diff --git "a/llama2_13b_peft/news_commentary_it/trainer_state.json" "b/llama2_13b_peft/news_commentary_it/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/llama2_13b_peft/news_commentary_it/trainer_state.json"
@@ -0,0 +1,8478 @@
+{
+  "best_metric": 0.6415141820907593,
+  "best_model_checkpoint": "ckpt/llama2_13b_fuze27_no_sys/news_commentary_it_no_sys/checkpoint-6000",
+  "epoch": 1.6842105263157894,
+  "eval_steps": 2000,
+  "global_step": 12000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0014035087719298245,
+      "grad_norm": 0.6813449859619141,
+      "learning_rate": 2.5e-05,
+      "loss": 1.3423,
+      "step": 10
+    },
+    {
+      "epoch": 0.002807017543859649,
+      "grad_norm": 0.9470943212509155,
+      "learning_rate": 5e-05,
+      "loss": 1.3855,
+      "step": 20
+    },
+    {
+      "epoch": 0.004210526315789474,
+      "grad_norm": 0.8929744958877563,
+      "learning_rate": 4.999999026832157e-05,
+      "loss": 0.9621,
+      "step": 30
+    },
+    {
+      "epoch": 0.005614035087719298,
+      "grad_norm": 1.383805274963379,
+      "learning_rate": 4.9999961073293845e-05,
+      "loss": 0.8217,
+      "step": 40
+    },
+    {
+      "epoch": 0.007017543859649123,
+      "grad_norm": 0.7758613228797913,
+      "learning_rate": 4.9999912414939555e-05,
+      "loss": 0.7743,
+      "step": 50
+    },
+    {
+      "epoch": 0.008421052631578947,
+      "grad_norm": 0.38530462980270386,
+      "learning_rate": 4.9999844293296585e-05,
+      "loss": 0.7671,
+      "step": 60
+    },
+    {
+      "epoch": 0.009824561403508772,
+      "grad_norm": 0.9287435412406921,
+      "learning_rate": 4.999975670841798e-05,
+      "loss": 0.7657,
+      "step": 70
+    },
+    {
+      "epoch": 0.011228070175438596,
+      "grad_norm": 0.5709918737411499,
+      "learning_rate": 4.9999649660371906e-05,
+      "loss": 0.6544,
+      "step": 80
+    },
+    {
+      "epoch": 0.01263157894736842,
+      "grad_norm": 0.6181680560112,
+      "learning_rate": 4.9999523149241714e-05,
+      "loss": 0.7627,
+      "step": 90
+    },
+    {
+      "epoch": 0.014035087719298246,
+      "grad_norm": 0.8074678182601929,
+      "learning_rate": 4.99993771751259e-05,
+      "loss": 0.7428,
+      "step": 100
+    },
+    {
+      "epoch": 0.015438596491228071,
+      "grad_norm": 0.7091221809387207,
+      "learning_rate": 4.999921173813812e-05,
+      "loss": 0.7024,
+      "step": 110
+    },
+    {
+      "epoch": 0.016842105263157894,
+      "grad_norm": 2.1647095680236816,
+      "learning_rate": 4.999902683840715e-05,
+      "loss": 0.8205,
+      "step": 120
+    },
+    {
+      "epoch": 0.018245614035087718,
+      "grad_norm": 1.178070068359375,
+      "learning_rate": 4.9998822476076955e-05,
+      "loss": 0.7359,
+      "step": 130
+    },
+    {
+      "epoch": 0.019649122807017545,
+      "grad_norm": 1.0926941633224487,
+      "learning_rate": 4.999859865130664e-05,
+      "loss": 0.6837,
+      "step": 140
+    },
+    {
+      "epoch": 0.021052631578947368,
+      "grad_norm": 1.5175189971923828,
+      "learning_rate": 4.9998355364270445e-05,
+      "loss": 0.7091,
+      "step": 150
+    },
+    {
+      "epoch": 0.02245614035087719,
+      "grad_norm": 0.9353613257408142,
+      "learning_rate": 4.999809261515779e-05,
+      "loss": 0.7608,
+      "step": 160
+    },
+    {
+      "epoch": 0.023859649122807018,
+      "grad_norm": 0.4437258839607239,
+      "learning_rate": 4.9997810404173234e-05,
+      "loss": 0.7725,
+      "step": 170
+    },
+    {
+      "epoch": 0.02526315789473684,
+      "grad_norm": 0.4320019781589508,
+      "learning_rate": 4.999750873153648e-05,
+      "loss": 0.7884,
+      "step": 180
+    },
+    {
+      "epoch": 0.02666666666666667,
+      "grad_norm": 0.8100196123123169,
+      "learning_rate": 4.9997187597482405e-05,
+      "loss": 0.7266,
+      "step": 190
+    },
+    {
+      "epoch": 0.028070175438596492,
+      "grad_norm": 1.1367573738098145,
+      "learning_rate": 4.9996847002261006e-05,
+      "loss": 0.6825,
+      "step": 200
+    },
+    {
+      "epoch": 0.029473684210526315,
+      "grad_norm": 0.9733144640922546,
+      "learning_rate": 4.999648694613746e-05,
+      "loss": 0.6162,
+      "step": 210
+    },
+    {
+      "epoch": 0.030877192982456142,
+      "grad_norm": 0.7170027494430542,
+      "learning_rate": 4.9996107429392083e-05,
+      "loss": 0.6696,
+      "step": 220
+    },
+    {
+      "epoch": 0.032280701754385965,
+      "grad_norm": 0.939182698726654,
+      "learning_rate": 4.9995708452320325e-05,
+      "loss": 0.7512,
+      "step": 230
+    },
+    {
+      "epoch": 0.03368421052631579,
+      "grad_norm": 0.7647657990455627,
+      "learning_rate": 4.999529001523282e-05,
+      "loss": 0.7137,
+      "step": 240
+    },
+    {
+      "epoch": 0.03508771929824561,
+      "grad_norm": 0.9428808093070984,
+      "learning_rate": 4.9994852118455335e-05,
+      "loss": 0.7676,
+      "step": 250
+    },
+    {
+      "epoch": 0.036491228070175435,
+      "grad_norm": 0.3808974325656891,
+      "learning_rate": 4.9994394762328786e-05,
+      "loss": 0.7208,
+      "step": 260
+    },
+    {
+      "epoch": 0.037894736842105266,
+      "grad_norm": 1.0278472900390625,
+      "learning_rate": 4.999391794720923e-05,
+      "loss": 0.7029,
+      "step": 270
+    },
+    {
+      "epoch": 0.03929824561403509,
+      "grad_norm": 0.8878808617591858,
+      "learning_rate": 4.9993421673467906e-05,
+      "loss": 0.6751,
+      "step": 280
+    },
+    {
+      "epoch": 0.04070175438596491,
+      "grad_norm": 0.5619615316390991,
+      "learning_rate": 4.9992905941491155e-05,
+      "loss": 0.7652,
+      "step": 290
+    },
+    {
+      "epoch": 0.042105263157894736,
+      "grad_norm": 1.1087744235992432,
+      "learning_rate": 4.9992370751680514e-05,
+      "loss": 0.7609,
+      "step": 300
+    },
+    {
+      "epoch": 0.04350877192982456,
+      "grad_norm": 0.7816822528839111,
+      "learning_rate": 4.999181610445263e-05,
+      "loss": 0.678,
+      "step": 310
+    },
+    {
+      "epoch": 0.04491228070175438,
+      "grad_norm": 1.0437147617340088,
+      "learning_rate": 4.9991242000239316e-05,
+      "loss": 0.7089,
+      "step": 320
+    },
+    {
+      "epoch": 0.04631578947368421,
+      "grad_norm": 0.7266655564308167,
+      "learning_rate": 4.9990648439487544e-05,
+      "loss": 0.7034,
+      "step": 330
+    },
+    {
+      "epoch": 0.047719298245614036,
+      "grad_norm": 0.8695891499519348,
+      "learning_rate": 4.999003542265941e-05,
+      "loss": 0.6789,
+      "step": 340
+    },
+    {
+      "epoch": 0.04912280701754386,
+      "grad_norm": 1.2530779838562012,
+      "learning_rate": 4.998940295023218e-05,
+      "loss": 0.6895,
+      "step": 350
+    },
+    {
+      "epoch": 0.05052631578947368,
+      "grad_norm": 0.9562914371490479,
+      "learning_rate": 4.9988751022698244e-05,
+      "loss": 0.7472,
+      "step": 360
+    },
+    {
+      "epoch": 0.051929824561403506,
+      "grad_norm": 1.5020138025283813,
+      "learning_rate": 4.9988079640565155e-05,
+      "loss": 0.7637,
+      "step": 370
+    },
+    {
+      "epoch": 0.05333333333333334,
+      "grad_norm": 1.3555861711502075,
+      "learning_rate": 4.998738880435561e-05,
+      "loss": 0.8042,
+      "step": 380
+    },
+    {
+      "epoch": 0.05473684210526316,
+      "grad_norm": 1.4689439535140991,
+      "learning_rate": 4.9986678514607434e-05,
+      "loss": 0.7878,
+      "step": 390
+    },
+    {
+      "epoch": 0.056140350877192984,
+      "grad_norm": 1.1399718523025513,
+      "learning_rate": 4.998594877187362e-05,
+      "loss": 0.6831,
+      "step": 400
+    },
+    {
+      "epoch": 0.05754385964912281,
+      "grad_norm": 0.9988260269165039,
+      "learning_rate": 4.998519957672232e-05,
+      "loss": 0.7905,
+      "step": 410
+    },
+    {
+      "epoch": 0.05894736842105263,
+      "grad_norm": 1.3424835205078125,
+      "learning_rate": 4.998443092973678e-05,
+      "loss": 0.6195,
+      "step": 420
+    },
+    {
+      "epoch": 0.060350877192982454,
+      "grad_norm": 1.3029276132583618,
+      "learning_rate": 4.998364283151542e-05,
+      "loss": 0.7603,
+      "step": 430
+    },
+    {
+      "epoch": 0.061754385964912284,
+      "grad_norm": 1.0647430419921875,
+      "learning_rate": 4.9982835282671816e-05,
+      "loss": 0.7099,
+      "step": 440
+    },
+    {
+      "epoch": 0.06315789473684211,
+      "grad_norm": 0.4545954763889313,
+      "learning_rate": 4.998200828383466e-05,
+      "loss": 0.6307,
+      "step": 450
+    },
+    {
+      "epoch": 0.06456140350877193,
+      "grad_norm": 0.9822194576263428,
+      "learning_rate": 4.99811618356478e-05,
+      "loss": 0.7084,
+      "step": 460
+    },
+    {
+      "epoch": 0.06596491228070175,
+      "grad_norm": 1.0566892623901367,
+      "learning_rate": 4.998029593877025e-05,
+      "loss": 0.6897,
+      "step": 470
+    },
+    {
+      "epoch": 0.06736842105263158,
+      "grad_norm": 0.9908930063247681,
+      "learning_rate": 4.9979410593876096e-05,
+      "loss": 0.7054,
+      "step": 480
+    },
+    {
+      "epoch": 0.0687719298245614,
+      "grad_norm": 0.5955024361610413,
+      "learning_rate": 4.997850580165464e-05,
+      "loss": 0.645,
+      "step": 490
+    },
+    {
+      "epoch": 0.07017543859649122,
+      "grad_norm": 1.570892095565796,
+      "learning_rate": 4.997758156281029e-05,
+      "loss": 0.6455,
+      "step": 500
+    },
+    {
+      "epoch": 0.07157894736842105,
+      "grad_norm": 0.9024527072906494,
+      "learning_rate": 4.997663787806259e-05,
+      "loss": 0.6797,
+      "step": 510
+    },
+    {
+      "epoch": 0.07298245614035087,
+      "grad_norm": 0.6280427575111389,
+      "learning_rate": 4.997567474814623e-05,
+      "loss": 0.7582,
+      "step": 520
+    },
+    {
+      "epoch": 0.07438596491228071,
+      "grad_norm": 1.0706899166107178,
+      "learning_rate": 4.997469217381105e-05,
+      "loss": 0.667,
+      "step": 530
+    },
+    {
+      "epoch": 0.07578947368421053,
+      "grad_norm": 0.8091099262237549,
+      "learning_rate": 4.997369015582201e-05,
+      "loss": 0.6878,
+      "step": 540
+    },
+    {
+      "epoch": 0.07719298245614035,
+      "grad_norm": 0.900131106376648,
+      "learning_rate": 4.9972668694959216e-05,
+      "loss": 0.7693,
+      "step": 550
+    },
+    {
+      "epoch": 0.07859649122807018,
+      "grad_norm": 0.791890025138855,
+      "learning_rate": 4.9971627792017915e-05,
+      "loss": 0.561,
+      "step": 560
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.3132946491241455,
+      "learning_rate": 4.997056744780848e-05,
+      "loss": 0.6739,
+      "step": 570
+    },
+    {
+      "epoch": 0.08140350877192983,
+      "grad_norm": 0.8182291984558105,
+      "learning_rate": 4.9969487663156434e-05,
+      "loss": 0.6561,
+      "step": 580
+    },
+    {
+      "epoch": 0.08280701754385965,
+      "grad_norm": 1.1820317506790161,
+      "learning_rate": 4.9968388438902415e-05,
+      "loss": 0.6056,
+      "step": 590
+    },
+    {
+      "epoch": 0.08421052631578947,
+      "grad_norm": 0.6508825421333313,
+      "learning_rate": 4.9967269775902204e-05,
+      "loss": 0.6962,
+      "step": 600
+    },
+    {
+      "epoch": 0.0856140350877193,
+      "grad_norm": 0.8885963559150696,
+      "learning_rate": 4.996613167502674e-05,
+      "loss": 0.6174,
+      "step": 610
+    },
+    {
+      "epoch": 0.08701754385964912,
+      "grad_norm": 0.8769521713256836,
+      "learning_rate": 4.996497413716205e-05,
+      "loss": 0.6806,
+      "step": 620
+    },
+    {
+      "epoch": 0.08842105263157894,
+      "grad_norm": 1.1168580055236816,
+      "learning_rate": 4.996379716320933e-05,
+      "loss": 0.7618,
+      "step": 630
+    },
+    {
+      "epoch": 0.08982456140350877,
+      "grad_norm": 0.6629518270492554,
+      "learning_rate": 4.996260075408489e-05,
+      "loss": 0.6796,
+      "step": 640
+    },
+    {
+      "epoch": 0.0912280701754386,
+      "grad_norm": 0.5513269901275635,
+      "learning_rate": 4.996138491072018e-05,
+      "loss": 0.6249,
+      "step": 650
+    },
+    {
+      "epoch": 0.09263157894736843,
+      "grad_norm": 0.8878002166748047,
+      "learning_rate": 4.996014963406177e-05,
+      "loss": 0.6905,
+      "step": 660
+    },
+    {
+      "epoch": 0.09403508771929825,
+      "grad_norm": 1.407973289489746,
+      "learning_rate": 4.9958894925071364e-05,
+      "loss": 0.7082,
+      "step": 670
+    },
+    {
+      "epoch": 0.09543859649122807,
+      "grad_norm": 2.0107500553131104,
+      "learning_rate": 4.995762078472581e-05,
+      "loss": 0.6751,
+      "step": 680
+    },
+    {
+      "epoch": 0.0968421052631579,
+      "grad_norm": 0.7563285827636719,
+      "learning_rate": 4.995632721401705e-05,
+      "loss": 0.6223,
+      "step": 690
+    },
+    {
+      "epoch": 0.09824561403508772,
+      "grad_norm": 0.7729387879371643,
+      "learning_rate": 4.995501421395219e-05,
+      "loss": 0.622,
+      "step": 700
+    },
+    {
+      "epoch": 0.09964912280701754,
+      "grad_norm": 0.9992890954017639,
+      "learning_rate": 4.995368178555343e-05,
+      "loss": 0.7565,
+      "step": 710
+    },
+    {
+      "epoch": 0.10105263157894737,
+      "grad_norm": 1.0641027688980103,
+      "learning_rate": 4.9952329929858125e-05,
+      "loss": 0.7486,
+      "step": 720
+    },
+    {
+      "epoch": 0.10245614035087719,
+      "grad_norm": 0.8268628716468811,
+      "learning_rate": 4.995095864791873e-05,
+      "loss": 0.6825,
+      "step": 730
+    },
+    {
+      "epoch": 0.10385964912280701,
+      "grad_norm": 0.7123477458953857,
+      "learning_rate": 4.994956794080285e-05,
+      "loss": 0.7342,
+      "step": 740
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 1.0346596240997314,
+      "learning_rate": 4.994815780959318e-05,
+      "loss": 0.6289,
+      "step": 750
+    },
+    {
+      "epoch": 0.10666666666666667,
+      "grad_norm": 0.7578685283660889,
+      "learning_rate": 4.994672825538757e-05,
+      "loss": 0.5675,
+      "step": 760
+    },
+    {
+      "epoch": 0.1080701754385965,
+      "grad_norm": 1.1263622045516968,
+      "learning_rate": 4.994527927929897e-05,
+      "loss": 0.7527,
+      "step": 770
+    },
+    {
+      "epoch": 0.10947368421052632,
+      "grad_norm": 0.8590745329856873,
+      "learning_rate": 4.9943810882455454e-05,
+      "loss": 0.6421,
+      "step": 780
+    },
+    {
+      "epoch": 0.11087719298245614,
+      "grad_norm": 0.7870830297470093,
+      "learning_rate": 4.994232306600023e-05,
+      "loss": 0.7016,
+      "step": 790
+    },
+    {
+      "epoch": 0.11228070175438597,
+      "grad_norm": 0.9499567747116089,
+      "learning_rate": 4.99408158310916e-05,
+      "loss": 0.6911,
+      "step": 800
+    },
+    {
+      "epoch": 0.11368421052631579,
+      "grad_norm": 1.1604363918304443,
+      "learning_rate": 4.9939289178903016e-05,
+      "loss": 0.697,
+      "step": 810
+    },
+    {
+      "epoch": 0.11508771929824561,
+      "grad_norm": 0.7308230400085449,
+      "learning_rate": 4.993774311062301e-05,
+      "loss": 0.5691,
+      "step": 820
+    },
+    {
+      "epoch": 0.11649122807017544,
+      "grad_norm": 1.0032395124435425,
+      "learning_rate": 4.993617762745526e-05,
+      "loss": 0.7744,
+      "step": 830
+    },
+    {
+      "epoch": 0.11789473684210526,
+      "grad_norm": 1.0617241859436035,
+      "learning_rate": 4.993459273061855e-05,
+      "loss": 0.7652,
+      "step": 840
+    },
+    {
+      "epoch": 0.11929824561403508,
+      "grad_norm": 1.207223653793335,
+      "learning_rate": 4.993298842134677e-05,
+      "loss": 0.6843,
+      "step": 850
+    },
+    {
+      "epoch": 0.12070175438596491,
+      "grad_norm": 0.6737737059593201,
+      "learning_rate": 4.993136470088894e-05,
+      "loss": 0.7147,
+      "step": 860
+    },
+    {
+      "epoch": 0.12210526315789473,
+      "grad_norm": 1.3904882669448853,
+      "learning_rate": 4.992972157050916e-05,
+      "loss": 0.641,
+      "step": 870
+    },
+    {
+      "epoch": 0.12350877192982457,
+      "grad_norm": 0.8821682929992676,
+      "learning_rate": 4.992805903148669e-05,
+      "loss": 0.6212,
+      "step": 880
+    },
+    {
+      "epoch": 0.12491228070175439,
+      "grad_norm": 1.214309811592102,
+      "learning_rate": 4.992637708511586e-05,
+      "loss": 0.6817,
+      "step": 890
+    },
+    {
+      "epoch": 0.12631578947368421,
+      "grad_norm": 1.2954894304275513,
+      "learning_rate": 4.9924675732706123e-05,
+      "loss": 0.7072,
+      "step": 900
+    },
+    {
+      "epoch": 0.12771929824561404,
+      "grad_norm": 0.8437069058418274,
+      "learning_rate": 4.992295497558204e-05,
+      "loss": 0.6221,
+      "step": 910
+    },
+    {
+      "epoch": 0.12912280701754386,
+      "grad_norm": 0.6401008367538452,
+      "learning_rate": 4.992121481508328e-05,
+      "loss": 0.6162,
+      "step": 920
+    },
+    {
+      "epoch": 0.13052631578947368,
+      "grad_norm": 1.1894147396087646,
+      "learning_rate": 4.9919455252564624e-05,
+      "loss": 0.7548,
+      "step": 930
+    },
+    {
+      "epoch": 0.1319298245614035,
+      "grad_norm": 0.9592342376708984,
+      "learning_rate": 4.991767628939594e-05,
+      "loss": 0.6377,
+      "step": 940
+    },
+    {
+      "epoch": 0.13333333333333333,
+      "grad_norm": 0.6419144868850708,
+      "learning_rate": 4.991587792696223e-05,
+      "loss": 0.6971,
+      "step": 950
+    },
+    {
+      "epoch": 0.13473684210526315,
+      "grad_norm": 1.0908111333847046,
+      "learning_rate": 4.991406016666356e-05,
+      "loss": 0.7929,
+      "step": 960
+    },
+    {
+      "epoch": 0.13614035087719298,
+      "grad_norm": 1.231597661972046,
+      "learning_rate": 4.9912223009915126e-05,
+      "loss": 0.7556,
+      "step": 970
+    },
+    {
+      "epoch": 0.1375438596491228,
+      "grad_norm": 0.7628648281097412,
+      "learning_rate": 4.991036645814722e-05,
+      "loss": 0.5883,
+      "step": 980
+    },
+    {
+      "epoch": 0.13894736842105262,
+      "grad_norm": 0.766953706741333,
+      "learning_rate": 4.9908490512805236e-05,
+      "loss": 0.6362,
+      "step": 990
+    },
+    {
+      "epoch": 0.14035087719298245,
+      "grad_norm": 1.070429801940918,
+      "learning_rate": 4.990659517534966e-05,
+      "loss": 0.7057,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14175438596491227,
+      "grad_norm": 0.8499042391777039,
+      "learning_rate": 4.990468044725606e-05,
+      "loss": 0.6051,
+      "step": 1010
+    },
+    {
+      "epoch": 0.1431578947368421,
+      "grad_norm": 1.1411361694335938,
+      "learning_rate": 4.990274633001514e-05,
+      "loss": 0.7434,
+      "step": 1020
+    },
+    {
+      "epoch": 0.14456140350877192,
+      "grad_norm": 1.3025455474853516,
+      "learning_rate": 4.990079282513266e-05,
+      "loss": 0.6681,
+      "step": 1030
+    },
+    {
+      "epoch": 0.14596491228070174,
+      "grad_norm": 0.9307923316955566,
+      "learning_rate": 4.9898819934129506e-05,
+      "loss": 0.6655,
+      "step": 1040
+    },
+    {
+      "epoch": 0.14736842105263157,
+      "grad_norm": 0.6463920474052429,
+      "learning_rate": 4.989682765854163e-05,
+      "loss": 0.7529,
+      "step": 1050
+    },
+    {
+      "epoch": 0.14877192982456142,
+      "grad_norm": 1.216407060623169,
+      "learning_rate": 4.989481599992009e-05,
+      "loss": 0.6249,
+      "step": 1060
+    },
+    {
+      "epoch": 0.15017543859649124,
+      "grad_norm": 0.747074544429779,
+      "learning_rate": 4.989278495983103e-05,
+      "loss": 0.6437,
+      "step": 1070
+    },
+    {
+      "epoch": 0.15157894736842106,
+      "grad_norm": 0.8777433633804321,
+      "learning_rate": 4.989073453985569e-05,
+      "loss": 0.6206,
+      "step": 1080
+    },
+    {
+      "epoch": 0.1529824561403509,
+      "grad_norm": 0.8588824272155762,
+      "learning_rate": 4.988866474159037e-05,
+      "loss": 0.6141,
+      "step": 1090
+    },
+    {
+      "epoch": 0.1543859649122807,
+      "grad_norm": 0.6369594931602478,
+      "learning_rate": 4.988657556664652e-05,
+      "loss": 0.6653,
+      "step": 1100
+    },
+    {
+      "epoch": 0.15578947368421053,
+      "grad_norm": 0.7276690006256104,
+      "learning_rate": 4.98844670166506e-05,
+      "loss": 0.6503,
+      "step": 1110
+    },
+    {
+      "epoch": 0.15719298245614036,
+      "grad_norm": 0.6937339305877686,
+      "learning_rate": 4.98823390932442e-05,
+      "loss": 0.6298,
+      "step": 1120
+    },
+    {
+      "epoch": 0.15859649122807018,
+      "grad_norm": 1.4779495000839233,
+      "learning_rate": 4.988019179808398e-05,
+      "loss": 0.6889,
+      "step": 1130
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.4205069541931152,
+      "learning_rate": 4.987802513284169e-05,
+      "loss": 0.7086,
+      "step": 1140
+    },
+    {
+      "epoch": 0.16140350877192983,
+      "grad_norm": 1.5097942352294922,
+      "learning_rate": 4.9875839099204134e-05,
+      "loss": 0.6727,
+      "step": 1150
+    },
+    {
+      "epoch": 0.16280701754385965,
+      "grad_norm": 0.8267427086830139,
+      "learning_rate": 4.987363369887324e-05,
+      "loss": 0.6993,
+      "step": 1160
+    },
+    {
+      "epoch": 0.16421052631578947,
+      "grad_norm": 1.1303791999816895,
+      "learning_rate": 4.987140893356597e-05,
+      "loss": 0.5671,
+      "step": 1170
+    },
+    {
+      "epoch": 0.1656140350877193,
+      "grad_norm": 0.9507080316543579,
+      "learning_rate": 4.986916480501438e-05,
+      "loss": 0.6929,
+      "step": 1180
+    },
+    {
+      "epoch": 0.16701754385964912,
+      "grad_norm": 1.0298510789871216,
+      "learning_rate": 4.986690131496561e-05,
+      "loss": 0.5368,
+      "step": 1190
+    },
+    {
+      "epoch": 0.16842105263157894,
+      "grad_norm": 1.0742335319519043,
+      "learning_rate": 4.986461846518186e-05,
+      "loss": 0.6473,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16982456140350877,
+      "grad_norm": 1.3724429607391357,
+      "learning_rate": 4.986231625744041e-05,
+      "loss": 0.6698,
+      "step": 1210
+    },
+    {
+      "epoch": 0.1712280701754386,
+      "grad_norm": 0.7210483551025391,
+      "learning_rate": 4.985999469353359e-05,
+      "loss": 0.6747,
+      "step": 1220
+    },
+    {
+      "epoch": 0.1726315789473684,
+      "grad_norm": 0.8128493428230286,
+      "learning_rate": 4.9857653775268853e-05,
+      "loss": 0.6509,
+      "step": 1230
+    },
+    {
+      "epoch": 0.17403508771929824,
+      "grad_norm": 0.9664400815963745,
+      "learning_rate": 4.985529350446865e-05,
+      "loss": 0.6895,
+      "step": 1240
+    },
+    {
+      "epoch": 0.17543859649122806,
+      "grad_norm": 1.0563639402389526,
+      "learning_rate": 4.985291388297055e-05,
+      "loss": 0.5882,
+      "step": 1250
+    },
+    {
+      "epoch": 0.17684210526315788,
+      "grad_norm": 0.7978933453559875,
+      "learning_rate": 4.985051491262716e-05,
+      "loss": 0.6688,
+      "step": 1260
+    },
+    {
+      "epoch": 0.1782456140350877,
+      "grad_norm": 1.0037199258804321,
+      "learning_rate": 4.984809659530617e-05,
+      "loss": 0.6135,
+      "step": 1270
+    },
+    {
+      "epoch": 0.17964912280701753,
+      "grad_norm": 1.0351414680480957,
+      "learning_rate": 4.9845658932890315e-05,
+      "loss": 0.6849,
+      "step": 1280
+    },
+    {
+      "epoch": 0.18105263157894738,
+      "grad_norm": 0.9015732407569885,
+      "learning_rate": 4.9843201927277407e-05,
+      "loss": 0.6036,
+      "step": 1290
+    },
+    {
+      "epoch": 0.1824561403508772,
+      "grad_norm": 1.1445683240890503,
+      "learning_rate": 4.984072558038031e-05,
+      "loss": 0.7348,
+      "step": 1300
+    },
+    {
+      "epoch": 0.18385964912280703,
+      "grad_norm": 1.2019379138946533,
+      "learning_rate": 4.983822989412693e-05,
+      "loss": 0.7679,
+      "step": 1310
+    },
+    {
+      "epoch": 0.18526315789473685,
+      "grad_norm": 0.6560442447662354,
+      "learning_rate": 4.983571487046026e-05,
+      "loss": 0.7083,
+      "step": 1320
+    },
+    {
+      "epoch": 0.18666666666666668,
+      "grad_norm": 0.8415977954864502,
+      "learning_rate": 4.9833180511338314e-05,
+      "loss": 0.6417,
+      "step": 1330
+    },
+    {
+      "epoch": 0.1880701754385965,
+      "grad_norm": 0.8725243210792542,
+      "learning_rate": 4.983062681873421e-05,
+      "loss": 0.6817,
+      "step": 1340
+    },
+    {
+      "epoch": 0.18947368421052632,
+      "grad_norm": 0.8865370154380798,
+      "learning_rate": 4.982805379463605e-05,
+      "loss": 0.6554,
+      "step": 1350
+    },
+    {
+      "epoch": 0.19087719298245615,
+      "grad_norm": 0.6979865431785583,
+      "learning_rate": 4.982546144104704e-05,
+      "loss": 0.6613,
+      "step": 1360
+    },
+    {
+      "epoch": 0.19228070175438597,
+      "grad_norm": 0.8604574203491211,
+      "learning_rate": 4.982284975998541e-05,
+      "loss": 0.6902,
+      "step": 1370
+    },
+    {
+      "epoch": 0.1936842105263158,
+      "grad_norm": 0.849172055721283,
+      "learning_rate": 4.982021875348445e-05,
+      "loss": 0.81,
+      "step": 1380
+    },
+    {
+      "epoch": 0.19508771929824562,
+      "grad_norm": 0.9217461347579956,
+      "learning_rate": 4.9817568423592484e-05,
+      "loss": 0.6796,
+      "step": 1390
+    },
+    {
+      "epoch": 0.19649122807017544,
+      "grad_norm": 1.1720378398895264,
+      "learning_rate": 4.981489877237288e-05,
+      "loss": 0.6109,
+      "step": 1400
+    },
+    {
+      "epoch": 0.19789473684210526,
+      "grad_norm": 0.8361873626708984,
+      "learning_rate": 4.9812209801904064e-05,
+      "loss": 0.7521,
+      "step": 1410
+    },
+    {
+      "epoch": 0.19929824561403509,
+      "grad_norm": 0.9124870896339417,
+      "learning_rate": 4.980950151427948e-05,
+      "loss": 0.6742,
+      "step": 1420
+    },
+    {
+      "epoch": 0.2007017543859649,
+      "grad_norm": 1.0720082521438599,
+      "learning_rate": 4.980677391160763e-05,
+      "loss": 0.659,
+      "step": 1430
+    },
+    {
+      "epoch": 0.20210526315789473,
+      "grad_norm": 0.7144408822059631,
+      "learning_rate": 4.980402699601205e-05,
+      "loss": 0.6392,
+      "step": 1440
+    },
+    {
+      "epoch": 0.20350877192982456,
+      "grad_norm": 0.8546087145805359,
+      "learning_rate": 4.98012607696313e-05,
+      "loss": 0.6674,
+      "step": 1450
+    },
+    {
+      "epoch": 0.20491228070175438,
+      "grad_norm": 0.8717739582061768,
+      "learning_rate": 4.979847523461898e-05,
+      "loss": 0.6772,
+      "step": 1460
+    },
+    {
+      "epoch": 0.2063157894736842,
+      "grad_norm": 0.9035875201225281,
+      "learning_rate": 4.9795670393143735e-05,
+      "loss": 0.6598,
+      "step": 1470
+    },
+    {
+      "epoch": 0.20771929824561403,
+      "grad_norm": 1.5168395042419434,
+      "learning_rate": 4.9792846247389214e-05,
+      "loss": 0.6784,
+      "step": 1480
+    },
+    {
+      "epoch": 0.20912280701754385,
+      "grad_norm": 1.3440768718719482,
+      "learning_rate": 4.979000279955413e-05,
