,
+      "learning_rate": 4.3192652891216326e-08,
+      "loss": 0.6773,
+      "step": 45190
+    },
+    {
+      "epoch": 0.9709154959831593,
+      "grad_norm": 0.5276766002981915,
+      "learning_rate": 4.25620990985931e-08,
+      "loss": 0.7053,
+      "step": 45200
+    },
+    {
+      "epoch": 0.97113030029643,
+      "grad_norm": 0.5299405568061919,
+      "learning_rate": 4.1936172008851806e-08,
+      "loss": 0.6931,
+      "step": 45210
+    },
+    {
+      "epoch": 0.9713451046097006,
+      "grad_norm": 0.5064992378132825,
+      "learning_rate": 4.131487191282757e-08,
+      "loss": 0.6806,
+      "step": 45220
+    },
+    {
+      "epoch": 0.9715599089229712,
+      "grad_norm": 0.5097429755248392,
+      "learning_rate": 4.069819909920614e-08,
+      "loss": 0.6799,
+      "step": 45230
+    },
+    {
+      "epoch": 0.9717747132362418,
+      "grad_norm": 0.4985251877587769,
+      "learning_rate": 4.008615385452275e-08,
+      "loss": 0.6891,
+      "step": 45240
+    },
+    {
+      "epoch": 0.9719895175495123,
+      "grad_norm": 0.5189916494654802,
+      "learning_rate": 3.947873646316325e-08,
+      "loss": 0.6916,
+      "step": 45250
+    },
+    {
+      "epoch": 0.972204321862783,
+      "grad_norm": 0.532252242471192,
+      "learning_rate": 3.88759472073641e-08,
+      "loss": 0.6869,
+      "step": 45260
+    },
+    {
+      "epoch": 0.9724191261760536,
+      "grad_norm": 0.5058615192775043,
+      "learning_rate": 3.827778636720791e-08,
+      "loss": 0.6909,
+      "step": 45270
+    },
+    {
+      "epoch": 0.9726339304893242,
+      "grad_norm": 0.4988140384119279,
+      "learning_rate": 3.7684254220630156e-08,
+      "loss": 0.6735,
+      "step": 45280
+    },
+    {
+      "epoch": 0.9728487348025948,
+      "grad_norm": 0.5031240100440125,
+      "learning_rate": 3.7095351043414665e-08,
+      "loss": 0.6965,
+      "step": 45290
+    },
+    {
+      "epoch": 0.9730635391158654,
+      "grad_norm": 0.5051450638071671,
+      "learning_rate": 3.651107710919366e-08,
+      "loss": 0.6921,
+      "step": 45300
+    },
+    {
+      "epoch": 0.9732783434291361,
+      "grad_norm": 0.50854818588177,
+      "learning_rate": 3.593143268944888e-08,
+      "loss": 0.6766,
+      "step": 45310
+    },
+    {
+      "epoch": 0.9734931477424067,
+      "grad_norm": 0.5209101220326824,
+      "learning_rate": 3.535641805351042e-08,
+      "loss": 0.6828,
+      "step": 45320
+    },
+    {
+      "epoch": 0.9737079520556773,
+      "grad_norm": 0.5185530290222669,
+      "learning_rate": 3.478603346855791e-08,
+      "loss": 0.6793,
+      "step": 45330
+    },
+    {
+      "epoch": 0.9739227563689479,
+      "grad_norm": 0.4981837565896543,
+      "learning_rate": 3.4220279199619346e-08,
+      "loss": 0.6813,
+      "step": 45340
+    },
+    {
+      "epoch": 0.9741375606822185,
+      "grad_norm": 0.6029095886431749,
+      "learning_rate": 3.365915550957222e-08,
+      "loss": 0.6776,
+      "step": 45350
+    },
+    {
+      "epoch": 0.9743523649954892,
+      "grad_norm": 0.529974027801158,
+      "learning_rate": 3.3102662659140195e-08,
+      "loss": 0.6956,
+      "step": 45360
+    },
+    {