+      "loss": 0.673,
+      "step": 1490
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.6575384140014648,
+      "learning_rate": 4.97871400518522e-05,
+      "loss": 0.6018,
+      "step": 1500
+    },
+    {
+      "epoch": 0.2119298245614035,
+      "grad_norm": 0.843136727809906,
+      "learning_rate": 4.978425800651216e-05,
+      "loss": 0.673,
+      "step": 1510
+    },
+    {
+      "epoch": 0.21333333333333335,
+      "grad_norm": 0.9389488101005554,
+      "learning_rate": 4.978135666577779e-05,
+      "loss": 0.6455,
+      "step": 1520
+    },
+    {
+      "epoch": 0.21473684210526317,
+      "grad_norm": 1.0860190391540527,
+      "learning_rate": 4.977843603190788e-05,
+      "loss": 0.6945,
+      "step": 1530
+    },
+    {
+      "epoch": 0.216140350877193,
+      "grad_norm": 0.923224925994873,
+      "learning_rate": 4.9775496107176245e-05,
+      "loss": 0.6441,
+      "step": 1540
+    },
+    {
+      "epoch": 0.21754385964912282,
+      "grad_norm": 0.9440721273422241,
+      "learning_rate": 4.977253689387172e-05,
+      "loss": 0.6399,
+      "step": 1550
+    },
+    {
+      "epoch": 0.21894736842105264,
+      "grad_norm": 1.225602626800537,
+      "learning_rate": 4.976955839429815e-05,
+      "loss": 0.7059,
+      "step": 1560
+    },
+    {
+      "epoch": 0.22035087719298246,
+      "grad_norm": 0.7701632380485535,
+      "learning_rate": 4.976656061077441e-05,
+      "loss": 0.7422,
+      "step": 1570
+    },
+    {
+      "epoch": 0.2217543859649123,
+      "grad_norm": 1.0036752223968506,
+      "learning_rate": 4.976354354563435e-05,
+      "loss": 0.7,
+      "step": 1580
+    },
+    {
+      "epoch": 0.2231578947368421,
+      "grad_norm": 1.2595415115356445,
+      "learning_rate": 4.976050720122688e-05,
+      "loss": 0.693,
+      "step": 1590
+    },
+    {
+      "epoch": 0.22456140350877193,
+      "grad_norm": 0.9705458283424377,
+      "learning_rate": 4.97574515799159e-05,
+      "loss": 0.6477,
+      "step": 1600
+    },
+    {
+      "epoch": 0.22596491228070176,
+      "grad_norm": 0.9339498281478882,
+      "learning_rate": 4.975437668408031e-05,
+      "loss": 0.6839,
+      "step": 1610
+    },
+    {
+      "epoch": 0.22736842105263158,
+      "grad_norm": 2.5165653228759766,
+      "learning_rate": 4.9751282516114024e-05,
+      "loss": 0.5796,
+      "step": 1620
+    },
+    {
+      "epoch": 0.2287719298245614,
+      "grad_norm": 1.2094191312789917,
+      "learning_rate": 4.9748169078425955e-05,
+      "loss": 0.6967,
+      "step": 1630
+    },
+    {
+      "epoch": 0.23017543859649123,
+      "grad_norm": 0.9400249719619751,
+      "learning_rate": 4.974503637344002e-05,
+      "loss": 0.5507,
+      "step": 1640
+    },
+    {
+      "epoch": 0.23157894736842105,
+      "grad_norm": 0.7801631093025208,
+      "learning_rate": 4.9741884403595135e-05,
+      "loss": 0.6792,
+      "step": 1650
+    },
+    {
+      "epoch": 0.23298245614035087,
+      "grad_norm": 0.8041971325874329,
+      "learning_rate": 4.9738713171345225e-05,
+      "loss": 0.616,
+      "step": 1660
+    },
+    {
+      "epoch": 0.2343859649122807,
+      "grad_norm": 0.9792094826698303,
+      "learning_rate": 4.9735522679159195e-05,
+      "loss": 0.635,
+      "step": 1670
+    },
+    {
+      "epoch": 0.23578947368421052,
+      "grad_norm": 0.8937766551971436,
+      "learning_rate": 4.9732312929520964e-05,
+      "loss": 0.6902,
+      "step": 1680
+    },
+    {
+      "epoch": 0.23719298245614034,
+      "grad_norm": 0.6050293445587158,
+      "learning_rate": 4.972908392492942e-05,
+      "loss": 0.6899,
+      "step": 1690
+    },
+    {
+      "epoch": 0.23859649122807017,
+      "grad_norm": 1.114696741104126,
+      "learning_rate": 4.9725835667898455e-05,
+      "loss": 0.7013,
+      "step": 1700
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7658337354660034,
+      "learning_rate": 4.972256816095695e-05,
+      "loss": 0.6379,
+      "step": 1710
+    },
+    {
+      "epoch": 0.24140350877192981,
+      "grad_norm": 1.0719423294067383,
+      "learning_rate": 4.971928140664878e-05,
+      "loss": 0.6819,
+      "step": 1720
+    },
+    {
+      "epoch": 0.24280701754385964,
+      "grad_norm": 0.5609824061393738,
+      "learning_rate": 4.971597540753279e-05,
+      "loss": 0.6888,
+      "step": 1730
+    },
+    {
+      "epoch": 0.24421052631578946,
+      "grad_norm": 0.8473712205886841,
+      "learning_rate": 4.971265016618281e-05,
+      "loss": 0.6761,
+      "step": 1740
+    },
+    {
+      "epoch": 0.24561403508771928,
+      "grad_norm": 1.023040533065796,
+      "learning_rate": 4.970930568518765e-05,
+      "loss": 0.5544,
+      "step": 1750
+    },
+    {
+      "epoch": 0.24701754385964914,
+      "grad_norm": 1.2763292789459229,
+      "learning_rate": 4.97059419671511e-05,
+      "loss": 0.7072,
+      "step": 1760
+    },
+    {
+      "epoch": 0.24842105263157896,
+      "grad_norm": 0.9501249194145203,
+      "learning_rate": 4.9702559014691965e-05,
+      "loss": 0.5992,
+      "step": 1770
+    },
+    {
+      "epoch": 0.24982456140350878,
+      "grad_norm": 0.714192271232605,
+      "learning_rate": 4.969915683044395e-05,
+      "loss": 0.6277,
+      "step": 1780
+    },
+    {
+      "epoch": 0.2512280701754386,
+      "grad_norm": 0.8613963723182678,
+      "learning_rate": 4.9695735417055776e-05,
+      "loss": 0.5501,
+      "step": 1790
+    },
+    {
+      "epoch": 0.25263157894736843,
+      "grad_norm": 0.7384011149406433,
+      "learning_rate": 4.969229477719116e-05,
+      "loss": 0.7619,
+      "step": 1800
+    },
+    {
+      "epoch": 0.2540350877192982,
+      "grad_norm": 0.8516148328781128,
+      "learning_rate": 4.9688834913528724e-05,
+      "loss": 0.6706,
+      "step": 1810
+    },
+    {
+      "epoch": 0.2554385964912281,
+      "grad_norm": 0.9726106524467468,
+      "learning_rate": 4.9685355828762115e-05,
+      "loss": 0.6825,
+      "step": 1820
+    },
+    {
+      "epoch": 0.25684210526315787,
+      "grad_norm": 0.9834999442100525,
+      "learning_rate": 4.96818575255999e-05,
+      "loss": 0.7195,
+      "step": 1830
+    },
+    {
+      "epoch": 0.2582456140350877,
+      "grad_norm": 0.6964922547340393,
+      "learning_rate": 4.967834000676564e-05,
+      "loss": 0.6196,
+      "step": 1840
+    },
+    {
+      "epoch": 0.2596491228070175,
+      "grad_norm": 1.0819238424301147,
+      "learning_rate": 4.967480327499785e-05,
+      "loss": 0.5768,
+      "step": 1850
+    },
+    {
+      "epoch": 0.26105263157894737,
+      "grad_norm": 0.7200153470039368,
+      "learning_rate": 4.9671247333049975e-05,
+      "loss": 0.6484,
+      "step": 1860
+    },
+    {
+      "epoch": 0.2624561403508772,
+      "grad_norm": 0.6098335385322571,
+      "learning_rate": 4.966767218369046e-05,
+      "loss": 0.6132,
+      "step": 1870
+    },
+    {
+      "epoch": 0.263859649122807,
+      "grad_norm": 1.1508702039718628,
+      "learning_rate": 4.966407782970267e-05,
+      "loss": 0.6435,
+      "step": 1880
+    },
+    {
+      "epoch": 0.26526315789473687,
+      "grad_norm": 0.9164888858795166,
+      "learning_rate": 4.966046427388494e-05,
+      "loss": 0.6581,
+      "step": 1890
+    },
+    {
+      "epoch": 0.26666666666666666,
+      "grad_norm": 0.7689521908760071,
+      "learning_rate": 4.965683151905054e-05,
+      "loss": 0.593,
+      "step": 1900
+    },
+    {
+      "epoch": 0.2680701754385965,
+      "grad_norm": 1.2253938913345337,
+      "learning_rate": 4.965317956802769e-05,
+      "loss": 0.7037,
+      "step": 1910
+    },
+    {
+      "epoch": 0.2694736842105263,
+      "grad_norm": 1.2759559154510498,
+      "learning_rate": 4.964950842365957e-05,
+      "loss": 0.7054,
+      "step": 1920
+    },
+    {
+      "epoch": 0.27087719298245616,
+      "grad_norm": 1.0961602926254272,
+      "learning_rate": 4.9645818088804284e-05,
+      "loss": 0.6463,
+      "step": 1930
+    },
+    {
+      "epoch": 0.27228070175438596,
+      "grad_norm": 1.0374549627304077,
+      "learning_rate": 4.964210856633489e-05,
+      "loss": 0.7222,
+      "step": 1940
+    },
+    {
+      "epoch": 0.2736842105263158,
+      "grad_norm": 1.228814721107483,
+      "learning_rate": 4.963837985913938e-05,
+      "loss": 0.7659,
+      "step": 1950
+    },
+    {
+      "epoch": 0.2750877192982456,
+      "grad_norm": 1.131882667541504,
+      "learning_rate": 4.963463197012067e-05,
+      "loss": 0.7388,
+      "step": 1960
+    },
+    {
+      "epoch": 0.27649122807017545,
+      "grad_norm": 0.6964682340621948,
+      "learning_rate": 4.9630864902196626e-05,
+      "loss": 0.6961,
+      "step": 1970
+    },
+    {
+      "epoch": 0.27789473684210525,
+      "grad_norm": 0.6383505463600159,
+      "learning_rate": 4.962707865830004e-05,
+      "loss": 0.5755,
+      "step": 1980
+    },
+    {
+      "epoch": 0.2792982456140351,
+      "grad_norm": 0.9402531981468201,
+      "learning_rate": 4.9623273241378636e-05,
+      "loss": 0.6845,
+      "step": 1990
+    },
+    {
+      "epoch": 0.2807017543859649,
+      "grad_norm": 0.8488597273826599,
+      "learning_rate": 4.9619448654395055e-05,
+      "loss": 0.7236,
+      "step": 2000
+    },
+    {
+      "epoch": 0.2807017543859649,
+      "eval_loss": 0.6709622740745544,
+      "eval_runtime": 44.146,
+      "eval_samples_per_second": 33.978,
+      "eval_steps_per_second": 8.495,
+      "step": 2000
+    },
+    {
+      "epoch": 0.28210526315789475,
+      "grad_norm": 1.1073328256607056,
+      "learning_rate": 4.9615604900326875e-05,
+      "loss": 0.5944,
+      "step": 2010
+    },
+    {
+      "epoch": 0.28350877192982454,
+      "grad_norm": 1.3910387754440308,
+      "learning_rate": 4.961174198216658e-05,
+      "loss": 0.6174,
+      "step": 2020
+    },
+    {
+      "epoch": 0.2849122807017544,
+      "grad_norm": 0.698826253414154,
+      "learning_rate": 4.9607859902921595e-05,
+      "loss": 0.6801,
+      "step": 2030
+    },
+    {
+      "epoch": 0.2863157894736842,
+      "grad_norm": 1.118665099143982,
+      "learning_rate": 4.960395866561425e-05,
+      "loss": 0.6657,
+      "step": 2040
+    },
+    {
+      "epoch": 0.28771929824561404,
+      "grad_norm": 1.1043261289596558,
+      "learning_rate": 4.960003827328179e-05,
+      "loss": 0.6536,
+      "step": 2050
+    },
+    {
+      "epoch": 0.28912280701754384,
+      "grad_norm": 0.7518707513809204,
+      "learning_rate": 4.959609872897637e-05,
+      "loss": 0.6361,
+      "step": 2060
+    },
+    {
+      "epoch": 0.2905263157894737,
+      "grad_norm": 1.0390689373016357,
+      "learning_rate": 4.959214003576507e-05,
+      "loss": 0.6369,
+      "step": 2070
+    },
+    {
+      "epoch": 0.2919298245614035,
+      "grad_norm": 1.122710108757019,
+      "learning_rate": 4.958816219672986e-05,
+      "loss": 0.7563,
+      "step": 2080
+    },
+    {
+      "epoch": 0.29333333333333333,
+      "grad_norm": 0.9062842726707458,
+      "learning_rate": 4.9584165214967634e-05,
+      "loss": 0.6575,
+      "step": 2090
+    },
+    {
+      "epoch": 0.29473684210526313,
+      "grad_norm": 1.4019687175750732,
+      "learning_rate": 4.9580149093590165e-05,
+      "loss": 0.6611,
+      "step": 2100
+    },
+    {
+      "epoch": 0.296140350877193,
+      "grad_norm": 0.9323289394378662,
+      "learning_rate": 4.957611383572415e-05,
+      "loss": 0.6456,
+      "step": 2110
+    },
+    {
+      "epoch": 0.29754385964912283,
+      "grad_norm": 1.0447218418121338,
+      "learning_rate": 4.9572059444511175e-05,
+      "loss": 0.7114,
+      "step": 2120
+    },
+    {
+      "epoch": 0.29894736842105263,
+      "grad_norm": 1.3190436363220215,
+      "learning_rate": 4.956798592310773e-05,
+      "loss": 0.6338,
+      "step": 2130
+    },
+    {
+      "epoch": 0.3003508771929825,
+      "grad_norm": 0.7944990396499634,
+      "learning_rate": 4.956389327468518e-05,
+      "loss": 0.6323,
+      "step": 2140
+    },
+    {
+      "epoch": 0.3017543859649123,
+      "grad_norm": 0.9921332001686096,
+      "learning_rate": 4.9559781502429784e-05,
+      "loss": 0.6231,
+      "step": 2150
+    },
+    {
+      "epoch": 0.3031578947368421,
+      "grad_norm": 1.0437482595443726,
+      "learning_rate": 4.955565060954272e-05,
+      "loss": 0.6515,
+      "step": 2160
+    },
+    {
+      "epoch": 0.3045614035087719,
+      "grad_norm": 0.8929722309112549,
+      "learning_rate": 4.9551500599240006e-05,
+      "loss": 0.6023,
+      "step": 2170
+    },
+    {
+      "epoch": 0.3059649122807018,
+      "grad_norm": 1.1816951036453247,
+      "learning_rate": 4.954733147475259e-05,
+      "loss": 0.678,
+      "step": 2180
+    },
+    {
+      "epoch": 0.30736842105263157,
+      "grad_norm": 1.4489054679870605,
+      "learning_rate": 4.954314323932627e-05,
+      "loss": 0.6307,
+      "step": 2190
+    },
+    {
+      "epoch": 0.3087719298245614,
+      "grad_norm": 0.6073512434959412,
+      "learning_rate": 4.953893589622172e-05,
+      "loss": 0.6543,
+      "step": 2200
+    },
+    {
+      "epoch": 0.3101754385964912,
+      "grad_norm": 0.8957934975624084,
+      "learning_rate": 4.9534709448714514e-05,
+      "loss": 0.7493,
+      "step": 2210
+    },
+    {
+      "epoch": 0.31157894736842107,
+      "grad_norm": 1.1038836240768433,
+      "learning_rate": 4.9530463900095084e-05,
+      "loss": 0.6856,
+      "step": 2220
+    },
+    {
+      "epoch": 0.31298245614035086,
+      "grad_norm": 1.2374224662780762,
+      "learning_rate": 4.952619925366873e-05,
+      "loss": 0.5721,
+      "step": 2230
+    },
+    {
+      "epoch": 0.3143859649122807,
+      "grad_norm": 0.9683862924575806,
+      "learning_rate": 4.9521915512755635e-05,
+      "loss": 0.7126,
+      "step": 2240
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 1.096661925315857,
+      "learning_rate": 4.951761268069082e-05,
+      "loss": 0.7522,
+      "step": 2250
+    },
+    {
+      "epoch": 0.31719298245614036,
+      "grad_norm": 0.9801945686340332,
+      "learning_rate": 4.95132907608242e-05,
+      "loss": 0.6825,
+      "step": 2260
+    },
+    {
+      "epoch": 0.31859649122807016,
+      "grad_norm": 0.8269819021224976,
+      "learning_rate": 4.950894975652055e-05,
+      "loss": 0.7592,
+      "step": 2270
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.7468457818031311,
+      "learning_rate": 4.950458967115946e-05,
+      "loss": 0.5719,
+      "step": 2280
+    },
+    {
+      "epoch": 0.3214035087719298,
+      "grad_norm": 0.9176953434944153,
+      "learning_rate": 4.9500210508135436e-05,
+      "loss": 0.6288,
+      "step": 2290
+    },
+    {
+      "epoch": 0.32280701754385965,
+      "grad_norm": 0.6870772838592529,
+      "learning_rate": 4.9495812270857786e-05,
+      "loss": 0.7081,
+      "step": 2300
+    },
+    {
+      "epoch": 0.32421052631578945,
+      "grad_norm": 0.8877288103103638,
+      "learning_rate": 4.94913949627507e-05,
+      "loss": 0.6371,
+      "step": 2310
+    },
+    {
+      "epoch": 0.3256140350877193,
+      "grad_norm": 0.9289653897285461,
+      "learning_rate": 4.9486958587253195e-05,
+      "loss": 0.6712,
+      "step": 2320
+    },
+    {
+      "epoch": 0.3270175438596491,
+      "grad_norm": 0.7378761172294617,
+      "learning_rate": 4.9482503147819156e-05,
+      "loss": 0.6232,
+      "step": 2330
+    },
+    {
+      "epoch": 0.32842105263157895,
+      "grad_norm": 0.7357892394065857,
+      "learning_rate": 4.947802864791727e-05,
+      "loss": 0.6519,
+      "step": 2340
+    },
+    {
+      "epoch": 0.3298245614035088,
+      "grad_norm": 1.509859323501587,
+      "learning_rate": 4.947353509103112e-05,
+      "loss": 0.7172,
+      "step": 2350
+    },
+    {
+      "epoch": 0.3312280701754386,
+      "grad_norm": 0.9467512369155884,
+      "learning_rate": 4.946902248065907e-05,
+      "loss": 0.6784,
+      "step": 2360
+    },
+    {
+      "epoch": 0.33263157894736844,
+      "grad_norm": 1.1108275651931763,
+      "learning_rate": 4.946449082031435e-05,
+      "loss": 0.612,
+      "step": 2370
+    },
+    {
+      "epoch": 0.33403508771929824,
+      "grad_norm": 1.0811039209365845,
+      "learning_rate": 4.9459940113525014e-05,
+      "loss": 0.7573,
+      "step": 2380
+    },
+    {
+      "epoch": 0.3354385964912281,
+      "grad_norm": 0.8881508708000183,
+      "learning_rate": 4.945537036383394e-05,
+      "loss": 0.7167,
+      "step": 2390
+    },
+    {
+      "epoch": 0.3368421052631579,
+      "grad_norm": 1.144106149673462,
+      "learning_rate": 4.945078157479884e-05,
+      "loss": 0.5797,
+      "step": 2400
+    },
+    {
+      "epoch": 0.33824561403508774,
+      "grad_norm": 1.1335030794143677,
+      "learning_rate": 4.944617374999224e-05,
+      "loss": 0.603,
+      "step": 2410
+    },
+    {
+      "epoch": 0.33964912280701753,
+      "grad_norm": 0.9101009368896484,
+      "learning_rate": 4.944154689300148e-05,
+      "loss": 0.6496,
+      "step": 2420
+    },
+    {
+      "epoch": 0.3410526315789474,
+      "grad_norm": 0.9584961533546448,
+      "learning_rate": 4.943690100742875e-05,
+      "loss": 0.6945,
+      "step": 2430
+    },
+    {
+      "epoch": 0.3424561403508772,
+      "grad_norm": 0.8912618160247803,
+      "learning_rate": 4.943223609689101e-05,
+      "loss": 0.6489,
+      "step": 2440
+    },
+    {
+      "epoch": 0.34385964912280703,
+      "grad_norm": 0.7363690733909607,
+      "learning_rate": 4.9427552165020066e-05,
+      "loss": 0.6066,
+      "step": 2450
+    },
+    {
+      "epoch": 0.3452631578947368,
+      "grad_norm": 1.2380393743515015,
+      "learning_rate": 4.9422849215462506e-05,
+      "loss": 0.6208,
+      "step": 2460
+    },
+    {
+      "epoch": 0.3466666666666667,
+      "grad_norm": 1.4995614290237427,
+      "learning_rate": 4.9418127251879756e-05,
+      "loss": 0.7249,
+      "step": 2470
+    },
+    {
+      "epoch": 0.3480701754385965,
+      "grad_norm": 1.0258910655975342,
+      "learning_rate": 4.9413386277948006e-05,
+      "loss": 0.7049,
+      "step": 2480
+    },
+    {
+      "epoch": 0.3494736842105263,
+      "grad_norm": 0.9672191143035889,
+      "learning_rate": 4.9408626297358286e-05,
+      "loss": 0.7138,
+      "step": 2490
+    },
+    {
+      "epoch": 0.3508771929824561,
+      "grad_norm": 0.9736180901527405,
+      "learning_rate": 4.940384731381639e-05,
+      "loss": 0.6047,
+      "step": 2500
+    },
+    {
+      "epoch": 0.35228070175438597,
+      "grad_norm": 0.7992679476737976,
+      "learning_rate": 4.9399049331042925e-05,
+      "loss": 0.6098,
+      "step": 2510
+    },
+    {
+      "epoch": 0.35368421052631577,
+      "grad_norm": 0.6984518766403198,
+      "learning_rate": 4.939423235277328e-05,
+      "loss": 0.6862,
+      "step": 2520
+    },
+    {
+      "epoch": 0.3550877192982456,
+      "grad_norm": 0.9038867354393005,
+      "learning_rate": 4.938939638275765e-05,
+      "loss": 0.7044,
+      "step": 2530
+    },
+    {
+      "epoch": 0.3564912280701754,
+      "grad_norm": 0.9274188280105591,
+      "learning_rate": 4.938454142476099e-05,
+      "loss": 0.6377,
+      "step": 2540
+    },
+    {
+      "epoch": 0.35789473684210527,
+      "grad_norm": 1.2159563302993774,
+      "learning_rate": 4.9379667482563066e-05,
+      "loss": 0.7172,
+      "step": 2550
+    },
+    {
+      "epoch": 0.35929824561403506,
+      "grad_norm": 0.8040406703948975,
+      "learning_rate": 4.937477455995839e-05,
+      "loss": 0.5563,
+      "step": 2560
+    },
+    {
+      "epoch": 0.3607017543859649,
+      "grad_norm": 0.9026057720184326,
+      "learning_rate": 4.9369862660756286e-05,
+      "loss": 0.7217,
+      "step": 2570
+    },
+    {
+      "epoch": 0.36210526315789476,
+      "grad_norm": 0.9877568483352661,
+      "learning_rate": 4.9364931788780835e-05,
+      "loss": 0.6424,
+      "step": 2580
+    },
+    {
+      "epoch": 0.36350877192982456,
+      "grad_norm": 0.8766788244247437,
+      "learning_rate": 4.9359981947870874e-05,
+      "loss": 0.6449,
+      "step": 2590
+    },
+    {
+      "epoch": 0.3649122807017544,
+      "grad_norm": 1.3229867219924927,
+      "learning_rate": 4.9355013141880045e-05,
+      "loss": 0.7963,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3663157894736842,
+      "grad_norm": 1.6900445222854614,
+      "learning_rate": 4.9350025374676725e-05,
+      "loss": 0.716,
+      "step": 2610
+    },
+    {
+      "epoch": 0.36771929824561406,
+      "grad_norm": 1.055550217628479,
+      "learning_rate": 4.934501865014405e-05,
+      "loss": 0.5228,
+      "step": 2620
+    },
+    {
+      "epoch": 0.36912280701754385,
+      "grad_norm": 0.8242397904396057,
+      "learning_rate": 4.933999297217994e-05,
+      "loss": 0.6206,
+      "step": 2630
+    },
+    {
+      "epoch": 0.3705263157894737,
+      "grad_norm": 0.9964637756347656,
+      "learning_rate": 4.933494834469706e-05,
+      "loss": 0.6324,
+      "step": 2640
+    },
+    {
+      "epoch": 0.3719298245614035,
+      "grad_norm": 1.4224967956542969,
+      "learning_rate": 4.9329884771622817e-05,
+      "loss": 0.7658,
+      "step": 2650
+    },
+    {
+      "epoch": 0.37333333333333335,
+      "grad_norm": 1.1131879091262817,
+      "learning_rate": 4.9324802256899385e-05,
+      "loss": 0.6711,
+      "step": 2660
+    },
+    {
+      "epoch": 0.37473684210526315,
+      "grad_norm": 0.7792202830314636,
+      "learning_rate": 4.931970080448366e-05,
+      "loss": 0.5751,
+      "step": 2670
+    },
+    {
+      "epoch": 0.376140350877193,
+      "grad_norm": 0.7242644429206848,
+      "learning_rate": 4.931458041834731e-05,
+      "loss": 0.6772,
+      "step": 2680
+    },
+    {
+      "epoch": 0.3775438596491228,
+      "grad_norm": 0.8322226405143738,
+      "learning_rate": 4.9309441102476734e-05,
+      "loss": 0.6141,
+      "step": 2690
+    },
+    {
+      "epoch": 0.37894736842105264,
+      "grad_norm": 1.1265790462493896,
+      "learning_rate": 4.930428286087306e-05,
+      "loss": 0.627,
+      "step": 2700
+    },
+    {
+      "epoch": 0.38035087719298244,
+      "grad_norm": 1.1249980926513672,
+      "learning_rate": 4.929910569755215e-05,
+      "loss": 0.6991,
+      "step": 2710
+    },
+    {
+      "epoch": 0.3817543859649123,
+      "grad_norm": 1.5213415622711182,
+      "learning_rate": 4.929390961654462e-05,
+      "loss": 0.6379,
+      "step": 2720
+    },
+    {
+      "epoch": 0.3831578947368421,
+      "grad_norm": 0.9948049783706665,
+      "learning_rate": 4.9288694621895776e-05,
+      "loss": 0.673,
+      "step": 2730
+    },
+    {
+      "epoch": 0.38456140350877194,
+      "grad_norm": 1.249971866607666,
+      "learning_rate": 4.928346071766569e-05,
+      "loss": 0.6562,
+      "step": 2740
+    },
+    {
+      "epoch": 0.38596491228070173,
+      "grad_norm": 1.5983259677886963,
+      "learning_rate": 4.927820790792912e-05,
+      "loss": 0.6517,
+      "step": 2750
+    },
+    {
+      "epoch": 0.3873684210526316,
+      "grad_norm": 1.1207720041275024,
+      "learning_rate": 4.9272936196775565e-05,
+      "loss": 0.6506,
+      "step": 2760
+    },
+    {
+      "epoch": 0.3887719298245614,
+      "grad_norm": 1.2459056377410889,
+      "learning_rate": 4.926764558830923e-05,
+      "loss": 0.7087,
+      "step": 2770
+    },
+    {
+      "epoch": 0.39017543859649123,
+      "grad_norm": 2.858981132507324,
+      "learning_rate": 4.926233608664904e-05,
+      "loss": 0.6409,
+      "step": 2780
+    },
+    {
+      "epoch": 0.391578947368421,
+      "grad_norm": 1.2133064270019531,
+      "learning_rate": 4.9257007695928624e-05,
+      "loss": 0.7131,
+      "step": 2790
+    },
+    {
+      "epoch": 0.3929824561403509,
+      "grad_norm": 1.264398455619812,
+      "learning_rate": 4.925166042029631e-05,
+      "loss": 0.7967,
+      "step": 2800
+    },
+    {
+      "epoch": 0.39438596491228073,
+      "grad_norm": 0.8172046542167664,
+      "learning_rate": 4.924629426391515e-05,
+      "loss": 0.6273,
+      "step": 2810
+    },
+    {
+      "epoch": 0.3957894736842105,
+      "grad_norm": 0.5741508603096008,
+      "learning_rate": 4.924090923096286e-05,
+      "loss": 0.6419,
+      "step": 2820
+    },
+    {
+      "epoch": 0.3971929824561404,
+      "grad_norm": 0.8728544116020203,
+      "learning_rate": 4.923550532563189e-05,
+      "loss": 0.6296,
+      "step": 2830
+    },
+    {
+      "epoch": 0.39859649122807017,
+      "grad_norm": 0.6913738250732422,
+      "learning_rate": 4.923008255212935e-05,
+      "loss": 0.6323,
+      "step": 2840
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.0395629405975342,
+      "learning_rate": 4.922464091467707e-05,
+      "loss": 0.6613,
+      "step": 2850
+    },
+    {
+      "epoch": 0.4014035087719298,
+      "grad_norm": 0.6149466633796692,
+      "learning_rate": 4.921918041751155e-05,
+      "loss": 0.6119,
+      "step": 2860
+    },
+    {
+      "epoch": 0.40280701754385967,
+      "grad_norm": 0.8594980239868164,
+      "learning_rate": 4.9213701064883966e-05,
+      "loss": 0.6575,
+      "step": 2870
+    },
+    {
+      "epoch": 0.40421052631578946,
+      "grad_norm": 1.0025339126586914,
+      "learning_rate": 4.9208202861060185e-05,
+      "loss": 0.7369,
+      "step": 2880
+    },
+    {
+      "epoch": 0.4056140350877193,
+      "grad_norm": 1.1241748332977295,
+      "learning_rate": 4.920268581032074e-05,
+      "loss": 0.6551,
+      "step": 2890
+    },
+    {
+      "epoch": 0.4070175438596491,
+      "grad_norm": 0.7128563523292542,
+      "learning_rate": 4.919714991696086e-05,
+      "loss": 0.6584,
+      "step": 2900
+    },
+    {
+      "epoch": 0.40842105263157896,
+      "grad_norm": 0.5740714073181152,
+      "learning_rate": 4.9191595185290414e-05,
+      "loss": 0.6674,
+      "step": 2910
+    },
+    {
+      "epoch": 0.40982456140350876,
+      "grad_norm": 0.9508911371231079,
+      "learning_rate": 4.918602161963396e-05,
+      "loss": 0.7091,
+      "step": 2920
+    },
+    {
+      "epoch": 0.4112280701754386,
+      "grad_norm": 1.1646149158477783,
+      "learning_rate": 4.9180429224330706e-05,
+      "loss": 0.5862,
+      "step": 2930
+    },
+    {
+      "epoch": 0.4126315789473684,
+      "grad_norm": 1.2261298894882202,
+      "learning_rate": 4.917481800373451e-05,
+      "loss": 0.731,
+      "step": 2940
+    },
+    {
+      "epoch": 0.41403508771929826,
+      "grad_norm": 0.6014220714569092,
+      "learning_rate": 4.916918796221393e-05,
+      "loss": 0.6716,
+      "step": 2950
+    },
+    {
+      "epoch": 0.41543859649122805,
+      "grad_norm": 1.0764710903167725,
+      "learning_rate": 4.9163539104152124e-05,
+      "loss": 0.6427,
+      "step": 2960
+    },
+    {
+      "epoch": 0.4168421052631579,
+      "grad_norm": 0.7629368901252747,
+      "learning_rate": 4.9157871433946925e-05,
+      "loss": 0.6184,
+      "step": 2970
+    },
+    {
+      "epoch": 0.4182456140350877,
+      "grad_norm": 0.8151566982269287,
+      "learning_rate": 4.9152184956010813e-05,
+      "loss": 0.6208,
+      "step": 2980
+    },
+    {
+      "epoch": 0.41964912280701755,
+      "grad_norm": 1.4884957075119019,
+      "learning_rate": 4.91464796747709e-05,
+      "loss": 0.6517,
+      "step": 2990
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 0.6417763233184814,
+      "learning_rate": 4.914075559466895e-05,
+      "loss": 0.6656,
+      "step": 3000
+    },
+    {
+      "epoch": 0.4224561403508772,
+      "grad_norm": 0.8164128065109253,
+      "learning_rate": 4.913501272016135e-05,
+      "loss": 0.6605,
+      "step": 3010
+    },
+    {
+      "epoch": 0.423859649122807,
+      "grad_norm": 0.9845851063728333,
+      "learning_rate": 4.9129251055719125e-05,
+      "loss": 0.6348,
+      "step": 3020
+    },
+    {
+      "epoch": 0.42526315789473684,
+      "grad_norm": 0.7174735069274902,
+      "learning_rate": 4.912347060582793e-05,
+      "loss": 0.6735,
+      "step": 3030
+    },
+    {
+      "epoch": 0.4266666666666667,
+      "grad_norm": 1.0722357034683228,
+      "learning_rate": 4.911767137498805e-05,
+      "loss": 0.604,
+      "step": 3040
+    },
+    {
+      "epoch": 0.4280701754385965,
+      "grad_norm": 0.8019692897796631,
+      "learning_rate": 4.911185336771437e-05,
+      "loss": 0.659,
+      "step": 3050
+    },
+    {
+      "epoch": 0.42947368421052634,
+      "grad_norm": 0.6950979232788086,
+      "learning_rate": 4.910601658853642e-05,
+      "loss": 0.6545,
+      "step": 3060
+    },
+    {
+      "epoch": 0.43087719298245614,
+      "grad_norm": 1.0000766515731812,
+      "learning_rate": 4.910016104199833e-05,
+      "loss": 0.6057,
+      "step": 3070
+    },
+    {
+      "epoch": 0.432280701754386,
+      "grad_norm": 1.0175904035568237,
+      "learning_rate": 4.909428673265884e-05,
+      "loss": 0.5503,
+      "step": 3080
+    },
+    {
+      "epoch": 0.4336842105263158,
+      "grad_norm": 1.158728003501892,
+      "learning_rate": 4.90883936650913e-05,
+      "loss": 0.6534,
+      "step": 3090
+    },
+    {
+      "epoch": 0.43508771929824563,
+      "grad_norm": 0.9984928369522095,
+      "learning_rate": 4.908248184388367e-05,
+      "loss": 0.6696,
+      "step": 3100
+    },
+    {
+      "epoch": 0.43649122807017543,
+      "grad_norm": 0.8490105867385864,
+      "learning_rate": 4.90765512736385e-05,
+      "loss": 0.5936,
+      "step": 3110
+    },
+    {
+      "epoch": 0.4378947368421053,
+      "grad_norm": 1.14065420627594,
+      "learning_rate": 4.907060195897296e-05,
+      "loss": 0.6154,
+      "step": 3120
+    },
+    {
+      "epoch": 0.4392982456140351,
+      "grad_norm": 1.0342949628829956,
+      "learning_rate": 4.906463390451878e-05,
+      "loss": 0.7975,
+      "step": 3130
+    },
+    {
+      "epoch": 0.44070175438596493,
+      "grad_norm": 1.2673470973968506,
+      "learning_rate": 4.9058647114922286e-05,
+      "loss": 0.6742,
+      "step": 3140
+    },
+    {
+      "epoch": 0.4421052631578947,
+      "grad_norm": 0.7902513146400452,
+      "learning_rate": 4.9052641594844416e-05,
+      "loss": 0.6221,
+      "step": 3150
+    },
+    {
+      "epoch": 0.4435087719298246,
+      "grad_norm": 0.813940167427063,
+      "learning_rate": 4.9046617348960666e-05,
+      "loss": 0.7789,
+      "step": 3160
+    },
+    {
+      "epoch": 0.44491228070175437,
+      "grad_norm": 0.9385407567024231,
+      "learning_rate": 4.904057438196111e-05,
+      "loss": 0.6668,
+      "step": 3170
+    },
+    {
+      "epoch": 0.4463157894736842,
+      "grad_norm": 1.005690574645996,
+      "learning_rate": 4.903451269855043e-05,
+      "loss": 0.6732,
+      "step": 3180
+    },
+    {
+      "epoch": 0.447719298245614,
+      "grad_norm": 1.1185845136642456,
+      "learning_rate": 4.9028432303447826e-05,
+      "loss": 0.652,
+      "step": 3190
+    },
+    {
+      "epoch": 0.44912280701754387,
+      "grad_norm": 1.234397292137146,
+      "learning_rate": 4.902233320138711e-05,
+      "loss": 0.7734,
+      "step": 3200
+    },
+    {
+      "epoch": 0.45052631578947366,
+      "grad_norm": 0.75343918800354,
+      "learning_rate": 4.901621539711664e-05,
+      "loss": 0.6524,
+      "step": 3210
+    },
+    {
+      "epoch": 0.4519298245614035,
+      "grad_norm": 0.7265051603317261,
+      "learning_rate": 4.901007889539933e-05,
+      "loss": 0.5631,
+      "step": 3220
+    },
+    {
+      "epoch": 0.4533333333333333,
+      "grad_norm": 0.8995214700698853,
+      "learning_rate": 4.900392370101266e-05,
+      "loss": 0.6949,
+      "step": 3230
+    },
+    {
+      "epoch": 0.45473684210526316,
+      "grad_norm": 1.1753424406051636,
+      "learning_rate": 4.899774981874867e-05,
+      "loss": 0.7445,
+      "step": 3240
+    },
+    {
+      "epoch": 0.45614035087719296,
+      "grad_norm": 0.9139629602432251,
+      "learning_rate": 4.8991557253413924e-05,
+      "loss": 0.6329,
+      "step": 3250
+    },
+    {
+      "epoch": 0.4575438596491228,
+      "grad_norm": 1.145979881286621,
+      "learning_rate": 4.8985346009829546e-05,
+      "loss": 0.6808,
+      "step": 3260
+    },
+    {
+      "epoch": 0.4589473684210526,
+      "grad_norm": 0.5931209921836853,
+      "learning_rate": 4.8979116092831223e-05,
+      "loss": 0.6464,
+      "step": 3270
+    },
+    {
+      "epoch": 0.46035087719298246,
+      "grad_norm": 0.9794625639915466,
+      "learning_rate": 4.897286750726913e-05,
+      "loss": 0.6997,
+      "step": 3280
+    },
+    {
+      "epoch": 0.4617543859649123,
+      "grad_norm": 1.121286153793335,
+      "learning_rate": 4.8966600258008024e-05,
+      "loss": 0.642,
+      "step": 3290
+    },
+    {
+      "epoch": 0.4631578947368421,
+      "grad_norm": 0.8356245160102844,
+      "learning_rate": 4.896031434992717e-05,
+      "loss": 0.651,
+      "step": 3300
+    },
+    {
+      "epoch": 0.46456140350877195,
+      "grad_norm": 0.8175771832466125,
+      "learning_rate": 4.8954009787920365e-05,
+      "loss": 0.6899,
+      "step": 3310
+    },
+    {
+      "epoch": 0.46596491228070175,
+      "grad_norm": 1.083617925643921,
+      "learning_rate": 4.894768657689592e-05,
+      "loss": 0.7559,
+      "step": 3320
+    },
+    {
+      "epoch": 0.4673684210526316,
+      "grad_norm": 1.0625582933425903,
+      "learning_rate": 4.8941344721776675e-05,
+      "loss": 0.6473,
+      "step": 3330
+    },
+    {
+      "epoch": 0.4687719298245614,
+      "grad_norm": 0.9211772680282593,
+      "learning_rate": 4.893498422749997e-05,
+      "loss": 0.726,
+      "step": 3340
+    },
+    {
+      "epoch": 0.47017543859649125,
+      "grad_norm": 0.8347317576408386,
+      "learning_rate": 4.8928605099017696e-05,
+      "loss": 0.6,
+      "step": 3350
+    },
+    {
+      "epoch": 0.47157894736842104,
+      "grad_norm": 1.115190029144287,
+      "learning_rate": 4.89222073412962e-05,
+      "loss": 0.6202,
+      "step": 3360
+    },
+    {
+      "epoch": 0.4729824561403509,
+      "grad_norm": 1.101366400718689,
+      "learning_rate": 4.8915790959316356e-05,
+      "loss": 0.6717,
+      "step": 3370
+    },
+    {
+      "epoch": 0.4743859649122807,
+      "grad_norm": 0.7661691308021545,
+      "learning_rate": 4.890935595807355e-05,
+      "loss": 0.6328,
+      "step": 3380
+    },
+    {
+      "epoch": 0.47578947368421054,
+      "grad_norm": 0.8245850205421448,
+      "learning_rate": 4.890290234257764e-05,
+      "loss": 0.7271,
+      "step": 3390
+    },
+    {
+      "epoch": 0.47719298245614034,
+      "grad_norm": 1.0110929012298584,
+      "learning_rate": 4.889643011785299e-05,
+      "loss": 0.582,
+      "step": 3400
+    },
+    {
+      "epoch": 0.4785964912280702,
+      "grad_norm": 0.7848758697509766,
+      "learning_rate": 4.888993928893846e-05,
+      "loss": 0.6851,
+      "step": 3410
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7310847640037537,
+      "learning_rate": 4.888342986088736e-05,
+      "loss": 0.583,
+      "step": 3420
+    },
+    {
+      "epoch": 0.48140350877192983,
+      "grad_norm": 1.3532679080963135,
+      "learning_rate": 4.887690183876752e-05,
+      "loss": 0.6261,
+      "step": 3430
+    },
+    {
+      "epoch": 0.48280701754385963,
+      "grad_norm": 1.0199493169784546,
+      "learning_rate": 4.887035522766122e-05,
+      "loss": 0.6563,
+      "step": 3440
+    },
+    {
+      "epoch": 0.4842105263157895,
+      "grad_norm": 0.7497562766075134,
+      "learning_rate": 4.886379003266523e-05,
+      "loss": 0.5178,
+      "step": 3450
+    },
+    {
+      "epoch": 0.4856140350877193,
+      "grad_norm": 0.8139173984527588,
+      "learning_rate": 4.885720625889078e-05,
+      "loss": 0.824,
+      "step": 3460
+    },
+    {
+      "epoch": 0.4870175438596491,
+      "grad_norm": 0.6662510633468628,
+      "learning_rate": 4.8850603911463556e-05,
+      "loss": 0.6821,
+      "step": 3470
+    },
+    {
+      "epoch": 0.4884210526315789,
+      "grad_norm": 0.9491138458251953,
+      "learning_rate": 4.8843982995523704e-05,
+      "loss": 0.6955,
+      "step": 3480
+    },
+    {
+      "epoch": 0.4898245614035088,
+      "grad_norm": 0.7988129258155823,
+      "learning_rate": 4.883734351622586e-05,
+      "loss": 0.6447,
+      "step": 3490
+    },
+    {
+      "epoch": 0.49122807017543857,
+      "grad_norm": 1.0620664358139038,
+      "learning_rate": 4.8830685478739057e-05,
+      "loss": 0.6454,
+      "step": 3500
+    },
+    {
+      "epoch": 0.4926315789473684,
+      "grad_norm": 0.7549204230308533,
+      "learning_rate": 4.8824008888246834e-05,
+      "loss": 0.6676,
+      "step": 3510
+    },
+    {
+      "epoch": 0.49403508771929827,
+      "grad_norm": 0.9870264530181885,
+      "learning_rate": 4.8817313749947115e-05,
+      "loss": 0.5715,
+      "step": 3520
+    },
+    {
+      "epoch": 0.49543859649122807,
+      "grad_norm": 0.7582098245620728,
+      "learning_rate": 4.881060006905232e-05,
+      "loss": 0.6479,
+      "step": 3530
+    },
+    {
+      "epoch": 0.4968421052631579,
+      "grad_norm": 0.6988912224769592,
+      "learning_rate": 4.880386785078925e-05,
+      "loss": 0.6208,
+      "step": 3540
+    },
+    {
+      "epoch": 0.4982456140350877,
+      "grad_norm": 0.7568824291229248,
+      "learning_rate": 4.87971171003992e-05,
+      "loss": 0.6503,
+      "step": 3550
+    },
+    {
+      "epoch": 0.49964912280701756,
+      "grad_norm": 1.2903584241867065,
+      "learning_rate": 4.879034782313786e-05,
+      "loss": 0.6525,
+      "step": 3560
+    },
+    {
+      "epoch": 0.5010526315789474,
+      "grad_norm": 0.7582905888557434,
+      "learning_rate": 4.878356002427532e-05,
+      "loss": 0.633,
+      "step": 3570
+    },
+    {
+      "epoch": 0.5024561403508772,
+      "grad_norm": 0.9976963400840759,
+      "learning_rate": 4.877675370909612e-05,
+      "loss": 0.6184,
+      "step": 3580
+    },
+    {
+      "epoch": 0.503859649122807,
+      "grad_norm": 0.8688436150550842,
+      "learning_rate": 4.876992888289923e-05,
+      "loss": 0.64,
+      "step": 3590
+    },
+    {
+      "epoch": 0.5052631578947369,
+      "grad_norm": 1.4120594263076782,
+      "learning_rate": 4.876308555099799e-05,
+      "loss": 0.6238,
+      "step": 3600
+    },
+    {
+      "epoch": 0.5066666666666667,
+      "grad_norm": 1.190382719039917,
+      "learning_rate": 4.875622371872017e-05,
+      "loss": 0.7433,
+      "step": 3610
+    },
+    {
+      "epoch": 0.5080701754385964,
+      "grad_norm": 0.8115689754486084,
+      "learning_rate": 4.874934339140795e-05,
+      "loss": 0.7031,
+      "step": 3620
+    },
+    {
+      "epoch": 0.5094736842105263,
+      "grad_norm": 0.8457335233688354,
+      "learning_rate": 4.8742444574417904e-05,
+      "loss": 0.5443,
+      "step": 3630
+    },
+    {
+      "epoch": 0.5108771929824562,
+      "grad_norm": 0.8754384517669678,
+      "learning_rate": 4.873552727312099e-05,
+      "loss": 0.6728,
+      "step": 3640
+    },
+    {
+      "epoch": 0.512280701754386,
+      "grad_norm": 1.2087777853012085,
+      "learning_rate": 4.872859149290256e-05,
+      "loss": 0.6321,
+      "step": 3650
+    },
+    {
+      "epoch": 0.5136842105263157,
+      "grad_norm": 1.0635002851486206,
+      "learning_rate": 4.872163723916237e-05,
+      "loss": 0.6301,
+      "step": 3660
+    },
+    {
+      "epoch": 0.5150877192982456,
+      "grad_norm": 1.1686186790466309,
+      "learning_rate": 4.871466451731453e-05,
+      "loss": 0.6991,
+      "step": 3670
+    },
+    {
+      "epoch": 0.5164912280701754,
+      "grad_norm": 1.1546950340270996,
+      "learning_rate": 4.870767333278755e-05,
+      "loss": 0.5503,
+      "step": 3680
+    },
+    {
+      "epoch": 0.5178947368421053,
+      "grad_norm": 0.8768120408058167,
+      "learning_rate": 4.87006636910243e-05,
+      "loss": 0.6342,
+      "step": 3690
+    },
+    {
+      "epoch": 0.519298245614035,
+      "grad_norm": 0.8353332281112671,
+      "learning_rate": 4.8693635597482045e-05,
+      "loss": 0.5933,
+      "step": 3700
+    },
+    {
+      "epoch": 0.5207017543859649,
+      "grad_norm": 0.8518616557121277,
+      "learning_rate": 4.868658905763238e-05,
+      "loss": 0.5878,
+      "step": 3710
+    },
+    {
+      "epoch": 0.5221052631578947,
+      "grad_norm": 0.8607089519500732,
+      "learning_rate": 4.8679524076961284e-05,
+      "loss": 0.5478,
+      "step": 3720
+    },
+    {
+      "epoch": 0.5235087719298246,
+      "grad_norm": 1.3177140951156616,
+      "learning_rate": 4.867244066096909e-05,
+      "loss": 0.6024,
+      "step": 3730
+    },
+    {
+      "epoch": 0.5249122807017544,
+      "grad_norm": 1.1247279644012451,
+      "learning_rate": 4.866533881517046e-05,
+      "loss": 0.6106,
+      "step": 3740
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 1.0166698694229126,
+      "learning_rate": 4.865821854509445e-05,
+      "loss": 0.602,
+      "step": 3750
+    },
+    {
+      "epoch": 0.527719298245614,
+      "grad_norm": 0.7537686824798584,
+      "learning_rate": 4.865107985628442e-05,
+      "loss": 0.7147,
+      "step": 3760
+    },
+    {
+      "epoch": 0.5291228070175439,
+      "grad_norm": 1.1428786516189575,
+      "learning_rate": 4.86439227542981e-05,
+      "loss": 0.6561,
+      "step": 3770
+    },
+    {
+      "epoch": 0.5305263157894737,
+      "grad_norm": 1.1645269393920898,
+      "learning_rate": 4.863674724470751e-05,
+      "loss": 0.7062,
+      "step": 3780
+    },
+    {
+      "epoch": 0.5319298245614035,
+      "grad_norm": 1.128609299659729,
+      "learning_rate": 4.862955333309905e-05,
+      "loss": 0.7019,
+      "step": 3790
+    },
+    {
+      "epoch": 0.5333333333333333,
+      "grad_norm": 1.0182465314865112,
+      "learning_rate": 4.8622341025073425e-05,
+      "loss": 0.5702,
+      "step": 3800
+    },
+    {
+      "epoch": 0.5347368421052632,
+      "grad_norm": 0.9816009998321533,
+      "learning_rate": 4.861511032624567e-05,
+      "loss": 0.6956,
+      "step": 3810
+    },
+    {
+      "epoch": 0.536140350877193,
+      "grad_norm": 0.7931702733039856,
+      "learning_rate": 4.860786124224512e-05,
+      "loss": 0.6266,
+      "step": 3820
+    },
+    {
+      "epoch": 0.5375438596491228,
+      "grad_norm": 1.3353627920150757,
+      "learning_rate": 4.860059377871544e-05,
+      "loss": 0.6758,
+      "step": 3830
+    },
+    {
+      "epoch": 0.5389473684210526,
+      "grad_norm": 1.1476149559020996,
+      "learning_rate": 4.85933079413146e-05,
+      "loss": 0.6559,
+      "step": 3840
+    },
+    {
+      "epoch": 0.5403508771929825,
+      "grad_norm": 0.9160752892494202,
+      "learning_rate": 4.858600373571487e-05,
+      "loss": 0.6052,
+      "step": 3850
+    },
+    {
+      "epoch": 0.5417543859649123,
+      "grad_norm": 1.0451756715774536,
+      "learning_rate": 4.8578681167602834e-05,
+      "loss": 0.6119,
+      "step": 3860
+    },
+    {
+      "epoch": 0.5431578947368421,
+      "grad_norm": 0.9673342108726501,
+      "learning_rate": 4.8571340242679354e-05,
+      "loss": 0.5872,
+      "step": 3870
+    },
+    {
+      "epoch": 0.5445614035087719,
+      "grad_norm": 1.24473237991333,
+      "learning_rate": 4.856398096665959e-05,
+      "loss": 0.7302,
+      "step": 3880
+    },
+    {
+      "epoch": 0.5459649122807018,
+      "grad_norm": 0.967494547367096,
+      "learning_rate": 4.8556603345273e-05,
+      "loss": 0.6889,
+      "step": 3890
+    },
+    {
+      "epoch": 0.5473684210526316,
+      "grad_norm": 0.9426731467247009,
+      "learning_rate": 4.8549207384263305e-05,
+      "loss": 0.6045,
+      "step": 3900
+    },
+    {
+      "epoch": 0.5487719298245614,
+      "grad_norm": 1.033600091934204,
+      "learning_rate": 4.854179308938852e-05,
+      "loss": 0.743,
+      "step": 3910
+    },
+    {
+      "epoch": 0.5501754385964912,
+      "grad_norm": 0.9784322381019592,
+      "learning_rate": 4.8534360466420926e-05,
+      "loss": 0.6416,
+      "step": 3920
+    },
+    {
+      "epoch": 0.5515789473684211,
+      "grad_norm": 1.0500706434249878,
+      "learning_rate": 4.852690952114708e-05,
+      "loss": 0.5975,
+      "step": 3930
+    },
+    {
+      "epoch": 0.5529824561403509,
+      "grad_norm": 1.6134823560714722,
+      "learning_rate": 4.851944025936779e-05,
+      "loss": 0.7975,
+      "step": 3940
+    },
+    {
+      "epoch": 0.5543859649122806,
+      "grad_norm": 0.785410463809967,
+      "learning_rate": 4.851195268689813e-05,
+      "loss": 0.6836,
+      "step": 3950
+    },
+    {
+      "epoch": 0.5557894736842105,
+      "grad_norm": 1.15956449508667,
+      "learning_rate": 4.850444680956745e-05,
+      "loss": 0.5265,
+      "step": 3960
+    },
+    {
+      "epoch": 0.5571929824561404,
+      "grad_norm": 1.0284963846206665,
+      "learning_rate": 4.8496922633219314e-05,
+      "loss": 0.687,
+      "step": 3970
+    },
+    {
+      "epoch": 0.5585964912280702,
+      "grad_norm": 0.5753929615020752,
+      "learning_rate": 4.8489380163711556e-05,
+      "loss": 0.5644,
+      "step": 3980
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.0494047403335571,
+      "learning_rate": 4.848181940691625e-05,
+      "loss": 0.6013,
+      "step": 3990
+    },
+    {
+      "epoch": 0.5614035087719298,
+      "grad_norm": 1.090614914894104,
+      "learning_rate": 4.8474240368719703e-05,
+      "loss": 0.724,
+      "step": 4000
+    },
+    {
+      "epoch": 0.5614035087719298,
+      "eval_loss": 0.6521075963973999,
+      "eval_runtime": 44.1632,
+      "eval_samples_per_second": 33.965,
+      "eval_steps_per_second": 8.491,
+      "step": 4000
+    },
+    {
+      "epoch": 0.5628070175438596,
+      "grad_norm": 1.593772530555725,
+      "learning_rate": 4.846664305502245e-05,
+      "loss": 0.6668,
+      "step": 4010
+    },
+    {
+      "epoch": 0.5642105263157895,
+      "grad_norm": 1.0096566677093506,
+      "learning_rate": 4.8459027471739284e-05,
+      "loss": 0.6898,
+      "step": 4020
+    },
+    {
+      "epoch": 0.5656140350877193,
+      "grad_norm": 1.126257061958313,
+      "learning_rate": 4.8451393624799165e-05,
+      "loss": 0.6639,
+      "step": 4030
+    },
+    {
+      "epoch": 0.5670175438596491,
+      "grad_norm": 1.0839751958847046,
+      "learning_rate": 4.844374152014532e-05,
+      "loss": 0.7336,
+      "step": 4040
+    },
+    {
+      "epoch": 0.5684210526315789,
+      "grad_norm": 0.4993619918823242,
+      "learning_rate": 4.843607116373518e-05,
+      "loss": 0.6233,
+      "step": 4050
+    },
+    {
+      "epoch": 0.5698245614035088,
+      "grad_norm": 1.6385512351989746,
+      "learning_rate": 4.8428382561540366e-05,
+      "loss": 0.6178,
+      "step": 4060
+    },
+    {
+      "epoch": 0.5712280701754386,
+      "grad_norm": 0.9295198321342468,
+      "learning_rate": 4.8420675719546723e-05,
+      "loss": 0.6121,
+      "step": 4070
+    },
+    {
+      "epoch": 0.5726315789473684,
+      "grad_norm": 1.2179811000823975,
+      "learning_rate": 4.8412950643754305e-05,
+      "loss": 0.6225,
+      "step": 4080
+    },
+    {
+      "epoch": 0.5740350877192982,
+      "grad_norm": 1.1477456092834473,
+      "learning_rate": 4.840520734017734e-05,
+      "loss": 0.6502,
+      "step": 4090
+    },
+    {
+      "epoch": 0.5754385964912281,
+      "grad_norm": 0.8792319297790527,
+      "learning_rate": 4.839744581484425e-05,
+      "loss": 0.6799,
+      "step": 4100
+    },
+    {
+      "epoch": 0.5768421052631579,
+      "grad_norm": 1.995977759361267,
+      "learning_rate": 4.8389666073797646e-05,
+      "loss": 0.7671,
+      "step": 4110
+    },
+    {
+      "epoch": 0.5782456140350877,
+      "grad_norm": 0.680174708366394,
+      "learning_rate": 4.8381868123094335e-05,
+      "loss": 0.6289,
+      "step": 4120
+    },
+    {
+      "epoch": 0.5796491228070175,
+      "grad_norm": 0.8312070369720459,
+      "learning_rate": 4.837405196880529e-05,
+      "loss": 0.6621,
+      "step": 4130
+    },
+    {
+      "epoch": 0.5810526315789474,
+      "grad_norm": 0.8448961973190308,
+      "learning_rate": 4.836621761701564e-05,
+      "loss": 0.601,
+      "step": 4140
+    },
+    {
+      "epoch": 0.5824561403508772,
+      "grad_norm": 1.1311395168304443,
+      "learning_rate": 4.835836507382471e-05,
+      "loss": 0.6818,
+      "step": 4150
+    },
+    {
+      "epoch": 0.583859649122807,
+      "grad_norm": 0.8135958313941956,
+      "learning_rate": 4.835049434534596e-05,
+      "loss": 0.6688,
+      "step": 4160
+    },
+    {
+      "epoch": 0.5852631578947368,
+      "grad_norm": 0.9292672276496887,
+      "learning_rate": 4.8342605437707034e-05,
+      "loss": 0.7652,
+      "step": 4170
+    },
+    {
+      "epoch": 0.5866666666666667,
+      "grad_norm": 1.1490682363510132,
+      "learning_rate": 4.8334698357049715e-05,
+      "loss": 0.5381,
+      "step": 4180
+    },
+    {
+      "epoch": 0.5880701754385965,
+      "grad_norm": 1.1863840818405151,
+      "learning_rate": 4.832677310952993e-05,
+      "loss": 0.6786,
+      "step": 4190
+    },
+    {
+      "epoch": 0.5894736842105263,
+      "grad_norm": 0.7175789475440979,
+      "learning_rate": 4.831882970131777e-05,
+      "loss": 0.629,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5908771929824561,
+      "grad_norm": 0.898485541343689,
+      "learning_rate": 4.831086813859743e-05,
+      "loss": 0.6021,
+      "step": 4210
+    },
+    {
+      "epoch": 0.592280701754386,
+      "grad_norm": 1.0772299766540527,
+      "learning_rate": 4.830288842756728e-05,
+      "loss": 0.5706,
+      "step": 4220
+    },
+    {
+      "epoch": 0.5936842105263158,
+      "grad_norm": 0.8830444812774658,
+      "learning_rate": 4.8294890574439784e-05,
+      "loss": 0.6716,
+      "step": 4230
+    },
+    {
+      "epoch": 0.5950877192982457,
+      "grad_norm": 1.12392258644104,
+      "learning_rate": 4.828687458544155e-05,
+      "loss": 0.6315,
+      "step": 4240
+    },
+    {
+      "epoch": 0.5964912280701754,
+      "grad_norm": 1.751460075378418,
+      "learning_rate": 4.82788404668133e-05,
+      "loss": 0.5633,
+      "step": 4250
+    },
+    {
+      "epoch": 0.5978947368421053,
+      "grad_norm": 1.024601697921753,
+      "learning_rate": 4.827078822480987e-05,
+      "loss": 0.6747,
+      "step": 4260
+    },
+    {
+      "epoch": 0.5992982456140351,
+      "grad_norm": 0.8278754949569702,
+      "learning_rate": 4.826271786570021e-05,
+      "loss": 0.6555,
+      "step": 4270
+    },
+    {
+      "epoch": 0.600701754385965,
+      "grad_norm": 0.9836990237236023,
+      "learning_rate": 4.825462939576737e-05,
+      "loss": 0.5987,
+      "step": 4280
+    },
+    {
+      "epoch": 0.6021052631578947,
+      "grad_norm": 0.5657834410667419,
+      "learning_rate": 4.8246522821308495e-05,
+      "loss": 0.6753,
+      "step": 4290
+    },
+    {
+      "epoch": 0.6035087719298246,
+      "grad_norm": 1.1341723203659058,
+      "learning_rate": 4.823839814863484e-05,
+      "loss": 0.5856,
+      "step": 4300
+    },
+    {
+      "epoch": 0.6049122807017544,
+      "grad_norm": 1.311997652053833,
+      "learning_rate": 4.823025538407173e-05,
+      "loss": 0.6204,
+      "step": 4310
+    },
+    {
+      "epoch": 0.6063157894736843,
+      "grad_norm": 0.8703358173370361,
+      "learning_rate": 4.82220945339586e-05,
+      "loss": 0.5866,
+      "step": 4320
+    },
+    {
+      "epoch": 0.607719298245614,
+      "grad_norm": 0.8117982149124146,
+      "learning_rate": 4.8213915604648944e-05,
+      "loss": 0.7384,
+      "step": 4330
+    },
+    {
+      "epoch": 0.6091228070175438,
+      "grad_norm": 1.2093411684036255,
+      "learning_rate": 4.820571860251034e-05,
+      "loss": 0.7113,
+      "step": 4340
+    },
+    {
+      "epoch": 0.6105263157894737,
+      "grad_norm": 0.895978569984436,
+      "learning_rate": 4.819750353392443e-05,
+      "loss": 0.6544,
+      "step": 4350
+    },
+    {
+      "epoch": 0.6119298245614035,
+      "grad_norm": 0.8177430629730225,
+      "learning_rate": 4.818927040528693e-05,
+      "loss": 0.6317,
+      "step": 4360
+    },
+    {
+      "epoch": 0.6133333333333333,
+      "grad_norm": 0.8065016865730286,
+      "learning_rate": 4.818101922300762e-05,
+      "loss": 0.5756,
+      "step": 4370
+    },
+    {
+      "epoch": 0.6147368421052631,
+      "grad_norm": 0.9234448075294495,
+      "learning_rate": 4.8172749993510315e-05,
+      "loss": 0.687,
+      "step": 4380
+    },
+    {
+      "epoch": 0.616140350877193,
+      "grad_norm": 1.0152438879013062,
+      "learning_rate": 4.81644627232329e-05,
+      "loss": 0.6573,
+      "step": 4390
+    },
+    {
+      "epoch": 0.6175438596491228,
+      "grad_norm": 0.8767795562744141,
+      "learning_rate": 4.81561574186273e-05,
+      "loss": 0.6787,
+      "step": 4400
+    },
+    {
+      "epoch": 0.6189473684210526,
+      "grad_norm": 0.8680139183998108,
+      "learning_rate": 4.814783408615948e-05,
+      "loss": 0.5503,
+      "step": 4410
+    },
+    {
+      "epoch": 0.6203508771929824,
+      "grad_norm": 0.9502211213111877,
+      "learning_rate": 4.813949273230944e-05,
+      "loss": 0.6495,
+      "step": 4420
+    },
+    {
+      "epoch": 0.6217543859649123,
+      "grad_norm": 0.8180057406425476,
+      "learning_rate": 4.8131133363571214e-05,
+      "loss": 0.5845,
+      "step": 4430
+    },
+    {
+      "epoch": 0.6231578947368421,
+      "grad_norm": 1.3863866329193115,
+      "learning_rate": 4.8122755986452845e-05,
+      "loss": 0.6093,
+      "step": 4440
+    },
+    {
+      "epoch": 0.624561403508772,
+      "grad_norm": 0.7499920129776001,
+      "learning_rate": 4.8114360607476416e-05,
+      "loss": 0.6465,
+      "step": 4450
+    },
+    {
+      "epoch": 0.6259649122807017,
+      "grad_norm": 0.7183496952056885,
+      "learning_rate": 4.810594723317801e-05,
+      "loss": 0.6228,
+      "step": 4460
+    },
+    {
+      "epoch": 0.6273684210526316,
+      "grad_norm": 1.3374441862106323,
+      "learning_rate": 4.809751587010774e-05,
+      "loss": 0.657,
+      "step": 4470
+    },
+    {
+      "epoch": 0.6287719298245614,
+      "grad_norm": 0.8970227837562561,
+      "learning_rate": 4.80890665248297e-05,
+      "loss": 0.6068,
+      "step": 4480
+    },
+    {
+      "epoch": 0.6301754385964913,
+      "grad_norm": 1.075203537940979,
+      "learning_rate": 4.808059920392201e-05,
+      "loss": 0.7177,
+      "step": 4490