+      "epoch": 0.9745671693087598,
+      "grad_norm": 0.5098466744359518,
+      "learning_rate": 3.255080090689644e-08,
+      "loss": 0.6872,
+      "step": 45370
+    },
+    {
+      "epoch": 0.9747819736220303,
+      "grad_norm": 0.5178231192461265,
+      "learning_rate": 3.200357050926361e-08,
+      "loss": 0.6765,
+      "step": 45380
+    },
+    {
+      "epoch": 0.9749967779353009,
+      "grad_norm": 0.507719206604319,
+      "learning_rate": 3.1460971720510544e-08,
+      "loss": 0.691,
+      "step": 45390
+    },
+    {
+      "epoch": 0.9752115822485715,
+      "grad_norm": 0.5012348553384962,
+      "learning_rate": 3.0923004792754454e-08,
+      "loss": 0.6947,
+      "step": 45400
+    },
+    {
+      "epoch": 0.9754263865618421,
+      "grad_norm": 0.5012135896224041,
+      "learning_rate": 3.0389669975959825e-08,
+      "loss": 0.6933,
+      "step": 45410
+    },
+    {
+      "epoch": 0.9756411908751128,
+      "grad_norm": 0.5171752738246117,
+      "learning_rate": 2.9860967517941766e-08,
+      "loss": 0.6924,
+      "step": 45420
+    },
+    {
+      "epoch": 0.9758559951883834,
+      "grad_norm": 0.5209513564127771,
+      "learning_rate": 2.933689766435932e-08,
+      "loss": 0.6994,
+      "step": 45430
+    },
+    {
+      "epoch": 0.976070799501654,
+      "grad_norm": 0.51299608455161,
+      "learning_rate": 2.881746065871993e-08,
+      "loss": 0.6846,
+      "step": 45440
+    },
+    {
+      "epoch": 0.9762856038149246,
+      "grad_norm": 0.5426181859963906,
+      "learning_rate": 2.830265674237942e-08,
+      "loss": 0.6972,
+      "step": 45450
+    },
+    {
+      "epoch": 0.9765004081281952,
+      "grad_norm": 0.5176632310591313,
+      "learning_rate": 2.7792486154540888e-08,
+      "loss": 0.6917,
+      "step": 45460
+    },
+    {
+      "epoch": 0.9767152124414659,
+      "grad_norm": 0.5069734550437351,
+      "learning_rate": 2.7286949132253605e-08,
+      "loss": 0.687,
+      "step": 45470
+    },
+    {
+      "epoch": 0.9769300167547365,
+      "grad_norm": 0.5134159936470577,
+      "learning_rate": 2.6786045910414117e-08,
+      "loss": 0.6829,
+      "step": 45480
+    },
+    {
+      "epoch": 0.977144821068007,
+      "grad_norm": 0.5314226916450205,
+      "learning_rate": 2.6289776721767356e-08,
+      "loss": 0.6875,
+      "step": 45490
+    },
+    {
+      "epoch": 0.9773596253812776,
+      "grad_norm": 0.5017572726059231,
+      "learning_rate": 2.57981417969011e-08,
+      "loss": 0.6891,
+      "step": 45500
+    },
+    {
+      "epoch": 0.9775744296945482,
+      "grad_norm": 0.5097143098042627,
+      "learning_rate": 2.5311141364254832e-08,
+      "loss": 0.6773,
+      "step": 45510
+    },
+    {
+      "epoch": 0.9777892340078189,
+      "grad_norm": 0.5094938311645063,
+      "learning_rate": 2.4828775650111993e-08,
+      "loss": 0.699,
+      "step": 45520
+    },
+    {
+      "epoch": 0.9780040383210895,
+      "grad_norm": 0.5041111422716359,
+      "learning_rate": 2.435104487860218e-08,
+      "loss": 0.689,
+      "step": 45530
+    },
+    {
+      "epoch": 0.9782188426343601,
+      "grad_norm": 0.49440540333290234,
+      "learning_rate": 2.3877949271702283e-08,