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 1.3227583169937134,
+      "learning_rate": 4.807211391397674e-05,
+      "loss": 0.6333,
+      "step": 4500
+    },
+    {
+      "epoch": 0.6329824561403509,
+      "grad_norm": 0.8684366941452026,
+      "learning_rate": 4.806361066160001e-05,
+      "loss": 0.6396,
+      "step": 4510
+    },
+    {
+      "epoch": 0.6343859649122807,
+      "grad_norm": 0.9884424209594727,
+      "learning_rate": 4.8055089453411875e-05,
+      "loss": 0.622,
+      "step": 4520
+    },
+    {
+      "epoch": 0.6357894736842106,
+      "grad_norm": 1.2879207134246826,
+      "learning_rate": 4.80465502960464e-05,
+      "loss": 0.6551,
+      "step": 4530
+    },
+    {
+      "epoch": 0.6371929824561403,
+      "grad_norm": 0.8450446724891663,
+      "learning_rate": 4.80379931961516e-05,
+      "loss": 0.6874,
+      "step": 4540
+    },
+    {
+      "epoch": 0.6385964912280702,
+      "grad_norm": 1.0679776668548584,
+      "learning_rate": 4.8029418160389484e-05,
+      "loss": 0.5982,
+      "step": 4550
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7384183406829834,
+      "learning_rate": 4.8020825195435994e-05,
+      "loss": 0.5541,
+      "step": 4560
+    },
+    {
+      "epoch": 0.6414035087719299,
+      "grad_norm": 0.8015978336334229,
+      "learning_rate": 4.8012214307981064e-05,
+      "loss": 0.7297,
+      "step": 4570
+    },
+    {
+      "epoch": 0.6428070175438596,
+      "grad_norm": 0.7276405692100525,
+      "learning_rate": 4.800358550472855e-05,
+      "loss": 0.7694,
+      "step": 4580
+    },
+    {
+      "epoch": 0.6442105263157895,
+      "grad_norm": 0.7692060470581055,
+      "learning_rate": 4.799493879239628e-05,
+      "loss": 0.6194,
+      "step": 4590
+    },
+    {
+      "epoch": 0.6456140350877193,
+      "grad_norm": 1.2254407405853271,
+      "learning_rate": 4.7986274177716024e-05,
+      "loss": 0.6358,
+      "step": 4600
+    },
+    {
+      "epoch": 0.6470175438596492,
+      "grad_norm": 1.0495854616165161,
+      "learning_rate": 4.797759166743346e-05,
+      "loss": 0.6828,
+      "step": 4610
+    },
+    {
+      "epoch": 0.6484210526315789,
+      "grad_norm": 0.9298211932182312,
+      "learning_rate": 4.7968891268308246e-05,
+      "loss": 0.7163,
+      "step": 4620
+    },
+    {
+      "epoch": 0.6498245614035087,
+      "grad_norm": 0.9762528538703918,
+      "learning_rate": 4.796017298711391e-05,
+      "loss": 0.5935,
+      "step": 4630
+    },
+    {
+      "epoch": 0.6512280701754386,
+      "grad_norm": 1.0231860876083374,
+      "learning_rate": 4.795143683063797e-05,
+      "loss": 0.5696,
+      "step": 4640
+    },
+    {
+      "epoch": 0.6526315789473685,
+      "grad_norm": 1.1608182191848755,
+      "learning_rate": 4.7942682805681797e-05,
+      "loss": 0.5665,
+      "step": 4650
+    },
+    {
+      "epoch": 0.6540350877192982,
+      "grad_norm": 0.6527351140975952,
+      "learning_rate": 4.79339109190607e-05,
+      "loss": 0.6242,
+      "step": 4660
+    },
+    {
+      "epoch": 0.655438596491228,
+      "grad_norm": 0.8694155812263489,
+      "learning_rate": 4.792512117760391e-05,
+      "loss": 0.6259,
+      "step": 4670
+    },
+    {
+      "epoch": 0.6568421052631579,
+      "grad_norm": 0.9847631454467773,
+      "learning_rate": 4.7916313588154514e-05,
+      "loss": 0.6757,
+      "step": 4680
+    },
+    {
+      "epoch": 0.6582456140350877,
+      "grad_norm": 0.5999444127082825,
+      "learning_rate": 4.790748815756954e-05,
+      "loss": 0.6324,
+      "step": 4690
+    },
+    {
+      "epoch": 0.6596491228070176,
+      "grad_norm": 1.4817160367965698,
+      "learning_rate": 4.78986448927199e-05,
+      "loss": 0.5834,
+      "step": 4700
+    },
+    {
+      "epoch": 0.6610526315789473,
+      "grad_norm": 1.3592370748519897,
+      "learning_rate": 4.788978380049036e-05,
+      "loss": 0.6985,
+      "step": 4710
+    },
+    {
+      "epoch": 0.6624561403508772,
+      "grad_norm": 0.9479141235351562,
+      "learning_rate": 4.78809048877796e-05,
+      "loss": 0.6595,
+      "step": 4720
+    },
+    {
+      "epoch": 0.663859649122807,
+      "grad_norm": 1.3383686542510986,
+      "learning_rate": 4.787200816150014e-05,
+      "loss": 0.7508,
+      "step": 4730
+    },
+    {
+      "epoch": 0.6652631578947369,
+      "grad_norm": 1.0097548961639404,
+      "learning_rate": 4.786309362857839e-05,
+      "loss": 0.6452,
+      "step": 4740
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 0.9222456812858582,
+      "learning_rate": 4.785416129595463e-05,
+      "loss": 0.6171,
+      "step": 4750
+    },
+    {
+      "epoch": 0.6680701754385965,
+      "grad_norm": 0.9993833303451538,
+      "learning_rate": 4.784521117058298e-05,
+      "loss": 0.654,
+      "step": 4760
+    },
+    {
+      "epoch": 0.6694736842105263,
+      "grad_norm": 0.6470888257026672,
+      "learning_rate": 4.7836243259431425e-05,
+      "loss": 0.6674,
+      "step": 4770
+    },
+    {
+      "epoch": 0.6708771929824562,
+      "grad_norm": 0.8498440980911255,
+      "learning_rate": 4.7827257569481776e-05,
+      "loss": 0.6319,
+      "step": 4780
+    },
+    {
+      "epoch": 0.6722807017543859,
+      "grad_norm": 0.9220410585403442,
+      "learning_rate": 4.781825410772972e-05,
+      "loss": 0.5856,
+      "step": 4790
+    },
+    {
+      "epoch": 0.6736842105263158,
+      "grad_norm": 1.065016746520996,
+      "learning_rate": 4.780923288118475e-05,
+      "loss": 0.5919,
+      "step": 4800
+    },
+    {
+      "epoch": 0.6750877192982456,
+      "grad_norm": 0.7213327288627625,
+      "learning_rate": 4.78001938968702e-05,
+      "loss": 0.6192,
+      "step": 4810
+    },
+    {
+      "epoch": 0.6764912280701755,
+      "grad_norm": 0.7141574025154114,
+      "learning_rate": 4.779113716182323e-05,
+      "loss": 0.6628,
+      "step": 4820
+    },
+    {
+      "epoch": 0.6778947368421052,
+      "grad_norm": 0.7694927453994751,
+      "learning_rate": 4.778206268309482e-05,
+      "loss": 0.6451,
+      "step": 4830
+    },
+    {
+      "epoch": 0.6792982456140351,
+      "grad_norm": 1.2971090078353882,
+      "learning_rate": 4.777297046774977e-05,
+      "loss": 0.6823,
+      "step": 4840
+    },
+    {
+      "epoch": 0.6807017543859649,
+      "grad_norm": 0.7971644401550293,
+      "learning_rate": 4.7763860522866665e-05,
+      "loss": 0.6916,
+      "step": 4850
+    },
+    {
+      "epoch": 0.6821052631578948,
+      "grad_norm": 0.8853887319564819,
+      "learning_rate": 4.775473285553792e-05,
+      "loss": 0.5936,
+      "step": 4860
+    },
+    {
+      "epoch": 0.6835087719298245,
+      "grad_norm": 1.3222453594207764,
+      "learning_rate": 4.774558747286973e-05,
+      "loss": 0.7202,
+      "step": 4870
+    },
+    {
+      "epoch": 0.6849122807017544,
+      "grad_norm": 1.187171220779419,
+      "learning_rate": 4.77364243819821e-05,
+      "loss": 0.6405,
+      "step": 4880
+    },
+    {
+      "epoch": 0.6863157894736842,
+      "grad_norm": 0.8649610280990601,
+      "learning_rate": 4.7727243590008806e-05,
+      "loss": 0.6704,
+      "step": 4890
+    },
+    {
+      "epoch": 0.6877192982456141,
+      "grad_norm": 0.9361883401870728,
+      "learning_rate": 4.771804510409741e-05,
+      "loss": 0.6304,
+      "step": 4900
+    },
+    {
+      "epoch": 0.6891228070175439,
+      "grad_norm": 0.7870001196861267,
+      "learning_rate": 4.7708828931409236e-05,
+      "loss": 0.6645,
+      "step": 4910
+    },
+    {
+      "epoch": 0.6905263157894737,
+      "grad_norm": 1.0028226375579834,
+      "learning_rate": 4.769959507911941e-05,
+      "loss": 0.7018,
+      "step": 4920
+    },
+    {
+      "epoch": 0.6919298245614035,
+      "grad_norm": 0.7500180602073669,
+      "learning_rate": 4.769034355441678e-05,
+      "loss": 0.5191,
+      "step": 4930
+    },
+    {
+      "epoch": 0.6933333333333334,
+      "grad_norm": 0.9766993522644043,
+      "learning_rate": 4.7681074364503995e-05,
+      "loss": 0.6723,
+      "step": 4940
+    },
+    {
+      "epoch": 0.6947368421052632,
+      "grad_norm": 1.3899115324020386,
+      "learning_rate": 4.767178751659743e-05,
+      "loss": 0.7069,
+      "step": 4950
+    },
+    {
+      "epoch": 0.696140350877193,
+      "grad_norm": 1.3812363147735596,
+      "learning_rate": 4.7662483017927215e-05,
+      "loss": 0.6333,
+      "step": 4960
+    },
+    {
+      "epoch": 0.6975438596491228,
+      "grad_norm": 0.6967772841453552,
+      "learning_rate": 4.765316087573722e-05,
+      "loss": 0.7116,
+      "step": 4970
+    },
+    {
+      "epoch": 0.6989473684210527,
+      "grad_norm": 1.235410213470459,
+      "learning_rate": 4.7643821097285044e-05,
+      "loss": 0.5517,
+      "step": 4980
+    },
+    {
+      "epoch": 0.7003508771929825,
+      "grad_norm": 1.0389471054077148,
+      "learning_rate": 4.763446368984205e-05,
+      "loss": 0.6856,
+      "step": 4990
+    },
+    {
+      "epoch": 0.7017543859649122,
+      "grad_norm": 0.9552194476127625,
+      "learning_rate": 4.762508866069327e-05,
+      "loss": 0.6119,
+      "step": 5000
+    },
+    {
+      "epoch": 0.7031578947368421,
+      "grad_norm": 0.8866641521453857,
+      "learning_rate": 4.7615696017137504e-05,
+      "loss": 0.5645,
+      "step": 5010
+    },
+    {
+      "epoch": 0.7045614035087719,
+      "grad_norm": 1.0465891361236572,
+      "learning_rate": 4.760628576648723e-05,
+      "loss": 0.6506,
+      "step": 5020
+    },
+    {
+      "epoch": 0.7059649122807018,
+      "grad_norm": 1.104183316230774,
+      "learning_rate": 4.759685791606868e-05,
+      "loss": 0.6092,
+      "step": 5030
+    },
+    {
+      "epoch": 0.7073684210526315,
+      "grad_norm": 0.8748829364776611,
+      "learning_rate": 4.758741247322174e-05,
+      "loss": 0.7659,
+      "step": 5040
+    },
+    {
+      "epoch": 0.7087719298245614,
+      "grad_norm": 0.9573276042938232,
+      "learning_rate": 4.7577949445300004e-05,
+      "loss": 0.5774,
+      "step": 5050
+    },
+    {
+      "epoch": 0.7101754385964912,
+      "grad_norm": 0.9269713759422302,
+      "learning_rate": 4.756846883967077e-05,
+      "loss": 0.6234,
+      "step": 5060
+    },
+    {
+      "epoch": 0.7115789473684211,
+      "grad_norm": 0.6953681111335754,
+      "learning_rate": 4.755897066371502e-05,
+      "loss": 0.6456,
+      "step": 5070
+    },
+    {
+      "epoch": 0.7129824561403508,
+      "grad_norm": 0.6628289818763733,
+      "learning_rate": 4.754945492482741e-05,
+      "loss": 0.54,
+      "step": 5080
+    },
+    {
+      "epoch": 0.7143859649122807,
+      "grad_norm": 0.7972025871276855,
+      "learning_rate": 4.7539921630416264e-05,
+      "loss": 0.5695,
+      "step": 5090
+    },
+    {
+      "epoch": 0.7157894736842105,
+      "grad_norm": 1.032006859779358,
+      "learning_rate": 4.7530370787903576e-05,
+      "loss": 0.6748,
+      "step": 5100
+    },
+    {
+      "epoch": 0.7171929824561404,
+      "grad_norm": 0.7866501212120056,
+      "learning_rate": 4.7520802404725007e-05,
+      "loss": 0.595,
+      "step": 5110
+    },
+    {
+      "epoch": 0.7185964912280701,
+      "grad_norm": 1.0693832635879517,
+      "learning_rate": 4.751121648832987e-05,
+      "loss": 0.641,
+      "step": 5120
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.0331542491912842,
+      "learning_rate": 4.750161304618114e-05,
+      "loss": 0.6345,
+      "step": 5130
+    },
+    {
+      "epoch": 0.7214035087719298,
+      "grad_norm": 0.6064502000808716,
+      "learning_rate": 4.749199208575541e-05,
+      "loss": 0.5997,
+      "step": 5140
+    },
+    {
+      "epoch": 0.7228070175438597,
+      "grad_norm": 1.1691397428512573,
+      "learning_rate": 4.748235361454293e-05,
+      "loss": 0.6529,
+      "step": 5150
+    },
+    {
+      "epoch": 0.7242105263157895,
+      "grad_norm": 0.7956925630569458,
+      "learning_rate": 4.7472697640047594e-05,
+      "loss": 0.5668,
+      "step": 5160
+    },
+    {
+      "epoch": 0.7256140350877193,
+      "grad_norm": 0.9746783971786499,
+      "learning_rate": 4.7463024169786895e-05,
+      "loss": 0.6433,
+      "step": 5170
+    },
+    {
+      "epoch": 0.7270175438596491,
+      "grad_norm": 1.2105709314346313,
+      "learning_rate": 4.745333321129197e-05,
+      "loss": 0.6749,
+      "step": 5180
+    },
+    {
+      "epoch": 0.728421052631579,
+      "grad_norm": 0.7860882878303528,
+      "learning_rate": 4.744362477210755e-05,
+      "loss": 0.7041,
+      "step": 5190
+    },
+    {
+      "epoch": 0.7298245614035088,
+      "grad_norm": 1.1629239320755005,
+      "learning_rate": 4.7433898859792e-05,
+      "loss": 0.5598,
+      "step": 5200
+    },
+    {
+      "epoch": 0.7312280701754386,
+      "grad_norm": 1.1319113969802856,
+      "learning_rate": 4.742415548191728e-05,
+      "loss": 0.6433,
+      "step": 5210
+    },
+    {
+      "epoch": 0.7326315789473684,
+      "grad_norm": 0.8640940189361572,
+      "learning_rate": 4.741439464606893e-05,
+      "loss": 0.6715,
+      "step": 5220
+    },
+    {
+      "epoch": 0.7340350877192983,
+      "grad_norm": 0.7730684280395508,
+      "learning_rate": 4.740461635984609e-05,
+      "loss": 0.6391,
+      "step": 5230
+    },
+    {
+      "epoch": 0.7354385964912281,
+      "grad_norm": 1.2042145729064941,
+      "learning_rate": 4.739482063086152e-05,
+      "loss": 0.5834,
+      "step": 5240
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.9434259533882141,
+      "learning_rate": 4.73850074667415e-05,
+      "loss": 0.7835,
+      "step": 5250
+    },
+    {
+      "epoch": 0.7382456140350877,
+      "grad_norm": 0.8331650495529175,
+      "learning_rate": 4.737517687512593e-05,
+      "loss": 0.6128,
+      "step": 5260
+    },
+    {
+      "epoch": 0.7396491228070176,
+      "grad_norm": 0.8063735365867615,
+      "learning_rate": 4.7365328863668256e-05,
+      "loss": 0.655,
+      "step": 5270
+    },
+    {
+      "epoch": 0.7410526315789474,
+      "grad_norm": 0.6377186179161072,
+      "learning_rate": 4.735546344003551e-05,
+      "loss": 0.6506,
+      "step": 5280
+    },
+    {
+      "epoch": 0.7424561403508771,
+      "grad_norm": 0.8817654252052307,
+      "learning_rate": 4.734558061190824e-05,
+      "loss": 0.6984,
+      "step": 5290
+    },
+    {
+      "epoch": 0.743859649122807,
+      "grad_norm": 1.2554540634155273,
+      "learning_rate": 4.733568038698057e-05,
+      "loss": 0.7401,
+      "step": 5300
+    },
+    {
+      "epoch": 0.7452631578947368,
+      "grad_norm": 1.0858135223388672,
+      "learning_rate": 4.732576277296017e-05,
+      "loss": 0.6432,
+      "step": 5310
+    },
+    {
+      "epoch": 0.7466666666666667,
+      "grad_norm": 1.1962653398513794,
+      "learning_rate": 4.731582777756825e-05,
+      "loss": 0.6687,
+      "step": 5320
+    },
+    {
+      "epoch": 0.7480701754385964,
+      "grad_norm": 1.0213031768798828,
+      "learning_rate": 4.730587540853954e-05,
+      "loss": 0.6489,
+      "step": 5330
+    },
+    {
+      "epoch": 0.7494736842105263,
+      "grad_norm": 0.8629382252693176,
+      "learning_rate": 4.729590567362228e-05,
+      "loss": 0.6149,
+      "step": 5340
+    },
+    {
+      "epoch": 0.7508771929824561,
+      "grad_norm": 0.6692180633544922,
+      "learning_rate": 4.728591858057827e-05,
+      "loss": 0.6227,
+      "step": 5350
+    },
+    {
+      "epoch": 0.752280701754386,
+      "grad_norm": 0.9368489980697632,
+      "learning_rate": 4.727591413718282e-05,
+      "loss": 0.712,
+      "step": 5360
+    },
+    {
+      "epoch": 0.7536842105263157,
+      "grad_norm": 1.1019880771636963,
+      "learning_rate": 4.7265892351224694e-05,
+      "loss": 0.7172,
+      "step": 5370
+    },
+    {
+      "epoch": 0.7550877192982456,
+      "grad_norm": 0.8168277144432068,
+      "learning_rate": 4.725585323050623e-05,
+      "loss": 0.6812,
+      "step": 5380
+    },
+    {
+      "epoch": 0.7564912280701754,
+      "grad_norm": 1.0383678674697876,
+      "learning_rate": 4.72457967828432e-05,
+      "loss": 0.6266,
+      "step": 5390
+    },
+    {
+      "epoch": 0.7578947368421053,
+      "grad_norm": 1.9418814182281494,
+      "learning_rate": 4.723572301606492e-05,
+      "loss": 0.6976,
+      "step": 5400
+    },
+    {
+      "epoch": 0.7592982456140351,
+      "grad_norm": 1.1380218267440796,
+      "learning_rate": 4.7225631938014134e-05,
+      "loss": 0.709,
+      "step": 5410
+    },
+    {
+      "epoch": 0.7607017543859649,
+      "grad_norm": 0.7876071333885193,
+      "learning_rate": 4.7215523556547116e-05,
+      "loss": 0.5956,
+      "step": 5420
+    },
+    {
+      "epoch": 0.7621052631578947,
+      "grad_norm": 0.9458256363868713,
+      "learning_rate": 4.720539787953357e-05,
+      "loss": 0.6943,
+      "step": 5430
+    },
+    {
+      "epoch": 0.7635087719298246,
+      "grad_norm": 0.6351762413978577,
+      "learning_rate": 4.71952549148567e-05,
+      "loss": 0.6322,
+      "step": 5440
+    },
+    {
+      "epoch": 0.7649122807017544,
+      "grad_norm": 0.8464050889015198,
+      "learning_rate": 4.7185094670413134e-05,
+      "loss": 0.6258,
+      "step": 5450
+    },
+    {
+      "epoch": 0.7663157894736842,
+      "grad_norm": 1.7159314155578613,
+      "learning_rate": 4.7174917154112984e-05,
+      "loss": 0.6347,
+      "step": 5460
+    },
+    {
+      "epoch": 0.767719298245614,
+      "grad_norm": 0.8159227967262268,
+      "learning_rate": 4.716472237387979e-05,
+      "loss": 0.6423,
+      "step": 5470
+    },
+    {
+      "epoch": 0.7691228070175439,
+      "grad_norm": 1.1517149209976196,
+      "learning_rate": 4.715451033765054e-05,
+      "loss": 0.6614,
+      "step": 5480
+    },
+    {
+      "epoch": 0.7705263157894737,
+      "grad_norm": 1.164534091949463,
+      "learning_rate": 4.714428105337565e-05,
+      "loss": 0.6326,
+      "step": 5490
+    },
+    {
+      "epoch": 0.7719298245614035,
+      "grad_norm": 1.0906124114990234,
+      "learning_rate": 4.713403452901898e-05,
+      "loss": 0.6146,
+      "step": 5500
+    },
+    {
+      "epoch": 0.7733333333333333,
+      "grad_norm": 0.7224928140640259,
+      "learning_rate": 4.7123770772557774e-05,
+      "loss": 0.6061,
+      "step": 5510
+    },
+    {
+      "epoch": 0.7747368421052632,
+      "grad_norm": 1.1344630718231201,
+      "learning_rate": 4.711348979198274e-05,
+      "loss": 0.7423,
+      "step": 5520
+    },
+    {
+      "epoch": 0.776140350877193,
+      "grad_norm": 1.0616703033447266,
+      "learning_rate": 4.710319159529798e-05,
+      "loss": 0.6648,
+      "step": 5530
+    },
+    {
+      "epoch": 0.7775438596491228,
+      "grad_norm": 0.8563722968101501,
+      "learning_rate": 4.709287619052098e-05,
+      "loss": 0.551,
+      "step": 5540
+    },
+    {
+      "epoch": 0.7789473684210526,
+      "grad_norm": 0.7541974186897278,
+      "learning_rate": 4.708254358568264e-05,
+      "loss": 0.7394,
+      "step": 5550
+    },
+    {
+      "epoch": 0.7803508771929825,
+      "grad_norm": 0.9201952815055847,
+      "learning_rate": 4.7072193788827236e-05,
+      "loss": 0.57,
+      "step": 5560
+    },
+    {
+      "epoch": 0.7817543859649123,
+      "grad_norm": 0.8615202307701111,
+      "learning_rate": 4.706182680801245e-05,
+      "loss": 0.6293,
+      "step": 5570
+    },
+    {
+      "epoch": 0.783157894736842,
+      "grad_norm": 0.8534351587295532,
+      "learning_rate": 4.705144265130934e-05,
+      "loss": 0.6007,
+      "step": 5580
+    },
+    {
+      "epoch": 0.7845614035087719,
+      "grad_norm": 0.8691478967666626,
+      "learning_rate": 4.704104132680231e-05,
+      "loss": 0.5963,
+      "step": 5590
+    },
+    {
+      "epoch": 0.7859649122807018,
+      "grad_norm": 1.209688663482666,
+      "learning_rate": 4.703062284258916e-05,
+      "loss": 0.7237,
+      "step": 5600
+    },
+    {
+      "epoch": 0.7873684210526316,
+      "grad_norm": 0.72704017162323,
+      "learning_rate": 4.702018720678103e-05,
+      "loss": 0.6452,
+      "step": 5610
+    },
+    {
+      "epoch": 0.7887719298245615,
+      "grad_norm": 1.3118873834609985,
+      "learning_rate": 4.7009734427502426e-05,
+      "loss": 0.6291,
+      "step": 5620
+    },
+    {
+      "epoch": 0.7901754385964912,
+      "grad_norm": 0.6223419308662415,
+      "learning_rate": 4.699926451289119e-05,
+      "loss": 0.5925,
+      "step": 5630
+    },
+    {
+      "epoch": 0.791578947368421,
+      "grad_norm": 1.0733870267868042,
+      "learning_rate": 4.698877747109852e-05,
+      "loss": 0.7342,
+      "step": 5640
+    },
+    {
+      "epoch": 0.7929824561403509,
+      "grad_norm": 0.7960459589958191,
+      "learning_rate": 4.697827331028893e-05,
+      "loss": 0.644,
+      "step": 5650
+    },
+    {
+      "epoch": 0.7943859649122808,
+      "grad_norm": 0.9189769625663757,
+      "learning_rate": 4.6967752038640264e-05,
+      "loss": 0.6567,
+      "step": 5660
+    },
+    {
+      "epoch": 0.7957894736842105,
+      "grad_norm": 1.1323273181915283,
+      "learning_rate": 4.695721366434369e-05,
+      "loss": 0.6873,
+      "step": 5670
+    },
+    {
+      "epoch": 0.7971929824561403,
+      "grad_norm": 0.8580273389816284,
+      "learning_rate": 4.694665819560371e-05,
+      "loss": 0.6733,
+      "step": 5680
+    },
+    {
+      "epoch": 0.7985964912280702,
+      "grad_norm": 1.3165494203567505,
+      "learning_rate": 4.693608564063811e-05,
+      "loss": 0.642,
+      "step": 5690
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7017198801040649,
+      "learning_rate": 4.692549600767798e-05,
+      "loss": 0.5438,
+      "step": 5700
+    },
+    {
+      "epoch": 0.8014035087719298,
+      "grad_norm": 0.8478591442108154,
+      "learning_rate": 4.6914889304967725e-05,
+      "loss": 0.6107,
+      "step": 5710
+    },
+    {
+      "epoch": 0.8028070175438596,
+      "grad_norm": 0.9716276526451111,
+      "learning_rate": 4.690426554076501e-05,
+      "loss": 0.5975,
+      "step": 5720
+    },
+    {
+      "epoch": 0.8042105263157895,
+      "grad_norm": 1.0631777048110962,
+      "learning_rate": 4.689362472334082e-05,
+      "loss": 0.6563,
+      "step": 5730
+    },
+    {
+      "epoch": 0.8056140350877193,
+      "grad_norm": 0.9736322164535522,
+      "learning_rate": 4.688296686097937e-05,
+      "loss": 0.6199,
+      "step": 5740
+    },
+    {
+      "epoch": 0.8070175438596491,
+      "grad_norm": 0.9049164652824402,
+      "learning_rate": 4.6872291961978195e-05,
+      "loss": 0.5772,
+      "step": 5750
+    },
+    {
+      "epoch": 0.8084210526315789,
+      "grad_norm": 1.2360827922821045,
+      "learning_rate": 4.6861600034648064e-05,
+      "loss": 0.6401,
+      "step": 5760
+    },
+    {
+      "epoch": 0.8098245614035088,
+      "grad_norm": 1.2036852836608887,
+      "learning_rate": 4.6850891087313e-05,
+      "loss": 0.6087,
+      "step": 5770
+    },
+    {
+      "epoch": 0.8112280701754386,
+      "grad_norm": 1.010108470916748,
+      "learning_rate": 4.6840165128310296e-05,
+      "loss": 0.6973,
+      "step": 5780
+    },
+    {
+      "epoch": 0.8126315789473684,
+      "grad_norm": 1.1753820180892944,
+      "learning_rate": 4.6829422165990475e-05,
+      "loss": 0.6509,
+      "step": 5790
+    },
+    {
+      "epoch": 0.8140350877192982,
+      "grad_norm": 1.0416866540908813,
+      "learning_rate": 4.6818662208717296e-05,
+      "loss": 0.6092,
+      "step": 5800
+    },
+    {
+      "epoch": 0.8154385964912281,
+      "grad_norm": 0.7539423108100891,
+      "learning_rate": 4.680788526486776e-05,
+      "loss": 0.5864,
+      "step": 5810
+    },
+    {
+      "epoch": 0.8168421052631579,
+      "grad_norm": 0.9227228164672852,
+      "learning_rate": 4.679709134283209e-05,
+      "loss": 0.5736,
+      "step": 5820
+    },
+    {
+      "epoch": 0.8182456140350877,
+      "grad_norm": 0.8869969844818115,
+      "learning_rate": 4.678628045101371e-05,
+      "loss": 0.5982,
+      "step": 5830
+    },
+    {
+      "epoch": 0.8196491228070175,
+      "grad_norm": 0.6802515387535095,
+      "learning_rate": 4.677545259782929e-05,
+      "loss": 0.6136,
+      "step": 5840
+    },
+    {
+      "epoch": 0.8210526315789474,
+      "grad_norm": 0.9065477848052979,
+      "learning_rate": 4.676460779170867e-05,
+      "loss": 0.6519,
+      "step": 5850
+    },
+    {
+      "epoch": 0.8224561403508772,
+      "grad_norm": 1.3136307001113892,
+      "learning_rate": 4.675374604109491e-05,
+      "loss": 0.6122,
+      "step": 5860
+    },
+    {
+      "epoch": 0.8238596491228071,
+      "grad_norm": 0.9648601412773132,
+      "learning_rate": 4.6742867354444256e-05,
+      "loss": 0.5582,
+      "step": 5870
+    },
+    {
+      "epoch": 0.8252631578947368,
+      "grad_norm": 1.085227370262146,
+      "learning_rate": 4.673197174022613e-05,
+      "loss": 0.6788,
+      "step": 5880
+    },
+    {
+      "epoch": 0.8266666666666667,
+      "grad_norm": 0.9866172075271606,
+      "learning_rate": 4.672105920692316e-05,
+      "loss": 0.647,
+      "step": 5890
+    },
+    {
+      "epoch": 0.8280701754385965,
+      "grad_norm": 1.5403311252593994,
+      "learning_rate": 4.6710129763031095e-05,
+      "loss": 0.6326,
+      "step": 5900
+    },
+    {
+      "epoch": 0.8294736842105264,
+      "grad_norm": 1.195082187652588,
+      "learning_rate": 4.669918341705891e-05,
+      "loss": 0.6205,
+      "step": 5910
+    },
+    {
+      "epoch": 0.8308771929824561,
+      "grad_norm": 0.9392557740211487,
+      "learning_rate": 4.66882201775287e-05,
+      "loss": 0.699,
+      "step": 5920
+    },
+    {
+      "epoch": 0.832280701754386,
+      "grad_norm": 1.280907392501831,
+      "learning_rate": 4.667724005297573e-05,
+      "loss": 0.6147,
+      "step": 5930
+    },
+    {
+      "epoch": 0.8336842105263158,
+      "grad_norm": 0.6876835823059082,
+      "learning_rate": 4.66662430519484e-05,
+      "loss": 0.5737,
+      "step": 5940
+    },
+    {
+      "epoch": 0.8350877192982457,
+      "grad_norm": 0.7067710161209106,
+      "learning_rate": 4.665522918300823e-05,
+      "loss": 0.6072,
+      "step": 5950
+    },
+    {
+      "epoch": 0.8364912280701754,
+      "grad_norm": 1.0336652994155884,
+      "learning_rate": 4.6644198454729933e-05,
+      "loss": 0.6296,
+      "step": 5960
+    },
+    {
+      "epoch": 0.8378947368421052,
+      "grad_norm": 1.3756647109985352,
+      "learning_rate": 4.663315087570128e-05,
+      "loss": 0.6489,
+      "step": 5970
+    },
+    {
+      "epoch": 0.8392982456140351,
+      "grad_norm": 1.0433988571166992,
+      "learning_rate": 4.662208645452321e-05,
+      "loss": 0.6742,
+      "step": 5980
+    },
+    {
+      "epoch": 0.840701754385965,
+      "grad_norm": 0.6354380249977112,
+      "learning_rate": 4.661100519980973e-05,
+      "loss": 0.573,
+      "step": 5990
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 1.112243890762329,
+      "learning_rate": 4.6599907120188005e-05,
+      "loss": 0.6455,
+      "step": 6000
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "eval_loss": 0.6415141820907593,
+      "eval_runtime": 44.3465,
+      "eval_samples_per_second": 33.825,
+      "eval_steps_per_second": 8.456,
+      "step": 6000
+    },
+    {
+      "epoch": 0.8435087719298245,
+      "grad_norm": 0.781201958656311,
+      "learning_rate": 4.658879222429825e-05,
+      "loss": 0.5362,
+      "step": 6010
+    },
+    {
+      "epoch": 0.8449122807017544,
+      "grad_norm": 1.069032073020935,
+      "learning_rate": 4.65776605207938e-05,
+      "loss": 0.6321,
+      "step": 6020
+    },
+    {
+      "epoch": 0.8463157894736842,
+      "grad_norm": 1.0449451208114624,
+      "learning_rate": 4.656651201834106e-05,
+      "loss": 0.6208,
+      "step": 6030
+    },
+    {
+      "epoch": 0.847719298245614,
+      "grad_norm": 1.9674957990646362,
+      "learning_rate": 4.655534672561953e-05,
+      "loss": 0.6529,
+      "step": 6040
+    },
+    {
+      "epoch": 0.8491228070175438,
+      "grad_norm": 0.9335805773735046,
+      "learning_rate": 4.654416465132177e-05,
+      "loss": 0.6515,
+      "step": 6050
+    },
+    {
+      "epoch": 0.8505263157894737,
+      "grad_norm": 0.8951327800750732,
+      "learning_rate": 4.6532965804153416e-05,
+      "loss": 0.613,
+      "step": 6060
+    },
+    {
+      "epoch": 0.8519298245614035,
+      "grad_norm": 1.1679803133010864,
+      "learning_rate": 4.652175019283314e-05,
+      "loss": 0.6215,
+      "step": 6070
+    },
+    {
+      "epoch": 0.8533333333333334,
+      "grad_norm": 1.6771854162216187,
+      "learning_rate": 4.6510517826092695e-05,
+      "loss": 0.7427,
+      "step": 6080
+    },
+    {
+      "epoch": 0.8547368421052631,
+      "grad_norm": 0.6942294836044312,
+      "learning_rate": 4.649926871267685e-05,
+      "loss": 0.58,
+      "step": 6090
+    },
+    {
+      "epoch": 0.856140350877193,
+      "grad_norm": 1.114723801612854,
+      "learning_rate": 4.6488002861343425e-05,
+      "loss": 0.6916,
+      "step": 6100
+    },
+    {
+      "epoch": 0.8575438596491228,
+      "grad_norm": 0.9489352107048035,
+      "learning_rate": 4.647672028086328e-05,
+      "loss": 0.6073,
+      "step": 6110
+    },
+    {
+      "epoch": 0.8589473684210527,
+      "grad_norm": 0.8159108757972717,
+      "learning_rate": 4.646542098002029e-05,
+      "loss": 0.6273,
+      "step": 6120
+    },
+    {
+      "epoch": 0.8603508771929824,
+      "grad_norm": 1.2675360441207886,
+      "learning_rate": 4.645410496761135e-05,
+      "loss": 0.6657,
+      "step": 6130
+    },
+    {
+      "epoch": 0.8617543859649123,
+      "grad_norm": 0.9706358313560486,
+      "learning_rate": 4.644277225244635e-05,
+      "loss": 0.6861,
+      "step": 6140
+    },
+    {
+      "epoch": 0.8631578947368421,
+      "grad_norm": 1.0342049598693848,
+      "learning_rate": 4.6431422843348216e-05,
+      "loss": 0.6834,
+      "step": 6150
+    },
+    {
+      "epoch": 0.864561403508772,
+      "grad_norm": 0.9016236066818237,
+      "learning_rate": 4.642005674915284e-05,
+      "loss": 0.6098,
+      "step": 6160
+    },
+    {
+      "epoch": 0.8659649122807017,
+      "grad_norm": 0.8684419989585876,
+      "learning_rate": 4.640867397870912e-05,
+      "loss": 0.6831,
+      "step": 6170
+    },
+    {
+      "epoch": 0.8673684210526316,
+      "grad_norm": 0.8743478059768677,
+      "learning_rate": 4.639727454087892e-05,
+      "loss": 0.5846,
+      "step": 6180
+    },
+    {
+      "epoch": 0.8687719298245614,
+      "grad_norm": 1.0925372838974,
+      "learning_rate": 4.638585844453711e-05,
+      "loss": 0.6436,
+      "step": 6190
+    },
+    {
+      "epoch": 0.8701754385964913,
+      "grad_norm": 1.0224460363388062,
+      "learning_rate": 4.6374425698571514e-05,
+      "loss": 0.7538,
+      "step": 6200
+    },
+    {
+      "epoch": 0.871578947368421,
+      "grad_norm": 0.8540046215057373,
+      "learning_rate": 4.63629763118829e-05,
+      "loss": 0.596,
+      "step": 6210
+    },
+    {
+      "epoch": 0.8729824561403509,
+      "grad_norm": 0.9685525298118591,
+      "learning_rate": 4.6351510293385026e-05,
+      "loss": 0.5844,
+      "step": 6220
+    },
+    {
+      "epoch": 0.8743859649122807,
+      "grad_norm": 0.9988105893135071,
+      "learning_rate": 4.634002765200456e-05,
+      "loss": 0.5785,
+      "step": 6230
+    },
+    {
+      "epoch": 0.8757894736842106,
+      "grad_norm": 0.7331526279449463,
+      "learning_rate": 4.632852839668115e-05,
+      "loss": 0.5728,
+      "step": 6240
+    },
+    {
+      "epoch": 0.8771929824561403,
+      "grad_norm": 1.0520068407058716,
+      "learning_rate": 4.6317012536367354e-05,
+      "loss": 0.6317,
+      "step": 6250
+    },
+    {
+      "epoch": 0.8785964912280702,
+      "grad_norm": 1.117604374885559,
+      "learning_rate": 4.630548008002866e-05,
+      "loss": 0.6152,
+      "step": 6260
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7635726928710938,
+      "learning_rate": 4.629393103664349e-05,
+      "loss": 0.64,
+      "step": 6270
+    },
+    {
+      "epoch": 0.8814035087719299,
+      "grad_norm": 1.1754323244094849,
+      "learning_rate": 4.6282365415203164e-05,
+      "loss": 0.5923,
+      "step": 6280
+    },
+    {
+      "epoch": 0.8828070175438596,
+      "grad_norm": 0.6220813989639282,
+      "learning_rate": 4.627078322471191e-05,
+      "loss": 0.6745,
+      "step": 6290
+    },
+    {
+      "epoch": 0.8842105263157894,
+      "grad_norm": 0.8440349698066711,
+      "learning_rate": 4.625918447418687e-05,
+      "loss": 0.5819,
+      "step": 6300
+    },
+    {
+      "epoch": 0.8856140350877193,
+      "grad_norm": 1.0416796207427979,
+      "learning_rate": 4.624756917265807e-05,
+      "loss": 0.5965,
+      "step": 6310
+    },
+    {
+      "epoch": 0.8870175438596491,
+      "grad_norm": 1.1395715475082397,
+      "learning_rate": 4.62359373291684e-05,
+      "loss": 0.5838,
+      "step": 6320
+    },
+    {
+      "epoch": 0.888421052631579,
+      "grad_norm": 1.3868945837020874,
+      "learning_rate": 4.622428895277367e-05,
+      "loss": 0.7304,
+      "step": 6330
+    },
+    {
+      "epoch": 0.8898245614035087,
+      "grad_norm": 1.0000405311584473,
+      "learning_rate": 4.621262405254253e-05,
+      "loss": 0.5938,
+      "step": 6340
+    },
+    {
+      "epoch": 0.8912280701754386,
+      "grad_norm": 0.754399836063385,
+      "learning_rate": 4.620094263755652e-05,
+      "loss": 0.6276,
+      "step": 6350
+    },
+    {
+      "epoch": 0.8926315789473684,
+      "grad_norm": 0.9784127473831177,
+      "learning_rate": 4.618924471691e-05,
+      "loss": 0.613,
+      "step": 6360
+    },
+    {
+      "epoch": 0.8940350877192983,
+      "grad_norm": 0.6419925689697266,
+      "learning_rate": 4.617753029971021e-05,
+      "loss": 0.599,
+      "step": 6370
+    },
+    {
+      "epoch": 0.895438596491228,
+      "grad_norm": 1.2562180757522583,
+      "learning_rate": 4.6165799395077236e-05,
+      "loss": 0.6358,
+      "step": 6380
+    },
+    {
+      "epoch": 0.8968421052631579,
+      "grad_norm": 1.1815166473388672,
+      "learning_rate": 4.615405201214398e-05,
+      "loss": 0.6747,
+      "step": 6390
+    },
+    {
+      "epoch": 0.8982456140350877,
+      "grad_norm": 1.5243850946426392,
+      "learning_rate": 4.614228816005618e-05,
+      "loss": 0.6082,
+      "step": 6400
+    },
+    {
+      "epoch": 0.8996491228070176,
+      "grad_norm": 0.894396960735321,
+      "learning_rate": 4.61305078479724e-05,
+      "loss": 0.5506,
+      "step": 6410
+    },
+    {
+      "epoch": 0.9010526315789473,
+      "grad_norm": 0.7782644629478455,
+      "learning_rate": 4.611871108506403e-05,
+      "loss": 0.5816,
+      "step": 6420
+    },
+    {
+      "epoch": 0.9024561403508772,
+      "grad_norm": 1.2209144830703735,
+      "learning_rate": 4.610689788051523e-05,
+      "loss": 0.6178,
+      "step": 6430
+    },
+    {
+      "epoch": 0.903859649122807,
+      "grad_norm": 0.8224475979804993,
+      "learning_rate": 4.6095068243523e-05,
+      "loss": 0.644,
+      "step": 6440
+    },
+    {
+      "epoch": 0.9052631578947369,
+      "grad_norm": 1.054763674736023,
+      "learning_rate": 4.608322218329711e-05,
+      "loss": 0.5564,
+      "step": 6450
+    },
+    {
+      "epoch": 0.9066666666666666,
+      "grad_norm": 0.9225585460662842,
+      "learning_rate": 4.607135970906014e-05,
+      "loss": 0.6534,
+      "step": 6460
+    },
+    {
+      "epoch": 0.9080701754385965,
+      "grad_norm": 0.7979352474212646,
+      "learning_rate": 4.605948083004741e-05,
+      "loss": 0.6671,
+      "step": 6470
+    },
+    {
+      "epoch": 0.9094736842105263,
+      "grad_norm": 1.223375916481018,
+      "learning_rate": 4.6047585555507045e-05,
+      "loss": 0.6996,
+      "step": 6480
+    },
+    {
+      "epoch": 0.9108771929824562,
+      "grad_norm": 0.803092360496521,
+      "learning_rate": 4.603567389469993e-05,
+      "loss": 0.5937,
+      "step": 6490
+    },
+    {
+      "epoch": 0.9122807017543859,
+      "grad_norm": 0.9455748796463013,
+      "learning_rate": 4.60237458568997e-05,
+      "loss": 0.555,
+      "step": 6500
+    },
+    {
+      "epoch": 0.9136842105263158,
+      "grad_norm": 1.0315808057785034,
+      "learning_rate": 4.6011801451392736e-05,
+      "loss": 0.6721,
+      "step": 6510
+    },
+    {
+      "epoch": 0.9150877192982456,
+      "grad_norm": 1.0508462190628052,
+      "learning_rate": 4.5999840687478167e-05,
+      "loss": 0.5849,
+      "step": 6520
+    },
+    {
+      "epoch": 0.9164912280701755,
+      "grad_norm": 0.8299797773361206,
+      "learning_rate": 4.598786357446786e-05,
+      "loss": 0.6013,
+      "step": 6530
+    },
+    {
+      "epoch": 0.9178947368421052,
+      "grad_norm": 0.8177257776260376,
+      "learning_rate": 4.5975870121686406e-05,
+      "loss": 0.6178,
+      "step": 6540
+    },
+    {
+      "epoch": 0.9192982456140351,
+      "grad_norm": 0.8297099471092224,
+      "learning_rate": 4.596386033847111e-05,
+      "loss": 0.5985,
+      "step": 6550
+    },
+    {
+      "epoch": 0.9207017543859649,
+      "grad_norm": 1.1290909051895142,
+      "learning_rate": 4.5951834234172025e-05,
+      "loss": 0.5878,
+      "step": 6560
+    },
+    {
+      "epoch": 0.9221052631578948,
+      "grad_norm": 0.7982479929924011,
+      "learning_rate": 4.593979181815187e-05,
+      "loss": 0.7004,
+      "step": 6570
+    },
+    {
+      "epoch": 0.9235087719298246,
+      "grad_norm": 0.9358506202697754,
+      "learning_rate": 4.5927733099786066e-05,
+      "loss": 0.615,
+      "step": 6580
+    },
+    {
+      "epoch": 0.9249122807017544,
+      "grad_norm": 0.8648248910903931,
+      "learning_rate": 4.591565808846276e-05,
+      "loss": 0.5789,
+      "step": 6590
+    },
+    {
+      "epoch": 0.9263157894736842,
+      "grad_norm": 1.1170403957366943,
+      "learning_rate": 4.5903566793582755e-05,
+      "loss": 0.5827,
+      "step": 6600
+    },
+    {
+      "epoch": 0.927719298245614,
+      "grad_norm": 1.5100983381271362,
+      "learning_rate": 4.589145922455954e-05,
+      "loss": 0.6176,
+      "step": 6610
+    },
+    {
+      "epoch": 0.9291228070175439,
+      "grad_norm": 1.2393382787704468,
+      "learning_rate": 4.587933539081927e-05,
+      "loss": 0.7056,
+      "step": 6620
+    },
+    {
+      "epoch": 0.9305263157894736,
+      "grad_norm": 0.8480477333068848,
+      "learning_rate": 4.586719530180075e-05,
+      "loss": 0.6418,
+      "step": 6630
+    },
+    {
+      "epoch": 0.9319298245614035,
+      "grad_norm": 1.1262218952178955,
+      "learning_rate": 4.585503896695549e-05,
+      "loss": 0.5231,
+      "step": 6640
+    },
+    {
+      "epoch": 0.9333333333333333,
+      "grad_norm": 1.1744680404663086,
+      "learning_rate": 4.584286639574758e-05,
+      "loss": 0.7463,
+      "step": 6650
+    },
+    {
+      "epoch": 0.9347368421052632,
+      "grad_norm": 1.0590097904205322,
+      "learning_rate": 4.58306775976538e-05,
+      "loss": 0.6373,
+      "step": 6660
+    },
+    {
+      "epoch": 0.9361403508771929,
+      "grad_norm": 1.1323457956314087,
+      "learning_rate": 4.581847258216355e-05,
+      "loss": 0.6609,
+      "step": 6670
+    },
+    {
+      "epoch": 0.9375438596491228,
+      "grad_norm": 1.043113350868225,
+      "learning_rate": 4.580625135877884e-05,
+      "loss": 0.6366,
+      "step": 6680
+    },
+    {
+      "epoch": 0.9389473684210526,
+      "grad_norm": 0.6503088474273682,
+      "learning_rate": 4.5794013937014326e-05,
+      "loss": 0.6359,
+      "step": 6690
+    },
+    {
+      "epoch": 0.9403508771929825,
+      "grad_norm": 1.6966040134429932,
+      "learning_rate": 4.578176032639724e-05,
+      "loss": 0.7239,
+      "step": 6700
+    },
+    {
+      "epoch": 0.9417543859649122,
+      "grad_norm": 0.7537420988082886,
+      "learning_rate": 4.5769490536467465e-05,
+      "loss": 0.5848,
+      "step": 6710
+    },
+    {
+      "epoch": 0.9431578947368421,
+      "grad_norm": 1.1561657190322876,
+      "learning_rate": 4.5757204576777437e-05,
+      "loss": 0.6155,
+      "step": 6720
+    },
+    {
+      "epoch": 0.9445614035087719,
+      "grad_norm": 0.6912992000579834,
+      "learning_rate": 4.574490245689219e-05,
+      "loss": 0.5817,
+      "step": 6730
+    },
+    {
+      "epoch": 0.9459649122807018,
+      "grad_norm": 1.0430986881256104,
+      "learning_rate": 4.573258418638936e-05,
+      "loss": 0.4903,
+      "step": 6740
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.9669928550720215,
+      "learning_rate": 4.572024977485914e-05,
+      "loss": 0.6289,
+      "step": 6750
+    },
+    {
+      "epoch": 0.9487719298245614,
+      "grad_norm": 0.9165218472480774,
+      "learning_rate": 4.5707899231904286e-05,
+      "loss": 0.6133,
+      "step": 6760
+    },
+    {
+      "epoch": 0.9501754385964912,
+      "grad_norm": 0.9709174036979675,
+      "learning_rate": 4.569553256714012e-05,
+      "loss": 0.638,
+      "step": 6770
+    },
+    {
+      "epoch": 0.9515789473684211,
+      "grad_norm": 1.2939212322235107,
+      "learning_rate": 4.5683149790194526e-05,
+      "loss": 0.7599,
+      "step": 6780
+    },
+    {
+      "epoch": 0.9529824561403509,
+      "grad_norm": 1.0535619258880615,
+      "learning_rate": 4.5670750910707903e-05,
+      "loss": 0.6906,
+      "step": 6790
+    },
+    {
+      "epoch": 0.9543859649122807,
+      "grad_norm": 0.9928086400032043,
+      "learning_rate": 4.565833593833321e-05,
+      "loss": 0.7387,
+      "step": 6800
+    },
+    {
+      "epoch": 0.9557894736842105,
+      "grad_norm": 0.7087190747261047,
+      "learning_rate": 4.5645904882735935e-05,
+      "loss": 0.5566,
+      "step": 6810
+    },
+    {
+      "epoch": 0.9571929824561404,
+      "grad_norm": 1.2111977338790894,
+      "learning_rate": 4.563345775359408e-05,
+      "loss": 0.5748,
+      "step": 6820
+    },
+    {
+      "epoch": 0.9585964912280702,
+      "grad_norm": 1.2516281604766846,
+      "learning_rate": 4.562099456059815e-05,
+      "loss": 0.6256,
+      "step": 6830
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.8081939220428467,
+      "learning_rate": 4.5608515313451186e-05,
+      "loss": 0.5826,
+      "step": 6840
+    },
+    {
+      "epoch": 0.9614035087719298,
+      "grad_norm": 1.453393578529358,
+      "learning_rate": 4.559602002186869e-05,
+      "loss": 0.5538,
+      "step": 6850
+    },
+    {
+      "epoch": 0.9628070175438597,
+      "grad_norm": 1.139618158340454,
+      "learning_rate": 4.558350869557868e-05,
+      "loss": 0.6514,
+      "step": 6860
+    },
+    {
+      "epoch": 0.9642105263157895,
+      "grad_norm": 0.9846227765083313,
+      "learning_rate": 4.557098134432167e-05,
+      "loss": 0.7813,
+      "step": 6870
+    },
+    {
+      "epoch": 0.9656140350877193,
+      "grad_norm": 0.8734840750694275,
+      "learning_rate": 4.555843797785061e-05,
+      "loss": 0.5993,
+      "step": 6880
+    },
+    {
+      "epoch": 0.9670175438596491,
+      "grad_norm": 1.172455072402954,
+      "learning_rate": 4.554587860593095e-05,
+      "loss": 0.5594,
+      "step": 6890
+    },
+    {
+      "epoch": 0.968421052631579,
+      "grad_norm": 0.5644223690032959,
+      "learning_rate": 4.553330323834059e-05,
+      "loss": 0.5581,
+      "step": 6900
+    },
+    {
+      "epoch": 0.9698245614035088,
+      "grad_norm": 0.7265453338623047,
+      "learning_rate": 4.552071188486989e-05,
+      "loss": 0.6295,
+      "step": 6910
+    },
+    {
+      "epoch": 0.9712280701754386,
+      "grad_norm": 0.8341143727302551,
+      "learning_rate": 4.550810455532164e-05,
+      "loss": 0.5697,
+      "step": 6920
+    },
+    {
+      "epoch": 0.9726315789473684,
+      "grad_norm": 0.7036447525024414,
+      "learning_rate": 4.5495481259511095e-05,
+      "loss": 0.5933,
+      "step": 6930
+    },
+    {
+      "epoch": 0.9740350877192983,
+      "grad_norm": 0.972158670425415,
+      "learning_rate": 4.54828420072659e-05,
+      "loss": 0.5723,
+      "step": 6940
+    },
+    {
+      "epoch": 0.9754385964912281,
+      "grad_norm": 1.3979262113571167,
+      "learning_rate": 4.547018680842616e-05,
+      "loss": 0.5749,
+      "step": 6950
+    },
+    {
+      "epoch": 0.9768421052631578,
+      "grad_norm": 1.3824502229690552,
+      "learning_rate": 4.545751567284439e-05,
+      "loss": 0.7053,
+      "step": 6960
+    },
+    {
+      "epoch": 0.9782456140350877,
+      "grad_norm": 1.1198428869247437,
+      "learning_rate": 4.5444828610385486e-05,
+      "loss": 0.629,
+      "step": 6970
+    },
+    {
+      "epoch": 0.9796491228070175,
+      "grad_norm": 0.7075201869010925,
+      "learning_rate": 4.543212563092677e-05,
+      "loss": 0.6647,
+      "step": 6980
+    },
+    {
+      "epoch": 0.9810526315789474,
+      "grad_norm": 1.0392223596572876,
+      "learning_rate": 4.541940674435794e-05,
+      "loss": 0.6921,
+      "step": 6990
+    },
+    {
+      "epoch": 0.9824561403508771,
+      "grad_norm": 1.027004599571228,
+      "learning_rate": 4.5406671960581096e-05,
+      "loss": 0.6187,
+      "step": 7000
+    },
+    {
+      "epoch": 0.983859649122807,
+      "grad_norm": 1.2175973653793335,
+      "learning_rate": 4.53939212895107e-05,
+      "loss": 0.6471,
+      "step": 7010
+    },
+    {
+      "epoch": 0.9852631578947368,
+      "grad_norm": 1.0904464721679688,
+      "learning_rate": 4.538115474107357e-05,
+      "loss": 0.5916,
+      "step": 7020
+    },
+    {
+      "epoch": 0.9866666666666667,
+      "grad_norm": 1.0575454235076904,
+      "learning_rate": 4.536837232520893e-05,
+      "loss": 0.6859,
+      "step": 7030
+    },
+    {
+      "epoch": 0.9880701754385965,
+      "grad_norm": 0.7100856304168701,
+      "learning_rate": 4.535557405186831e-05,
+      "loss": 0.64,
+      "step": 7040
+    },
+    {
+      "epoch": 0.9894736842105263,
+      "grad_norm": 0.9754576683044434,
+      "learning_rate": 4.534275993101561e-05,
+      "loss": 0.5839,
+      "step": 7050
+    },
+    {
+      "epoch": 0.9908771929824561,
+      "grad_norm": 0.8776949644088745,
+      "learning_rate": 4.532992997262706e-05,
+      "loss": 0.577,
+      "step": 7060
+    },
+    {
+      "epoch": 0.992280701754386,
+      "grad_norm": 1.569716215133667,
+      "learning_rate": 4.531708418669122e-05,
+      "loss": 0.7155,
+      "step": 7070
+    },
+    {
+      "epoch": 0.9936842105263158,
+      "grad_norm": 0.7174299955368042,
+      "learning_rate": 4.5304222583208983e-05,
+      "loss": 0.6713,
+      "step": 7080
+    },
+    {
+      "epoch": 0.9950877192982456,
+      "grad_norm": 1.0695897340774536,
+      "learning_rate": 4.5291345172193546e-05,
+      "loss": 0.6528,
+      "step": 7090
+    },
+    {
+      "epoch": 0.9964912280701754,
+      "grad_norm": 1.0685267448425293,
+      "learning_rate": 4.5278451963670403e-05,
+      "loss": 0.5705,
+      "step": 7100
+    },
+    {
+      "epoch": 0.9978947368421053,
+      "grad_norm": 1.2662453651428223,
+      "learning_rate": 4.526554296767738e-05,
+      "loss": 0.6763,
+      "step": 7110
+    },
+    {
+      "epoch": 0.9992982456140351,
+      "grad_norm": 1.1944515705108643,
+      "learning_rate": 4.525261819426455e-05,
+      "loss": 0.5855,
+      "step": 7120
+    },
+    {
+      "epoch": 1.0007017543859649,
+      "grad_norm": 0.7510038614273071,
+      "learning_rate": 4.5239677653494305e-05,
+      "loss": 0.5631,
+      "step": 7130
+    },
+    {
+      "epoch": 1.0021052631578948,
+      "grad_norm": 0.7383008003234863,
+      "learning_rate": 4.5226721355441306e-05,
+      "loss": 0.493,
+      "step": 7140
+    },
+    {
+      "epoch": 1.0035087719298246,
+      "grad_norm": 1.3719711303710938,
+      "learning_rate": 4.5213749310192455e-05,
+      "loss": 0.5049,
+      "step": 7150
+    },
+    {
+      "epoch": 1.0049122807017543,
+      "grad_norm": 0.7755573987960815,
+      "learning_rate": 4.520076152784695e-05,
+      "loss": 0.5204,
+      "step": 7160
+    },
+    {
+      "epoch": 1.0063157894736843,
+      "grad_norm": 1.0142574310302734,
+      "learning_rate": 4.518775801851622e-05,
+      "loss": 0.5395,
+      "step": 7170
+    },
+    {
+      "epoch": 1.007719298245614,
+      "grad_norm": 1.0278340578079224,
+      "learning_rate": 4.517473879232395e-05,
+      "loss": 0.5231,
+      "step": 7180
+    },
+    {
+      "epoch": 1.0091228070175438,
+      "grad_norm": 1.4633328914642334,
+      "learning_rate": 4.516170385940603e-05,
+      "loss": 0.5764,
+      "step": 7190
+    },
+    {
+      "epoch": 1.0105263157894737,
+      "grad_norm": 1.5323199033737183,
+      "learning_rate": 4.514865322991063e-05,
+      "loss": 0.5339,
+      "step": 7200
+    },
+    {
+      "epoch": 1.0119298245614035,
+      "grad_norm": 1.379055380821228,
+      "learning_rate": 4.51355869139981e-05,
+      "loss": 0.5684,
+      "step": 7210
+    },
+    {
+      "epoch": 1.0133333333333334,
+      "grad_norm": 0.9581233859062195,
+      "learning_rate": 4.512250492184101e-05,
+      "loss": 0.539,
+      "step": 7220
+    },
+    {
+      "epoch": 1.0147368421052632,
+      "grad_norm": 1.1807743310928345,
+      "learning_rate": 4.510940726362416e-05,
+      "loss": 0.5348,
+      "step": 7230
+    },
+    {
+      "epoch": 1.016140350877193,
+      "grad_norm": 1.2164653539657593,
+      "learning_rate": 4.50962939495445e-05,
+      "loss": 0.5795,
+      "step": 7240
+    },
+    {
+      "epoch": 1.0175438596491229,
+      "grad_norm": 0.7895627617835999,
+      "learning_rate": 4.50831649898112e-05,
+      "loss": 0.5894,
+      "step": 7250
+    },
+    {
+      "epoch": 1.0189473684210526,
+      "grad_norm": 1.4003938436508179,
+      "learning_rate": 4.507002039464562e-05,
+      "loss": 0.5825,
+      "step": 7260
+    },
+    {
+      "epoch": 1.0203508771929826,
+      "grad_norm": 0.8824437856674194,
+      "learning_rate": 4.505686017428127e-05,
+      "loss": 0.5513,
+      "step": 7270
+    },
+    {
+      "epoch": 1.0217543859649123,
+      "grad_norm": 1.0241267681121826,
+      "learning_rate": 4.504368433896382e-05,
+      "loss": 0.6675,
+      "step": 7280
+    },
+    {
+      "epoch": 1.023157894736842,
+      "grad_norm": 1.1416174173355103,
+      "learning_rate": 4.5030492898951134e-05,
+      "loss": 0.5146,
+      "step": 7290
+    },
+    {
+      "epoch": 1.024561403508772,
+      "grad_norm": 1.4306304454803467,
+      "learning_rate": 4.501728586451318e-05,
+      "loss": 0.6254,
+      "step": 7300
+    },
+    {
+      "epoch": 1.0259649122807017,
+      "grad_norm": 0.7919867634773254,
+      "learning_rate": 4.5004063245932097e-05,
+      "loss": 0.4688,
+      "step": 7310
+    },
+    {
+      "epoch": 1.0273684210526315,
+      "grad_norm": 1.0270862579345703,
+      "learning_rate": 4.4990825053502136e-05,
+      "loss": 0.5227,
+      "step": 7320
+    },
+    {
+      "epoch": 1.0287719298245614,
+      "grad_norm": 1.332261085510254,
+      "learning_rate": 4.497757129752969e-05,
+      "loss": 0.5219,
+      "step": 7330
+    },
+    {
+      "epoch": 1.0301754385964912,
+      "grad_norm": 1.1045178174972534,
+      "learning_rate": 4.496430198833327e-05,
+      "loss": 0.5006,
+      "step": 7340
+    },
+    {
+      "epoch": 1.0315789473684212,
+      "grad_norm": 1.069557547569275,
+      "learning_rate": 4.495101713624348e-05,
+      "loss": 0.519,
+      "step": 7350
+    },
+    {
+      "epoch": 1.032982456140351,
+      "grad_norm": 1.0084444284439087,
+      "learning_rate": 4.493771675160303e-05,
+      "loss": 0.6042,
+      "step": 7360
+    },
+    {
+      "epoch": 1.0343859649122806,
+      "grad_norm": 1.1198923587799072,
+      "learning_rate": 4.4924400844766734e-05,
+      "loss": 0.5092,
+      "step": 7370
+    },
+    {
+      "epoch": 1.0357894736842106,
+      "grad_norm": 0.7310131788253784,
+      "learning_rate": 4.491106942610147e-05,
+      "loss": 0.6178,
+      "step": 7380
+    },
+    {
+      "epoch": 1.0371929824561403,
+      "grad_norm": 1.1642857789993286,
+      "learning_rate": 4.489772250598622e-05,
+      "loss": 0.6226,
+      "step": 7390
+    },
+    {
+      "epoch": 1.03859649122807,
+      "grad_norm": 1.194806456565857,
+      "learning_rate": 4.488436009481201e-05,
+      "loss": 0.5821,
+      "step": 7400
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.3842540979385376,
+      "learning_rate": 4.487098220298193e-05,
+      "loss": 0.5265,
+      "step": 7410
+    },
+    {
+      "epoch": 1.0414035087719298,
+      "grad_norm": 1.546705722808838,
+      "learning_rate": 4.485758884091113e-05,
+      "loss": 0.5617,
+      "step": 7420
+    },
+    {
+      "epoch": 1.0428070175438597,
+      "grad_norm": 1.5181100368499756,
+      "learning_rate": 4.4844180019026805e-05,
+      "loss": 0.5468,
+      "step": 7430
+    },
+    {
+      "epoch": 1.0442105263157895,
+      "grad_norm": 1.8832321166992188,
+      "learning_rate": 4.483075574776819e-05,
+      "loss": 0.5048,
+      "step": 7440
+    },
+    {
+      "epoch": 1.0456140350877192,
+      "grad_norm": 1.2133930921554565,
+      "learning_rate": 4.4817316037586524e-05,
+      "loss": 0.5684,
+      "step": 7450
+    },
+    {
+      "epoch": 1.0470175438596492,
+      "grad_norm": 1.6424169540405273,
+      "learning_rate": 4.480386089894509e-05,
+      "loss": 0.5851,
+      "step": 7460
+    },
+    {
+      "epoch": 1.048421052631579,
+      "grad_norm": 1.2695761919021606,
+      "learning_rate": 4.479039034231918e-05,
+      "loss": 0.5308,
+      "step": 7470
+    },
+    {
+      "epoch": 1.0498245614035087,
+      "grad_norm": 1.3531373739242554,
+      "learning_rate": 4.477690437819607e-05,
+      "loss": 0.5904,
+      "step": 7480
+    },
+    {
+      "epoch": 1.0512280701754386,
+      "grad_norm": 1.6086102724075317,
+      "learning_rate": 4.476340301707507e-05,
+      "loss": 0.4894,
+      "step": 7490
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.7145791053771973,
+      "learning_rate": 4.4749886269467416e-05,
+      "loss": 0.4906,
+      "step": 7500
+    },
+    {
+      "epoch": 1.0540350877192983,
+      "grad_norm": 1.5852210521697998,
+      "learning_rate": 4.473635414589639e-05,
+      "loss": 0.5399,
+      "step": 7510
+    },
+    {
+      "epoch": 1.055438596491228,
+      "grad_norm": 1.3225674629211426,
+      "learning_rate": 4.47228066568972e-05,