+      "loss": 0.6787,
+      "step": 45540
+    },
+    {
+      "epoch": 0.9784336469476307,
+      "grad_norm": 0.5075734688713911,
+      "learning_rate": 2.3409489049235347e-08,
+      "loss": 0.6876,
+      "step": 45550
+    },
+    {
+      "epoch": 0.9786484512609013,
+      "grad_norm": 0.4937175102422703,
+      "learning_rate": 2.2945664428870583e-08,
+      "loss": 0.6766,
+      "step": 45560
+    },
+    {
+      "epoch": 0.978863255574172,
+      "grad_norm": 0.5195853543250676,
+      "learning_rate": 2.2486475626122274e-08,
+      "loss": 0.6893,
+      "step": 45570
+    },
+    {
+      "epoch": 0.9790780598874426,
+      "grad_norm": 0.5066981294356002,
+      "learning_rate": 2.203192285435196e-08,
+      "loss": 0.6866,
+      "step": 45580
+    },
+    {
+      "epoch": 0.9792928642007132,
+      "grad_norm": 0.5106780154162585,
+      "learning_rate": 2.158200632476626e-08,
+      "loss": 0.6806,
+      "step": 45590
+    },
+    {
+      "epoch": 0.9795076685139837,
+      "grad_norm": 0.5216459413562595,
+      "learning_rate": 2.1136726246419048e-08,
+      "loss": 0.6838,
+      "step": 45600
+    },
+    {
+      "epoch": 0.9797224728272543,
+      "grad_norm": 0.5244173549621101,
+      "learning_rate": 2.0696082826209273e-08,
+      "loss": 0.6783,
+      "step": 45610
+    },
+    {
+      "epoch": 0.979937277140525,
+      "grad_norm": 0.508672858772139,
+      "learning_rate": 2.0260076268878716e-08,
+      "loss": 0.6991,
+      "step": 45620
+    },
+    {
+      "epoch": 0.9801520814537956,
+      "grad_norm": 0.5316971841084902,
+      "learning_rate": 1.9828706777017535e-08,
+      "loss": 0.6776,
+      "step": 45630
+    },
+    {
+      "epoch": 0.9803668857670662,
+      "grad_norm": 0.5278618608963893,
+      "learning_rate": 1.9401974551062075e-08,
+      "loss": 0.6895,
+      "step": 45640
+    },
+    {
+      "epoch": 0.9805816900803368,
+      "grad_norm": 0.5178835527189292,
+      "learning_rate": 1.89798797892915e-08,
+      "loss": 0.6877,
+      "step": 45650
+    },
+    {
+      "epoch": 0.9807964943936074,
+      "grad_norm": 0.5107328830474321,
+      "learning_rate": 1.856242268783226e-08,
+      "loss": 0.6806,
+      "step": 45660
+    },
+    {
+      "epoch": 0.9810112987068781,
+      "grad_norm": 0.5178035321000685,
+      "learning_rate": 1.814960344065364e-08,
+      "loss": 0.6884,
+      "step": 45670
+    },
+    {
+      "epoch": 0.9812261030201487,
+      "grad_norm": 0.5209369501476074,
+      "learning_rate": 1.7741422239572203e-08,
+      "loss": 0.7015,
+      "step": 45680
+    },
+    {
+      "epoch": 0.9814409073334193,
+      "grad_norm": 0.5145085242189749,
+      "learning_rate": 1.733787927424735e-08,
+      "loss": 0.692,
+      "step": 45690
+    },
+    {
+      "epoch": 0.9816557116466899,
+      "grad_norm": 0.5135454075520934,
+      "learning_rate": 1.6938974732185754e-08,
+      "loss": 0.6877,
+      "step": 45700
+    },
+    {
+      "epoch": 0.9818705159599604,
+      "grad_norm": 0.5111760775067833,
+      "learning_rate": 1.6544708798736932e-08,
+      "loss": 0.6897,
+      "step": 45710
+    },
+    {