+      "loss": 0.5168,
+      "step": 7520
+    },
+    {
+      "epoch": 1.0568421052631578,
+      "grad_norm": 1.3213186264038086,
+      "learning_rate": 4.470924381301704e-05,
+      "loss": 0.4888,
+      "step": 7530
+    },
+    {
+      "epoch": 1.0582456140350878,
+      "grad_norm": 1.4983114004135132,
+      "learning_rate": 4.469566562481503e-05,
+      "loss": 0.4909,
+      "step": 7540
+    },
+    {
+      "epoch": 1.0596491228070175,
+      "grad_norm": 1.3175050020217896,
+      "learning_rate": 4.4682072102862286e-05,
+      "loss": 0.5369,
+      "step": 7550
+    },
+    {
+      "epoch": 1.0610526315789475,
+      "grad_norm": 1.14377760887146,
+      "learning_rate": 4.466846325774179e-05,
+      "loss": 0.5046,
+      "step": 7560
+    },
+    {
+      "epoch": 1.0624561403508772,
+      "grad_norm": 0.6691097021102905,
+      "learning_rate": 4.4654839100048535e-05,
+      "loss": 0.5201,
+      "step": 7570
+    },
+    {
+      "epoch": 1.063859649122807,
+      "grad_norm": 1.4467300176620483,
+      "learning_rate": 4.464119964038937e-05,
+      "loss": 0.5238,
+      "step": 7580
+    },
+    {
+      "epoch": 1.065263157894737,
+      "grad_norm": 0.8880655169487,
+      "learning_rate": 4.462754488938309e-05,
+      "loss": 0.5074,
+      "step": 7590
+    },
+    {
+      "epoch": 1.0666666666666667,
+      "grad_norm": 2.043294906616211,
+      "learning_rate": 4.4613874857660384e-05,
+      "loss": 0.5297,
+      "step": 7600
+    },
+    {
+      "epoch": 1.0680701754385964,
+      "grad_norm": 1.054681420326233,
+      "learning_rate": 4.460018955586384e-05,
+      "loss": 0.5585,
+      "step": 7610
+    },
+    {
+      "epoch": 1.0694736842105264,
+      "grad_norm": 1.256369709968567,
+      "learning_rate": 4.458648899464793e-05,
+      "loss": 0.4944,
+      "step": 7620
+    },
+    {
+      "epoch": 1.070877192982456,
+      "grad_norm": 1.0441490411758423,
+      "learning_rate": 4.457277318467903e-05,
+      "loss": 0.5736,
+      "step": 7630
+    },
+    {
+      "epoch": 1.072280701754386,
+      "grad_norm": 0.885286271572113,
+      "learning_rate": 4.4559042136635345e-05,
+      "loss": 0.6152,
+      "step": 7640
+    },
+    {
+      "epoch": 1.0736842105263158,
+      "grad_norm": 1.8804951906204224,
+      "learning_rate": 4.4545295861206975e-05,
+      "loss": 0.4936,
+      "step": 7650
+    },
+    {
+      "epoch": 1.0750877192982455,
+      "grad_norm": 1.5045465230941772,
+      "learning_rate": 4.453153436909587e-05,
+      "loss": 0.5547,
+      "step": 7660
+    },
+    {
+      "epoch": 1.0764912280701755,
+      "grad_norm": 1.7368062734603882,
+      "learning_rate": 4.4517757671015826e-05,
+      "loss": 0.537,
+      "step": 7670
+    },
+    {
+      "epoch": 1.0778947368421052,
+      "grad_norm": 1.3677830696105957,
+      "learning_rate": 4.4503965777692456e-05,
+      "loss": 0.5131,
+      "step": 7680
+    },
+    {
+      "epoch": 1.079298245614035,
+      "grad_norm": 1.2926596403121948,
+      "learning_rate": 4.449015869986325e-05,
+      "loss": 0.4782,
+      "step": 7690
+    },
+    {
+      "epoch": 1.080701754385965,
+      "grad_norm": 2.191722869873047,
+      "learning_rate": 4.447633644827747e-05,
+      "loss": 0.4962,
+      "step": 7700
+    },
+    {
+      "epoch": 1.0821052631578947,
+      "grad_norm": 1.8317209482192993,
+      "learning_rate": 4.446249903369621e-05,
+      "loss": 0.5025,
+      "step": 7710
+    },
+    {
+      "epoch": 1.0835087719298246,
+      "grad_norm": 1.2881171703338623,
+      "learning_rate": 4.444864646689239e-05,
+      "loss": 0.4816,
+      "step": 7720
+    },
+    {
+      "epoch": 1.0849122807017544,
+      "grad_norm": 1.1918405294418335,
+      "learning_rate": 4.443477875865071e-05,
+      "loss": 0.4762,
+      "step": 7730
+    },
+    {
+      "epoch": 1.0863157894736841,
+      "grad_norm": 1.1728036403656006,
+      "learning_rate": 4.4420895919767626e-05,
+      "loss": 0.4501,
+      "step": 7740
+    },
+    {
+      "epoch": 1.087719298245614,
+      "grad_norm": 1.505370855331421,
+      "learning_rate": 4.440699796105143e-05,
+      "loss": 0.4855,
+      "step": 7750
+    },
+    {
+      "epoch": 1.0891228070175438,
+      "grad_norm": 1.6580755710601807,
+      "learning_rate": 4.439308489332215e-05,
+      "loss": 0.5558,
+      "step": 7760
+    },
+    {
+      "epoch": 1.0905263157894738,
+      "grad_norm": 1.496596097946167,
+      "learning_rate": 4.437915672741158e-05,
+      "loss": 0.5219,
+      "step": 7770
+    },
+    {
+      "epoch": 1.0919298245614035,
+      "grad_norm": 1.2828936576843262,
+      "learning_rate": 4.43652134741633e-05,
+      "loss": 0.4643,
+      "step": 7780
+    },
+    {
+      "epoch": 1.0933333333333333,
+      "grad_norm": 1.2443900108337402,
+      "learning_rate": 4.435125514443258e-05,
+      "loss": 0.562,
+      "step": 7790
+    },
+    {
+      "epoch": 1.0947368421052632,
+      "grad_norm": 1.5212448835372925,
+      "learning_rate": 4.4337281749086477e-05,
+      "loss": 0.5022,
+      "step": 7800
+    },
+    {
+      "epoch": 1.096140350877193,
+      "grad_norm": 2.2898428440093994,
+      "learning_rate": 4.432329329900375e-05,
+      "loss": 0.5462,
+      "step": 7810
+    },
+    {
+      "epoch": 1.0975438596491227,
+      "grad_norm": 1.3173396587371826,
+      "learning_rate": 4.4309289805074895e-05,
+      "loss": 0.5103,
+      "step": 7820
+    },
+    {
+      "epoch": 1.0989473684210527,
+      "grad_norm": 1.3435895442962646,
+      "learning_rate": 4.42952712782021e-05,
+      "loss": 0.5588,
+      "step": 7830
+    },
+    {
+      "epoch": 1.1003508771929824,
+      "grad_norm": 1.1860660314559937,
+      "learning_rate": 4.428123772929928e-05,
+      "loss": 0.5107,
+      "step": 7840
+    },
+    {
+      "epoch": 1.1017543859649124,
+      "grad_norm": 1.2616344690322876,
+      "learning_rate": 4.426718916929202e-05,
+      "loss": 0.463,
+      "step": 7850
+    },
+    {
+      "epoch": 1.1031578947368421,
+      "grad_norm": 0.8766297101974487,
+      "learning_rate": 4.425312560911762e-05,
+      "loss": 0.4801,
+      "step": 7860
+    },
+    {
+      "epoch": 1.1045614035087719,
+      "grad_norm": 1.4563919305801392,
+      "learning_rate": 4.4239047059725035e-05,
+      "loss": 0.4755,
+      "step": 7870
+    },
+    {
+      "epoch": 1.1059649122807018,
+      "grad_norm": 1.3378584384918213,
+      "learning_rate": 4.422495353207491e-05,
+      "loss": 0.4449,
+      "step": 7880
+    },
+    {
+      "epoch": 1.1073684210526316,
+      "grad_norm": 1.259637713432312,
+      "learning_rate": 4.4210845037139525e-05,
+      "loss": 0.4613,
+      "step": 7890
+    },
+    {
+      "epoch": 1.1087719298245613,
+      "grad_norm": 1.8832120895385742,
+      "learning_rate": 4.419672158590282e-05,
+      "loss": 0.6132,
+      "step": 7900
+    },
+    {
+      "epoch": 1.1101754385964913,
+      "grad_norm": 1.7765206098556519,
+      "learning_rate": 4.4182583189360415e-05,
+      "loss": 0.5235,
+      "step": 7910
+    },
+    {
+      "epoch": 1.111578947368421,
+      "grad_norm": 1.5252950191497803,
+      "learning_rate": 4.416842985851951e-05,
+      "loss": 0.5066,
+      "step": 7920
+    },
+    {
+      "epoch": 1.112982456140351,
+      "grad_norm": 1.145727515220642,
+      "learning_rate": 4.415426160439897e-05,
+      "loss": 0.5148,
+      "step": 7930
+    },
+    {
+      "epoch": 1.1143859649122807,
+      "grad_norm": 1.413393259048462,
+      "learning_rate": 4.414007843802927e-05,
+      "loss": 0.4731,
+      "step": 7940
+    },
+    {
+      "epoch": 1.1157894736842104,
+      "grad_norm": 1.227738618850708,
+      "learning_rate": 4.412588037045248e-05,
+      "loss": 0.4657,
+      "step": 7950
+    },
+    {
+      "epoch": 1.1171929824561404,
+      "grad_norm": 1.1237843036651611,
+      "learning_rate": 4.411166741272228e-05,
+      "loss": 0.5292,
+      "step": 7960
+    },
+    {
+      "epoch": 1.1185964912280701,
+      "grad_norm": 1.5785701274871826,
+      "learning_rate": 4.4097439575903964e-05,
+      "loss": 0.5086,
+      "step": 7970
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.8202313780784607,
+      "learning_rate": 4.408319687107437e-05,
+      "loss": 0.4074,
+      "step": 7980
+    },
+    {
+      "epoch": 1.1214035087719298,
+      "grad_norm": 1.6983180046081543,
+      "learning_rate": 4.406893930932195e-05,
+      "loss": 0.5302,
+      "step": 7990
+    },
+    {
+      "epoch": 1.1228070175438596,
+      "grad_norm": 1.3259834051132202,
+      "learning_rate": 4.4054666901746685e-05,
+      "loss": 0.5533,
+      "step": 8000
+    },
+    {
+      "epoch": 1.1228070175438596,
+      "eval_loss": 0.6547604203224182,
+      "eval_runtime": 43.9015,
+      "eval_samples_per_second": 34.167,
+      "eval_steps_per_second": 8.542,
+      "step": 8000
+    },
+    {
+      "epoch": 1.1242105263157895,
+      "grad_norm": 1.5617778301239014,
+      "learning_rate": 4.404037965946015e-05,
+      "loss": 0.5533,
+      "step": 8010
+    },
+    {
+      "epoch": 1.1256140350877193,
+      "grad_norm": 1.3925784826278687,
+      "learning_rate": 4.402607759358545e-05,
+      "loss": 0.573,
+      "step": 8020
+    },
+    {
+      "epoch": 1.127017543859649,
+      "grad_norm": 2.007066011428833,
+      "learning_rate": 4.401176071525722e-05,
+      "loss": 0.6002,
+      "step": 8030
+    },
+    {
+      "epoch": 1.128421052631579,
+      "grad_norm": 1.2479066848754883,
+      "learning_rate": 4.399742903562166e-05,
+      "loss": 0.5412,
+      "step": 8040
+    },
+    {
+      "epoch": 1.1298245614035087,
+      "grad_norm": 2.116882562637329,
+      "learning_rate": 4.3983082565836454e-05,
+      "loss": 0.5516,
+      "step": 8050
+    },
+    {
+      "epoch": 1.1312280701754387,
+      "grad_norm": 1.0991559028625488,
+      "learning_rate": 4.3968721317070835e-05,
+      "loss": 0.5142,
+      "step": 8060
+    },
+    {
+      "epoch": 1.1326315789473684,
+      "grad_norm": 1.5136295557022095,
+      "learning_rate": 4.395434530050553e-05,
+      "loss": 0.4974,
+      "step": 8070
+    },
+    {
+      "epoch": 1.1340350877192982,
+      "grad_norm": 1.6304662227630615,
+      "learning_rate": 4.393995452733274e-05,
+      "loss": 0.5921,
+      "step": 8080
+    },
+    {
+      "epoch": 1.1354385964912281,
+      "grad_norm": 1.1499663591384888,
+      "learning_rate": 4.392554900875619e-05,
+      "loss": 0.5516,
+      "step": 8090
+    },
+    {
+      "epoch": 1.1368421052631579,
+      "grad_norm": 1.14556884765625,
+      "learning_rate": 4.3911128755991085e-05,
+      "loss": 0.4377,
+      "step": 8100
+    },
+    {
+      "epoch": 1.1382456140350876,
+      "grad_norm": 2.168900728225708,
+      "learning_rate": 4.3896693780264054e-05,
+      "loss": 0.5489,
+      "step": 8110
+    },
+    {
+      "epoch": 1.1396491228070176,
+      "grad_norm": 1.8360158205032349,
+      "learning_rate": 4.388224409281324e-05,
+      "loss": 0.4883,
+      "step": 8120
+    },
+    {
+      "epoch": 1.1410526315789473,
+      "grad_norm": 1.3180638551712036,
+      "learning_rate": 4.3867779704888225e-05,
+      "loss": 0.5316,
+      "step": 8130
+    },
+    {
+      "epoch": 1.1424561403508773,
+      "grad_norm": 1.194568157196045,
+      "learning_rate": 4.385330062775001e-05,
+      "loss": 0.5961,
+      "step": 8140
+    },
+    {
+      "epoch": 1.143859649122807,
+      "grad_norm": 1.7998569011688232,
+      "learning_rate": 4.383880687267107e-05,
+      "loss": 0.5839,
+      "step": 8150
+    },
+    {
+      "epoch": 1.1452631578947368,
+      "grad_norm": 1.313109040260315,
+      "learning_rate": 4.3824298450935284e-05,
+      "loss": 0.4834,
+      "step": 8160
+    },
+    {
+      "epoch": 1.1466666666666667,
+      "grad_norm": 1.082961916923523,
+      "learning_rate": 4.380977537383796e-05,
+      "loss": 0.5543,
+      "step": 8170
+    },
+    {
+      "epoch": 1.1480701754385965,
+      "grad_norm": 1.8646924495697021,
+      "learning_rate": 4.37952376526858e-05,
+      "loss": 0.5548,
+      "step": 8180
+    },
+    {
+      "epoch": 1.1494736842105264,
+      "grad_norm": 0.8285521268844604,
+      "learning_rate": 4.378068529879693e-05,
+      "loss": 0.4596,
+      "step": 8190
+    },
+    {
+      "epoch": 1.1508771929824562,
+      "grad_norm": 1.6364754438400269,
+      "learning_rate": 4.376611832350085e-05,
+      "loss": 0.5165,
+      "step": 8200
+    },
+    {
+      "epoch": 1.152280701754386,
+      "grad_norm": 1.4383785724639893,
+      "learning_rate": 4.3751536738138454e-05,
+      "loss": 0.5085,
+      "step": 8210
+    },
+    {
+      "epoch": 1.1536842105263159,
+      "grad_norm": 1.0523866415023804,
+      "learning_rate": 4.3736940554062e-05,
+      "loss": 0.5485,
+      "step": 8220
+    },
+    {
+      "epoch": 1.1550877192982456,
+      "grad_norm": 2.191441535949707,
+      "learning_rate": 4.372232978263513e-05,
+      "loss": 0.5049,
+      "step": 8230
+    },
+    {
+      "epoch": 1.1564912280701753,
+      "grad_norm": 1.923846960067749,
+      "learning_rate": 4.3707704435232816e-05,
+      "loss": 0.4833,
+      "step": 8240
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 1.855549693107605,
+      "learning_rate": 4.36930645232414e-05,
+      "loss": 0.5616,
+      "step": 8250
+    },
+    {
+      "epoch": 1.159298245614035,
+      "grad_norm": 1.620718240737915,
+      "learning_rate": 4.367841005805855e-05,
+      "loss": 0.5448,
+      "step": 8260
+    },
+    {
+      "epoch": 1.1607017543859648,
+      "grad_norm": 1.5233041048049927,
+      "learning_rate": 4.366374105109327e-05,
+      "loss": 0.4714,
+      "step": 8270
+    },
+    {
+      "epoch": 1.1621052631578948,
+      "grad_norm": 1.2673170566558838,
+      "learning_rate": 4.364905751376589e-05,
+      "loss": 0.5994,
+      "step": 8280
+    },
+    {
+      "epoch": 1.1635087719298245,
+      "grad_norm": 1.4488414525985718,
+      "learning_rate": 4.3634359457508046e-05,
+      "loss": 0.5633,
+      "step": 8290
+    },
+    {
+      "epoch": 1.1649122807017545,
+      "grad_norm": 2.300537586212158,
+      "learning_rate": 4.3619646893762675e-05,
+      "loss": 0.5566,
+      "step": 8300
+    },
+    {
+      "epoch": 1.1663157894736842,
+      "grad_norm": 0.8948672413825989,
+      "learning_rate": 4.360491983398402e-05,
+      "loss": 0.4631,
+      "step": 8310
+    },
+    {
+      "epoch": 1.167719298245614,
+      "grad_norm": 1.453062653541565,
+      "learning_rate": 4.3590178289637585e-05,
+      "loss": 0.4525,
+      "step": 8320
+    },
+    {
+      "epoch": 1.169122807017544,
+      "grad_norm": 1.201952576637268,
+      "learning_rate": 4.357542227220019e-05,
+      "loss": 0.4501,
+      "step": 8330
+    },
+    {
+      "epoch": 1.1705263157894736,
+      "grad_norm": 1.3226593732833862,
+      "learning_rate": 4.356065179315988e-05,
+      "loss": 0.561,
+      "step": 8340
+    },
+    {
+      "epoch": 1.1719298245614036,
+      "grad_norm": 0.9465067386627197,
+      "learning_rate": 4.354586686401599e-05,
+      "loss": 0.513,
+      "step": 8350
+    },
+    {
+      "epoch": 1.1733333333333333,
+      "grad_norm": 1.12758207321167,
+      "learning_rate": 4.353106749627909e-05,
+      "loss": 0.4915,
+      "step": 8360
+    },
+    {
+      "epoch": 1.174736842105263,
+      "grad_norm": 2.4800631999969482,
+      "learning_rate": 4.3516253701471e-05,
+      "loss": 0.5193,
+      "step": 8370
+    },
+    {
+      "epoch": 1.176140350877193,
+      "grad_norm": 0.8092995285987854,
+      "learning_rate": 4.350142549112476e-05,
+      "loss": 0.5881,
+      "step": 8380
+    },
+    {
+      "epoch": 1.1775438596491228,
+      "grad_norm": 1.0919562578201294,
+      "learning_rate": 4.348658287678465e-05,
+      "loss": 0.5378,
+      "step": 8390
+    },
+    {
+      "epoch": 1.1789473684210527,
+      "grad_norm": 1.3849503993988037,
+      "learning_rate": 4.347172587000614e-05,
+      "loss": 0.5356,
+      "step": 8400
+    },
+    {
+      "epoch": 1.1803508771929825,
+      "grad_norm": 2.6931231021881104,
+      "learning_rate": 4.345685448235594e-05,
+      "loss": 0.5849,
+      "step": 8410
+    },
+    {
+      "epoch": 1.1817543859649122,
+      "grad_norm": 1.188615083694458,
+      "learning_rate": 4.3441968725411905e-05,
+      "loss": 0.5157,
+      "step": 8420
+    },
+    {
+      "epoch": 1.1831578947368422,
+      "grad_norm": 2.472364902496338,
+      "learning_rate": 4.342706861076313e-05,
+      "loss": 0.6508,
+      "step": 8430
+    },
+    {
+      "epoch": 1.184561403508772,
+      "grad_norm": 1.9964373111724854,
+      "learning_rate": 4.341215415000987e-05,
+      "loss": 0.525,
+      "step": 8440
+    },
+    {
+      "epoch": 1.1859649122807017,
+      "grad_norm": 1.2414706945419312,
+      "learning_rate": 4.339722535476353e-05,
+      "loss": 0.6218,
+      "step": 8450
+    },
+    {
+      "epoch": 1.1873684210526316,
+      "grad_norm": 1.69329035282135,
+      "learning_rate": 4.3382282236646684e-05,
+      "loss": 0.5375,
+      "step": 8460
+    },
+    {
+      "epoch": 1.1887719298245614,
+      "grad_norm": 0.9698866605758667,
+      "learning_rate": 4.336732480729306e-05,
+      "loss": 0.5454,
+      "step": 8470
+    },
+    {
+      "epoch": 1.190175438596491,
+      "grad_norm": 1.450108528137207,
+      "learning_rate": 4.335235307834755e-05,
+      "loss": 0.5507,
+      "step": 8480
+    },
+    {
+      "epoch": 1.191578947368421,
+      "grad_norm": 1.593243956565857,
+      "learning_rate": 4.333736706146615e-05,
+      "loss": 0.5172,
+      "step": 8490
+    },
+    {
+      "epoch": 1.1929824561403508,
+      "grad_norm": 0.9896023869514465,
+      "learning_rate": 4.332236676831598e-05,
+      "loss": 0.5401,
+      "step": 8500
+    },
+    {
+      "epoch": 1.1943859649122808,
+      "grad_norm": 1.9976292848587036,
+      "learning_rate": 4.330735221057529e-05,
+      "loss": 0.4734,
+      "step": 8510
+    },
+    {
+      "epoch": 1.1957894736842105,
+      "grad_norm": 0.9883520007133484,
+      "learning_rate": 4.329232339993342e-05,
+      "loss": 0.4783,
+      "step": 8520
+    },
+    {
+      "epoch": 1.1971929824561403,
+      "grad_norm": 1.6024073362350464,
+      "learning_rate": 4.327728034809082e-05,
+      "loss": 0.5843,
+      "step": 8530
+    },
+    {
+      "epoch": 1.1985964912280702,
+      "grad_norm": 1.5246341228485107,
+      "learning_rate": 4.326222306675902e-05,
+      "loss": 0.4922,
+      "step": 8540
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.8065810203552246,
+      "learning_rate": 4.324715156766064e-05,
+      "loss": 0.6196,
+      "step": 8550
+    },
+    {
+      "epoch": 1.20140350877193,
+      "grad_norm": 1.25635826587677,
+      "learning_rate": 4.3232065862529334e-05,
+      "loss": 0.4713,
+      "step": 8560
+    },
+    {
+      "epoch": 1.2028070175438597,
+      "grad_norm": 1.874711036682129,
+      "learning_rate": 4.321696596310987e-05,
+      "loss": 0.5015,
+      "step": 8570
+    },
+    {
+      "epoch": 1.2042105263157894,
+      "grad_norm": 1.4795438051223755,
+      "learning_rate": 4.3201851881158004e-05,
+      "loss": 0.569,
+      "step": 8580
+    },
+    {
+      "epoch": 1.2056140350877194,
+      "grad_norm": 1.1996725797653198,
+      "learning_rate": 4.31867236284406e-05,
+      "loss": 0.5079,
+      "step": 8590
+    },
+    {
+      "epoch": 1.207017543859649,
+      "grad_norm": 1.1284021139144897,
+      "learning_rate": 4.31715812167355e-05,
+      "loss": 0.5132,
+      "step": 8600
+    },
+    {
+      "epoch": 1.208421052631579,
+      "grad_norm": 1.3568930625915527,
+      "learning_rate": 4.3156424657831596e-05,
+      "loss": 0.5907,
+      "step": 8610
+    },
+    {
+      "epoch": 1.2098245614035088,
+      "grad_norm": 2.9363083839416504,
+      "learning_rate": 4.3141253963528795e-05,
+      "loss": 0.6086,
+      "step": 8620
+    },
+    {
+      "epoch": 1.2112280701754385,
+      "grad_norm": 1.58176589012146,
+      "learning_rate": 4.3126069145637987e-05,
+      "loss": 0.4966,
+      "step": 8630
+    },
+    {
+      "epoch": 1.2126315789473685,
+      "grad_norm": 1.1019052267074585,
+      "learning_rate": 4.3110870215981095e-05,
+      "loss": 0.5713,
+      "step": 8640
+    },
+    {
+      "epoch": 1.2140350877192982,
+      "grad_norm": 1.8327674865722656,
+      "learning_rate": 4.309565718639098e-05,
+      "loss": 0.5538,
+      "step": 8650
+    },
+    {
+      "epoch": 1.215438596491228,
+      "grad_norm": 1.4098116159439087,
+      "learning_rate": 4.308043006871153e-05,
+      "loss": 0.5065,
+      "step": 8660
+    },
+    {
+      "epoch": 1.216842105263158,
+      "grad_norm": 1.7343579530715942,
+      "learning_rate": 4.306518887479758e-05,
+      "loss": 0.495,
+      "step": 8670
+    },
+    {
+      "epoch": 1.2182456140350877,
+      "grad_norm": 1.1002309322357178,
+      "learning_rate": 4.3049933616514895e-05,
+      "loss": 0.5217,
+      "step": 8680
+    },
+    {
+      "epoch": 1.2196491228070174,
+      "grad_norm": 1.6965640783309937,
+      "learning_rate": 4.303466430574024e-05,
+      "loss": 0.5196,
+      "step": 8690
+    },
+    {
+      "epoch": 1.2210526315789474,
+      "grad_norm": 2.227039337158203,
+      "learning_rate": 4.301938095436129e-05,
+      "loss": 0.4687,
+      "step": 8700
+    },
+    {
+      "epoch": 1.2224561403508771,
+      "grad_norm": 1.9886293411254883,
+      "learning_rate": 4.300408357427666e-05,
+      "loss": 0.6043,
+      "step": 8710
+    },
+    {
+      "epoch": 1.223859649122807,
+      "grad_norm": 1.9546360969543457,
+      "learning_rate": 4.298877217739587e-05,
+      "loss": 0.5359,
+      "step": 8720
+    },
+    {
+      "epoch": 1.2252631578947368,
+      "grad_norm": 1.1686962842941284,
+      "learning_rate": 4.29734467756394e-05,
+      "loss": 0.4502,
+      "step": 8730
+    },
+    {
+      "epoch": 1.2266666666666666,
+      "grad_norm": 1.1630245447158813,
+      "learning_rate": 4.2958107380938564e-05,
+      "loss": 0.4823,
+      "step": 8740
+    },
+    {
+      "epoch": 1.2280701754385965,
+      "grad_norm": 1.3308658599853516,
+      "learning_rate": 4.294275400523564e-05,
+      "loss": 0.4295,
+      "step": 8750
+    },
+    {
+      "epoch": 1.2294736842105263,
+      "grad_norm": 1.694773554801941,
+      "learning_rate": 4.2927386660483726e-05,
+      "loss": 0.4983,
+      "step": 8760
+    },
+    {
+      "epoch": 1.2308771929824562,
+      "grad_norm": 0.8299292325973511,
+      "learning_rate": 4.291200535864684e-05,
+      "loss": 0.5405,
+      "step": 8770
+    },
+    {
+      "epoch": 1.232280701754386,
+      "grad_norm": 1.7378555536270142,
+      "learning_rate": 4.289661011169986e-05,
+      "loss": 0.5094,
+      "step": 8780
+    },
+    {
+      "epoch": 1.2336842105263157,
+      "grad_norm": 1.8789598941802979,
+      "learning_rate": 4.28812009316285e-05,
+      "loss": 0.5622,
+      "step": 8790
+    },
+    {
+      "epoch": 1.2350877192982457,
+      "grad_norm": 1.2347322702407837,
+      "learning_rate": 4.286577783042934e-05,
+      "loss": 0.4577,
+      "step": 8800
+    },
+    {
+      "epoch": 1.2364912280701754,
+      "grad_norm": 1.610954761505127,
+      "learning_rate": 4.285034082010981e-05,
+      "loss": 0.6015,
+      "step": 8810
+    },
+    {
+      "epoch": 1.2378947368421054,
+      "grad_norm": 0.8974846005439758,
+      "learning_rate": 4.2834889912688126e-05,
+      "loss": 0.5716,
+      "step": 8820
+    },
+    {
+      "epoch": 1.2392982456140351,
+      "grad_norm": 1.1207072734832764,
+      "learning_rate": 4.281942512019336e-05,
+      "loss": 0.5634,
+      "step": 8830
+    },
+    {
+      "epoch": 1.2407017543859649,
+      "grad_norm": 2.1318647861480713,
+      "learning_rate": 4.2803946454665376e-05,
+      "loss": 0.4982,
+      "step": 8840
+    },
+    {
+      "epoch": 1.2421052631578948,
+      "grad_norm": 1.3747590780258179,
+      "learning_rate": 4.2788453928154855e-05,
+      "loss": 0.5006,
+      "step": 8850
+    },
+    {
+      "epoch": 1.2435087719298246,
+      "grad_norm": 0.8502065539360046,
+      "learning_rate": 4.2772947552723266e-05,
+      "loss": 0.4901,
+      "step": 8860
+    },
+    {
+      "epoch": 1.2449122807017543,
+      "grad_norm": 1.888156771659851,
+      "learning_rate": 4.275742734044283e-05,
+      "loss": 0.4847,
+      "step": 8870
+    },
+    {
+      "epoch": 1.2463157894736843,
+      "grad_norm": 2.0071113109588623,
+      "learning_rate": 4.274189330339658e-05,
+      "loss": 0.5224,
+      "step": 8880
+    },
+    {
+      "epoch": 1.247719298245614,
+      "grad_norm": 1.0914371013641357,
+      "learning_rate": 4.272634545367831e-05,
+      "loss": 0.4698,
+      "step": 8890
+    },
+    {
+      "epoch": 1.2491228070175437,
+      "grad_norm": 1.4466750621795654,
+      "learning_rate": 4.271078380339252e-05,
+      "loss": 0.5801,
+      "step": 8900
+    },
+    {
+      "epoch": 1.2505263157894737,
+      "grad_norm": 1.5080820322036743,
+      "learning_rate": 4.269520836465452e-05,
+      "loss": 0.6584,
+      "step": 8910
+    },
+    {
+      "epoch": 1.2519298245614034,
+      "grad_norm": 1.510321855545044,
+      "learning_rate": 4.2679619149590304e-05,
+      "loss": 0.5752,
+      "step": 8920
+    },
+    {
+      "epoch": 1.2533333333333334,
+      "grad_norm": 1.527969479560852,
+      "learning_rate": 4.266401617033662e-05,
+      "loss": 0.4829,
+      "step": 8930
+    },
+    {
+      "epoch": 1.2547368421052632,
+      "grad_norm": 1.2593231201171875,
+      "learning_rate": 4.264839943904091e-05,
+      "loss": 0.5411,
+      "step": 8940
+    },
+    {
+      "epoch": 1.256140350877193,
+      "grad_norm": 1.490929365158081,
+      "learning_rate": 4.2632768967861345e-05,
+      "loss": 0.5089,
+      "step": 8950
+    },
+    {
+      "epoch": 1.2575438596491229,
+      "grad_norm": 0.7822336554527283,
+      "learning_rate": 4.261712476896679e-05,
+      "loss": 0.6257,
+      "step": 8960
+    },
+    {
+      "epoch": 1.2589473684210526,
+      "grad_norm": 1.331175446510315,
+      "learning_rate": 4.2601466854536774e-05,
+      "loss": 0.5403,
+      "step": 8970
+    },
+    {
+      "epoch": 1.2603508771929826,
+      "grad_norm": 1.4372813701629639,
+      "learning_rate": 4.2585795236761526e-05,
+      "loss": 0.5305,
+      "step": 8980
+    },
+    {
+      "epoch": 1.2617543859649123,
+      "grad_norm": 1.873630166053772,
+      "learning_rate": 4.257010992784194e-05,
+      "loss": 0.5776,
+      "step": 8990
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 1.2255460023880005,
+      "learning_rate": 4.255441093998956e-05,
+      "loss": 0.5772,
+      "step": 9000
+    },
+    {
+      "epoch": 1.264561403508772,
+      "grad_norm": 1.3667577505111694,
+      "learning_rate": 4.253869828542659e-05,
+      "loss": 0.589,
+      "step": 9010
+    },
+    {
+      "epoch": 1.2659649122807017,
+      "grad_norm": 1.189122200012207,
+      "learning_rate": 4.2522971976385876e-05,
+      "loss": 0.5012,
+      "step": 9020
+    },
+    {
+      "epoch": 1.2673684210526317,
+      "grad_norm": 2.0150930881500244,
+      "learning_rate": 4.250723202511089e-05,
+      "loss": 0.4813,
+      "step": 9030
+    },
+    {
+      "epoch": 1.2687719298245614,
+      "grad_norm": 1.83956778049469,
+      "learning_rate": 4.2491478443855704e-05,
+      "loss": 0.513,
+      "step": 9040
+    },
+    {
+      "epoch": 1.2701754385964912,
+      "grad_norm": 1.8281301259994507,
+      "learning_rate": 4.247571124488504e-05,
+      "loss": 0.6229,
+      "step": 9050
+    },
+    {
+      "epoch": 1.271578947368421,
+      "grad_norm": 1.5498483180999756,
+      "learning_rate": 4.2459930440474194e-05,
+      "loss": 0.5493,
+      "step": 9060
+    },
+    {
+      "epoch": 1.2729824561403509,
+      "grad_norm": 1.680643081665039,
+      "learning_rate": 4.2444136042909064e-05,
+      "loss": 0.4845,
+      "step": 9070
+    },
+    {
+      "epoch": 1.2743859649122806,
+      "grad_norm": 1.4468814134597778,
+      "learning_rate": 4.2428328064486134e-05,
+      "loss": 0.5174,
+      "step": 9080
+    },
+    {
+      "epoch": 1.2757894736842106,
+      "grad_norm": 1.8637295961380005,
+      "learning_rate": 4.2412506517512456e-05,
+      "loss": 0.501,
+      "step": 9090
+    },
+    {
+      "epoch": 1.2771929824561403,
+      "grad_norm": 1.8078296184539795,
+      "learning_rate": 4.239667141430564e-05,
+      "loss": 0.6422,
+      "step": 9100
+    },
+    {
+      "epoch": 1.27859649122807,
+      "grad_norm": 1.8999830484390259,
+      "learning_rate": 4.238082276719387e-05,
+      "loss": 0.5323,
+      "step": 9110
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.8832138776779175,
+      "learning_rate": 4.236496058851585e-05,
+      "loss": 0.4542,
+      "step": 9120
+    },
+    {
+      "epoch": 1.2814035087719298,
+      "grad_norm": 1.2980352640151978,
+      "learning_rate": 4.234908489062083e-05,
+      "loss": 0.5697,
+      "step": 9130
+    },
+    {
+      "epoch": 1.2828070175438597,
+      "grad_norm": 1.667039394378662,
+      "learning_rate": 4.233319568586859e-05,
+      "loss": 0.5108,
+      "step": 9140
+    },
+    {
+      "epoch": 1.2842105263157895,
+      "grad_norm": 1.6664785146713257,
+      "learning_rate": 4.231729298662942e-05,
+      "loss": 0.4472,
+      "step": 9150
+    },
+    {
+      "epoch": 1.2856140350877192,
+      "grad_norm": 1.4384082555770874,
+      "learning_rate": 4.230137680528411e-05,
+      "loss": 0.62,
+      "step": 9160
+    },
+    {
+      "epoch": 1.2870175438596492,
+      "grad_norm": 1.851901888847351,
+      "learning_rate": 4.228544715422395e-05,
+      "loss": 0.5226,
+      "step": 9170
+    },
+    {
+      "epoch": 1.288421052631579,
+      "grad_norm": 1.372755527496338,
+      "learning_rate": 4.2269504045850744e-05,
+      "loss": 0.5492,
+      "step": 9180
+    },
+    {
+      "epoch": 1.2898245614035089,
+      "grad_norm": 1.7806882858276367,
+      "learning_rate": 4.225354749257673e-05,
+      "loss": 0.5359,
+      "step": 9190
+    },
+    {
+      "epoch": 1.2912280701754386,
+      "grad_norm": 1.4967597723007202,
+      "learning_rate": 4.2237577506824624e-05,
+      "loss": 0.535,
+      "step": 9200
+    },
+    {
+      "epoch": 1.2926315789473684,
+      "grad_norm": 1.220828890800476,
+      "learning_rate": 4.222159410102761e-05,
+      "loss": 0.4581,
+      "step": 9210
+    },
+    {
+      "epoch": 1.2940350877192983,
+      "grad_norm": 1.9541898965835571,
+      "learning_rate": 4.220559728762933e-05,
+      "loss": 0.5109,
+      "step": 9220
+    },
+    {
+      "epoch": 1.295438596491228,
+      "grad_norm": 0.9027903079986572,
+      "learning_rate": 4.2189587079083846e-05,
+      "loss": 0.4501,
+      "step": 9230
+    },
+    {
+      "epoch": 1.296842105263158,
+      "grad_norm": 2.4572014808654785,
+      "learning_rate": 4.217356348785565e-05,
+      "loss": 0.5574,
+      "step": 9240
+    },
+    {
+      "epoch": 1.2982456140350878,
+      "grad_norm": 1.5705862045288086,
+      "learning_rate": 4.215752652641967e-05,
+      "loss": 0.5558,
+      "step": 9250
+    },
+    {
+      "epoch": 1.2996491228070175,
+      "grad_norm": 1.5693955421447754,
+      "learning_rate": 4.21414762072612e-05,
+      "loss": 0.4734,
+      "step": 9260
+    },
+    {
+      "epoch": 1.3010526315789472,
+      "grad_norm": 1.4699418544769287,
+      "learning_rate": 4.2125412542876e-05,
+      "loss": 0.574,
+      "step": 9270
+    },
+    {
+      "epoch": 1.3024561403508772,
+      "grad_norm": 1.8956423997879028,
+      "learning_rate": 4.210933554577016e-05,
+      "loss": 0.4505,
+      "step": 9280
+    },
+    {
+      "epoch": 1.303859649122807,
+      "grad_norm": 1.8722734451293945,
+      "learning_rate": 4.209324522846018e-05,
+      "loss": 0.5021,
+      "step": 9290
+    },
+    {
+      "epoch": 1.305263157894737,
+      "grad_norm": 1.3624267578125,
+      "learning_rate": 4.207714160347292e-05,
+      "loss": 0.4925,
+      "step": 9300
+    },
+    {
+      "epoch": 1.3066666666666666,
+      "grad_norm": 2.2316009998321533,
+      "learning_rate": 4.206102468334561e-05,
+      "loss": 0.6289,
+      "step": 9310
+    },
+    {
+      "epoch": 1.3080701754385964,
+      "grad_norm": 2.3992788791656494,
+      "learning_rate": 4.2044894480625825e-05,
+      "loss": 0.5387,
+      "step": 9320
+    },
+    {
+      "epoch": 1.3094736842105263,
+      "grad_norm": 1.9581996202468872,
+      "learning_rate": 4.202875100787147e-05,
+      "loss": 0.5788,
+      "step": 9330
+    },
+    {
+      "epoch": 1.310877192982456,
+      "grad_norm": 1.5940258502960205,
+      "learning_rate": 4.201259427765081e-05,
+      "loss": 0.5313,
+      "step": 9340
+    },
+    {
+      "epoch": 1.312280701754386,
+      "grad_norm": 1.2207392454147339,
+      "learning_rate": 4.1996424302542404e-05,
+      "loss": 0.5948,
+      "step": 9350
+    },
+    {
+      "epoch": 1.3136842105263158,
+      "grad_norm": 1.743915319442749,
+      "learning_rate": 4.198024109513512e-05,
+      "loss": 0.4913,
+      "step": 9360
+    },
+    {
+      "epoch": 1.3150877192982455,
+      "grad_norm": 1.9989562034606934,
+      "learning_rate": 4.196404466802816e-05,
+      "loss": 0.4895,
+      "step": 9370
+    },
+    {
+      "epoch": 1.3164912280701755,
+      "grad_norm": 1.8793307542800903,
+      "learning_rate": 4.194783503383098e-05,
+      "loss": 0.5537,
+      "step": 9380
+    },
+    {
+      "epoch": 1.3178947368421052,
+      "grad_norm": 1.9246269464492798,
+      "learning_rate": 4.193161220516334e-05,
+      "loss": 0.5641,
+      "step": 9390
+    },
+    {
+      "epoch": 1.3192982456140352,
+      "grad_norm": 1.5612519979476929,
+      "learning_rate": 4.191537619465529e-05,
+      "loss": 0.464,
+      "step": 9400
+    },
+    {
+      "epoch": 1.320701754385965,
+      "grad_norm": 0.9451802968978882,
+      "learning_rate": 4.189912701494709e-05,
+      "loss": 0.4657,
+      "step": 9410
+    },
+    {
+      "epoch": 1.3221052631578947,
+      "grad_norm": 1.790861964225769,
+      "learning_rate": 4.1882864678689296e-05,
+      "loss": 0.5113,
+      "step": 9420
+    },
+    {
+      "epoch": 1.3235087719298246,
+      "grad_norm": 1.9305384159088135,
+      "learning_rate": 4.186658919854269e-05,
+      "loss": 0.5593,
+      "step": 9430
+    },
+    {
+      "epoch": 1.3249122807017544,
+      "grad_norm": 2.051849603652954,
+      "learning_rate": 4.1850300587178304e-05,
+      "loss": 0.4578,
+      "step": 9440
+    },
+    {
+      "epoch": 1.3263157894736843,
+      "grad_norm": 1.7359352111816406,
+      "learning_rate": 4.183399885727737e-05,
+      "loss": 0.5637,
+      "step": 9450
+    },
+    {
+      "epoch": 1.327719298245614,
+      "grad_norm": 1.373119592666626,
+      "learning_rate": 4.181768402153135e-05,
+      "loss": 0.5491,
+      "step": 9460
+    },
+    {
+      "epoch": 1.3291228070175438,
+      "grad_norm": 1.0692249536514282,
+      "learning_rate": 4.1801356092641886e-05,
+      "loss": 0.5558,
+      "step": 9470
+    },
+    {
+      "epoch": 1.3305263157894736,
+      "grad_norm": 1.0997167825698853,
+      "learning_rate": 4.178501508332085e-05,
+      "loss": 0.4543,
+      "step": 9480
+    },
+    {
+      "epoch": 1.3319298245614035,
+      "grad_norm": 1.5097479820251465,
+      "learning_rate": 4.176866100629027e-05,
+      "loss": 0.5832,
+      "step": 9490
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 1.5708390474319458,
+      "learning_rate": 4.175229387428235e-05,
+      "loss": 0.5378,
+      "step": 9500
+    },
+    {
+      "epoch": 1.3347368421052632,
+      "grad_norm": 1.6936376094818115,
+      "learning_rate": 4.1735913700039477e-05,
+      "loss": 0.5046,
+      "step": 9510
+    },
+    {
+      "epoch": 1.336140350877193,
+      "grad_norm": 0.9409717321395874,
+      "learning_rate": 4.171952049631416e-05,
+      "loss": 0.5171,
+      "step": 9520
+    },
+    {
+      "epoch": 1.3375438596491227,
+      "grad_norm": 1.803077220916748,
+      "learning_rate": 4.170311427586908e-05,
+      "loss": 0.5939,
+      "step": 9530
+    },
+    {
+      "epoch": 1.3389473684210527,
+      "grad_norm": 1.1349605321884155,
+      "learning_rate": 4.168669505147705e-05,
+      "loss": 0.5768,
+      "step": 9540
+    },
+    {
+      "epoch": 1.3403508771929824,
+      "grad_norm": 1.6885027885437012,
+      "learning_rate": 4.1670262835920996e-05,
+      "loss": 0.5029,
+      "step": 9550
+    },
+    {
+      "epoch": 1.3417543859649124,
+      "grad_norm": 1.278064489364624,
+      "learning_rate": 4.1653817641993936e-05,
+      "loss": 0.4611,
+      "step": 9560
+    },
+    {
+      "epoch": 1.343157894736842,
+      "grad_norm": 1.7429572343826294,
+      "learning_rate": 4.163735948249905e-05,
+      "loss": 0.5701,
+      "step": 9570
+    },
+    {
+      "epoch": 1.3445614035087718,
+      "grad_norm": 2.2477900981903076,
+      "learning_rate": 4.162088837024956e-05,
+      "loss": 0.5356,
+      "step": 9580
+    },
+    {
+      "epoch": 1.3459649122807018,
+      "grad_norm": 1.617583990097046,
+      "learning_rate": 4.16044043180688e-05,
+      "loss": 0.4985,
+      "step": 9590
+    },
+    {
+      "epoch": 1.3473684210526315,
+      "grad_norm": 1.5791269540786743,
+      "learning_rate": 4.158790733879017e-05,
+      "loss": 0.5036,
+      "step": 9600
+    },
+    {
+      "epoch": 1.3487719298245615,
+      "grad_norm": 1.9323596954345703,
+      "learning_rate": 4.1571397445257124e-05,
+      "loss": 0.5212,
+      "step": 9610
+    },
+    {
+      "epoch": 1.3501754385964913,
+      "grad_norm": 1.3054085969924927,
+      "learning_rate": 4.155487465032319e-05,
+      "loss": 0.5225,
+      "step": 9620
+    },
+    {
+      "epoch": 1.351578947368421,
+      "grad_norm": 1.5751895904541016,
+      "learning_rate": 4.153833896685193e-05,
+      "loss": 0.4985,
+      "step": 9630
+    },
+    {
+      "epoch": 1.352982456140351,
+      "grad_norm": 1.8643230199813843,
+      "learning_rate": 4.1521790407716936e-05,
+      "loss": 0.5386,
+      "step": 9640
+    },
+    {
+      "epoch": 1.3543859649122807,
+      "grad_norm": 1.0118595361709595,
+      "learning_rate": 4.150522898580183e-05,
+      "loss": 0.5283,
+      "step": 9650
+    },
+    {
+      "epoch": 1.3557894736842107,
+      "grad_norm": 1.9065098762512207,
+      "learning_rate": 4.148865471400024e-05,
+      "loss": 0.5684,
+      "step": 9660
+    },
+    {
+      "epoch": 1.3571929824561404,
+      "grad_norm": 2.0933990478515625,
+      "learning_rate": 4.147206760521582e-05,
+      "loss": 0.525,
+      "step": 9670
+    },
+    {
+      "epoch": 1.3585964912280701,
+      "grad_norm": 2.0099165439605713,
+      "learning_rate": 4.145546767236219e-05,
+      "loss": 0.5258,
+      "step": 9680
+    },
+    {
+      "epoch": 1.3599999999999999,
+      "grad_norm": 2.2704153060913086,
+      "learning_rate": 4.143885492836297e-05,
+      "loss": 0.5159,
+      "step": 9690
+    },
+    {
+      "epoch": 1.3614035087719298,
+      "grad_norm": 1.3344398736953735,
+      "learning_rate": 4.1422229386151754e-05,
+      "loss": 0.5656,
+      "step": 9700
+    },
+    {
+      "epoch": 1.3628070175438596,
+      "grad_norm": 2.356660842895508,
+      "learning_rate": 4.140559105867209e-05,
+      "loss": 0.488,
+      "step": 9710
+    },
+    {
+      "epoch": 1.3642105263157895,
+      "grad_norm": 1.0358322858810425,
+      "learning_rate": 4.1388939958877495e-05,
+      "loss": 0.457,
+      "step": 9720
+    },
+    {
+      "epoch": 1.3656140350877193,
+      "grad_norm": 1.4958525896072388,
+      "learning_rate": 4.137227609973141e-05,
+      "loss": 0.459,
+      "step": 9730
+    },
+    {
+      "epoch": 1.367017543859649,
+      "grad_norm": 1.9942265748977661,
+      "learning_rate": 4.135559949420723e-05,
+      "loss": 0.4794,
+      "step": 9740
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 1.7793415784835815,
+      "learning_rate": 4.133891015528826e-05,
+      "loss": 0.5903,
+      "step": 9750
+    },
+    {
+      "epoch": 1.3698245614035087,
+      "grad_norm": 1.069421410560608,
+      "learning_rate": 4.132220809596772e-05,
+      "loss": 0.5521,
+      "step": 9760
+    },
+    {
+      "epoch": 1.3712280701754387,
+      "grad_norm": 0.8958350419998169,
+      "learning_rate": 4.1305493329248734e-05,
+      "loss": 0.4667,
+      "step": 9770
+    },
+    {
+      "epoch": 1.3726315789473684,
+      "grad_norm": 1.314070701599121,
+      "learning_rate": 4.128876586814433e-05,
+      "loss": 0.4307,
+      "step": 9780
+    },
+    {
+      "epoch": 1.3740350877192982,
+      "grad_norm": 1.3073476552963257,
+      "learning_rate": 4.127202572567741e-05,
+      "loss": 0.5016,
+      "step": 9790
+    },
+    {
+      "epoch": 1.3754385964912281,
+      "grad_norm": 1.695670247077942,
+      "learning_rate": 4.1255272914880735e-05,
+      "loss": 0.5489,
+      "step": 9800
+    },
+    {
+      "epoch": 1.3768421052631579,
+      "grad_norm": 1.6946247816085815,
+      "learning_rate": 4.1238507448796945e-05,
+      "loss": 0.488,
+      "step": 9810
+    },
+    {
+      "epoch": 1.3782456140350878,
+      "grad_norm": 1.3960559368133545,
+      "learning_rate": 4.122172934047855e-05,
+      "loss": 0.5739,
+      "step": 9820
+    },
+    {
+      "epoch": 1.3796491228070176,
+      "grad_norm": 1.4782212972640991,
+      "learning_rate": 4.120493860298786e-05,
+      "loss": 0.5036,
+      "step": 9830
+    },
+    {
+      "epoch": 1.3810526315789473,
+      "grad_norm": 1.7010905742645264,
+      "learning_rate": 4.1188135249397056e-05,
+      "loss": 0.4737,
+      "step": 9840
+    },
+    {
+      "epoch": 1.3824561403508773,
+      "grad_norm": 1.8230018615722656,
+      "learning_rate": 4.117131929278811e-05,
+      "loss": 0.5341,
+      "step": 9850
+    },
+    {
+      "epoch": 1.383859649122807,
+      "grad_norm": 1.5947978496551514,
+      "learning_rate": 4.1154490746252825e-05,
+      "loss": 0.4567,
+      "step": 9860
+    },
+    {
+      "epoch": 1.385263157894737,
+      "grad_norm": 2.077136516571045,
+      "learning_rate": 4.113764962289281e-05,
+      "loss": 0.5586,
+      "step": 9870
+    },
+    {
+      "epoch": 1.3866666666666667,
+      "grad_norm": 1.0836787223815918,
+      "learning_rate": 4.112079593581944e-05,
+      "loss": 0.5065,
+      "step": 9880
+    },
+    {
+      "epoch": 1.3880701754385965,
+      "grad_norm": 1.8127710819244385,
+      "learning_rate": 4.110392969815391e-05,
+      "loss": 0.5335,
+      "step": 9890
+    },
+    {
+      "epoch": 1.3894736842105262,
+      "grad_norm": 1.7939358949661255,
+      "learning_rate": 4.108705092302715e-05,
+      "loss": 0.5445,
+      "step": 9900
+    },
+    {
+      "epoch": 1.3908771929824562,
+      "grad_norm": 1.5963224172592163,
+      "learning_rate": 4.1070159623579855e-05,
+      "loss": 0.5146,
+      "step": 9910
+    },
+    {
+      "epoch": 1.392280701754386,
+      "grad_norm": 1.4976569414138794,
+      "learning_rate": 4.105325581296251e-05,
+      "loss": 0.4938,
+      "step": 9920
+    },
+    {
+      "epoch": 1.3936842105263159,
+      "grad_norm": 1.3835642337799072,
+      "learning_rate": 4.103633950433528e-05,
+      "loss": 0.5353,
+      "step": 9930
+    },
+    {
+      "epoch": 1.3950877192982456,
+      "grad_norm": 1.504701018333435,
+      "learning_rate": 4.1019410710868115e-05,
+      "loss": 0.4869,
+      "step": 9940
+    },
+    {
+      "epoch": 1.3964912280701753,
+      "grad_norm": 1.6695371866226196,
+      "learning_rate": 4.100246944574064e-05,
+      "loss": 0.4858,
+      "step": 9950
+    },
+    {
+      "epoch": 1.3978947368421053,
+      "grad_norm": 1.033554196357727,
+      "learning_rate": 4.098551572214223e-05,
+      "loss": 0.5173,
+      "step": 9960
+    },
+    {
+      "epoch": 1.399298245614035,
+      "grad_norm": 2.1895320415496826,
+      "learning_rate": 4.0968549553271926e-05,
+      "loss": 0.5862,
+      "step": 9970
+    },
+    {
+      "epoch": 1.400701754385965,
+      "grad_norm": 2.323758363723755,
+      "learning_rate": 4.095157095233848e-05,
+      "loss": 0.5312,
+      "step": 9980
+    },
+    {
+      "epoch": 1.4021052631578947,
+      "grad_norm": 1.7012853622436523,
+      "learning_rate": 4.093457993256031e-05,
+      "loss": 0.4668,
+      "step": 9990
+    },
+    {
+      "epoch": 1.4035087719298245,
+      "grad_norm": 1.5631529092788696,
+      "learning_rate": 4.0917576507165514e-05,
+      "loss": 0.5192,
+      "step": 10000
+    },
+    {
+      "epoch": 1.4035087719298245,
+      "eval_loss": 0.6501449942588806,
+      "eval_runtime": 43.9031,
+      "eval_samples_per_second": 34.166,
+      "eval_steps_per_second": 8.542,
+      "step": 10000
+    },
+    {
+      "epoch": 1.4049122807017544,
+      "grad_norm": 1.5768696069717407,
+      "learning_rate": 4.090056068939183e-05,
+      "loss": 0.6369,
+      "step": 10010
+    },
+    {
+      "epoch": 1.4063157894736842,
+      "grad_norm": 2.0955562591552734,
+      "learning_rate": 4.088353249248667e-05,
+      "loss": 0.4765,
+      "step": 10020
+    },
+    {
+      "epoch": 1.4077192982456141,
+      "grad_norm": 1.7173198461532593,
+      "learning_rate": 4.0866491929707064e-05,
+      "loss": 0.4858,
+      "step": 10030
+    },
+    {
+      "epoch": 1.4091228070175439,
+      "grad_norm": 1.091640591621399,
+      "learning_rate": 4.084943901431966e-05,
+      "loss": 0.4502,
+      "step": 10040
+    },
+    {
+      "epoch": 1.4105263157894736,
+      "grad_norm": 0.9591197967529297,
+      "learning_rate": 4.083237375960075e-05,
+      "loss": 0.5036,
+      "step": 10050
+    },
+    {
+      "epoch": 1.4119298245614036,
+      "grad_norm": 1.6289422512054443,
+      "learning_rate": 4.081529617883622e-05,
+      "loss": 0.5185,
+      "step": 10060
+    },
+    {
+      "epoch": 1.4133333333333333,
+      "grad_norm": 1.0051218271255493,
+      "learning_rate": 4.079820628532155e-05,
+      "loss": 0.4701,
+      "step": 10070
+    },
+    {
+      "epoch": 1.4147368421052633,
+      "grad_norm": 1.5907773971557617,
+      "learning_rate": 4.0781104092361813e-05,
+      "loss": 0.5406,
+      "step": 10080
+    },
+    {
+      "epoch": 1.416140350877193,
+      "grad_norm": 1.4398341178894043,
+      "learning_rate": 4.0763989613271635e-05,
+      "loss": 0.4963,
+      "step": 10090
+    },
+    {
+      "epoch": 1.4175438596491228,
+      "grad_norm": 1.8131810426712036,
+      "learning_rate": 4.0746862861375245e-05,
+      "loss": 0.5931,
+      "step": 10100
+    },
+    {
+      "epoch": 1.4189473684210525,
+      "grad_norm": 1.3968654870986938,
+      "learning_rate": 4.07297238500064e-05,
+      "loss": 0.4908,
+      "step": 10110
+    },
+    {
+      "epoch": 1.4203508771929825,
+      "grad_norm": 1.4002443552017212,
+      "learning_rate": 4.0712572592508394e-05,
+      "loss": 0.5732,
+      "step": 10120
+    },
+    {
+      "epoch": 1.4217543859649122,
+      "grad_norm": 1.8497573137283325,
+      "learning_rate": 4.069540910223409e-05,
+      "loss": 0.5323,
+      "step": 10130
+    },
+    {
+      "epoch": 1.4231578947368422,
+      "grad_norm": 1.6966348886489868,
+      "learning_rate": 4.067823339254584e-05,
+      "loss": 0.5727,
+      "step": 10140
+    },
+    {
+      "epoch": 1.424561403508772,
+      "grad_norm": 1.1128507852554321,
+      "learning_rate": 4.066104547681553e-05,
+      "loss": 0.5295,
+      "step": 10150
+    },
+    {
+      "epoch": 1.4259649122807017,
+      "grad_norm": 1.2667880058288574,
+      "learning_rate": 4.0643845368424545e-05,
+      "loss": 0.554,
+      "step": 10160
+    },
+    {
+      "epoch": 1.4273684210526316,
+      "grad_norm": 2.0188159942626953,
+      "learning_rate": 4.062663308076374e-05,
+      "loss": 0.5138,
+      "step": 10170
+    },
+    {
+      "epoch": 1.4287719298245614,
+      "grad_norm": 1.113797903060913,
+      "learning_rate": 4.0609408627233494e-05,
+      "loss": 0.543,
+      "step": 10180
+    },
+    {
+      "epoch": 1.4301754385964913,
+      "grad_norm": 1.800862193107605,
+      "learning_rate": 4.059217202124361e-05,
+      "loss": 0.5094,
+      "step": 10190
+    },
+    {
+      "epoch": 1.431578947368421,
+      "grad_norm": 1.7962946891784668,
+      "learning_rate": 4.0574923276213405e-05,
+      "loss": 0.5468,
+      "step": 10200
+    },
+    {
+      "epoch": 1.4329824561403508,
+      "grad_norm": 2.258661985397339,
+      "learning_rate": 4.0557662405571595e-05,
+      "loss": 0.6082,
+      "step": 10210
+    },
+    {
+      "epoch": 1.4343859649122808,
+      "grad_norm": 2.073396921157837,
+      "learning_rate": 4.054038942275637e-05,
+      "loss": 0.5164,
+      "step": 10220
+    },
+    {
+      "epoch": 1.4357894736842105,
+      "grad_norm": 1.6468226909637451,
+      "learning_rate": 4.052310434121533e-05,
+      "loss": 0.5451,
+      "step": 10230
+    },
+    {
+      "epoch": 1.4371929824561405,
+      "grad_norm": 2.4547080993652344,
+      "learning_rate": 4.050580717440552e-05,
+      "loss": 0.5821,
+      "step": 10240
+    },
+    {
+      "epoch": 1.4385964912280702,
+      "grad_norm": 1.6606553792953491,
+      "learning_rate": 4.048849793579337e-05,
+      "loss": 0.5088,
+      "step": 10250
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.6139086484909058,
+      "learning_rate": 4.04711766388547e-05,
+      "loss": 0.5441,
+      "step": 10260
+    },
+    {
+      "epoch": 1.4414035087719297,
+      "grad_norm": 1.6367645263671875,
+      "learning_rate": 4.0453843297074756e-05,
+      "loss": 0.494,
+      "step": 10270
+    },
+    {
+      "epoch": 1.4428070175438596,
+      "grad_norm": 1.6214492321014404,
+      "learning_rate": 4.043649792394812e-05,
+      "loss": 0.571,
+      "step": 10280
+    },
+    {
+      "epoch": 1.4442105263157896,
+      "grad_norm": 1.8566523790359497,
+      "learning_rate": 4.041914053297878e-05,
+      "loss": 0.5845,
+      "step": 10290
+    },
+    {
+      "epoch": 1.4456140350877194,
+      "grad_norm": 1.105668067932129,
+      "learning_rate": 4.0401771137680046e-05,
+      "loss": 0.4655,
+      "step": 10300
+    },
+    {
+      "epoch": 1.447017543859649,
+      "grad_norm": 1.5444446802139282,
+      "learning_rate": 4.038438975157458e-05,
+      "loss": 0.4939,
+      "step": 10310
+    },
+    {
+      "epoch": 1.4484210526315788,
+      "grad_norm": 2.6764674186706543,
+      "learning_rate": 4.036699638819441e-05,
+      "loss": 0.6172,
+      "step": 10320
+    },
+    {
+      "epoch": 1.4498245614035088,
+      "grad_norm": 1.574623942375183,
+      "learning_rate": 4.0349591061080846e-05,
+      "loss": 0.4888,
+      "step": 10330
+    },
+    {
+      "epoch": 1.4512280701754385,
+      "grad_norm": 2.2457685470581055,
+      "learning_rate": 4.0332173783784536e-05,
+      "loss": 0.4427,
+      "step": 10340
+    },
+    {
+      "epoch": 1.4526315789473685,
+      "grad_norm": 1.68437659740448,
+      "learning_rate": 4.031474456986543e-05,
+      "loss": 0.4867,
+      "step": 10350
+    },
+    {
+      "epoch": 1.4540350877192982,
+      "grad_norm": 1.4421491622924805,
+      "learning_rate": 4.0297303432892775e-05,
+      "loss": 0.4401,
+      "step": 10360
+    },
+    {
+      "epoch": 1.455438596491228,
+      "grad_norm": 2.1541783809661865,
+      "learning_rate": 4.027985038644507e-05,
+      "loss": 0.546,
+      "step": 10370
+    },
+    {
+      "epoch": 1.456842105263158,
+      "grad_norm": 1.7601039409637451,
+      "learning_rate": 4.026238544411014e-05,
+      "loss": 0.5211,
+      "step": 10380
+    },
+    {
+      "epoch": 1.4582456140350877,
+      "grad_norm": 1.2212331295013428,
+      "learning_rate": 4.024490861948503e-05,
+      "loss": 0.4633,
+      "step": 10390
+    },
+    {
+      "epoch": 1.4596491228070176,
+      "grad_norm": 1.688339114189148,
+      "learning_rate": 4.022741992617603e-05,
+      "loss": 0.5898,
+      "step": 10400
+    },
+    {
+      "epoch": 1.4610526315789474,
+      "grad_norm": 0.9874732494354248,
+      "learning_rate": 4.020991937779872e-05,
+      "loss": 0.4944,
+      "step": 10410
+    },
+    {
+      "epoch": 1.4624561403508771,
+      "grad_norm": 0.9479324817657471,
+      "learning_rate": 4.019240698797785e-05,
+      "loss": 0.55,
+      "step": 10420
+    },
+    {
+      "epoch": 1.463859649122807,
+      "grad_norm": 2.4362101554870605,
+      "learning_rate": 4.017488277034742e-05,
+      "loss": 0.5103,
+      "step": 10430
+    },
+    {
+      "epoch": 1.4652631578947368,
+      "grad_norm": 1.659631371498108,
+      "learning_rate": 4.015734673855065e-05,
+      "loss": 0.5073,
+      "step": 10440
+    },
+    {
+      "epoch": 1.4666666666666668,
+      "grad_norm": 1.9145371913909912,
+      "learning_rate": 4.013979890623992e-05,
+      "loss": 0.5588,
+      "step": 10450
+    },
+    {
+      "epoch": 1.4680701754385965,
+      "grad_norm": 1.3370527029037476,
+      "learning_rate": 4.0122239287076834e-05,
+      "loss": 0.5984,
+      "step": 10460
+    },
+    {
+      "epoch": 1.4694736842105263,
+      "grad_norm": 1.9445977210998535,
+      "learning_rate": 4.010466789473215e-05,
+      "loss": 0.5437,
+      "step": 10470
+    },
+    {
+      "epoch": 1.470877192982456,
+      "grad_norm": 1.197405457496643,
+      "learning_rate": 4.008708474288581e-05,
+      "loss": 0.4573,
+      "step": 10480
+    },
+    {
+      "epoch": 1.472280701754386,
+      "grad_norm": 1.8886786699295044,
+      "learning_rate": 4.006948984522687e-05,
+      "loss": 0.5319,
+      "step": 10490
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 1.8042242527008057,
+      "learning_rate": 4.00518832154536e-05,
+      "loss": 0.4559,
+      "step": 10500
+    },
+    {
+      "epoch": 1.4750877192982457,
+      "grad_norm": 1.6905839443206787,
+      "learning_rate": 4.003426486727335e-05,
+      "loss": 0.4776,
+      "step": 10510
+    },
+    {
+      "epoch": 1.4764912280701754,
+      "grad_norm": 1.3357295989990234,
+      "learning_rate": 4.00166348144026e-05,
+      "loss": 0.5672,
+      "step": 10520
+    },
+    {
+      "epoch": 1.4778947368421052,
+      "grad_norm": 1.399383783340454,
+      "learning_rate": 3.9998993070566954e-05,
+      "loss": 0.5926,