+      "epoch": 0.982085320273231,
+      "grad_norm": 0.5073586927391305,
+      "learning_rate": 1.615508165709545e-08,
+      "loss": 0.6936,
+      "step": 45720
+    },
+    {
+      "epoch": 0.9823001245865017,
+      "grad_norm": 0.5311468459249292,
+      "learning_rate": 1.577009348830205e-08,
+      "loss": 0.7014,
+      "step": 45730
+    },
+    {
+      "epoch": 0.9825149288997723,
+      "grad_norm": 0.5133173645780131,
+      "learning_rate": 1.5389744471238087e-08,
+      "loss": 0.6858,
+      "step": 45740
+    },
+    {
+      "epoch": 0.9827297332130429,
+      "grad_norm": 0.5177357908350075,
+      "learning_rate": 1.5014034782635523e-08,
+      "loss": 0.7051,
+      "step": 45750
+    },
+    {
+      "epoch": 0.9829445375263135,
+      "grad_norm": 0.5334859426374919,
+      "learning_rate": 1.4642964597064713e-08,
+      "loss": 0.691,
+      "step": 45760
+    },
+    {
+      "epoch": 0.9831593418395841,
+      "grad_norm": 0.5104405442693613,
+      "learning_rate": 1.4276534086943295e-08,
+      "loss": 0.6878,
+      "step": 45770
+    },
+    {
+      "epoch": 0.9833741461528548,
+      "grad_norm": 0.5062353640908156,
+      "learning_rate": 1.391474342253174e-08,
+      "loss": 0.6876,
+      "step": 45780
+    },
+    {
+      "epoch": 0.9835889504661254,
+      "grad_norm": 0.5032985620842442,
+      "learning_rate": 1.3557592771935579e-08,
+      "loss": 0.6852,
+      "step": 45790
+    },
+    {
+      "epoch": 0.983803754779396,
+      "grad_norm": 0.5095670609317529,
+      "learning_rate": 1.3205082301105399e-08,
+      "loss": 0.6862,
+      "step": 45800
+    },
+    {
+      "epoch": 0.9840185590926666,
+      "grad_norm": 0.5048150094794633,
+      "learning_rate": 1.2857212173833512e-08,
+      "loss": 0.6619,
+      "step": 45810
+    },
+    {
+      "epoch": 0.9842333634059371,
+      "grad_norm": 0.5367881117018972,
+      "learning_rate": 1.2513982551756176e-08,
+      "loss": 0.6991,
+      "step": 45820
+    },
+    {
+      "epoch": 0.9844481677192078,
+      "grad_norm": 0.5139161341851894,
+      "learning_rate": 1.2175393594355822e-08,
+      "loss": 0.6819,
+      "step": 45830
+    },
+    {
+      "epoch": 0.9846629720324784,
+      "grad_norm": 0.5099451512210587,
+      "learning_rate": 1.1841445458956601e-08,
+      "loss": 0.6815,
+      "step": 45840
+    },
+    {
+      "epoch": 0.984877776345749,
+      "grad_norm": 0.5153723004890068,
+      "learning_rate": 1.1512138300726615e-08,
+      "loss": 0.6928,
+      "step": 45850
+    },
+    {
+      "epoch": 0.9850925806590196,
+      "grad_norm": 0.5169126964230165,
+      "learning_rate": 1.1187472272677913e-08,
+      "loss": 0.6861,
+      "step": 45860
+    },
+    {
+      "epoch": 0.9853073849722902,
+      "grad_norm": 0.5111747109332822,
+      "learning_rate": 1.0867447525665376e-08,
+      "loss": 0.6846,
+      "step": 45870
+    },
+    {
+      "epoch": 0.9855221892855609,
+      "grad_norm": 0.5237477989382217,
+      "learning_rate": 1.0552064208388946e-08,
+      "loss": 0.6892,
+      "step": 45880
+    },
+    {
+      "epoch": 0.9857369935988315,
+      "grad_norm": 0.5252395338889467,
+      "learning_rate": 1.0241322467389181e-08,
+      "loss": 0.7028,
+      "step": 45890
+    },
+    {
+      "epoch": 0.9859517979121021,
+      "grad_norm": 0.5050689116075816,
+      "learning_rate": 9.935222447052805e-09,
+      "loss": 0.6945,
+      "step": 45900
+    },
+    {
+      "epoch": 0.9861666022253727,
+      "grad_norm": 0.5177719740515985,
+      "learning_rate": 9.633764289608272e-09,
+      "loss": 0.6848,
+      "step": 45910
+    },
+    {
+      "epoch": 0.9863814065386433,
+      "grad_norm": 0.5082442912081105,
+      "learning_rate": 9.336948135127976e-09,
+      "loss": 0.6884,
+      "step": 45920
+    },
+    {
+      "epoch": 0.986596210851914,
+      "grad_norm": 0.5106716496528746,
+      "learning_rate": 9.044774121526045e-09,
+      "loss": 0.6841,
+      "step": 45930
+    },
+    {
+      "epoch": 0.9868110151651845,
+      "grad_norm": 0.5203087673402492,
+      "learning_rate": 8.75724238456166e-09,
+      "loss": 0.683,
+      "step": 45940
+    },
+    {
+      "epoch": 0.9870258194784551,
+      "grad_norm": 0.5157554968748806,
+      "learning_rate": 8.474353057834616e-09,
+      "loss": 0.6789,
+      "step": 45950
+    },
+    {
+      "epoch": 0.9872406237917257,
+      "grad_norm": 0.5085732376060498,
+      "learning_rate": 8.196106272789772e-09,
+      "loss": 0.6825,
+      "step": 45960
+    },
+    {
+      "epoch": 0.9874554281049963,
+      "grad_norm": 0.5199228182558453,
+      "learning_rate": 7.922502158713708e-09,
+      "loss": 0.6964,
+      "step": 45970
+    },
+    {
+      "epoch": 0.987670232418267,
+      "grad_norm": 0.5170481222004207,
+      "learning_rate": 7.65354084273473e-09,
+      "loss": 0.6866,
+      "step": 45980
+    },
+    {
+      "epoch": 0.9878850367315376,
+      "grad_norm": 0.5057263382088416,
+      "learning_rate": 7.389222449827316e-09,
+      "loss": 0.6828,
+      "step": 45990
+    },
+    {
+      "epoch": 0.9880998410448082,
+      "grad_norm": 0.5189108042110895,
+      "learning_rate": 7.1295471028054455e-09,
+      "loss": 0.6798,
+      "step": 46000
+    },
+    {
+      "epoch": 0.9883146453580788,
+      "grad_norm": 0.517177846632246,
+      "learning_rate": 6.874514922325937e-09,
+      "loss": 0.6847,
+      "step": 46010
+    },
+    {
+      "epoch": 0.9885294496713494,
+      "grad_norm": 0.5334596524865981,
+      "learning_rate": 6.624126026890665e-09,
+      "loss": 0.7139,
+      "step": 46020
+    },
+    {
+      "epoch": 0.98874425398462,
+      "grad_norm": 0.5017669931684683,
+      "learning_rate": 6.378380532839901e-09,
+      "loss": 0.6784,
+      "step": 46030
+    },
+    {
+      "epoch": 0.9889590582978907,
+      "grad_norm": 0.5157157762964487,
+      "learning_rate": 6.137278554361192e-09,
+      "loss": 0.6891,
+      "step": 46040
+    },
+    {
+      "epoch": 0.9891738626111612,
+      "grad_norm": 0.5324764096551247,
+      "learning_rate": 5.900820203481594e-09,
+      "loss": 0.683,
+      "step": 46050
+    },
+    {
+      "epoch": 0.9893886669244318,
+      "grad_norm": 0.5040931566397142,
+      "learning_rate": 5.669005590069887e-09,
+      "loss": 0.6853,
+      "step": 46060
+    },
+    {
+      "epoch": 0.9896034712377024,