+      "step": 10530
+    },
+    {
+      "epoch": 1.4792982456140351,
+      "grad_norm": 1.8520910739898682,
+      "learning_rate": 3.998133964950112e-05,
+      "loss": 0.4975,
+      "step": 10540
+    },
+    {
+      "epoch": 1.4807017543859649,
+      "grad_norm": 0.865352988243103,
+      "learning_rate": 3.9963674564948886e-05,
+      "loss": 0.5229,
+      "step": 10550
+    },
+    {
+      "epoch": 1.4821052631578948,
+      "grad_norm": 2.3526158332824707,
+      "learning_rate": 3.9945997830663126e-05,
+      "loss": 0.4995,
+      "step": 10560
+    },
+    {
+      "epoch": 1.4835087719298246,
+      "grad_norm": 2.1046128273010254,
+      "learning_rate": 3.992830946040579e-05,
+      "loss": 0.5464,
+      "step": 10570
+    },
+    {
+      "epoch": 1.4849122807017543,
+      "grad_norm": 2.0516717433929443,
+      "learning_rate": 3.9910609467947866e-05,
+      "loss": 0.5691,
+      "step": 10580
+    },
+    {
+      "epoch": 1.4863157894736843,
+      "grad_norm": 1.386889100074768,
+      "learning_rate": 3.989289786706942e-05,
+      "loss": 0.5982,
+      "step": 10590
+    },
+    {
+      "epoch": 1.487719298245614,
+      "grad_norm": 1.7038406133651733,
+      "learning_rate": 3.987517467155954e-05,
+      "loss": 0.5173,
+      "step": 10600
+    },
+    {
+      "epoch": 1.489122807017544,
+      "grad_norm": 1.3720016479492188,
+      "learning_rate": 3.985743989521633e-05,
+      "loss": 0.4406,
+      "step": 10610
+    },
+    {
+      "epoch": 1.4905263157894737,
+      "grad_norm": 1.5871185064315796,
+      "learning_rate": 3.9839693551846924e-05,
+      "loss": 0.5417,
+      "step": 10620
+    },
+    {
+      "epoch": 1.4919298245614034,
+      "grad_norm": 1.440131664276123,
+      "learning_rate": 3.982193565526747e-05,
+      "loss": 0.6226,
+      "step": 10630
+    },
+    {
+      "epoch": 1.4933333333333334,
+      "grad_norm": 1.91805100440979,
+      "learning_rate": 3.9804166219303086e-05,
+      "loss": 0.5337,
+      "step": 10640
+    },
+    {
+      "epoch": 1.4947368421052631,
+      "grad_norm": 1.9551035165786743,
+      "learning_rate": 3.9786385257787886e-05,
+      "loss": 0.5027,
+      "step": 10650
+    },
+    {
+      "epoch": 1.496140350877193,
+      "grad_norm": 1.93511962890625,
+      "learning_rate": 3.9768592784564974e-05,
+      "loss": 0.5676,
+      "step": 10660
+    },
+    {
+      "epoch": 1.4975438596491228,
+      "grad_norm": 1.6610795259475708,
+      "learning_rate": 3.975078881348638e-05,
+      "loss": 0.4542,
+      "step": 10670
+    },
+    {
+      "epoch": 1.4989473684210526,
+      "grad_norm": 1.6202747821807861,
+      "learning_rate": 3.9732973358413115e-05,
+      "loss": 0.5563,
+      "step": 10680
+    },
+    {
+      "epoch": 1.5003508771929823,
+      "grad_norm": 1.8837049007415771,
+      "learning_rate": 3.971514643321513e-05,
+      "loss": 0.4128,
+      "step": 10690
+    },
+    {
+      "epoch": 1.5017543859649123,
+      "grad_norm": 1.5047054290771484,
+      "learning_rate": 3.969730805177129e-05,
+      "loss": 0.4909,
+      "step": 10700
+    },
+    {
+      "epoch": 1.5031578947368422,
+      "grad_norm": 1.5055679082870483,
+      "learning_rate": 3.967945822796938e-05,
+      "loss": 0.5664,
+      "step": 10710
+    },
+    {
+      "epoch": 1.504561403508772,
+      "grad_norm": 1.2587159872055054,
+      "learning_rate": 3.9661596975706104e-05,
+      "loss": 0.5827,
+      "step": 10720
+    },
+    {
+      "epoch": 1.5059649122807017,
+      "grad_norm": 1.2444645166397095,
+      "learning_rate": 3.9643724308887065e-05,
+      "loss": 0.5105,
+      "step": 10730
+    },
+    {
+      "epoch": 1.5073684210526315,
+      "grad_norm": 2.214508295059204,
+      "learning_rate": 3.962584024142675e-05,
+      "loss": 0.5455,
+      "step": 10740
+    },
+    {
+      "epoch": 1.5087719298245614,
+      "grad_norm": 1.7483428716659546,
+      "learning_rate": 3.96079447872485e-05,
+      "loss": 0.5419,
+      "step": 10750
+    },
+    {
+      "epoch": 1.5101754385964914,
+      "grad_norm": 2.044471263885498,
+      "learning_rate": 3.9590037960284546e-05,
+      "loss": 0.5861,
+      "step": 10760
+    },
+    {
+      "epoch": 1.5115789473684211,
+      "grad_norm": 2.063427209854126,
+      "learning_rate": 3.9572119774475975e-05,
+      "loss": 0.5692,
+      "step": 10770
+    },
+    {
+      "epoch": 1.5129824561403509,
+      "grad_norm": 2.0187671184539795,
+      "learning_rate": 3.95541902437727e-05,
+      "loss": 0.5345,
+      "step": 10780
+    },
+    {
+      "epoch": 1.5143859649122806,
+      "grad_norm": 1.9781345129013062,
+      "learning_rate": 3.953624938213348e-05,
+      "loss": 0.5212,
+      "step": 10790
+    },
+    {
+      "epoch": 1.5157894736842106,
+      "grad_norm": 1.9213941097259521,
+      "learning_rate": 3.95182972035259e-05,
+      "loss": 0.4838,
+      "step": 10800
+    },
+    {
+      "epoch": 1.5171929824561403,
+      "grad_norm": 2.619076728820801,
+      "learning_rate": 3.950033372192633e-05,
+      "loss": 0.5011,
+      "step": 10810
+    },
+    {
+      "epoch": 1.5185964912280703,
+      "grad_norm": 1.8988882303237915,
+      "learning_rate": 3.948235895131997e-05,
+      "loss": 0.5043,
+      "step": 10820
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.4304900169372559,
+      "learning_rate": 3.946437290570078e-05,
+      "loss": 0.5062,
+      "step": 10830
+    },
+    {
+      "epoch": 1.5214035087719298,
+      "grad_norm": 3.060408353805542,
+      "learning_rate": 3.944637559907152e-05,
+      "loss": 0.6164,
+      "step": 10840
+    },
+    {
+      "epoch": 1.5228070175438595,
+      "grad_norm": 0.9590080976486206,
+      "learning_rate": 3.9428367045443704e-05,
+      "loss": 0.5159,
+      "step": 10850
+    },
+    {
+      "epoch": 1.5242105263157895,
+      "grad_norm": 2.1461575031280518,
+      "learning_rate": 3.941034725883762e-05,
+      "loss": 0.6505,
+      "step": 10860
+    },
+    {
+      "epoch": 1.5256140350877194,
+      "grad_norm": 1.578477144241333,
+      "learning_rate": 3.939231625328229e-05,
+      "loss": 0.4808,
+      "step": 10870
+    },
+    {
+      "epoch": 1.5270175438596492,
+      "grad_norm": 1.8510093688964844,
+      "learning_rate": 3.9374274042815465e-05,
+      "loss": 0.5194,
+      "step": 10880
+    },
+    {
+      "epoch": 1.528421052631579,
+      "grad_norm": 1.9132167100906372,
+      "learning_rate": 3.935622064148361e-05,
+      "loss": 0.5079,
+      "step": 10890
+    },
+    {
+      "epoch": 1.5298245614035086,
+      "grad_norm": 1.3352388143539429,
+      "learning_rate": 3.9338156063341946e-05,
+      "loss": 0.4808,
+      "step": 10900
+    },
+    {
+      "epoch": 1.5312280701754386,
+      "grad_norm": 2.102167844772339,
+      "learning_rate": 3.932008032245434e-05,
+      "loss": 0.429,
+      "step": 10910
+    },
+    {
+      "epoch": 1.5326315789473686,
+      "grad_norm": 1.9585574865341187,
+      "learning_rate": 3.930199343289339e-05,
+      "loss": 0.489,
+      "step": 10920
+    },
+    {
+      "epoch": 1.5340350877192983,
+      "grad_norm": 1.905050277709961,
+      "learning_rate": 3.9283895408740355e-05,
+      "loss": 0.4881,
+      "step": 10930
+    },
+    {
+      "epoch": 1.535438596491228,
+      "grad_norm": 1.964416742324829,
+      "learning_rate": 3.926578626408517e-05,
+      "loss": 0.5913,
+      "step": 10940
+    },
+    {
+      "epoch": 1.5368421052631578,
+      "grad_norm": 2.3363118171691895,
+      "learning_rate": 3.924766601302642e-05,
+      "loss": 0.4719,
+      "step": 10950
+    },
+    {
+      "epoch": 1.5382456140350877,
+      "grad_norm": 1.7316786050796509,
+      "learning_rate": 3.9229534669671344e-05,
+      "loss": 0.5445,
+      "step": 10960
+    },
+    {
+      "epoch": 1.5396491228070175,
+      "grad_norm": 2.2813808917999268,
+      "learning_rate": 3.9211392248135815e-05,
+      "loss": 0.4989,
+      "step": 10970
+    },
+    {
+      "epoch": 1.5410526315789475,
+      "grad_norm": 0.9021309018135071,
+      "learning_rate": 3.9193238762544325e-05,
+      "loss": 0.5321,
+      "step": 10980
+    },
+    {
+      "epoch": 1.5424561403508772,
+      "grad_norm": 2.614776134490967,
+      "learning_rate": 3.9175074227029996e-05,
+      "loss": 0.4765,
+      "step": 10990
+    },
+    {
+      "epoch": 1.543859649122807,
+      "grad_norm": 2.1491498947143555,
+      "learning_rate": 3.915689865573454e-05,
+      "loss": 0.5748,
+      "step": 11000
+    },
+    {
+      "epoch": 1.545263157894737,
+      "grad_norm": 1.7609467506408691,
+      "learning_rate": 3.913871206280824e-05,
+      "loss": 0.7091,
+      "step": 11010
+    },
+    {
+      "epoch": 1.5466666666666666,
+      "grad_norm": 2.5683560371398926,
+      "learning_rate": 3.912051446241001e-05,
+      "loss": 0.507,
+      "step": 11020
+    },
+    {
+      "epoch": 1.5480701754385966,
+      "grad_norm": 1.3458114862442017,
+      "learning_rate": 3.910230586870729e-05,
+      "loss": 0.5738,
+      "step": 11030
+    },
+    {
+      "epoch": 1.5494736842105263,
+      "grad_norm": 1.7107462882995605,
+      "learning_rate": 3.90840862958761e-05,
+      "loss": 0.4437,
+      "step": 11040
+    },
+    {
+      "epoch": 1.550877192982456,
+      "grad_norm": 1.6637877225875854,
+      "learning_rate": 3.9065855758101e-05,
+      "loss": 0.4859,
+      "step": 11050
+    },
+    {
+      "epoch": 1.5522807017543858,
+      "grad_norm": 2.1268763542175293,
+      "learning_rate": 3.904761426957509e-05,
+      "loss": 0.5433,
+      "step": 11060
+    },
+    {
+      "epoch": 1.5536842105263158,
+      "grad_norm": 1.8485718965530396,
+      "learning_rate": 3.902936184449999e-05,
+      "loss": 0.5938,
+      "step": 11070
+    },
+    {
+      "epoch": 1.5550877192982457,
+      "grad_norm": 1.9369820356369019,
+      "learning_rate": 3.901109849708585e-05,
+      "loss": 0.4484,
+      "step": 11080
+    },
+    {
+      "epoch": 1.5564912280701755,
+      "grad_norm": 1.5223256349563599,
+      "learning_rate": 3.8992824241551295e-05,
+      "loss": 0.4353,
+      "step": 11090
+    },
+    {
+      "epoch": 1.5578947368421052,
+      "grad_norm": 1.2845451831817627,
+      "learning_rate": 3.897453909212348e-05,
+      "loss": 0.4497,
+      "step": 11100
+    },
+    {
+      "epoch": 1.559298245614035,
+      "grad_norm": 1.2751349210739136,
+      "learning_rate": 3.895624306303799e-05,
+      "loss": 0.4648,
+      "step": 11110
+    },
+    {
+      "epoch": 1.560701754385965,
+      "grad_norm": 1.6384958028793335,
+      "learning_rate": 3.893793616853894e-05,
+      "loss": 0.5921,
+      "step": 11120
+    },
+    {
+      "epoch": 1.5621052631578949,
+      "grad_norm": 1.817355990409851,
+      "learning_rate": 3.891961842287886e-05,
+      "loss": 0.4611,
+      "step": 11130
+    },
+    {
+      "epoch": 1.5635087719298246,
+      "grad_norm": 1.7115503549575806,
+      "learning_rate": 3.890128984031876e-05,
+      "loss": 0.4745,
+      "step": 11140
+    },
+    {
+      "epoch": 1.5649122807017544,
+      "grad_norm": 1.7166131734848022,
+      "learning_rate": 3.888295043512804e-05,
+      "loss": 0.5716,
+      "step": 11150
+    },
+    {
+      "epoch": 1.566315789473684,
+      "grad_norm": 1.8528428077697754,
+      "learning_rate": 3.886460022158458e-05,
+      "loss": 0.5193,
+      "step": 11160
+    },
+    {
+      "epoch": 1.567719298245614,
+      "grad_norm": 1.9985193014144897,
+      "learning_rate": 3.884623921397463e-05,
+      "loss": 0.4974,
+      "step": 11170
+    },
+    {
+      "epoch": 1.5691228070175438,
+      "grad_norm": 1.4072109460830688,
+      "learning_rate": 3.882786742659289e-05,
+      "loss": 0.4418,
+      "step": 11180
+    },
+    {
+      "epoch": 1.5705263157894738,
+      "grad_norm": 1.3553410768508911,
+      "learning_rate": 3.880948487374241e-05,
+      "loss": 0.5278,
+      "step": 11190
+    },
+    {
+      "epoch": 1.5719298245614035,
+      "grad_norm": 1.6441354751586914,
+      "learning_rate": 3.8791091569734625e-05,
+      "loss": 0.476,
+      "step": 11200
+    },
+    {
+      "epoch": 1.5733333333333333,
+      "grad_norm": 1.4078179597854614,
+      "learning_rate": 3.8772687528889385e-05,
+      "loss": 0.581,
+      "step": 11210
+    },
+    {
+      "epoch": 1.5747368421052632,
+      "grad_norm": 2.096179723739624,
+      "learning_rate": 3.875427276553485e-05,
+      "loss": 0.5076,
+      "step": 11220
+    },
+    {
+      "epoch": 1.576140350877193,
+      "grad_norm": 1.3418902158737183,
+      "learning_rate": 3.873584729400753e-05,
+      "loss": 0.5177,
+      "step": 11230
+    },
+    {
+      "epoch": 1.577543859649123,
+      "grad_norm": 2.1806328296661377,
+      "learning_rate": 3.8717411128652304e-05,
+      "loss": 0.5348,
+      "step": 11240
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 1.6312358379364014,
+      "learning_rate": 3.869896428382236e-05,
+      "loss": 0.4699,
+      "step": 11250
+    },
+    {
+      "epoch": 1.5803508771929824,
+      "grad_norm": 2.4073407649993896,
+      "learning_rate": 3.8680506773879184e-05,
+      "loss": 0.5403,
+      "step": 11260
+    },
+    {
+      "epoch": 1.5817543859649121,
+      "grad_norm": 2.2572133541107178,
+      "learning_rate": 3.8662038613192596e-05,
+      "loss": 0.4982,
+      "step": 11270
+    },
+    {
+      "epoch": 1.583157894736842,
+      "grad_norm": 1.9021124839782715,
+      "learning_rate": 3.8643559816140685e-05,
+      "loss": 0.5025,
+      "step": 11280
+    },
+    {
+      "epoch": 1.584561403508772,
+      "grad_norm": 1.5274001359939575,
+      "learning_rate": 3.862507039710982e-05,
+      "loss": 0.4716,
+      "step": 11290
+    },
+    {
+      "epoch": 1.5859649122807018,
+      "grad_norm": 1.0260459184646606,
+      "learning_rate": 3.860657037049466e-05,
+      "loss": 0.5378,
+      "step": 11300
+    },
+    {
+      "epoch": 1.5873684210526315,
+      "grad_norm": 2.2560298442840576,
+      "learning_rate": 3.85880597506981e-05,
+      "loss": 0.5017,
+      "step": 11310
+    },
+    {
+      "epoch": 1.5887719298245613,
+      "grad_norm": 1.3712495565414429,
+      "learning_rate": 3.856953855213131e-05,
+      "loss": 0.4612,
+      "step": 11320
+    },
+    {
+      "epoch": 1.5901754385964912,
+      "grad_norm": 1.3681050539016724,
+      "learning_rate": 3.855100678921365e-05,
+      "loss": 0.5077,
+      "step": 11330
+    },
+    {
+      "epoch": 1.5915789473684212,
+      "grad_norm": 2.0733203887939453,
+      "learning_rate": 3.8532464476372765e-05,
+      "loss": 0.5643,
+      "step": 11340
+    },
+    {
+      "epoch": 1.592982456140351,
+      "grad_norm": 2.131699323654175,
+      "learning_rate": 3.851391162804445e-05,
+      "loss": 0.4939,
+      "step": 11350
+    },
+    {
+      "epoch": 1.5943859649122807,
+      "grad_norm": 1.787874698638916,
+      "learning_rate": 3.849534825867275e-05,
+      "loss": 0.5191,
+      "step": 11360
+    },
+    {
+      "epoch": 1.5957894736842104,
+      "grad_norm": 1.7359322309494019,
+      "learning_rate": 3.847677438270988e-05,
+      "loss": 0.5361,
+      "step": 11370
+    },
+    {
+      "epoch": 1.5971929824561404,
+      "grad_norm": 1.7936285734176636,
+      "learning_rate": 3.845819001461625e-05,
+      "loss": 0.5005,
+      "step": 11380
+    },
+    {
+      "epoch": 1.5985964912280701,
+      "grad_norm": 0.9876174330711365,
+      "learning_rate": 3.8439595168860406e-05,
+      "loss": 0.491,
+      "step": 11390
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.5902925729751587,
+      "learning_rate": 3.842098985991909e-05,
+      "loss": 0.5636,
+      "step": 11400
+    },
+    {
+      "epoch": 1.6014035087719298,
+      "grad_norm": 2.157257556915283,
+      "learning_rate": 3.840237410227717e-05,
+      "loss": 0.5482,
+      "step": 11410
+    },
+    {
+      "epoch": 1.6028070175438596,
+      "grad_norm": 2.4455907344818115,
+      "learning_rate": 3.838374791042764e-05,
+      "loss": 0.4854,
+      "step": 11420
+    },
+    {
+      "epoch": 1.6042105263157893,
+      "grad_norm": 2.3983774185180664,
+      "learning_rate": 3.8365111298871645e-05,
+      "loss": 0.5535,
+      "step": 11430
+    },
+    {
+      "epoch": 1.6056140350877193,
+      "grad_norm": 1.6250687837600708,
+      "learning_rate": 3.834646428211841e-05,
+      "loss": 0.5493,
+      "step": 11440
+    },
+    {
+      "epoch": 1.6070175438596492,
+      "grad_norm": 0.9640924334526062,
+      "learning_rate": 3.83278068746853e-05,
+      "loss": 0.5126,
+      "step": 11450
+    },
+    {
+      "epoch": 1.608421052631579,
+      "grad_norm": 1.3491884469985962,
+      "learning_rate": 3.830913909109772e-05,
+      "loss": 0.5692,
+      "step": 11460
+    },
+    {
+      "epoch": 1.6098245614035087,
+      "grad_norm": 1.2153112888336182,
+      "learning_rate": 3.8290460945889186e-05,
+      "loss": 0.4367,
+      "step": 11470
+    },
+    {
+      "epoch": 1.6112280701754385,
+      "grad_norm": 1.869314193725586,
+      "learning_rate": 3.827177245360129e-05,
+      "loss": 0.5275,
+      "step": 11480
+    },
+    {
+      "epoch": 1.6126315789473684,
+      "grad_norm": 2.1873159408569336,
+      "learning_rate": 3.825307362878364e-05,
+      "loss": 0.5663,
+      "step": 11490
+    },
+    {
+      "epoch": 1.6140350877192984,
+      "grad_norm": 1.1323574781417847,
+      "learning_rate": 3.823436448599393e-05,
+      "loss": 0.4986,
+      "step": 11500
+    },
+    {
+      "epoch": 1.6154385964912281,
+      "grad_norm": 1.424310326576233,
+      "learning_rate": 3.8215645039797874e-05,
+      "loss": 0.5401,
+      "step": 11510
+    },
+    {
+      "epoch": 1.6168421052631579,
+      "grad_norm": 1.8369444608688354,
+      "learning_rate": 3.8196915304769184e-05,
+      "loss": 0.534,
+      "step": 11520
+    },
+    {
+      "epoch": 1.6182456140350876,
+      "grad_norm": 1.404891848564148,
+      "learning_rate": 3.817817529548962e-05,
+      "loss": 0.457,
+      "step": 11530
+    },
+    {
+      "epoch": 1.6196491228070176,
+      "grad_norm": 1.5708239078521729,
+      "learning_rate": 3.815942502654889e-05,
+      "loss": 0.5023,
+      "step": 11540
+    },
+    {
+      "epoch": 1.6210526315789475,
+      "grad_norm": 1.4378736019134521,
+      "learning_rate": 3.8140664512544746e-05,
+      "loss": 0.4885,
+      "step": 11550
+    },
+    {
+      "epoch": 1.6224561403508773,
+      "grad_norm": 2.0573270320892334,
+      "learning_rate": 3.8121893768082896e-05,
+      "loss": 0.5204,
+      "step": 11560
+    },
+    {
+      "epoch": 1.623859649122807,
+      "grad_norm": 1.4034109115600586,
+      "learning_rate": 3.8103112807776986e-05,
+      "loss": 0.4611,
+      "step": 11570
+    },
+    {
+      "epoch": 1.6252631578947367,
+      "grad_norm": 1.8642430305480957,
+      "learning_rate": 3.8084321646248654e-05,
+      "loss": 0.4999,
+      "step": 11580
+    },
+    {
+      "epoch": 1.6266666666666667,
+      "grad_norm": 2.0132107734680176,
+      "learning_rate": 3.806552029812747e-05,
+      "loss": 0.5241,
+      "step": 11590
+    },
+    {
+      "epoch": 1.6280701754385964,
+      "grad_norm": 1.3874375820159912,
+      "learning_rate": 3.804670877805091e-05,
+      "loss": 0.5275,
+      "step": 11600
+    },
+    {
+      "epoch": 1.6294736842105264,
+      "grad_norm": 1.296034574508667,
+      "learning_rate": 3.802788710066439e-05,
+      "loss": 0.4517,
+      "step": 11610
+    },
+    {
+      "epoch": 1.6308771929824561,
+      "grad_norm": 1.7629979848861694,
+      "learning_rate": 3.800905528062123e-05,
+      "loss": 0.4437,
+      "step": 11620
+    },
+    {
+      "epoch": 1.6322807017543859,
+      "grad_norm": 1.788439393043518,
+      "learning_rate": 3.7990213332582665e-05,
+      "loss": 0.5334,
+      "step": 11630
+    },
+    {
+      "epoch": 1.6336842105263156,
+      "grad_norm": 1.6043287515640259,
+      "learning_rate": 3.7971361271217775e-05,
+      "loss": 0.5915,
+      "step": 11640
+    },
+    {
+      "epoch": 1.6350877192982456,
+      "grad_norm": 1.1127432584762573,
+      "learning_rate": 3.7952499111203544e-05,
+      "loss": 0.633,
+      "step": 11650
+    },
+    {
+      "epoch": 1.6364912280701756,
+      "grad_norm": 1.6562187671661377,
+      "learning_rate": 3.793362686722483e-05,
+      "loss": 0.523,
+      "step": 11660
+    },
+    {
+      "epoch": 1.6378947368421053,
+      "grad_norm": 2.0622971057891846,
+      "learning_rate": 3.7914744553974284e-05,
+      "loss": 0.5025,
+      "step": 11670
+    },
+    {
+      "epoch": 1.639298245614035,
+      "grad_norm": 2.0334134101867676,
+      "learning_rate": 3.789585218615246e-05,
+      "loss": 0.5153,
+      "step": 11680
+    },
+    {
+      "epoch": 1.6407017543859648,
+      "grad_norm": 1.7830958366394043,
+      "learning_rate": 3.787694977846771e-05,
+      "loss": 0.5783,
+      "step": 11690
+    },
+    {
+      "epoch": 1.6421052631578947,
+      "grad_norm": 2.2819151878356934,
+      "learning_rate": 3.78580373456362e-05,
+      "loss": 0.5333,
+      "step": 11700
+    },
+    {
+      "epoch": 1.6435087719298247,
+      "grad_norm": 1.7580994367599487,
+      "learning_rate": 3.783911490238191e-05,
+      "loss": 0.574,
+      "step": 11710
+    },
+    {
+      "epoch": 1.6449122807017544,
+      "grad_norm": 1.012489676475525,
+      "learning_rate": 3.782018246343661e-05,
+      "loss": 0.5028,
+      "step": 11720
+    },
+    {
+      "epoch": 1.6463157894736842,
+      "grad_norm": 0.9846917390823364,
+      "learning_rate": 3.780124004353987e-05,
+      "loss": 0.5425,
+      "step": 11730
+    },
+    {
+      "epoch": 1.647719298245614,
+      "grad_norm": 1.3875446319580078,
+      "learning_rate": 3.778228765743898e-05,
+      "loss": 0.4961,
+      "step": 11740
+    },
+    {
+      "epoch": 1.6491228070175439,
+      "grad_norm": 0.912992537021637,
+      "learning_rate": 3.776332531988903e-05,
+      "loss": 0.5135,
+      "step": 11750
+    },
+    {
+      "epoch": 1.6505263157894738,
+      "grad_norm": 1.4657293558120728,
+      "learning_rate": 3.774435304565288e-05,
+      "loss": 0.5917,
+      "step": 11760
+    },
+    {
+      "epoch": 1.6519298245614036,
+      "grad_norm": 1.4134496450424194,
+      "learning_rate": 3.772537084950106e-05,
+      "loss": 0.6529,
+      "step": 11770
+    },
+    {
+      "epoch": 1.6533333333333333,
+      "grad_norm": 1.0357835292816162,
+      "learning_rate": 3.770637874621189e-05,
+      "loss": 0.4853,
+      "step": 11780
+    },
+    {
+      "epoch": 1.654736842105263,
+      "grad_norm": 1.1166404485702515,
+      "learning_rate": 3.7687376750571347e-05,
+      "loss": 0.5509,
+      "step": 11790
+    },
+    {
+      "epoch": 1.656140350877193,
+      "grad_norm": 1.819243311882019,
+      "learning_rate": 3.7668364877373154e-05,
+      "loss": 0.5083,
+      "step": 11800
+    },
+    {
+      "epoch": 1.6575438596491228,
+      "grad_norm": 2.0793590545654297,
+      "learning_rate": 3.764934314141869e-05,
+      "loss": 0.5239,
+      "step": 11810
+    },
+    {
+      "epoch": 1.6589473684210527,
+      "grad_norm": 1.4497408866882324,
+      "learning_rate": 3.763031155751705e-05,
+      "loss": 0.5295,
+      "step": 11820
+    },
+    {
+      "epoch": 1.6603508771929825,
+      "grad_norm": 1.8401798009872437,
+      "learning_rate": 3.7611270140484956e-05,
+      "loss": 0.3987,
+      "step": 11830
+    },
+    {
+      "epoch": 1.6617543859649122,
+      "grad_norm": 1.0776817798614502,
+      "learning_rate": 3.759221890514681e-05,
+      "loss": 0.5236,
+      "step": 11840
+    },
+    {
+      "epoch": 1.663157894736842,
+      "grad_norm": 1.6151456832885742,
+      "learning_rate": 3.757315786633465e-05,
+      "loss": 0.4783,
+      "step": 11850
+    },
+    {
+      "epoch": 1.664561403508772,
+      "grad_norm": 2.149061679840088,
+      "learning_rate": 3.7554087038888155e-05,
+      "loss": 0.6304,
+      "step": 11860
+    },
+    {
+      "epoch": 1.6659649122807019,
+      "grad_norm": 1.848923683166504,
+      "learning_rate": 3.753500643765461e-05,
+      "loss": 0.4951,
+      "step": 11870
+    },
+    {
+      "epoch": 1.6673684210526316,
+      "grad_norm": 1.3279706239700317,
+      "learning_rate": 3.751591607748891e-05,
+      "loss": 0.5195,
+      "step": 11880
+    },
+    {
+      "epoch": 1.6687719298245614,
+      "grad_norm": 2.280778646469116,
+      "learning_rate": 3.749681597325357e-05,
+      "loss": 0.6116,
+      "step": 11890
+    },
+    {
+      "epoch": 1.670175438596491,
+      "grad_norm": 1.8069521188735962,
+      "learning_rate": 3.7477706139818683e-05,
+      "loss": 0.5038,
+      "step": 11900
+    },
+    {
+      "epoch": 1.671578947368421,
+      "grad_norm": 1.8922659158706665,
+      "learning_rate": 3.745858659206188e-05,
+      "loss": 0.5671,
+      "step": 11910
+    },
+    {
+      "epoch": 1.672982456140351,
+      "grad_norm": 1.6917041540145874,
+      "learning_rate": 3.743945734486841e-05,
+      "loss": 0.5559,
+      "step": 11920
+    },
+    {
+      "epoch": 1.6743859649122808,
+      "grad_norm": 1.2357120513916016,
+      "learning_rate": 3.742031841313103e-05,
+      "loss": 0.5069,
+      "step": 11930
+    },
+    {
+      "epoch": 1.6757894736842105,
+      "grad_norm": 1.5987924337387085,
+      "learning_rate": 3.7401169811750066e-05,
+      "loss": 0.5431,
+      "step": 11940
+    },
+    {
+      "epoch": 1.6771929824561402,
+      "grad_norm": 1.3475733995437622,
+      "learning_rate": 3.7382011555633365e-05,
+      "loss": 0.5636,
+      "step": 11950
+    },
+    {
+      "epoch": 1.6785964912280702,
+      "grad_norm": 1.4985164403915405,
+      "learning_rate": 3.736284365969627e-05,
+      "loss": 0.4871,
+      "step": 11960
+    },
+    {
+      "epoch": 1.6800000000000002,
+      "grad_norm": 1.2310782670974731,
+      "learning_rate": 3.7343666138861646e-05,
+      "loss": 0.6245,
+      "step": 11970
+    },
+    {
+      "epoch": 1.68140350877193,
+      "grad_norm": 1.7251943349838257,
+      "learning_rate": 3.7324479008059865e-05,
+      "loss": 0.5126,
+      "step": 11980
+    },
+    {
+      "epoch": 1.6828070175438596,
+      "grad_norm": 2.389265775680542,
+      "learning_rate": 3.7305282282228756e-05,
+      "loss": 0.5669,
+      "step": 11990
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 2.130988359451294,
+      "learning_rate": 3.728607597631363e-05,
+      "loss": 0.4796,
+      "step": 12000
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "eval_loss": 0.6500447392463684,
+      "eval_runtime": 119.5925,
+      "eval_samples_per_second": 12.543,
+      "eval_steps_per_second": 3.136,
+      "step": 12000
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "step": 12000,
+      "total_flos": 6.933368738955264e+17,
+      "train_loss": 0.6038338423768679,
+      "train_runtime": 5861.7175,
+      "train_samples_per_second": 24.31,
+      "train_steps_per_second": 6.078
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 35625,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 2000,
+  "total_flos": 6.933368738955264e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}