+      "grad_norm": 0.5144642042437939,
+      "learning_rate": 5.4418348218387985e-09,
+      "loss": 0.6855,
+      "step": 46070
+    },
+    {
+      "epoch": 0.989818275550973,
+      "grad_norm": 0.5086245276544579,
+      "learning_rate": 5.219308004343893e-09,
+      "loss": 0.6973,
+      "step": 46080
+    },
+    {
+      "epoch": 0.9900330798642437,
+      "grad_norm": 0.5106173051210142,
+      "learning_rate": 5.00142524097913e-09,
+      "loss": 0.6809,
+      "step": 46090
+    },
+    {
+      "epoch": 0.9902478841775143,
+      "grad_norm": 0.5044758804491363,
+      "learning_rate": 4.788186632985747e-09,
+      "loss": 0.6866,
+      "step": 46100
+    },
+    {
+      "epoch": 0.9904626884907849,
+      "grad_norm": 0.5011346072824133,
+      "learning_rate": 4.579592279444489e-09,
+      "loss": 0.6773,
+      "step": 46110
+    },
+    {
+      "epoch": 0.9906774928040555,
+      "grad_norm": 0.5104576978811137,
+      "learning_rate": 4.375642277276715e-09,
+      "loss": 0.6896,
+      "step": 46120
+    },
+    {
+      "epoch": 0.9908922971173261,
+      "grad_norm": 0.5061972545009272,
+      "learning_rate": 4.1763367212477316e-09,
+      "loss": 0.6916,
+      "step": 46130
+    },
+    {
+      "epoch": 0.9911071014305968,
+      "grad_norm": 0.5114242229614572,
+      "learning_rate": 3.981675703965681e-09,
+      "loss": 0.686,
+      "step": 46140
+    },
+    {
+      "epoch": 0.9913219057438674,
+      "grad_norm": 0.5034691443686983,
+      "learning_rate": 3.791659315878216e-09,
+      "loss": 0.6991,
+      "step": 46150
+    },
+    {
+      "epoch": 0.991536710057138,
+      "grad_norm": 0.5237369943432371,
+      "learning_rate": 3.606287645276929e-09,
+      "loss": 0.7013,
+      "step": 46160
+    },
+    {
+      "epoch": 0.9917515143704085,
+      "grad_norm": 0.5050943270330421,
+      "learning_rate": 3.4255607782940348e-09,
+      "loss": 0.6686,
+      "step": 46170
+    },
+    {
+      "epoch": 0.9919663186836791,
+      "grad_norm": 0.5207022192351678,
+      "learning_rate": 3.2494787989034715e-09,
+      "loss": 0.6932,
+      "step": 46180
+    },
+    {
+      "epoch": 0.9921811229969498,
+      "grad_norm": 0.5210263524410033,
+      "learning_rate": 3.078041788922015e-09,
+      "loss": 0.6781,
+      "step": 46190
+    },
+    {
+      "epoch": 0.9923959273102204,
+      "grad_norm": 0.5144052422535814,
+      "learning_rate": 2.9112498280070565e-09,
+      "loss": 0.6888,
+      "step": 46200
+    },
+    {
+      "epoch": 0.992610731623491,
+      "grad_norm": 0.5123530950333133,
+      "learning_rate": 2.7491029936588252e-09,
+      "loss": 0.6871,
+      "step": 46210
+    },
+    {
+      "epoch": 0.9928255359367616,
+      "grad_norm": 0.5017887855233025,
+      "learning_rate": 2.5916013612170555e-09,
+      "loss": 0.6838,
+      "step": 46220
+    },
+    {
+      "epoch": 0.9930403402500322,
+      "grad_norm": 0.5059792773955643,
+      "learning_rate": 2.4387450038665385e-09,
+      "loss": 0.6769,
+      "step": 46230
+    },
+    {
+      "epoch": 0.9932551445633029,
+      "grad_norm": 0.5090551052637086,
+      "learning_rate": 2.2905339926293513e-09,
+      "loss": 0.6725,
+      "step": 46240
+    },
+    {
+      "epoch": 0.9934699488765735,
+      "grad_norm": 0.5044393490597461,
+      "learning_rate": 2.1469683963737386e-09,
+      "loss": 0.6938,
+      "step": 46250
+    },
+    {
+      "epoch": 0.9936847531898441,
+      "grad_norm": 0.5023690129044163,
+      "learning_rate": 2.008048281806341e-09,
+      "loss": 0.6813,
+      "step": 46260
+    },
+    {
+      "epoch": 0.9938995575031147,
+      "grad_norm": 0.5241055668792824,
+      "learning_rate": 1.8737737134744138e-09,
+      "loss": 0.6848,
+      "step": 46270
+    },
+    {
+      "epoch": 0.9941143618163852,
+      "grad_norm": 0.5136221543314958,
+      "learning_rate": 1.7441447537713817e-09,
+      "loss": 0.6737,
+      "step": 46280
+    },
+    {
+      "epoch": 0.9943291661296558,
+      "grad_norm": 0.510186886967427,
+      "learning_rate": 1.6191614629268437e-09,
+      "loss": 0.6827,
+      "step": 46290
+    },
+    {
+      "epoch": 0.9945439704429265,
+      "grad_norm": 0.5042541482836803,
+      "learning_rate": 1.4988238990143456e-09,
+      "loss": 0.6836,
+      "step": 46300
+    },
+    {
+      "epoch": 0.9947587747561971,
+      "grad_norm": 0.5162834914933954,
+      "learning_rate": 1.3831321179491596e-09,
+      "loss": 0.6976,
+      "step": 46310
+    },
+    {
+      "epoch": 0.9949735790694677,
+      "grad_norm": 0.5229889038934948,
+      "learning_rate": 1.2720861734871749e-09,
+      "loss": 0.698,
+      "step": 46320
+    },
+    {
+      "epoch": 0.9951883833827383,
+      "grad_norm": 0.5140297849222798,
+      "learning_rate": 1.1656861172248957e-09,
+      "loss": 0.6909,
+      "step": 46330
+    },
+    {
+      "epoch": 0.9954031876960089,
+      "grad_norm": 0.5211761716193927,
+      "learning_rate": 1.0639319986016639e-09,
+      "loss": 0.7059,
+      "step": 46340
+    },
+    {
+      "epoch": 0.9956179920092796,
+      "grad_norm": 0.5166206612511635,
+      "learning_rate": 9.66823864897437e-10,
+      "loss": 0.6799,
+      "step": 46350
+    },
+    {
+      "epoch": 0.9958327963225502,
+      "grad_norm": 0.5253510685541853,
+      "learning_rate": 8.74361761231679e-10,
+      "loss": 0.6795,
+      "step": 46360
+    },
+    {
+      "epoch": 0.9960476006358208,
+      "grad_norm": 0.5175002276710958,
+      "learning_rate": 7.865457305689106e-10,
+      "loss": 0.7023,
+      "step": 46370
+    },
+    {
+      "epoch": 0.9962624049490914,
+      "grad_norm": 0.5230288452217193,
+      "learning_rate": 7.033758137120483e-10,
+      "loss": 0.6874,
+      "step": 46380
+    },
+    {
+      "epoch": 0.9964772092623619,
+      "grad_norm": 0.5224400052978022,
+      "learning_rate": 6.24852049304625e-10,
+      "loss": 0.681,
+      "step": 46390
+    },
+    {
+      "epoch": 0.9966920135756326,
+      "grad_norm": 0.48862639002586716,
+      "learning_rate": 5.509744738341205e-10,
+      "loss": 0.6711,
+      "step": 46400
+    },
+    {
+      "epoch": 0.9969068178889032,
+      "grad_norm": 0.5192250692618456,
+      "learning_rate": 4.817431216264101e-10,
+      "loss": 0.6856,
+      "step": 46410
+    },
+    {
+      "epoch": 0.9971216222021738,
+      "grad_norm": 0.5154856759457314,
+      "learning_rate": 4.1715802485020604e-10,
+      "loss": 0.6921,
+      "step": 46420
+    },
+    {
+      "epoch": 0.9973364265154444,
+      "grad_norm": 0.5105045772126395,
+      "learning_rate": 3.5721921351483666e-10,
+      "loss": 0.6938,
+      "step": 46430
+    },
+    {
+      "epoch": 0.997551230828715,
+      "grad_norm": 0.508481663732828,
+      "learning_rate": 3.0192671547024655e-10,
+      "loss": 0.702,
+      "step": 46440
+    },
+    {
+      "epoch": 0.9977660351419857,
+      "grad_norm": 0.5050542413161292,
+      "learning_rate": 2.5128055640921687e-10,
+      "loss": 0.6901,
+      "step": 46450
+    },
+    {
+      "epoch": 0.9979808394552563,
+      "grad_norm": 0.5337828816495631,
+      "learning_rate": 2.0528075986292473e-10,
+      "loss": 0.69,
+      "step": 46460
+    },
+    {
+      "epoch": 0.9981956437685269,
+      "grad_norm": 0.49911609656939643,
+      "learning_rate": 1.6392734720649395e-10,
+      "loss": 0.6754,
+      "step": 46470
+    },
+    {
+      "epoch": 0.9984104480817975,
+      "grad_norm": 0.5450726921486923,
+      "learning_rate": 1.272203376534442e-10,
+      "loss": 0.6854,
+      "step": 46480
+    },
+    {
+      "epoch": 0.998625252395068,
+      "grad_norm": 0.5271282020213263,
+      "learning_rate": 9.515974826013186e-11,
+      "loss": 0.6966,
+      "step": 46490
+    },
+    {
+      "epoch": 0.9988400567083388,
+      "grad_norm": 0.5071464320894464,
+      "learning_rate": 6.774559392352942e-11,
+      "loss": 0.6808,
+      "step": 46500
+    },
+    {
+      "epoch": 0.9990548610216093,
+      "grad_norm": 0.5184516291604048,
+      "learning_rate": 4.497788738122566e-11,
+      "loss": 0.6894,
+      "step": 46510
+    },
+    {
+      "epoch": 0.9992696653348799,
+      "grad_norm": 0.5043819815091191,
+      "learning_rate": 2.6856639212535783e-11,
+      "loss": 0.6941,
+      "step": 46520
+    },
+    {
+      "epoch": 0.9994844696481505,
+      "grad_norm": 0.5554025858919428,
+      "learning_rate": 1.3381857838501433e-11,
+      "loss": 0.6999,
+      "step": 46530
+    },
+    {
+      "epoch": 0.9996992739614211,
+      "grad_norm": 0.5088148890810692,
+      "learning_rate": 4.553549517449795e-12,
+      "loss": 0.6761,
+      "step": 46540
+    },
+    {
+      "epoch": 0.9999140782746918,
+      "grad_norm": 0.5035234752449383,
+      "learning_rate": 3.7171835387539434e-13,
+      "loss": 0.69,
+      "step": 46550
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.6645376086235046,
+      "eval_runtime": 16.1599,
+      "eval_samples_per_second": 44.245,
+      "eval_steps_per_second": 0.743,
+      "step": 46554
+    },
+    {
+      "epoch": 1.0,
+      "step": 46554,
+      "total_flos": 6.49330829599703e+16,
+      "train_loss": 0.6052397525345854,
+      "train_runtime": 238477.2919,
+      "train_samples_per_second": 12.494,
+      "train_steps_per_second": 0.195
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 46554,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 4000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 6.49330829599703e+16,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}